JP2000001115A - Air conditioning system for vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a twin type air conditioning system for one box vehicle facilitating adjustment of the control logic and control of the temperatures in the cabin with high accurately by accurately presuming the temperatures in the cabin at front and rear portions that are the main elements of the control logic in the air conditioning system. SOLUTION: In this system, two temperature presumption units 12a, 12b constituted by neural network are provided. In this neural network, each of measurement temperatures Ti1, Ti2 by inner temperature sensors 6a, 6b at the front and rear portions in the cabin, and the vehicle environmental factors including an outer temperature TA, an insolation amounts TS, and a blower duty ratio and the state of the air conditioning system is assumed to be an input signal. Further, each of presumed values TN1, TN2 of temperature in the cabin at the front and rear portions is assumed to be an output signal. Control is made in control logic 11a, 11b on the basis that the presumed values TN1, TN2 are assumed to be feedback values.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle air conditioner for properly controlling the temperature in a passenger compartment to a target temperature, and more particularly to a vehicle air conditioner used for a one-box twin type vehicle. Things.

[0002]

BACKGROUND ART FIG. 9 is a diagram showing a control flow of a conventional vehicle air conditioner, a conventional air conditioner for a vehicle, with respect to the target temperature T _Z that is set in the temperature setting unit 4, the vehicle interior the located control logic 1p was the feedback value of the measured temperature T _i measured by the inside air temperature sensor 6 as the vehicle interior temperature detection means, so as to control the temperature and air volume, etc. of the conditioned air to be blown into the passenger compartment . However, the inside air temperature sensor 6 is normally therefore are installed below the front panel, due to the effect of such ambient temperature and solar radiation, different from the temperature (vehicle interior temperature) T ₀ of the occupant's seating position near I have.
Therefore, the feedback value is set to the actual vehicle interior temperature T _0.
, A vehicle is provided with an outside air temperature sensor and a solar radiation sensor, and the measured temperature T _{i is} determined based on the output of each sensor.
Or corrected, and further, based on the blowing mode and the like of the conditioned air was correcting the measured temperature T _i.

[0003]

However, the internal temperature
Measurement temperature T measured by sensor_iAnd the actual vehicle interior temperature T₀
Is used in the above control logic to correct the difference
Because of the large number of parameters, it takes time to match
There was a problem. Also, JP-A-6-195323
In the official gazette, as shown in FIG. _Z， Car interior
Temperature (measured temperature of internal air temperature sensor) T_i, Outside temperature T_A,Day
Radiation T_SNeural network type with input signals
Using the additional learning device to control the amount of conditioned air
Techniques are disclosed. This is a neural network
Target blow temperature, blow mode status, blow
Controlling the final air flow by learning and calculating each formula such as air volume
The input / output function of the neural network
Because many teacher signals are used to learn
The problem is that the time is long and the learning is difficult to converge.
Was. In particular, the setting for each of the front and rear
Temperature T_Z1, T_Z2And air volume W_Z1, W_Z2In the settings
Air-conditioning system for one-box twin type vehicles
Indicates the condition of the conditioned air at the front and rear in the passenger compartment.
Learning is further converged because it affects the other conditioned air
It is difficult to learn and the learning time will be long
There was a problem.

SUMMARY OF THE INVENTION The present invention has been made in view of the conventional problems, and is directed to a vehicle air conditioner of a type in which a set temperature and an air volume of a front portion and a rear portion in a vehicle cabin can be respectively set. A vehicle air-conditioning system capable of easily adjusting a control logic and accurately controlling a vehicle interior temperature by accurately estimating a vehicle interior temperature of a front part and a rear part which are main elements. Aim.

[0005]

According to a first aspect of the present invention, there is provided an air conditioner for a vehicle, wherein at least two vehicle interior temperature detecting means for detecting a temperature of a front portion and a rear portion of the vehicle, Neural circuits that use the respective detected temperatures from the detecting means, the environmental factors of the vehicle, and the state of the air conditioning equipment as the respective input signals, and use the estimated value of the temperature at the front of the passenger compartment and the estimated value of the temperature at the rear as the respective output values. It has two temperature estimators constituted by a network, and controls the temperature in the vehicle interior based on the two estimated values.

