JP2005327118A - Multivariable control optimization program and volumetric feeder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a multivariable control optimization program capable of making optimal adjustment about PI control at the appropriate time and a volumetric feeder. <P>SOLUTION: The volumetric feeder includes a hopper 2 for dumping powder and grain 1 serving as the raw materials of a synthetic resin product; a feed screw 3 provided in the bottom of the hopper 2 for discharging the powder and grain 1 from the hopper 2 in a mass proportional to the number of revolutions; a measuring means 4 for measuring the amount of discharge by the feed screw 3; and a controller 5 that performs negative feedback control of the number of revolutions of the feed screw 3 on the basis of measurement data sampled by the measuring means 4 in accordance with the multivariable control optimization program. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a multivariable control optimization program for adjusting a supply amount, a discharge amount, a flow rate, or the like of a raw material or the like, and a quantitative supply device.

  The following patent documents each disclose a fixed-quantity supply device that controls the granular material 1. Conventionally, proportional integral control (hereinafter referred to as “PI control”) is applied in order to maintain the supply amount or the discharge amount of the object at the target value.

According to a known quantitative supply device, powder particles are accumulated in a weighing hopper, and a feed screw provided at a lower portion of the hopper is rotated to discharge the powder particles from the hopper. Thereby, since the total weight of the hopper including the granular material is reduced, the reduced amount is measured as the mass of the granular material discharged by the rotation of the feed screw. Then, if the mass of the powder discharged within a predetermined time is small in light of the target value, the rotational speed of the feed screw is increased, or if it is large in light of the target value, the rotational speed of the feed screw is decreased. .
JP 2000-55722 A JP 2002-318152 A

  In order to appropriately execute the above PI control, it is essential to initially set a control constant such as a proportionality constant for calculating the rotation speed of the feed screw and an integration time. Therefore, immediately after changing the so-called raw material, a trial operation for grasping the relationship between the rotation speed of the feed screw and the amount of the powder actually discharged from the hopper is performed. In this process, the bulk specific gravity of the granular material accumulated in the hopper fluctuates from when the hopper is full to when it is full. Therefore, considering that this is reflected in the PI control, the hopper is full. Continue the trial run from the state until it is full. During this time, the operation of the quantitative supply device must be interrupted.

  An object of the present invention is to provide a multivariable control optimization program and a quantitative supply device that can quickly make an optimal adjustment related to PI control.

  The optimization program for multivariable control according to the present invention provides a control amount calculated based on a deviation between an actual measurement value obtained by measuring a control object proportional to an output of the adjustment unit and a target value. A plurality of measured values obtained by repeatedly measuring the control object at a constant time interval are added to a computer that performs negative feedback control of the control object by adding to the output in the order in which the control object is measured. A function of recording, a function of recording a multivariable control amount calculation expression including a plurality of variables and giving a control amount based on the deviation as a function, and a numerical value group including a plurality of numerical values that can be respectively substituted into the plurality of variables. , A function of recording the plurality of numerical values different from each other, and sequentially selecting a numerical value group one by one from the plurality of numerical value groups, the plurality of numerical values included in the one numerical value group, Many A function for substituting each of a plurality of variables in a numerical control amount calculation formula, and a correction value obtained by adding a new control amount given by the multivariable control amount calculation formula based on the deviation to the output of the adjusting means; A function that repeats each time a numerical value group is selected from the plurality of numerical value groups, a calculation process that calculates a simulation value that is proportional to the product of the correction value and each of the actual measured values is selected. And by recording the plurality of simulation values calculated by the calculation process for each time as one data group, the plurality of simulation values calculated by repeating the calculation process are converted into a plurality of data groups. And the standard deviation for each data group of the plurality of data groups, and the standard deviation of the plurality of data groups are compared. A function for specifying a data group from which a small standard deviation is obtained, and the plurality of numerical values respectively assigned to a plurality of variables of the multivariable control amount calculation formula for calculating the data group from which the minimum standard deviation is obtained. And a function of recognizing it as an appropriate value.

  With the following i and j as arbitrary integers, the multivariable control optimization program according to the present invention is based on a deviation between a measured value obtained by measuring a control object proportional to the output of the adjusting means and a target value. A computer that performs negative feedback control of the control object by adding the calculated control amount to the output of the adjusting means, and a plurality of actual measurement values obtained by repeatedly measuring the control object at regular time intervals, A function of recording n pieces in the order in which the control object is measured, a function of calculating a deviation α (i) between the i-th actual measurement value W (i) of the n actual measurement values and the target value; A function of recording a multivariable control amount arithmetic expression including a plurality of variables in a proportional term and an integral term and giving a control amount based on the deviation as a function, and a numerical value group including a plurality of numerical values that can be substituted for the plurality of variables, respectively Different from each other A function of recording a plurality of values, and selecting one numerical value group from the plurality of numerical value groups, and converting the plurality of numerical values included in the one numerical value group into a plurality of expressions of the multivariable control amount arithmetic expression A control amount S (i) based on the deviation α (i) is calculated by the function of substituting each into a variable and the multivariable control amount arithmetic expression into which a plurality of numerical values of the one numerical value group are respectively substituted, A correction value R (i + 1) obtained by adding the control amount S (i) to the output of the adjusting means is obtained, and is proportional to the product of the correction value R (i + 1) and the (i + 1) th actual measurement value W (i + 1). A calculation function for repeating the trial calculation for obtaining the simulation value Ws (i + 1) until the nth simulation value Ws (n) is obtained, and one method other than the numerical value group already selected from the plurality of numerical value groups. A process of sequentially selecting a numerical group of Until all of the numerical value groups are selected, and the calculation process is repeated for each of the plurality of numerical value groups, and the simulation values Ws (i + 1) to Ws (n) sequentially calculated by the calculation process. A function of accumulating a plurality of the data groups by recording each time the calculation process is repeated for each of the plurality of numerical values groups, and a plurality of the data groups. Calculating a standard deviation σ for each data group of the plurality of data groups, comparing the standard deviation σ of each of the plurality of data groups, and identifying a data group from which the minimum standard deviation σ is obtained, and the minimum And a function of recognizing the plurality of numerical values respectively assigned to a plurality of variables included in the multivariable control amount calculation formula obtained by trial calculation of a data group from which a standard deviation σ is obtained as an appropriate value It is characterized by that.

