WO1999036280A1

WO1999036280A1 - Air-conditioning device for vehicles and method for controlling the device

Info

Publication number: WO1999036280A1
Application number: PCT/JP1998/003528
Authority: WO
Inventors: Yoshihiro Adachi; Shigenobu Itoh; Mamoru Masauji
Original assignee: Zexel Corporation
Priority date: 1998-01-19
Filing date: 1998-08-05
Publication date: 1999-07-22

Abstract

A control logic (1a) for an air conditioner for vehicles, comprising a temperature-estimating unit (1b) for estimating the room temperature of the vehicle which is a main factor of the control logic. The temperature-estimating unit (1b) is constituted of a neural network which receives input signals representing environmental factors of the vehicle, such as temperature Ti measured with an internal temperature sensor (6), the external air temperature TA, the opening P(%) of an air-mixing door (2e), the blower duty ratio D(%), etc., and the state of the air-conditioner, and outputs an estimated value TN of the room temperature. The air-conditioner is controlled by using the estimated value TN as a feedback value, and the room temperature is quickly regulated to a target temperature TZ.

Description

Description Vehicle air conditioner and control method

The present invention relates to an air conditioner, and more particularly to an air conditioner for a vehicle for appropriately controlling the temperature in a passenger compartment to a target temperature and a control method thereof. Background art

The first 1 Figure is a diagram showing a control flow of a conventional vehicle weight air conditioner, a conventional air conditioning apparatus for a vehicle, and the target temperature T _z set by the temperature setting unit 4, the installed car cabin The control logic 1 p is used to control the measured temperature T i measured by the inside air temperature sensor 6 as a room temperature detection means as a parameter, and the control logic 1 p controls the temperature of the air-conditioned air blown into the vehicle compartment. Control the air volume etc. to control the vehicle interior temperature T. To keep it properly. However, since the inside air temperature sensor 6 is usually installed below the front panel, the measured temperature Ί \ is the temperature near the occupant's seating position due to the effects of outside air temperature and solar radiation. T. Is not necessarily the same. Therefore, in order to bring the measured temperature Ti of the inside air temperature sensor 6 closer to the actual vehicle interior temperature T0, the vehicle is provided with an outside air temperature sensor and a solar radiation sensor, and the measured temperature Τ \ is determined based on the output of each sensor. The measurement temperature T i is corrected or corrected according to the air-conditioning air blowing mode.

However, in order to correct the difference between the measured temperature T i measured by the inside air temperature sensor 6 and the actual vehicle interior temperature T 0, there are many parameters used in the above control logic, so that the matching work takes time. There was a point. Furthermore, in a vehicle air-conditioning system that blows conditioned air to the front and rear portions of the vehicle, the condition of the front and rear conditioned air affects the other conditioned air. The task of matching the control logic becomes more difficult.

Japanese Patent Application Laid-Open No. 6-1955323 also discloses that, as shown in FIG. 12, a target temperature T ₇ , a measured temperature of the internal air temperature sensor 6 外, an outside air temperature T _A , and a solar radiation amount T _s. And the input signal and There is disclosed a technology for controlling the amount of air-conditioned air to be blown using the neural network type additional learning device described above. In the neural network, the final airflow is controlled by learning and calculating each arithmetic expression such as the target air outlet temperature, air outlet mode status, and blower air volume. However, there are problems that the learning is difficult to converge and that the learning time becomes longer because of the large number of teacher signals used in the training.

The present invention has been made in view of the conventional problems, and accurately adjusts a control logic by accurately estimating a vehicle interior temperature which is a main element of a control logic of an air conditioner. It is an object of the present invention to provide a vehicle air conditioner capable of controlling the temperature with high accuracy and a control method thereof. Disclosure of the invention

The control method for a vehicle air conditioner according to the present invention includes: setting a temperature model for estimating a vehicle interior temperature from an output value of a vehicle interior temperature detecting means for detecting a vehicle interior temperature; The temperature and air volume of the conditioned air blown into the cabin are adjusted based on the estimated value to control the temperature in the cabin. That is, instead of the output value of the vehicle interior temperature detecting means itself, an estimated value of the vehicle interior temperature that is very close to the actual temperature of the vehicle interior is obtained, and the temperature in the vehicle interior is controlled using the above estimated value. The temperature in the passenger compartment is quickly controlled to the target temperature.