According to a second aspect of the present invention, the temperature of the vehicle interior is adjusted by adjusting the temperature and the amount of conditioned air to be blown to the front and rear portions of the vehicle interior using the two estimated values as feedback values. It is intended to be controlled.

According to a third aspect 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 determined based on the blow mode, the mix door opening, and the blow air flow at the front and rear of the vehicle compartment. Or any combination of the above information.

According to a fourth aspect of the present invention, the number of inputs and the structure of the two neural networks are the same.

According to a fifth aspect of the present invention, there is provided a vehicle air conditioner, comprising: a neural network for obtaining an estimated value of a temperature in a front portion of a vehicle interior; The average temperature of the four points at the position corresponding to the part.

According to a sixth aspect of the present invention, there is provided a vehicle air conditioner, comprising: a neural network for obtaining an estimated value of a temperature in a rear portion of a vehicle compartment;
It is the average temperature at the position corresponding to the head and foot at the center and left.

According to a seventh aspect of the present invention, the input state of the teacher signal used for learning the neural network is normalized from 0.02 to 0.98.

[0012] In the vehicle air conditioner according to the present invention, the sigmoid function used for the neural network is constituted by a linear function having a coefficient different depending on an 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 is 3%.
Within the range, the input range and the coefficient of the linear function are approximated by a set function.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. Embodiment 1 FIG. FIG. 1 is a diagram showing a configuration of a vehicle air conditioner according to Embodiment 1 of the present invention, wherein 1 is a control device, 2A is a front air duct, 2B is a rear air duct,
3 is a setting panel for setting the temperature and air volume of the conditioned air,
Reference numerals 4a and 4b _denote temperature setting devices that set the front and rear target temperatures T _Z1 and T _Z2 of the vehicle interior input from the setting panel 3 and output the same to the control device 1, and 5a and 5b denote the air ducts 2A and 2B. A plurality of driving devices for respectively adjusting the opening degree of the mix door 2e in the vehicle, 6a is an inside air temperature sensor as a temperature detecting means at the front of the vehicle cabin installed below the front panel, and 6b is below the front seat 9. With rear seat 10
, An outside air temperature sensor installed near the bumper of the vehicle, and an insolation sensor 8 installed above the front panel. The air duct 2A includes an inside / outside air switching door 2a that adjusts a ratio of inside air and outside air introduced into the air duct 2A, a blower 2b that blows air taken in from the inside / outside air switching door 2a, and an evaporator 2 that cools the blowing air.
c, a heater 2d for warming the blown air, a mix door 2e for controlling the amount of blown air cooled by the evaporator 2c and passing through the heater 2d to adjust the temperature of the blown air, and An outlet switching door 2h for distributing the adjusted air to the upper outlet 2f and the lower outlet 2g. The upper outlet 2f is provided with an upper outlet temperature sensor 2m for detecting the temperature of the upper outlet 2f, and the lower outlet 2g is provided with a lower outlet temperature sensor 2n for detecting the temperature of the lower outlet 2g. Is installed. Also, the driving device 5a
A plurality of actuators for opening and closing the inside / outside air switching door 2a, the mixing door 2e, and the outlet switching door 2h;
And a blower driving circuit for driving the blower 2b. The air duct 2B has the same structure as the air duct 2A. The control device 1 controls the opening degree and the like of the front or rear mix door 2e according to the environmental factors such as the outside air temperature and the states (control factors) of the front and rear air conditioners such as the air-conditioning air blowing mode. Control logic 11a and rear seat control logic 11b for outputting a control signal to be transmitted to the driving device 5a or the driving device 5b, the measured temperatures T _i1 and T _i2 from the inside air temperature sensors 6a and 6b, and the control logic 11a, 1.
The neural network is constituted by inputting information on the environmental factors and the state of the air conditioners from 1b as input signals, and using the estimated values _TN1 and _TN2 of the temperature in the passenger compartment output to the control logics 11a and 11b as output values. And two temperature estimators 12a and 12b (in the figure, a front temperature estimator A and a rear temperature estimator B). The control factors include, for example, outputs from the outside air temperature sensor 7 and the solar radiation sensor 8, the degree of opening and closing of the inside / outside air switching door 2a, the mix door 2e, and the outlet switching door 2h, and the drive voltage of the blower 2b.