  Further, the arithmetic processing is performed within a time required for measuring the n actual measured values by dividing the n actual measured values by τ according to the order in which the n actual measured values are measured. A function for assigning a control cycle that is repeated once, and the multivariable control amount calculation formula into which a plurality of numerical values of one numerical value group selected from the plurality of numerical value groups are respectively substituted, The control amount calculation process for calculating the control amount S (1) based on the deviation α (1) between the first measured value W (1) belonging to the control cycle T (1) and the target value is executed, and the control A correction value R (2) obtained by adding the amount S (1) to the output of the adjusting means is obtained, and a second simulation value proportional to the product of the correction value R (2) and the second actual measurement value W (2). A function for obtaining Ws (2) and a plurality of numerical values in the one numerical value group are substituted. Then, the control variable S (i + 1) based on the deviation α (i + 1) between the second and subsequent simulation values Ws (i + 1) and the target value is calculated by the multivariable control variable calculation formula in which the proportional term is zero. The control value calculation process is executed to obtain a correction value R (i + 2) obtained by adding the control quantity S (i + 1) to the output of the adjusting means, and the correction value R (i + 2) and the (i + 2) th actual measurement value W are obtained. A function of an initial period calculation process that repeats trial calculation to obtain the (i + 2) th simulation value Ws (i + 2) proportional to the product of (i + 2) until the (τ + 1) th simulation value Ws (τ + 1) is obtained; The last simulation belonging to the first control cycle T (1) or the second and subsequent control cycles T (j) by the multivariable control amount calculation formula into which a plurality of numerical values of the one numerical value group are respectively substituted. A control amount calculation process for calculating a control amount S (jτ + 1) based on a deviation α (jτ + 1) between the control value Ws (jτ + 1) and the target value is executed, and the control amount S (jτ + 1) is output from the adjusting means. A function of obtaining a correction value R (jτ + 2) added to the above and obtaining a simulation value Ws (jτ + 2) proportional to the product of the correction value R (jτ + 2) and the (jτ + 2) -th actual measurement value W (jτ + 2); A control amount based on a deviation α (jτ + 2) between the simulation value Ws (jτ + 2) and the target value by the multivariable control amount arithmetic expression in which a plurality of numerical values of the numerical value group are respectively substituted and the proportional term is zero. A control amount calculation process for calculating S (jτ + 2) is executed, a correction value R (jτ + 3) obtained by adding the control amount S (jτ + 2) to the output of the adjusting means is obtained, and the correction values R (jτ + 3) and (jτ + 3) are obtained. ) The function of the subsequent cycle calculation processing is repeated until the (jτ + τ + 1) -th simulation value Ws (jτ + τ + 1) is obtained by trial calculation for obtaining the simulation value Ws (jτ + 3) proportional to the product of the actual measurement value W (jτ + 3) of the eye. It may be characterized by being realized by a computer.

  The control amount calculation process includes a proportional term and an integral term of the multivariable control amount calculation expression including a proportional constant κ, a trend constant kh, and an integration time t as variables {α (i) × κ}, α The function of recording as {(i) × κ / t} and calculating S (i) = [{α (i) × κ} + {α (i) × κ / t}] × kh The computer may be realized with a function of deriving a function S (i) given to the value of the deviation α (i) as the control amount.

  Furthermore, the optimization program for multivariable control according to the present invention records the values of the proportionality constant κ, the integration time t, and the control period T one by one so that one numerical value group is recorded. The computer may have a function of forming and recording a plurality of numerical values so as to be mutually distinguishable.

  Further, the quantitative supply device according to the present invention provides a control amount calculated based on a deviation between a measured value obtained by measuring a control object proportional to the output of the adjusting unit and a target value as an output of the adjusting unit. A computer that performs negative feedback control of the controlled object by adding, a hopper into which the granular material is charged as the controlled object, and a rotation speed that increases and decreases in response to the output of the adjusting means provided at the bottom of the hopper A feed screw that rotates together with an electric motor, a weighing means that measures the mass of the control target discharged from the hopper in proportion to the rotation speed of the feed screw, and an optimization program for the multivariable control And a recording medium on which is recorded.

  According to the optimization program for multivariable control according to the present invention, an actual measurement value obtained by measuring a physical quantity such as temperature, pressure, specific gravity, voltage, or flow rate related to any control object proportional to the output of the adjusting means, and a desired value By calculating the control amount based on the deviation from the target value that can be set to this value and adding this to the output of the adjusting means, it is possible to perform negative feedback control on any controlled object.

  For example, if this optimization program is applied to a quantitative supply device, even if the bulk specific gravity of the granular material accumulated in the hopper varies from the full hopper to the full hopper, the rotation of the feed screw It is possible to eliminate the complication that the operator resets the proportionality constant, integration time, etc. for calculating the number in small increments. In addition, since the proportional constant, integration time, and other settings can be achieved without trial operation, there is no need to stop production lines that provide powder particles, which greatly contributes to improved productivity. Will do.