In addition, the control method of the vehicle air conditioner of the present invention is characterized in that the temperature model is formed by a dual network in which an environmental factor of the vehicle and a state of the air conditioner are input signals and an estimated value of the temperature in the vehicle compartment is an output value. The system estimates the temperature of the cabin temperature with high accuracy. Further, in the control method of the vehicle air conditioner according to the present invention, the environmental factors may be the outside air temperature and the amount of solar radiation, and any one of the information of the state of the air conditioner, the blow mode, the blow temperature, the blow air volume, and the position of the suction port. Estimate the vehicle interior temperature with a small number of input signals.

The vehicle air conditioner of the present invention includes a temperature estimator constituted by a neural network, and adjusts the temperature and the amount of air-conditioned air blown into the vehicle interior based on the estimated value of the vehicle interior temperature from the temperature estimator. Thus, the temperature in the vehicle compartment is controlled. this thing As a result, an estimated value of the vehicle interior temperature that is extremely close to the actual vehicle interior temperature is obtained, and the vehicle interior temperature is controlled based on the estimated value, so that the vehicle interior temperature quickly reaches the target temperature.

In addition, the vehicle air conditioner of the present invention controls the temperature in the vehicle cabin by adjusting the temperature and air volume of the air-conditioned air blown into the vehicle cabin using the above-mentioned estimated value as a feedback value, thereby achieving accurate and quick temperature control. Is what you do.

Further, in the vehicle air conditioner of the present invention, the environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is any combination of the following information: a blow mode, a blow temperature, a blow air volume, and a suction port position. The entire system is designed to reliably estimate the temperature in the cabin.

Still further, in the air conditioner for a vehicle according to the present invention, the temperature data of the outlet other than the outlet set in the outlet mode is fixed to a predetermined value in the outlet temperature data. is there. As a result, unnecessary temperature fluctuations were suppressed, and the estimated value of the temperature in the vehicle interior was obtained more accurately.

In addition, the vehicle air conditioner of the present invention includes a temperature estimator configured as a neural network that outputs an estimated value of the temperature at the front part and the rear part of the vehicle interior as an output value. It controls the temperature in the cabin by adjusting the temperature, air volume, etc. of the air-conditioned air sent to the rear and rear sections. As a result, even when the set temperature and air flow are different between the front and rear seats, control is performed in consideration of the mutual influences, and the temperatures at the front and rear in the passenger compartment are quickly adjusted to each. Set the target temperature. Furthermore, the vehicle air conditioner of the present invention controls the temperature of the vehicle interior by adjusting the temperature and air volume of the conditioned air blown to the front and rear portions of the vehicle interior using the above two estimated values as feedback values, The temperature at the front and rear of the vehicle compartment is accurately and quickly set to the target temperature.

Further, in the vehicle air conditioner of the present invention, the environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is the front and rear blow modes, the blow temperature, the blow air volume, and the suction port position. Any combination or all of the information is used to ensure that control is performed in consideration of the settings of the front and rear seats. Further, in the vehicle air conditioner of the present invention, the teacher signal used at the time of learning of the neural network is set as an average temperature of a position corresponding to a head and a foot of a driver's seat and a passenger's seat, and a vehicle interior temperature is obtained. The difference between the estimated value and the actual cabin temperature is made extremely small.

Further, the vehicle air conditioner of the present invention is configured such that the input state of the teacher signal used during learning of the neural network is changed from 0.02 to 0.98, and the neural network learning is performed. Learning efficiency and safety are improved. Further, in the vehicle air conditioner of the present invention, the absolute value of the error between the output value in each input range of the linear function and the output value of the sigmoid function is set to 3%. The approximation is made by a function in which the input range and the coefficient of the linear function are set so as to be within the range. As a result, the number of required memories can be reduced, and the calculation time can be significantly reduced, and the temperature in the vehicle interior can be more quickly set to the target temperature. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a diagram illustrating a configuration of a vehicle air conditioner according to a first embodiment of the present invention, and FIG. 2 is a diagram illustrating a configuration of a neural network of a temperature estimator according to the first embodiment. It is. FIG. 3 is a diagram showing a sigmoid function used in the neural network. FIG. 4 is a diagram showing a relationship between an estimated value and a measured value of the vehicle interior temperature by the temperature estimator of the first embodiment, and FIG. 5 is a control flow of the vehicle air conditioner according to the first embodiment. FIG.