FIG. 2A shows a neural network (Front Inc.s) in the front temperature estimator 12a.
FIG. 3 is a diagram showing a configuration of a front neural network 12F. The front neural network 12F includes an input layer, a hidden layer, and an output layer.
It is a hierarchical network of layers. FIG. 2 (b)
It is a figure showing composition of a neural network (Inc.s model for rear) in temperature estimator 12b of the back, and rear neural network 12R has the same layer structure as above-mentioned neural network 12F for front. The input signals of the front neural network 12F include a measured temperature T _i1 (° C.) from the inside air temperature sensor (Inc.s) 6 a and an outside air temperature T from the outside air temperature sensor 7.
_A (° C.), the amount of solar radiation from the solar radiation sensor 8 (Kcal / m ²⁾
hour), the blow mode (1, 2, 3, 4, 5) set on the setting panel 3, the opening degree (%) of the front mix door 2e, and the front blower motor duty ratio (%) corresponding to the drive voltage of the front blower 2b. ), And the output value is an estimated value T _N1 (° C.) of the temperature in the front part of the vehicle interior.
It is. The input signal of the rear neural network 12R is the same as that of the front neural network 1R.
Like the input signal of the 2F, the input signal of the front neural network 12F is configured by replacing the control factor of the front air conditioner with the control factor of the rear air conditioner. That is, the rear neural network 12 </ b> R is the front neural network 1.
The number and structure of input signals are the same as 2F.

The neural networks 12F, 12F
In R, a sigmoid function such as the following equation (1) is used. _{y j = 1 / (1 +} exp (- | x j |) ‥‥‥ (1) where, _{x j} is the weight to each input signal _{v i} to the layer i w
_ij multiplied by the bias b _j (x _j
= In _{Σv i · w ij -b j)} , the _{y j} is the output signal from layer i (input signal to the layer j). As shown in FIG. 3, this sigmoid function outputs 0 to 1 when the input variable is -∞ to + ∞. Here, the temperature estimator 12a at the front uses the four-point average temperatures at positions corresponding to the head and foot of the driver's seat and the passenger's seat as teacher signals at the time of learning of the neural network 12F, The weight w _ij and the bias b _j for each input signal of the arithmetic expression for calculating the estimated value T _N1 of the temperature at the front portion are obtained by learning, and at the time of control, the temperature of the temperature at the front of the vehicle compartment is determined for each of the input signals. and outputs an estimated value _{T N1} to the control logic 11a. On the other hand, the rear temperature estimator 12b uses the average temperature at the position corresponding to the head, foot, and left of the right, center, and left portions of the rear seat as the teacher signal, and estimates the temperature T _{N2 at the} rear of the vehicle compartment. determined by learning the value of the weight w _ij and the bias b _j for each input signal of the operational expression for calculating the, at the time of control for each input signal, outputs the estimated value T _N2 in the vehicle interior rear of temperature control logic 11b I do. As an 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, and the weight w for the type of each input signal is obtained.
_ij can be evaluated in the same way. On the other hand, the teacher signal normalizes the minimum value to the maximum value 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. That is, in the learning of the neural network using the sigmoid function as the input function of the neuron, in consideration of the learning efficiency and the safety, the training signal has a slightly narrower range of 0.02 than that of normalizing the teacher signal from 0 to 1.
Is more effective when the convergence speed is increased.

By the way, as described above, the rear
The neural network 12R is a neural network for the front.
The number and structure of input signals are the same as those of the network 12F.
Therefore, using a calculation program for estimating the indoor temperature, FIG.
Inc. as shown in FIG. Flow for calculating s model
-Estimate the temperature at the front and rear of the cabin according to the chart
Value T_N1, T_N2Can be requested. First, the front
In the temperature estimator 12a, the passenger compartment is
Weight w for each input signal at the inner front_ijYaba
Ias b_jIs determined by learning and a constant is set (step
Step S1), the rear estimator 12b uses the rear teacher
Signals provide a signal for each input signal at the rear of the cabin.
Eight w_ijAnd bias b _jIs determined by learning
The number is set (step S2). Next, estimate the temperature to be calculated.
A fixed place is input (step S3). Temperature estimation location
If is the front of the cabin, the calculation program
Weight w for ram_ijAnd bias b_jTo the value of
Substitute the front constant set in step S1 above.
(Step S4), Inc. Calculate the s model (S
Step S5). On the other hand, in step S3,
If the temperature estimation location is at the rear of the vehicle interior,
Weight w used in the calculation program_ijAnd vias
B_jTo the value of the rear constant set in step S2 above.
Is substituted (Step S6), and Inc. s model calculation
Perform (Step S6).