  The multivariable control optimization program and the quantitative supply device according to the present embodiment will be described first as an overview and then as an example.

  FIG. 1 shows an outline of a quantitative supply device. This quantitative supply device has a hopper 2 into which the powder body 1 as a raw material of the synthetic resin product is charged, and a mass (hereinafter referred to as “discharge amount”) provided at the bottom of the hopper 2 and proportional to the rotational speed. A feed screw 3 that discharges the granular material 1 from the hopper 2, a measuring unit 4 that measures the discharge amount, and a control device 5 that performs negative feedback control of the rotation speed of the feed screw 3 based on the discharge amount measured by the measuring unit 4. Is provided.

  The control device 5 executes the following calculation steps based on this optimization program. That is, the characteristic unique to the feed screw 3 is reflected in the relationship between the rotation speed of the feed screw 3 and the discharge amount. The discharge amount [g / s] at which the granular material 1 is discharged to the outside of the hopper 2 during one rotation of the feed screw 3 is a target value (“volume per pitch” × “bulk of the granular material 1” in advance. Specific gravity ”ד filling rate ”). However, the actual discharge amount is affected by the change in the bulk specific gravity for each lot of the granular material 1 and the change in the filling rate due to the change in the height of the granular material 1 in the hopper 2. Must not.

  Therefore, as shown in FIG. 2, the discharge amount during the actual R rotation of the feed screw 3 per second is measured for 5 minutes at regular time intervals, and 300 actually measured values W (1), W ( 2)... W (300) is sampled. In this process, the discharge amount is converted into the discharge amount per one rotation of the feed screw 3 to obtain converted values Wr (1), Wr (2),... Wr (300) and the first actually measured value W (1 ) And a target value Q that is a set discharge amount that can be arbitrarily set by the operator, a control amount S (1) is calculated.

  Further, a correction rotation speed R (2) is calculated as a correction value obtained by adding the control amount S (1) to the rotation speed Rm (1) of the first feed screw 3 at the time of sampling, and the correction rotation speed R (2) is calculated. Assuming that the feed screw 3 is rotated, the discharge amount Ws (2) for 1 second is estimated. From the next time, based on the deviation α (2) between the estimated discharge amount Ws (2) and the target value Q, “proportional constant κ”, “trend constant kh”, and “integration time t” described later are used as variables. The discharge amount Ws (3) for 1 second is estimated according to the multivariable control amount calculation formula including it. The discharge amounts Ws (3), Ws (4),... Ws (11) calculated in this way are hereinafter referred to as simulation values.

  Further, τ = 5 in FIG. 2 indicates that five (n = 300) measured values W (1) to W (300) are divided into five pieces in accordance with the order in which they are measured. Illustrated. In addition, since the time length of each control cycle allocated to 5 minutes required for measurement of 300 actually measured values W (1) to W (300) is 5 seconds in total, in the following explanation, The time length of the control cycle is represented by an algebra T, and T (1), T (2), and T (j) are used as indices for distinguishing individual control cycles.

  In this way, when the control cycle T is set to 5 seconds, the first control amount S (1) of the first control cycle T (1) is “proportional” prepared as a correction factor in advance for the deviation α (2). It is calculated by multiplying by a constant κ, but the second to fifth corrected rotation speeds R (3),... R (6) of the first control cycle T (1) are multivariable control amount arithmetic expressions described later. And the feed screw 3 is rotated at these corrected rotational speeds R (3),... R (6), based on the converted values Wr (2),. Fifth simulation values Ws (3),... Ws (6) are calculated.

  At the beginning of the second control cycle T (2), in other words, the sixth control amount S (6), the simulation value Ws (6) obtained at the end of the first control cycle T (1), the target value Q, Is calculated by multiplying the deviation α (6) by a “proportional constant κ” prepared in advance. The seventh to tenth corrected rotational speeds R (8),... R (11) are multivariables described later. Sequentially calculated according to the control amount calculation formula and assuming that the feed screw 3 is rotated at these correction rotational speeds, the seventh to tenth simulation values Ws are based on the converted values Wr (7),... Wr (10). (8),... Ws (11) is calculated.

  By repeating the above trial calculation for each control cycle T (1), T (2), a data group consisting of 300 simulation values Ws (2),... Ws (300) can be finally obtained. .

  Further, the optimization program sequentially changes the values of “proportional constant κ”, “integration time t”, and “control cycle T” related to a multivariable control amount calculation formula, which will be described later, and calculates, for example, 153 kinds of the above data group. For each of these data groups, the calculation step for obtaining the standard deviation for each of the simulation values Ws (2),... Ws (300) is repeated. Then, the control device 5 recognizes “control cycle T”, “proportional constant κ” and “integration time t” at which the minimum standard deviation is obtained as appropriate values in the calculation step repeated 153 times.

  An embodiment according to the present invention will be described with reference to FIGS. Hereinafter, in addition to the hardware resources indispensable for the execution of the optimization program for multivariable control, the description or illustration of well-known techniques is omitted. Moreover, the English letter attached to the head of the sentence is an index for dividing the calculation step of the simulation by the optimization program.

  The weighing means 4 may be a load cell that constantly monitors the total weight including the granular material 1, the hopper 2, and the feed screw 3. In this case, when the control device 5 opens the electromagnetic opening / closing valve 7 provided at the raw material supply port of the raw material supply device 6, the granular material 1 is put into the hopper 2. Thereafter, when the full sensor 8 disposed on the upper portion of the hopper 2 detects the powder 1, the control device 5 determines that the hopper 2 has reached full and closes the electromagnetic switching valve 7. The control device 5 records the total weight at this time as an initial value, and the total weight that decreases every time from the initial value in the process in which the powder 1 is discharged from the hopper 2 to the outside as the feed screw 3 rotates. Is recognized as a discharge amount.