FIG. 6 is a diagram showing the configuration of a neural network of a temperature estimator according to the second embodiment of the present invention. FIG. 7 is a diagram showing the estimation of the vehicle interior temperature by the temperature estimator according to the second embodiment. It is a figure showing the relation between a value and an actual measurement value.

FIG. 8 is a diagram showing an approximation method of a sigmoid function according to the third embodiment of the present invention, and FIG. 9 is a diagram showing an error between the sigmoid function and a linear function. FIG. 10 is a diagram showing a control flow of the vehicle air conditioner according to Embodiment 3 of the present invention.

FIG. 11 is a diagram showing a control flow of a conventional vehicle air conditioner, and FIG. FIG. 2 is a diagram showing a configuration of a conventional neural network of a vehicle air conditioner. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings.

Best mode 1.

FIG. 1 is a diagram showing a configuration of a vehicle air conditioner according to a first embodiment of the present invention, wherein 1 is a control device, 2 is an air duct, 3 is a setting panel for setting the temperature and air volume of the conditioned air, 4 temperature settings for outputting to the control unit 1 sets the target temperature T _z of the vehicle interior which is input from the setting panel 3, 5 or perform opening degree adjustment of the air air mix door 2 e in the air duct 2, the inside and outside air A driving device for switching the switching door 2a and the outlet switching door, and for driving the blower 2b, 6 is an inside air temperature sensor as room temperature detecting means usually installed below the front panel, and 7 is The outside air temperature sensor is installed near the bumper of the vehicle, and the solar radiation sensor 8 is installed above the front panel.

The air duct 2 includes an inside / outside air switching door 2a that adjusts a ratio of inside air and outside air introduced into the air duct, a blower 2b that blows air taken in from the inside / outside air switching door 2a, and cools the blown air. The temperature of the blown air is controlled by controlling the amount of air that passes through the heater 2d of the blown air cooled by the evaporator 2p, based on the evaporator 2p, the heater 2d that warms the blast air, and the opening / closing degree of the heater. An air mix door 2e to be used and an outlet switching door 2h for distributing the temperature-controlled air to the upper outlet 2f and the lower outlet 2g. The upper outlet 2 f is provided with an upper outlet temperature sensor 2 m for detecting the temperature of the air blown from the upper outlet 2 f, and the lower outlet 2 g is provided with a lower outlet 2 g from the lower outlet 2 g. A lower outlet temperature sensor 2n that detects the temperature of the blast air is installed.

The driving device 5 includes a control device of the actuator 1c that drives the inside / outside air switching door 2a, a control device of the control device 2c for controlling the rotation speed of the blower 2b, and a laser door 2e. Actuator drive 3c control device and outlet switching door

(Blow-up mode switching door) Actuator to control 2h 4c with control device Or a control device for controlling other parts. The illustration of each control device is omitted.

The control device 1 transmits a control signal for controlling the opening degree of the air mix door 2 e according to the environmental factors such as the outside air temperature and the condition (control factors) of the air conditioning equipment such as the air-conditioning air blowing mode. Control logic 1a to be output to 5 and measured temperature 1 from inside air temperature sensor 6 and information on environmental factors and air conditioner status from control logic 1a as input signals and output to control logic 1a above A temperature estimator 1b configured by a neural network that outputs an estimated value 1 of the temperature in the vehicle compartment as an output value. The above control factors include, for example, the output from the outside air temperature sensor 7 and the solar radiation sensor 8, the switching mode of the inside / outside air switching door 2a, the opening degree of the air mixing door 2e, and the switching mode of the outlet switching door 2h. The drive voltage for speed control of the blower 2b, etc. The switching mode of the inside / outside air switching door 2a has two modes, an inside air introduction mode and an outside air introduction mode.The switching mode of the outlet switching door 2h includes a vent mode, a foot blowing mode, There are four modes, bi-level mode and defrost mode. Fig. 2 is a diagram showing the configuration (temperature model) of the neural network in the temperature estimator 1b. The neural network is a three-layer hierarchical network consisting of an input layer, a hidden layer, and an output layer.

The input signals are the outside air temperature T _A (° C) from the outside air temperature sensor 7, the measured temperature 1 \ (° C) from the inside air temperature sensor (Inc.s) 6, and the opening P of the air mix door 2 e. (%), Blower duty ratio D (%) corresponding to the drive voltage that controls the speed of blower 2b, foot outlet temperature _TF (° C) from lower outlet temperature sensor 2n, upper outlet temperature sensor 2m Venting temperature T _B (° C) from the outlet, a blowing mode switching signal M _x indicating the switching mode of the outlet 2 h set on the setting panel 3, indicating the inside / outside air switching mode of the inside / outside air switching door 2 a made from Intel Ichiku signal M _y, and insolation _{^{T s (Kc al / m 2}} hour) from solar radiation sensor 8, the output value is an estimate of the temperature of the passenger compartment T _N (° C).