FIGS. 5 (a) and 5 (b) show temperature estimators 12a and 12b constituted by the above-described neural network.
5 is a graph in which outside temperature data is input and temperatures T ₁ and T ₂ at the front and rear portions of the vehicle compartment are estimated after sufficient learning under the following learning conditions. The temperature T at the front of the vehicle compartment
₁ has an average value of each measurement temperature measured at a position corresponding to each of the head and foot of the driver's seat and the front passenger seat, temperature T ₂ of the passenger compartment rear portion, the rear seat right, center, left It is the average value of each measured temperature measured at the position corresponding to each head and foot. Learning conditions Outside temperature ‥‥ -10 to 35 (° C.) Solar radiation ‥‥ 0 to 660 (Kcal / m ² hour) Vehicle speed ‥‥ Idle to 40 (equivalent to Km / h) In FIGS. 5A and 5B, solid lines Looking at the changes in the output values (estimated values T _N1 and T _N2 ) of the temperature estimators 12a and 12b shown in the above, the maximum error is 2.2 ° C. at the front and 2.8 ° C. at the rear, and the absolute value of the error The average of the.
The temperature is as small as 5 ° C. and 1.3 ° C. at the rear, almost follows the changes in the vehicle interior temperatures T ₁ and T ₂ shown by ○, and I shown by the broken line.
nc. s (internal temperature sensor) Measured temperature T from 6a, 6b
_i1, compared to the change of _{T i 2,} much cabin temperature (T
₁ , T ₂ ).

Next, the temperature of the cabin of the air conditioner for a vehicle is measured.
The control method will be described based on the control flow of FIG.
You. The front-seat control logic 11a determines whether each control
Estimated value T from the temperature estimator 12a_N1And from the cabin
Front temperature T₁Is set by the temperature setting device 4a and the target temperature T
_Z1Control the front air duct 2A so that
You. Similarly, the rear seat control logic 11 b
Factor and estimate T from rear temperature estimator 12b _N2Or
The temperature T at the rear of the cabin₂Is set by the temperature setting device 4b.
Target temperature T_Z2Control the rear air duct 2B so that
I will. The front seat control logic 11a determines the input outside air temperature.
Environmental factors such as air conditioning equipment
The input of the temperature estimator 12a is obtained from each control factor comprising the state.
Extract control factors to be signals and output to temperature estimator 12a
I do. The temperature estimator 12a calculates the extracted control factor.
And Inc. s (internal temperature sensor) 6a measured temperature T
_i1Is used as an input signal.
And the estimated value T of the temperature in the front of the passenger compartment_N1Control log in search of
Output to the memory 11a. The control logic 11a
Control factor and the above estimated value T_N1From the vehicle interior temperature T₁But
Target temperature T set by temperature setting device 4a_Z1To be
And control of the air duct 2A such as the opening degree of the mix door 2e.
Elements are adjusted to the front of the cabin according to preset control logic.
Temperature feedback control. Rear seat control logic 1
Similarly, the vehicle interior temperature T₂Is set by the temperature setting device 4b.
Set target temperature T_Z2Mix door 2e
The control elements of the air duct 2B such as the opening of the air duct are set in advance.
Feedback the temperature in the rear of the cabin according to the control logic.
Control. That is, the front seat control logic 11a, the rear seat
The control logic 11b receives signals from the internal temperature sensors 6a and 6b.
Measurement temperature T_i1, T_i2But not the front and rear of the cabin
Temperature T₁, T₂Temperature estimator 12 very close to
a, 12b, the estimated value T of the temperature at the front and rear of the vehicle compartment
_N1, T_N2Control with the feedback value
To accurately and quickly target the temperature at the front and rear of the cabin
Temperature T_Z1, T_Z2Can be