  Further, when the bulk of the granular material 1 is reduced to a height at which the empty particle sensor 9 is discharged and the empty cup sensor 9 disposed below the hopper 2 does not detect the granular material 1, the control device 5 causes the hopper 2 to The electromagnetic on-off valve 7 is reopened when it is determined that it is full, and the above operation is repeated.

  The control device 5 is capable of timely reading various information associated with the execution of the optimization program, a recording medium 10 including a memory chip or a hard disk in which the optimization program is recorded, a calculation unit including a CPU, and the like. The computer includes a recording unit and an output interface for setting or changing various variables related to a control operation of an adjusting unit described later based on the appropriate value.

  The adjusting means applied in the present embodiment may be a general-purpose product that is commercially available as an industrial process controller. Further, in the present embodiment, such adjusting means exclusively increases or decreases the rotation speed of the electric motor 11 in response to the strength of the output signal, and controls (increases or decreases) the rotation speed of the feed screw 3 rotating together with the electric motor 11. However, the present invention can be widely applied to PI control of all control objects.

  A: To execute the optimization program, sampling is performed at certain time intervals. That is, the rotation speed R [rpm] of the feed screw 3 is kept constant, and the discharge amount [g / s] of the powdered particles 1 is sampled by the measuring means 4 at time intervals of, for example, 0.1 seconds. If this duration is set to about 5 minutes (300 seconds), for example, 3,000 actually measured data d1, d2,... D3000 described as algebra in the following equation are obtained. These actually measured data are transmitted from the measuring means 4 to the recording unit of the control device 5 and stored in the recording unit.

  The duration time of about 5 minutes is approximately equal to the time from when the hopper 2 is full until it becomes full. During this time, the bulk specific gravity and the filling rate of the granular material 1 are changed, and the influence on the discharge amount is as follows. It is intended to be reflected in the measured data d1, d2,... D3000. Therefore, it is desirable to appropriately increase or decrease the duration time according to the capacity of the hopper 2. Also, the above time interval is not limited to 0.1 seconds, and may be increased or decreased to about 0.1 seconds to 0.5 seconds.

B: The discharge amount [g / s] per second is calculated according to the following equation. Thereby, 300 actual measurement data described in FIG. 2 as the discharge amounts W (1), W (2),... W (300) can be obtained.

C: Further, for the discharge amounts W (1), W (2),... W (300), W (1) ′, W (2) ′,. Although [g / s] may be calculated according to the following equation, this calculation process is omitted in this embodiment. Here, the moving average value is an average value of five data (corresponding to a time width of 5 seconds) including the four data following the first data W (1) in the data, and then An average value of five data including four data following W (2) is calculated, and finally an average value of five data including W (300) following W (296) is obtained. The same calculation is repeated until it is obtained.

  Actually, the discharge amount of the granular material 1 varies little by little. In consideration of the case where PI control cannot be performed properly due to this, it is desirable to derive the moving average value of the actual measurement data by repeating the calculation of the moving average several times also in the above step A or B. . Further, the time width of 5 seconds for calculating the moving average of step C may be changed as appropriate in consideration of the sensitivity or accuracy of the measuring means 4. For example, when the load cell detection accuracy does not reach the controllable accuracy with the moving average value of the time width of 5 seconds, or when it is recognized that the data detected by the load cell is unstable and the control based on it is difficult The moving average time width may be extended to 6 seconds or more.

  D: In step A, the rotational speed R of the feed screw 3 actually varies slightly, but the control device 5 recognizes the rotational speed R as the initial rotational speed Rm (1). “M” added to the sign of the initial rotational speed is an index that means that the rotational speed of the feed screw 3 is actually measured by, for example, a tachometer not shown in the figure.

E: Discharge amounts W (1), W (2),... W (300) are converted into discharge amounts per one rotation of the feed screw 3 according to the following formula, and converted values Wr (1), Wr (2 ),... Wr (300) is obtained.

F: target value Q of the discharge amount of the granular material 1 discharged by the feed screw 3 rotating at the rotation speed Rm (1) and the discharge amount W (1 of the granular material 1 actually discharged by the feed screw 3 ) Is calculated according to the following equation: α (1) [g / s].

G: The first control amount S (1) in the first control cycle T (1) is calculated according to the following equation. The following expression is a multivariable control amount calculation expression that gives the control amount S (i) as a function with respect to the deviation α (i), and uses “proportional constant κ”, “trend constant kh”, and “integration time t” as variables. Contains. The proportional term of this multivariable control amount calculation expression is {α (i) × κ}, and the integral term is {α (i) × κ / t}. Here, “i” means an arbitrary integer of 1 or more, and matches the order in which the actual measurement values W (1) to W (300) are measured.

  For example, if the proportionality constant κ = 0.1 (10%), the tendency constant kh = 1, and the integration time t = 1, S (1) = α (1) × 0.1 + α (1) × 0.1 Become. Alternatively, if the proportionality constant κ = 0.2, the tendency constant kh = 1, and the integration time t = ∞, S (1) = α (1) × 0.2.

H: Further, when the control amount S (1) is added to the first rotational speed Rm (1) of the feed screw 3, a corrected rotational speed R (2) is obtained.

I: Assuming that the feed screw 3 is rotated at the correction rotational speed R (2), a simulation value Ws (2) is estimated according to the following equation.