In this neural network, a sigmoid function as shown in the following equation (1) is used.

y _j = l / (1 + exp (-| χ, |) …… (1) Here, χ ”is the value obtained by subtracting the bias bj from the value obtained by multiplying each input signal V i to the layer i by the weight w (X j = ∑V i · w _u —b”), and yj is The output signal from layer i (the input signal to layer :) '). Note that this sigmoid function outputs 0 to 1 when the input variable is (1∞ to + ∞) as shown in FIG.

The temperature estimator lb uses the four-point average temperatures of the driver's seat and the passenger's seat at positions corresponding to the head and feet as teacher signals when learning the neural network, and estimates the temperature in the above-mentioned cabin. The value of weight w and bias bj for each input signal in the formula for calculating the value _TN is determined by learning, and at the time of control, the estimated value _TN of the temperature in the vehicle compartment for each of the input signals is controlled by control logic. Output to 1a.

As the input state to the neural network, each input signal is normalized from the minimum value to the maximum value of the measurement data from 0 to 1 so that the weight Wij for each of the above input signal types can be evaluated equally. ing. On the other hand, the teacher signal takes the maximum and minimum values of the measurement data from 0.02 to 0.98 in consideration of the fact that 0 and 1 are saturation output values as output characteristics of the sigmoid function. Is becoming In other words, in learning a neural network using the above sigmoid function as an input function of a neuron, considering the learning efficiency and safety, a range of 0 which is slightly narrower than that of normalizing the teacher signal from 0 to 1 is considered. The convergence speed is faster and more effective when the normalization is performed from .02 to 0.98.

Graph Figure 4 is have your temperature estimator 1 b made of a neural network as described above, was thoroughly learned by the learning under the following conditions, estimating the vehicle interior temperature T ₀ by entering Isseki outside air temperature de It is. The above vehicle interior temperature T. Is the average value of the measured temperatures measured at the positions corresponding to the head and feet of the driver's seat and the passenger's seat. Learning conditions

Outside air temperature 10 ~ 35 (.C)

Insolation 0 to 660 (Kcal / m ² hour) Vehicle speed ... Idle to 40 (Km / h) equivalent

Looking at the change in the output value (estimated value T _N ) of the temperature estimator 1b shown by the solid line, the maximum error is 4.9 ° C when the outside air temperature changes suddenly, but the average of the absolute value of the error is 0. The temperature inside the vehicle, T, which is as small as 83 ° C and indicated by 〇. Almost follow the change in c. Compared to the change in measured temperature Ti from s (internal temperature sensor) 6, the temperature T in the vehicle compartment is much higher. It can be seen that the value is close to.

The average of the maximum error and the absolute value of the error when the input signal shown in Fig. 2 is learned by deleting each signal is described below.

(1) Air mix door opening signal removal

The maximum error is 2.8 ° C, and the average absolute error is 0.64 ° C

(2) Intake signal removal

The maximum error is 2.4 ° C and the average absolute error is 0.90 ° C

(3) Removal of blow-out temperature signal

The maximum error is 4.0 ° C, and the average absolute error is 0.69 ° C

(4) Air mix door opening signal and intake signal removal

The maximum error is 2.8 and the average absolute error is 1.04 ° C

From the above results, the vehicle interior temperature T even when the input signal is reduced. The estimated value T _N close to can be obtained. For example, for the input signal shown in FIG. 2, the foot air temperature T _F (° C), the vent blowout temperature T _B (° C), to remove three of Intel Ichiku signal M _y, the number of input signal Even if the number is set to 5, the maximum error is 1.5 ° C and the average of the absolute error values is 0.5 ° C, which is enough to meet the specifications.

