JPS642498B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a control method for controlling an injection molding machine to operate under optimal molding conditions. In an injection molding machine, for example, the holding pressure (secondary pressure) of the injection pressure is responsible for insufficient filling and the occurrence of burrs, and the injection temperature, injection pressure (primary pressure), and injection speed are responsible for changes in the internal pressure of the molten resin. , injection speed, mold temperature, etc. are closely related to each other. Product weight and wall thickness are related to injection speed, injection pressure, cylinder temperature, and mold temperature. Sink marks are affected by holding pressure after injection, injection speed, and mold temperature. Many factors, such as those determined by mold temperature and other factors, are intricately intertwined to determine whether a molded product is good or bad. Therefore, it is extremely difficult to select the optimal conditions based on these factors, but at present, the molding conditions are set by the operator based on experience, depending on the type of resin material, the weight, thickness, etc. of the product to be molded. Keep it
Workers visually inspect the condition of the actually molded product and make corrections to the molding conditions based on intuition, and the above corrections are repeated every time a defective product occurs, resulting in inefficiency and loss of work time and materials. It is usually done using a large scale method. For this reason, in recent years, computers have also begun to be applied to control injection molding machines. As this control method, for example, in response to different combinations of material, weight, and wall thickness ratio of the molded product to be injection molded,
A plurality of programs that set a number of operating conditions for the molding machine are stored in the computer, and an appropriate program is selected from the program group according to the material, weight, and wall thickness ratio of the molded product to perform molding. A method is known in which the actual operating values detected by sensors installed in each part of the machine are fed back and controlled so that the operating values are as close as possible to the setting conditions of the selected program. It is being However, in this case, it is difficult to create a program and requires a program selection operation, and basically, each center detects obvious abnormalities in which the operating value exceeds the upper or lower limit. However, since this control method feeds back this detection signal to the control system, highly accurate control with a small fluctuation range cannot be expected. Therefore, in preparation for the occurrence of defective products, multiple correction programs are stored to change the operating conditions of the molding machine according to the defective state, and appropriate correction programs can be applied depending on the defective item of the molded product. The circuit configuration is complicated and the operability is poor. In view of the current situation, the present invention proposes a control method that can improve the precision and quality of molded products by controlling an injection molding machine using an algorithm (an abnormality prediction method). It is. Here, the algorithm is a method that attempts to identify abnormalities at the stage of their signs based on the movement of data over time. Twenty types of this method have already been established by making full use of advanced statistics, but the basic data for all of them is temporal changes in process data. A determination is made as to whether or not there is an abnormality with respect to this time-based process data according to a pre-established algorithm. For example, in a typical case where it is necessary to judge that there is a sign of an abnormality when something over a certain value occurs more than a certain number of times, the algorithm must refer to the hourly data. Not only process data, but also predetermined "setting values (code setting values)" such as "certain value" and "certain number of times," and "intermediate progress (code setting value)" that shows how many times such an event has already occurred. code information)”
is necessary. Cases other than this example can also be summarized in the same way. In short, all algorithms refer to time-series data, code setting values, and code information, and determine whether or not there are signs of abnormality while updating the code information from time to time. Furthermore, the time-series data referred to by the algorithm is assumed to be several types of process data, but can basically be divided into the following three types.