J: Deviation α (2) between target value Q and simulation value Ws (2) is calculated according to the following equation.

K: The second control amount S (2) in the first control cycle T (1) is set to zero, that is, {α (i) × κ} = 0. Calculated according to the following equation. That is, the second control amount S (2) depends only on the integral term of the multivariable control amount calculation formula. The value of kh described in the following equation will be described later.

L: Further, according to the following equation, the control amount S (2) is added to the correction rotation speed R (2) to calculate the correction rotation speed R (3), and then the feed screw 3 is corrected to the correction rotation speed R (3 ), The simulation value Ws (3) is estimated.

M: When the “control period T” is set to 5 seconds as described above, the trial calculation of Ws (i + 1) is repeated four times until i = 5 = τ is reached, so that the simulation values Ws (4) to Ws ( 6) are calculated sequentially. This is expressed in the following equation.

  In the steps J to M, a control amount S (i + 1) based on a deviation α (i + 1) between the second and subsequent simulation values Ws (i + 1) and the target value Q is calculated, and this control amount S (i + 1) is calculated. Is added to the output of the adjusting means to obtain the corrected rotational speed R (i + 2), and the feed screw 3 is rotated at the corrected rotational speed R (i + 2), and a new simulation value Ws (i + 2) is estimated. By repeating the process, the simulation value Ws (τ + 1), that is, Ws (6) is finally obtained.

N: The beginning of the second control cycle T (2), in other words, the sixth control amount S (6) is the simulation value Ws (6) and target value obtained at the end of the first control cycle T (1). Based on the deviation α (6) from Q, it is obtained by repeating substantially the same calculation as the above step G. Supposing that the proportional constant κ = 0.2, the trend constant kh = 1, and the integration time t = ∞, following the above step G, S (6) = α (1) × 0.2. Further, based on the control amount S (6), the calculation substantially similar to the above step L is repeated to obtain the sixth simulation value Ws (7), and then the seventh and subsequent simulation values Ws (8 ),... Ws (11) is obtained sequentially. This is expressed by the following equation.

  O: Deviation α (11) between the simulation value Ws (11) obtained at the end of the second control cycle T (2) and the target value Q by repeating again the operation substantially similar to the above step L. After obtaining the first control amount S (11) of the third control cycle T (3) and further obtaining the eleventh simulation value Ws (12), the twelfth and subsequent simulation values Ws (13),. Ws (16) is obtained sequentially.

  In step O, when the number of control cycles is represented by “j”, the deviation between the simulation value Ws (11) belonging to the end of the second control cycle T (2), that is, Ws (jτ + 1) and the target value Q. A control amount S (jτ + 1) based on α (jτ + 1) is calculated, and further, a correction value R (jτ + 2) obtained by adding the control amount S (jτ + 1) to the output of the adjusting means is obtained, and correction values R (jτ + 2) and (jτ + 2) are obtained. ) The simulation value Ws (jτ + 2) proportional to the product of the first actual measurement value W (jτ + 2) and then the trial calculation for obtaining Ws (jτ + 3) are repeated, so that the simulation value Ws (jτ + τ + 1), that is, Ws (16) is finally obtained. Is included in the subsequent intra-cycle calculation process.

  The simulation value Ws (300) can be finally obtained by repeating the above arithmetic processing for every control cycle.

  Although the trend constant kh = 1 is set in the above step G, the trend constant kh is determined step by step based on Table 1 when the second and subsequent control amounts S (i ≧ 2) are calculated. Table 1 shows integers corresponding to the order of discharge amounts in which k in the parentheses is an actual measurement value. A1 in Table 1 is the difference between the kth discharge amount Ws (k) and the previous discharge amount Ws (k-1), and A2 in Table 1 is the discharge amount Ws (k-1). And the discharge amount Ws (k−2) immediately before that.

For example, when the control device 5 selects the trend weight kw corresponding to the change in the values of A1 and A2 in the above step O, the trend weight kw is determined in five stages from 1 to 5, so the control device 5 The tendency constant kh can be calculated by substituting such tendency weight kw into the conversion formula described in the right column of the table. However, when the value of k is 2 or less and there is no data corresponding to Ws (k−1) or Ws (k−2), that is, the first control cycle T (1) of the simulation as in step G above. Only when it corresponds to the trend constant kh = 1.

The tendency constant kh is determined based on the simulation value, and the control device 5 confirms the value of the tendency constant kh at the first time of each control cycle. Further, the tendency constant kh serves as an index for observing the tendency of the change in the discharge amount to be reflected in the next control by observing the result of the control already performed. In this case, the second and subsequent control amounts S (i) in the first control cycle T (1) can be expressed as a general expression as follows.

  P: The control device 5 considers the above W (1) as the simulation value Ws (1), and the 300 simulation values Ws (1),. Ws (300) is stored as one data group.

  Q: While repeating the values of “proportional constant κ” “integration time t” exemplified in step G and “control cycle T” exemplified in step M above, the steps G to O are repeated, and the data Multiple groups are formed.

  For this, first, one value group is formed by recording the values of “proportional constant κ”, “integration time t” and “control cycle T” one by one in the recording unit of the control device 5. To do. For example, the first numerical group is formed with the control cycle T = 2 seconds; the proportionality constant κ = 10%; and the integration time t = 10 seconds. Subsequently, the control cycle T = 3 seconds; the proportionality constant κ = 20%; the integration time t = 20 seconds, and the second numerical group is formed. The control cycle T = 4 seconds; the proportionality constant κ = 30%; integration time A third numerical value group is formed at t = 30 seconds.