Next, a method of controlling the cabin temperature of the vehicle air conditioner will be described with reference to the control flow of FIG. The control logic la is a control factor that becomes an input signal of the temperature estimator 1b from each control factor consisting of an environmental factor such as the input outside air temperature T _A and an air conditioner state such as an air conditioning air blowing mode. Is extracted and output to the temperature estimator 1b. The temperature estimator lb receives the control factor extracted above and the measured temperature 1 \ from Inc. s (internal temperature sensor) 6 as input signals, and estimates the interior temperature T _N of the vehicle interior temperature using a neural network. And outputs it to the control logic 1a. Control logic 1 a, said from the control factors and the estimated value T _N, as vehicle interior temperature T 0 is set target temperature T _z at a temperature setter 4, the opening degree of the air mixing door 2 e [rho The feedback control of the temperature in the passenger compartment of the control element of the air duct 2 such as () is performed according to the preset control logic 1a. That is, the temperature in the cabin is measured by the inside air temperature sensor 6 and then input to the temperature estimator 1b, and is sent to the control logic _1a as the estimated value _TN. It is eavesdropped. Therefore, the control logic la is not the measured temperature Ί \ from the internal air temperature sensor 6 but the vehicle interior temperature T. Since the estimated value _TN of the cabin temperature from the temperature estimator 1b, which is very close to the above, is used as the feedback value, it is possible to accurately and quickly bring the cabin temperature to the target temperature _Tz. it can.

Thus, according to the best mode 1 of the present invention, the measured temperature T i from the inside air temperature sensor 6, the outside air temperature T _A , the opening degree P (%) of the air mixing door 2 e, and the blow duty ratio D (%) As input signals, and a temperature estimator lb composed of a neural network that uses the estimated value _TN of the cabin temperature as an output value as input signals, and a control logic 1a In the above, control is performed using the above-mentioned estimated value T _N as a feedback value, so that the temperature in the vehicle compartment can be accurately and quickly set to the target temperature T _z .

Furthermore, learning of the neural network is easy because learning of the temperature estimator lb is easy because the teacher signal is only the four-point average temperature at the positions corresponding to the head and foot of the driver's seat and the passenger's seat. The learning of the estimated value T _N can be performed in a short time. The estimated value T _N of the vehicle interior temperature used in the control logic 1 a is the vehicle interior temperature T. Since the value is very close to, not only does the matching accuracy of the control logic 1a greatly improve, but also the time required for the matching process and the number of wind tunnel experiments in the matching process can be significantly reduced. Furthermore, even in the design of air conditioners of different types of vehicles, the control temperature of the internal temperature sensor 6 as a control factor is simply changed to the estimated value 1 of the vehicle interior temperature at the control port jack 1a. Since there is no need to adjust the temperature, there is no need to change the control logic 1a for each vehicle type, and only the temperature model for each vehicle type needs to be adjusted, thus significantly improving development efficiency.

In the best mode 1 of the present invention, the temperature estimator 1b is obtained from the environmental factors such as the outside air temperature input to the control logic 1a and the control factors such as the state of the air conditioner such as the air-conditioning air blowing mode. Although an element to be an input signal is selected and output to the temperature estimator 1b, each of the control factors may be independently input to the control logic 1a and the temperature estimator 1b. Also, the control factors input to the control logic 1a do not need to be all the parameters used in the control logic 1a, and the parameters used in the control logic 1a are not necessarily the temperature estimator 1b. Need not include all of the input signals Needless to say.

Best mode 2.

FIG. 6 is a diagram showing a configuration (temperature model) of a neural network according to the second embodiment of the present invention. As in the first embodiment, the neural network has an input layer, a hidden layer, This is a three-layer hierarchical network consisting of an output layer.

The input signals are the measured temperature Ί \ (° C) from the inside air temperature sensor (Inc. s) 6, the outside air temperature T _A (° C) from the outside air temperature sensor 7, and the amount of solar radiation T _s ( Kc al / m ² hour), blower duty ratio D (% corresponding to a drive voltage of the blower 2 b), the set blowing mode switching signal M _x at setting panel 3, foot from the lower air outlet temperature sensor 2 n It consists of the outlet temperature T _F (° C) and the vent outlet temperature T _B (° C) from the upper outlet temperature sensor 2 m, and the output value is the estimated value T _N (° C) of the cabin temperature. . However, the foot outlet temperature _TF (° C) and the vent outlet temperature T _B (° C) are defined as a predetermined value (for example, the temperature data of outlets other than the outlets set in the outlet mode). , 25 ° C).

In addition, as the input state to the neural network, each input signal is normalized from the minimum value of the measurement data to the maximum value from 0 to 1 so that the Eight for each type of the input signal can be evaluated in the same manner. I have to. On the other hand, in consideration of the fact that 0 and 1 are saturation output values as output characteristics of the sigmoid function, the teacher signal is normalized from the minimum value to the maximum value of the measured data from 0.02 to 0.98. are doing.