[Table] Process data In all of these cases (1) to (3), the basic idea of abnormality prediction is to make full use of various statistical achievements,
This is pattern recognition, just like a human looking at the data chart of a measuring instrument and determining whether something is abnormal or not. Embodiments of the present invention will be described below based on the drawings. FIG. 1 shows an injection molding machine controlled by the control method of the present invention, and FIG. 2 shows a control panel. A
1 is an injection device having a cylinder 1, a raw material supply screw 2, a heater 3, etc.; B is a molding mold consisting of a fixed mold 4 having a molten resin injection port 4a and a movable mold 5; The molded product inside the mold B is automatically taken out at a certain timing by an owner's hand (not shown) or the like. 7 is a mold temperature adjustment means, which includes a hot water tank 8, a cold water tank 9, a hot water circulation pump Pa, a cold water circulation pump Pb,
It is composed of three-way valves Va, Vb and solenoid valves V ₁ , V ₂ , V ₃ , V ₄ and the like for switching hot water and cold water flow paths. S ₁ and S ₂ are infrared sensors of known structure each using a pyroelectric detector, and one infrared sensor S ₁ is for detecting the residue 10 in the mold B,
The injection port 4a of the fixture 4, where residue 10 tends to occur, is placed in the detection field of view, and the other infrared sensor _S2 is for measuring the mold temperature, and is placed at an appropriate point on the inner surface of the movable mold 5. It is placed within the detection field of view. S ₃ is a sensor for detecting the injection temperature, and S ₄ is a sensor for detecting the temperature of cylinder 1.
S ₃ and S ₄ are composed of thermocouples or other temperature sensing elements. S ₅ is a sensor for detecting injection pressure (primary pressure), S ₆ is a sensor for detecting holding pressure (secondary pressure) after injection, and S ₇ is a sensor for detecting injection speed, which is linked to the rotation of the screw 2. It consists of a potentiometer. S ₈ is a sensor for detecting the injection speed, S ₉ is a sensor for detecting mold clamping pressure of mold B, and S ₁₀ is a sensor for detecting damage to mold B (damage due to twisting, distortion, etc.). It consists of a microphone etc. that collects the sounds generated by the system. _S11 is a sensor that emits a synchronizing signal as the mold B opens and closes, and is composed of a phototransistor. Detection signals from each of the sensors are input to the microcomputer CPU. The microcomputer CPU has a learning function and a monitoring function, and in the learning mode, when it is visually confirmed that the molded product is good, it learns the optimal conditions at the detection points by each of the sensors. In other words, when performing molding work,
Power push button 11, start push button 12, learning push button 1
3 in this order, perform molding for about 20 cycles, visually check the condition of the molded product, and if it is good, press the learning result registration push button 14,
The molding conditions at that time are stored, and then the monitoring push button 15 is pressed to switch to the monitoring mode. In the monitoring mode, the process data or composite process data based on the detection signals of each sensor is processed using an appropriate algorithm stored in ROM in advance to predict abnormalities, and based on the abnormality detection, the control output from the microcomputer CPU is used to The molding machine is controlled so that the detection signal is as close as possible to the optimal conditions stored in the learning mode. For example, regarding insufficient filling and burrs, which are defective elements of molded products, the microcomputer CPU analyzes the detection signal of sensor S ₆ using a certain algorithm, and when signs of abnormality appear, it issues a control signal and controls the proportional control valve. , operates a servo valve, etc. to control the holding pressure after injection. Sensors are used to detect changes in mold pressure of molten resin.
Composite process data from detection signals such as S ₂ , S ₃ , S ₅ , S ₇ , S ₈ etc. is analyzed using an appropriate algorithm, and when signs of abnormality appear, appropriate signals are output from the microcomputer CPU. control. For example, the mold temperature is controlled by the refrigerant flow rate and temperature, and the injection temperature is controlled by the heater 3.
The injection pressure is controlled by proportional control valves, servo valves, etc., and the injection speed is controlled by the rotation of the screw 2. For the weight and dimensions of the molded product, the sensor
Composite process data from detection signals such as S ₂ , S ₄ , S ₅ , and S ₇ is analyzed using an algorithm, and the mold temperature, cylinder temperature, injection pressure, and injection speed are controlled when signs of abnormality appear. Regarding sink marks, we use an algorithm to analyze combined process data from the detection signals of sensors S ₂ , S ₆ , and S ₈ , and when signs of abnormality appear, we adjust the mold temperature, post-injection holding pressure, and injection speed. Control. Reference numeral 16 denotes a printer for recording data stored in the learning mode and data stored in the monitoring mode, and a printing operation is performed by pressing an operation end push button 17. 18... is the sensor S ₁ ~
Display section for digitally displaying the detection signal of S ₁₀ , 19
...is the channel switch. The control system for the injection molding machine according to the present invention has the above-mentioned configuration. It visually confirms that molding is being performed normally, and learns the molding conditions at that time. In the monitoring mode, each molding machine In order to control the detection signal from the sensor in the body part to be as close as possible to the optimal conditions stored in the learning mode, a number of types of programs are stored in advance, and the necessary program is selected from these. There is no need to create a program as in the case of controlling a molding machine based on. In particular, we use algorithms to analyze the detection signals from each sensor and identify abnormalities at their earliest signs.
Since control signals are issued based on this, high-precision control with small fluctuation range is possible, improving the precision of molded products.
High quality can be achieved, and better operability can be obtained compared to the case where multiple types of correction programs are stored and the appropriate correction program is selected depending on the defective item of the molded product. There is an effect.

[Brief explanation of drawings]

The drawings show an embodiment of the present invention, with FIG. 1 being a configuration diagram of an injection molding machine, and FIG. 2 being a front view of a control panel. S ₁ to S ₁₀ ...Sensor, CPU...Microcomputer.

Claims (1)

[Claims] 1 Each part of the injection molding machine has cylinder temperature, injection temperature, injection pressure, holding pressure after injection, injection speed, mold temperature, residue, mold clamping,
Sensors are installed to detect mold clamping pressure, etc., and the detection signals from these centers are input to a computer that has learning and monitoring functions.In learning mode, it is visually confirmed that molding is being performed normally. At that point, the computer is made to learn the optimal conditions for each part, and then in monitoring mode,
Detection signals from each sensor are processed by an algorithm to predict an abnormality, and a control signal output from the computer based on this abnormality prediction is used to
A control method for an injection molding machine characterized by controlling the injection molding machine so that detection signals of each sensor are as close as possible to optimal conditions stored in a learning mode.