  In this manner, nine values of the control cycle T are prepared in increments of 1 second in the range of 2 seconds to 10 seconds, and 12 values of the value of the proportionality constant κ are prepared in increments of 10% in the range of 10% to 120%. If the integration time t is prepared in the range of 10 seconds to 450 seconds in 45 increments every 10 seconds, 153 (= 9 × 12 + 45) combinations can be made for the reason described later. Further, the numerical value groups 1 to 153 are assigned to the numerical value groups for each numerical value group so that the individual numerical value groups can be distinguished from each other in the recording unit of the control device 5. Record.

R: 153 standard deviations σ (1), σ (2),... Σ (153) for each of the simulation values Ws (2),. ) These standard deviations are stored one by one in the recording unit of the control device 5 together with the values of “proportional constant κ”, “integration time t” and “control cycle T” corresponding to each.

  S: The control device 5 sets σ (32) as a data group in which the standard deviation σ is the smallest among the standard deviations σ (1), σ (2),... Σ (153) as shown in FIG. And the values of “control cycle T”, “proportional constant κ”, and “integration time t” corresponding to this are called from the recording unit and recognized as appropriate values.

  Preferably, the “control period T”, “proportional constant κ”, and “integration time t” corresponding to σ (32) may be further simulated. For example, the 30% proportionality constant κ from which σ (32) was obtained is changed in increments of 1% within a range of 5% above and below, as shown in FIG. 3, and 10 new data groups are calculated. Of these, a data group having the minimum standard deviation σ may be selected.

  Further, the control device 5 adjusts with an operation signal (adjustment means output) so that the target value of the adjustment means of the quantitative supply device shown in FIG. 1 matches the appropriate value. Thereby, the optimal adjustment according to the adjustment means can be performed in a short time of several tens of seconds required for the information processing by the calculation unit without bothering the operator. Further, there is an advantage that it is not necessary to interrupt the operation of the quantitative supply device.

  In the present embodiment, the steps K to O in the first embodiment described above are changed, and only this change will be described below.

K ′: The proportional term of the multivariable control amount calculation expression is {α (i) × κ} = 0, and the integral term is α (i) = α (1). The control amounts S (2), S (3),... S (i) for the second and subsequent control cycles T (1) are sequentially calculated. For example, when the control cycle T is set to 5 seconds, the control amount S is calculated up to S (5).

Here, the values of kh are distinguished as kh (2), kh (3),... Kh (i), respectively, because the expressions of control amounts S (2), S (3),. This is to clarify that the value of kh is determined sequentially based on Table 1 each time it is obtained. In this case, the second and subsequent control amounts S (i) in the first control cycle T (1) can be expressed as a general expression as follows.

  L ′: The control amount S (2) is added to the corrected rotational speed R (2) to calculate the corrected rotational speed R (3), and the simulation is performed assuming that the feed screw 3 is rotated at the corrected rotational speed R (3). The value Ws (3) = Wr (3) × R (3) is estimated.

  The steps K ′ and L ′ are the same as those in the first embodiment described above in that i is an integer and each deviation value α (i) between each simulation value Ws (i) and the target value Q is calculated. There is a difference in that the control amount S (i) is calculated based on the initial deviation α (1) and kh (i).

O ′: The beginning of the second control cycle T (2), in other words, the sixth control amount S (6) is the simulation value Ws (6) obtained at the end of the first control cycle T (1) and the target. Based on the deviation α (6) from the value Q, it is obtained by repeating substantially the same calculation as the above step G. Further, the sixth simulation value Ws (7) is obtained by repeating substantially the same calculation as the step K ′ based on the control amount S (6).

  Further, the calculation substantially the same as the step of L ′ is repeated to obtain the seventh and subsequent simulation values Ws (8),... Ws (11) sequentially, and finally the simulation value Ws (300) is obtained. Thereafter, the control device 5 executes the above step O.

  Next, a description will be given of the case where the appropriate value is reset during the operation by the PI control of the quantitative supply device shown in FIG.

  T: Sampling is continued in a state where the rotational speed of the feed screw 3 is already set to the rotational speed R by the control device 5. This sampling is performed when the value of the discharge amount W (i) is obtained in synchronization with this measurement while measuring the discharge amount W (i) discharged by the feed screw 3 per unit time (1 second). The rotation speed Rm (i) of the feed screw 3 is measured. “M” added to the sign of the rotational speed is an index for distinguishing that the rotational speed of the feed screw 3 is actually measured with respect to R (2),... .

  The numbers of the discharge amount W (i) and the rotation speed Rm (i) obtained by the above sampling may be set so as not to exceed 300, for example. This is because the capacity of the recording unit of the control device 5 is limited. When 5 minutes have passed since the sampling was started, the measured data and the rotational speed data already stored in the recording unit are respectively stored in the oldest order. This is for updating with the latest measured data and rotation speed data. Accordingly, when the calculation unit of the control device 5 accesses the recording unit in response to an operator's command to reset the appropriate value, the measured data W (i) and the rotation speed data sampled during five minutes from that point in time The calculation unit reads Rm (i).

The above i is an integer representing the sampling order, and the discharge amount Wr (i) per rotation of the feed screw 3 is given according to the following equation.

Therefore, the control device 5 converts the discharge amount per one rotation of the feed screw 3 according to the following formula to obtain the converted values Wr (1), Wr (2),... Wr (300).

  U: The control device 5 replaces the rotation speed R described in step D above with R = Rm (i) this time, executes the steps F to S described above, and resets the appropriate value. .