Figure 7 is have your temperature estimator 1 b made of a neural network as described above, was thoroughly learned by the learning conditions below, is a graph estimating the vehicle interior temperature T ₀ by entering the outside air temperature data . The above vehicle interior temperature T. Are the average values of the measured temperatures measured at the positions corresponding to the head and feet of the driver's seat and front passenger seat, respectively. Learning conditions

Outside air temperature 10 to 3.5 (° C)

Looking at the change in the output value (estimated value T _N ) of the temperature estimator 1 b shown by the solid line, the maximum error is 1.9. C, the average of the absolute value of the error is 0.5 ° C, Temperature is the input signal (maximum error; 4.9 ° C, average of absolute value of error; 0.83 ° C)), and the vehicle interior temperature T indicated by ② in the figure. Good follow-up to changes in This is because not only the factor used as the input signal is properly selected, but also the temperature data of the outlets other than the outlets set in the blowout mode is fixed to a predetermined value (for example, 25 ° C). As a result, unnecessary temperature fluctuations have been eliminated, and the estimated value T _N of the vehicle interior temperature can be obtained more accurately. Note that the broken line in the figure is the detection temperature of Inc. s (internal temperature sensor) 6.

Best mode 3.

In the above first and second best modes, the sigmoid function shown in equation (1) was used for the neural network. However, approximation of this sigmoid function with a plurality of straight lines required the temperature estimator 1b. Since the number of memories can be reduced and the calculation time can be significantly reduced, the temperature in the vehicle compartment can be more quickly brought to the target temperature _Tz .

In other words, when the sigmoid function shown in equation (1) is programmed with, for example, an 8-bit built-in microcomputer, an exponential function and a floating-point arithmetic library are required to maintain accuracy, so that ROM capacity and Calculation time increases. On the other hand, since the linear function does not include division, there is no need for floating-point arithmetic, so that even if integer arithmetic is used, the arithmetic accuracy can be maintained, and the ROM capacity and calculation time can be reduced. Therefore, when the neural network learns, the sigmoid function of the above equation (1) is used, and when the trained network is programmed into an actual embedded microcomputer, as shown in FIG. A function approximating the sigmoid function of the equation with 17 straight lines (linear function; yi = a iX + bi (i = 1 to 17)) whose error is within ± 0.005, for example, By substituting the sigmoid function in equation (1), the ROM capacity and calculation time can be reduced, and the calculation accuracy can be maintained.

Fig. 9 shows the error between the sigmoid function of equation (1) and the function obtained by approximating the sigmoid function with the linear function. If there are 17 straight lines, the magnitude of the error is The range can be within ± 0.005.

In the above example, the sigmoid function is set so that the error is within ± 0.005. Although 17 straight lines were approximated, there is no practical problem if the error is within ± 0.03 (± 3%).

In addition, the function approximated above does not necessarily have to be a broken line, and in some cases, it may be better to make the function discontinuous in order to reduce the number of straight lines (the number of divisions of the input range) and increase the calculation speed.

In the first, second, and third embodiments, the case where the logarithmic sigmoid function of Equation (1) is used has been described. However, a tanh sigmoid function as shown in Equation (2) below is used. Of course, other types of sigmoid functions may be used.

Best mode 4.

In the above-described best mode 1, 2, has been described ordinary type air conditioner for a vehicle, in each of the front seat and a rear seat, the set temperature T _.zeta.1, T _{?? 2} and air volume W _z! , W _Z2 can also be set in a one-box twin type vehicle air conditioner, such as the environmental factors such as the outside air temperature, the state of the air conditioning equipment such as the air-conditioning air blowing mode, and the inside air temperature sensor of the front seat (front seat temperature) sensor) 6 a and an inside air temperature sensor (rear temperature sensor in the rear seat) as a measurement temperature i \ have the input signal and T _i2 from 6 b, the front seat estimated temperature T _N i and rear estimated temperature T _N2 by controlling the vehicle dual air conditioner based on providing the temperature estimator 1 B configured in a neural network to an output value in the above front seat estimated temperature T _N i and rear estimated temperature T _{N 2} the passenger compartment The temperature of the front and rear seats can be accurately and quickly set to the respective target temperatures Ζ _Ζ1 , Τ _{それ} ₂ .

FIG. 10 is a control flow chart showing a method of controlling the temperature in the passenger compartment of the vehicle air conditioner according to the fourth embodiment of the present invention. However, in order to avoid complexity, the control factors input to the front seat control logic 11 and the rear seat control logic 12 have been omitted.

The temperature estimator 1B calculates the outside air temperature Τ _日 , the amount of solar radiation T _s , the amount of air blown by the front seat 9 and the rear seat 10, the air mix door opening of each of the front and rear air ducts (not shown), the front and rear Each of the blowout modes and the measured temperatures T and _Ti2 from the front seat temperature sensor 6a and the rear seat temperature sensor 6b installed near the front and rear seats, respectively, are used as input signals, and a neural network is used. The estimated values T _N1 and T _N2 of the temperatures at the front and rear of the vehicle cabin are obtained and output to the control outlet jacks 11 and 12, respectively. . At this time, the estimated value T _N1 of the temperature at the front part is the amount of air blown to the rear seat 10, the opening degree of the air mix door at the rear part, the blowing mode at the rear part, and the measured temperature from the rear seat temperature sensor 6b. It is learned as a value taking into account T _i2, and the estimated value T _N2 of the rear part is learned as a value taking into account each of the above-mentioned data of the front seat 9, so that the above-mentioned estimated values T _N1 , T _N _{Since N2} can be obtained as an estimated value including the interaction between the front and rear air conditioners, the control logic 11 that can obtain values very close to the actual temperatures T i and T ₂ at the front of the passenger compartment is From the above control factors and the above estimated value T _N1 , the temperature T! So it becomes the target temperature T _z i is set at a temperature setter 4 a, to control the front of Eadaku bets. Similarly, the control logic 12, the from the regulator and the estimated value 1 _2, urchin by the temperature T ₂ of the cabin front reaches a target temperature T _{z 2} is set at a temperature setter 4 b, the rear of the air duct Control. Therefore, by performing control using the estimated values T _N1 and T _N2 of the front and rear temperatures in the passenger compartment as feedback values, the temperatures in the front and rear of the passenger compartment can be accurately and promptly set to the respective target temperatures. T _z Ζ _Ζ2 .

In the fourth embodiment of the present invention, the outside air temperature Τ _日 , the amount of solar radiation T _s , the front and rear air flow rates, the opening of the air mix door, the blowing mode, the front seat temperature sensor 6 a and the rear seat The measured temperatures T ii and T i _{2 of the} temperature sensor 6b were used as input signals, but are not limited thereto. For example, it was used as the input signal in the above best mode 1, may be added to the input signal a foot blow-out temperature T _F and the vent outlet air temperature T _B, and the like. Further, one or more of the input signals may be appropriately deleted. Industrial applicability

INDUSTRIAL APPLICABILITY As described above, the present invention is excellent as a vehicle air conditioner control method and a vehicle air conditioner for appropriately controlling the temperature in a vehicle cabin to a target temperature. The control method is effective in improving the development efficiency of vehicle air conditioners.

Claims

The scope of the claims

1. A temperature model for estimating the cabin temperature from the output value of the cabin temperature detecting means for detecting the cabin temperature is set, and the cabin temperature is controlled based on the estimated temperature from the temperature model. A method for controlling a vehicle air conditioner.

2. The temperature model according to claim 1, wherein the temperature model is configured by a neural network using an environmental factor of the vehicle and a state of the air conditioner as input signals and an estimated value of the temperature in the passenger compartment as an output value. The control method of the vehicle air conditioner according to the above.

3. The environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is any combination or all of the following information: blow mode, blow temperature, blow air volume, and inlet position. 3. The control method for a vehicle air conditioner according to claim 2, wherein:

4. Vehicle interior temperature detecting means for detecting the temperature in the vehicle interior, and the detected temperature of the vehicle interior temperature detecting means, the environmental factors of the vehicle, and the state of the air conditioner as input signals, and output the estimated value of the vehicle interior temperature as an output value. And a temperature estimator configured by a neural network, wherein the temperature in the vehicle interior is controlled based on the estimated value.

5. The vehicle air conditioner according to claim 4, wherein the temperature in the passenger compartment is controlled using the estimated value as a feedback value.

6. The above environmental factors are the outside air temperature and the amount of solar radiation, and the condition of the air conditioner is the combination of any one of the information of the blow mode, blow temperature, blow air volume, and suction port position. The vehicle according to claim 4 or 5, wherein

7. The vehicle air conditioner according to claim 6, wherein the temperature of the air outlets other than the air outlets set in the air outlet mode among the air outlet temperatures is fixed to a predetermined value. .