  It should be noted that the present invention can be implemented in a mode in which various improvements, modifications, and variations are added based on the knowledge of those skilled in the art without departing from the spirit of the present invention. For example, for the sampling time interval, the number of various data, or the variables related to the calculation, only the numerical values considering the convenience of explanation are listed, and the conditions for implementing the present invention by the numerical values exemplified above, Or it should not be understood that the operating environment of the optimization program is limited. The present invention also integrates the control device 5, the recording medium 10 on which the optimization program is recorded, and the adjusting means applied to the present embodiment, and performs PI control of any object by the output signal of this integrated device. Comprehensive technical ideas.

  The reason why the numerical value group is set to 153 is based on the following. That is, if the calculation processing of the control device 5 can be speeded up, 4,860 is a combination of 9 control cycles T, 12 “proportional constants κ”, and 45 “control cycles T”. (= 9 × 12 × 45) different data groups may be calculated. Alternatively, if the “integration time t” is set to 10 seconds in the range of 10 seconds to 100 seconds and changed to 50 seconds in the range of 100 seconds to 450 seconds, the “integration time t” is reduced to 16 ways. Thus, 1,728 (= 9 × 12 × 16) data groups can be calculated.

  However, in consideration of the performance of a CPU such as a general programmable controller or a personal computer, it is preferable to set the data group to about 150 ways. Therefore, the steps Q to S are executed under the condition fixed at the integration time t = ∞ (integral term = 0) while changing only the combination of “proportional constant κ” and “control cycle T”, and the standard deviation σ A data group having a minimum value is specified, and the values of “control cycle T” and “proportional constant κ” corresponding to the data group are called as appropriate values from the recording unit, and “control cycle T” and “ The value of “proportional constant κ” is fixed, the steps Q to S are executed while only the “integration time t” is replaced, and the step of specifying the data group that minimizes the standard deviation σ again is performed. When appropriate values are determined for all of “t”, “proportional constant κ”, and “control cycle T”, the numerical value group is substantially 153 combinations.

  The present invention is not limited to the control of the rotational speed of an electric motor or the like, and can be widely applied to control such as opening / closing of a regulating valve, heating, or temperature adjustment of a refrigeration apparatus. In addition, the present invention incorporates an automatic optimization setting device, that is, an optimization program in the adjustment means (controller), and automatically performs simulations at regular or regular intervals regardless of operator's command, as well as when changing materials. By doing so, it is possible to realize the full automation of the optimal adjustment related to the PI control.

The block diagram which shows the structure of the control apparatus which concerns on Example 1 of this invention. The conceptual diagram which shows progress of the calculation step of the optimization program of the multivariable control which concerns on Example 1 of this invention. The conceptual diagram which shows the procedure of the simulation by the optimization program of the multivariable control which concerns on Example 1 of this invention. The conceptual diagram which shows progress of the calculation step of the optimization program of the multivariable control which concerns on Example 2 of this invention.

Explanation of symbols

1: Powder body 2: Hopper 3: Feed screw 4: Metering means 5: Control device 6: Raw material supply device 7: Electromagnetic switching valve 8: Full sensor 9: Empty sensor 10: Recording medium 11: Electric motor

Claims (6)

Negative feedback control of the control object by adding to the output of the adjustment means a control amount calculated based on the deviation between the actual value obtained by measuring the control object proportional to the output of the adjustment means and the target value To the computer
A function of recording a plurality of actual measurement values obtained by repeatedly measuring the control object at regular time intervals in the order in which the control object is measured;
A function of recording a multivariable control amount calculation formula including a plurality of variables and giving a control amount based on the deviation as a function;
A numerical value group including a plurality of numerical values that can be substituted for each of the plurality of variables, a function of recording a plurality of different numerical values from each other, and
A function of sequentially selecting one numerical value group from each of the plurality of numerical value groups, and substituting the plurality of numerical values included in the one numerical value group for a plurality of variables of the multivariable control amount arithmetic expression, respectively. When,
A correction value obtained by adding a new control amount given by the multivariable control amount calculation formula to the output of the adjustment means based on the deviation is obtained, and a product of the correction value and each actual measurement value in the actual measurement value is obtained. A function that repeats the calculation process for each of the simulation values proportional to each time one numerical value group is selected from the plurality of numerical value groups, and
By recording the plurality of simulation values calculated by the calculation process for each time as one data group, the plurality of simulation values calculated by repeating the calculation process are stored as a plurality of data groups. Function to
A function for calculating a standard deviation for each data group in the plurality of data groups and comparing the standard deviations of the plurality of data groups to identify a data group from which the minimum standard deviation is obtained. When,
A function for recognizing the plurality of numerical values respectively assigned to a plurality of variables of the multivariable control amount calculation formula for which a trial calculation of the data group from which the minimum standard deviation is obtained as an appropriate value;
Multivariable control optimization program characterized by realizing Negative feedback control of the control object by adding to the output of the adjustment means a control amount calculated based on the deviation between the actual value obtained by measuring the control object proportional to the output of the adjustment means and the target value To the computer
A function of recording a plurality of actual measurement values obtained by repeatedly measuring the control object at regular time intervals in the order in which the control object is measured;
A function of calculating a deviation α (i) between the i-th actual measurement value W (i) of the n actual measurement values and the target value;
A function of recording a multivariable control amount calculation formula including a plurality of variables in a proportional term and an integral term and giving a control amount based on the deviation as a function;
A numerical value group including a plurality of numerical values that can be substituted for each of the plurality of variables, and a function of recording a plurality of different numerical values from each other;
A function of selecting one numerical value group from the plurality of numerical value groups, and substituting the plurality of numerical values included in the one numerical value group into a plurality of variables of the multivariable control amount arithmetic expression, respectively; ,
A control variable S (i) based on the deviation α (i) is calculated by the multivariable control variable calculation formula into which a plurality of numerical values of the one numerical value group are substituted, and the control variable S (i) is calculated. A correction value R (i + 1) added to the output of the adjusting means is obtained, and a simulation value Ws (i + 1) proportional to the product of the correction value R (i + 1) and the (i + 1) th actual measurement value W (i + 1) is obtained. A function of calculation processing for repeating the trial calculation until the nth simulation value Ws (n) is obtained;
A process of sequentially selecting one numerical value group other than the numerical value group already selected from the plurality of numerical value groups is repeated until all of the multiple numerical value groups are selected, and the arithmetic processing is repeated. A function that repeats each street numerical value group,
The simulation values Ws (i + 1) to Ws (n) sequentially calculated by the calculation process are set as one data group, and the one data group is repeated each time the calculation process is repeated for the plurality of numerical value groups. A function of storing a plurality of the data groups by recording;
The standard deviation σ is calculated for each data group in the plurality of data groups, and the standard deviation σ of each of the plurality of data groups is compared to identify the data group from which the minimum standard deviation σ is obtained. Function and
A function of recognizing the plurality of numerical values respectively assigned to a plurality of variables included in the multivariable control amount calculation formula for trial calculation of a data group from which the minimum standard deviation σ is obtained as an appropriate value;
Multivariable control optimization program characterized by realizing The arithmetic processing is as follows:
A function of allocating a control cycle that is repeated a plurality of times within the time required for measuring the n actual measured values by dividing the n actual measured values by τ according to the order in which the n actual measured values are measured. When,
It belongs to the first control cycle T (1) of the control cycle by the multi-variable control amount arithmetic expression into which a plurality of values of one value group selected from the plurality of value groups are respectively substituted. A control amount calculation process for calculating a control amount S (1) based on a deviation α (1) between the first actually measured value W (1) and the target value is executed, and the control amount S (1) is calculated by the adjusting means. A correction value R (2) added to the output of the output, and a second simulation value Ws (2) proportional to the product of the correction value R (2) and the second actual measurement value W (2);
A deviation α between the second and subsequent simulation values Ws (i + 1) and the target value is obtained by the multivariable control amount calculation formula in which a plurality of numerical values of the one numerical value group are respectively substituted and the proportional term is zero. A control amount calculation process for calculating a control amount S (i + 1) based on (i + 1) is executed, a correction value R (i + 2) obtained by adding the control amount S (i + 1) to the output of the adjusting means is obtained, and the correction value The trial calculation for obtaining the (i + 2) th simulation value Ws (i + 2) that is proportional to the product of R (i + 2) and the (i + 2) th actual measurement value W (i + 2) is the (τ + 1) th simulation value Ws (τ + 1). The function of the calculation process in the first cycle to be repeated until it is obtained,
The last simulation belonging to the first control cycle T (1) or the second and subsequent control cycles T (j) by the multi-variable control amount calculation formula into which a plurality of numerical values of the one numerical value group are respectively substituted. Control amount calculation processing for calculating a control amount S (jτ + 1) based on a deviation α (jτ + 1) between the value Ws (jτ + 1) and the target value is executed, and the control amount S (jτ + 1) is added to the output of the adjusting means. A correction value R (jτ + 2) obtained, and a simulation value Ws (jτ + 2) proportional to the product of the correction value R (jτ + 2) and the (jτ + 2) th actual measurement value W (jτ + 2);
A deviation α (jτ + 2) between the simulation value Ws (jτ + 2) and the target value is calculated by the multivariable control amount calculation formula in which a plurality of numerical values of the one numerical value group are respectively substituted and the proportional term is zero. A control amount calculation process for calculating a control amount S (jτ + 2) based on the control amount is executed to obtain a correction value R (jτ + 3) obtained by adding the control amount S (jτ + 2) to the output of the adjusting means, and the correction value R (jτ + 3) In the subsequent cycle, the trial calculation for obtaining the simulation value Ws (jτ + 3) proportional to the product of the (jτ + 3) th actual measurement value W (jτ + 3) is repeated until the (jτ + τ + 1) th simulation value Ws (jτ + τ + 1) is obtained. Function and
The multivariable control optimization program according to claim 1 or 2, wherein the computer is implemented. The control amount calculation process includes a proportional term and an integral term of the multivariable control amount calculation expression including a proportional constant κ, a trend constant kh, and an integration time t as variables {α (i) × κ}, α A function of recording as {(i) × κ / t};
By calculating S (i) = [{α (i) × κ} + {α (i) × κ / t}] × kh, a function S (given to the value of the deviation α (i) a function of deriving i) as the control amount;
The multivariable control optimization program according to claim 3, wherein the computer is realized by the computer.   By recording the values of the proportionality constant κ, the integration time t, and the control cycle T one by one, one set of numerical values can be formed, and the one set of numerical values can be distinguished from each other. The multivariable control optimization program according to claim 1, wherein the computer has a function of recording a plurality of types of recording. Negative feedback control of the control object by adding to the output of the adjustment means a control amount calculated based on the deviation between the actual value obtained by measuring the control object proportional to the output of the adjustment means and the target value And a computer to
A hopper into which the granular material is charged as the control object;
A feed screw that is provided at the bottom of the hopper and rotates together with an electric motor whose rotational speed increases or decreases in response to the output of the adjusting means;
A weighing unit that measures the mass of the control target discharged from the hopper in proportion to the rotation speed of the feed screw as the actual measurement value;
A quantitative supply apparatus comprising: a recording medium on which the multivariable control optimization program according to claim 1, 2, 3, 4, or 5 is recorded.

Images