8. At least two vehicle interior temperature detecting means for detecting the temperatures of the front and rear parts of the vehicle interior, and the detected temperatures from the vehicle interior temperature detecting means, the environmental factors of the vehicle, and the state of the air conditioning equipment. Outputs estimated values of front and rear temperatures in the passenger compartment as input signals And a temperature estimator configured by a neural network for controlling the temperature of the vehicle interior based on the above two estimated values.

9. The vehicle air conditioner according to claim 8, wherein the temperature in the passenger compartment is controlled using the two estimated values as feedback values.

10. The above environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is any combination of the following information on the blow mode, blow temperature, blow air volume, and suction port position at the front and rear of the passenger compartment. 10. The vehicle air conditioner according to claim 8 or 9, wherein the vehicle air conditioner is a whole.

11. The teacher signal used in learning of the neural network is an average temperature at a position corresponding to a head and a foot of a driver's seat and a passenger's seat, respectively. Item 8. The vehicle air conditioner according to item 8.

12. The input state of a teacher signal used at the time of learning of the neural network is set to a normal value from 0.02 to 0.98. Vehicle air conditioner.

1 3. The sigmoid function used for the neural network is composed of a linear function having different coefficients depending on the input range, and the absolute value of the error between the output value of the linear function and the output value of the sigmoid function in each input range. 9. The vehicle air conditioner according to claim 4, wherein the input range and the coefficient of the linear function are approximated by a function in which the value is within 3%.

Claims of amendment

[11.11.99.99 (11.0.1.99) Accepted by the International Bureau: Claim 2 originally filed was withdrawn; Claim 1 originally filed was amended The other claims remain unchanged.

(Page 2)]

1. (After correction) A neural network that uses the temperature detected by the vehicle interior temperature detection means, the environmental factors of the vehicle, and the state of the air conditioner as input signals, and outputs the estimated value of ^ as in the vehicle interior as the output value. A vehicle air conditioner control method comprising: estimating a vehicle interior temperature using the estimator; and controlling the vehicle interior temperature based on the estimated value of the vehicle interior temperature.

2. (Delete)

3. The environmental factors are the outside air temperature and the amount of solar radiation, and the condition of the air conditioner is any combination or all of the information of the blow mode, the blow air flow, and the position of the suction port. 3. The control method for a vehicle air conditioner according to claim 2, wherein

4. Vehicle interior temperature detecting means for detecting the temperature in the vehicle interior, detection of the vehicle interior temperature detecting means, the environmental factors of the vehicle and the state of the air conditioner as input signals, and an estimated value of the vehicle interior temperature as an output value. And a temperature estimator configured by a neural network. The vehicle air conditioner is configured to control a temperature in a vehicle cabin based on the estimated value.

6. The environmental factors are the outside air temperature and the amount of solar radiation, and the condition of the air conditioning equipment is any combination or all of the following information: blowout mode, blowout ^, blowout airflow, and suction port position. 6. The vehicle air conditioner according to claim 4 or claim 5, wherein

8. At least two vehicle interior temperature detecting means for detecting the temperatures of the front and rear parts of the vehicle interior, and the detected temperatures from the vehicle interior temperature detecting means, the environmental factors of the vehicle, and the state of the air conditioning equipment. Outputs estimated values of front and rear temperatures in the passenger compartment as input signals

Amended paper (Article 19 of the Convention) And a temperature estimator configured by a neural network for controlling the temperature of the vehicle interior based on the above two estimated values.

9. The vehicle air conditioner according to claim 8, wherein the temperature in the vehicle cabin is controlled using the two estimated values as feedback values.

11. The teacher signal used in learning of the neural network is an average temperature at a position corresponding to a head and a foot of a driver seat and a passenger seat, respectively. An air conditioner for a vehicle according to the item.

12. The input state of a teacher signal used in learning of a neural network is changed from 0.02 to 0.98 in a regular manner. 9. The vehicle air conditioner according to claim 8.

Amended paper (Article 19 of the Convention) Statement based on Article 9 of Convention 1 The original claim 1 controls the temperature in the cabin from the estimated indoor temperature, but the contents of the original claim 2, that is, the neural network configuration, Claim 1 has been amended. Claim 1 is merely limiting and does not change its gist.