As described above, according to the first embodiment, the measured temperatures T _i1 and T _i2 from the inside and outside temperature sensors 6a and 6b in the vehicle interior, the outside temperature T _A and the amount of solar radiation T _{S are obtained.} , A blower duty ratio and other environmental factors of the vehicle and the state of the air conditioner are used as input signals, and the estimated values T _N1 and T _N2 of the front and rear portions of the passenger compartment are output values, respectively. Temperature estimators 12
a, 12b are provided, and the control logics 11a, 11b perform control using the estimated values T _N1 , T _N2 as feedback values, respectively. Therefore, the temperature in the vehicle compartment is accurately and quickly set to the target temperature T _Z1 , _TZ2 . Further, the learning of the neural network is based on the four points average temperature at the position where the front teacher signal corresponds to the head and foot of the driver's seat and the passenger's seat, respectively, and the right, center and left of the rear seat respectively. 6 at positions corresponding to the head and feet
Since only the point average temperature is used, the learning of the temperature estimators 12a and 12b is easy, and the learning of the estimated values T _N1 and T _N2 can be performed in a short time. Also, the control logics 11a, 1
Estimated value T of temperature in front and rear of vehicle compartment used in 1b
_N1, since _{T N2} is very close to the vehicle interior front and rear portions of the temperatures _T 1, _{T 2,} the control logic 11a, 11b
Not only can the matching accuracy of the system be greatly improved, but also the time for matching work and the number of wind tunnel experiments in the matching process can be significantly reduced. Furthermore, since the front neural network 12F and the rear neural network 12R have the same number of input signals and the same structure, the calculation of the room temperature estimation can be performed using one program, and the development efficiency is significantly improved.

In the first embodiment, the teacher signal used in the neural network at the rear is converted into a six-point average temperature at positions corresponding to the right, center, and left heads and feet of the rear seat. However, in a vehicle having two rear seats including a two-row part and a three-row part, when the two rear seats are air-conditioned by one rear air conditioner, the teacher signal is transmitted to, for example, the second row. Six points at positions corresponding to the subsequent right, center, and left heads and feet, and six points at positions corresponding to the right, center, and left heads and feet in the third row
If the average temperature of the 12 points in total is used, the estimated value T _N2 of the temperature at the rear of the vehicle compartment can be obtained more accurately, and the temperature in the vehicle compartment can be accurately and quickly converted to the target temperatures T _Z1 and T _Z2 . can do.

Embodiment 2 FIG. In the first embodiment, the sigmoid function shown in equation (1) is used for the neural network. However, by approximating this sigmoid function with a plurality of straight lines, the number of required memories of the temperature estimators 11a and 11b can be reduced. At the same time, since the calculation time can be greatly reduced, the temperature in the vehicle compartment can be more quickly brought to the target temperature Tz. That is, when the sigmoid function shown in the equation (1) is programmed by, for example, an 8-bit embedded microcomputer, an exponential function and a floating-point arithmetic library are required to maintain accuracy.
The capacity and calculation time increase. On the other hand, since the linear function does not include division, there is no need for a floating-point operation, so that the operation accuracy can be maintained even when an integer operation is used, and the ROM capacity and the calculation time can be reduced. Therefore, when the neural network learns, the sigmoid function of the above equation (1) is used. When the learned network is programmed in an actual embedded microcomputer, as shown in FIG. 7, the sigmoid function of the above equation (1) is used. Such that the error is within ± 0.005, for example, 1
The function approximated by seven straight lines (linear function) is calculated by the above (1)
By substituting the sigmoid function for the equation, the ROM capacity and calculation time can be reduced, and the calculation accuracy can be maintained. FIG. 8 is a diagram showing an error between the sigmoid function of the equation (1) and a function obtained by approximating the sigmoid function by the linear function. The magnitude of the error is ± 0.0
It can be seen that it is within 05.

In the above example, the sigmoid function is approximated to 17 straight lines such that the error is within ± 0.005. However, if the error is within ± 0.03 (± 3%), there is a practical problem. Absent. In addition, the approximated function does not necessarily need 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 and second embodiments, the case where the logarithmic sigmoid function of the equation (1) is used has been described. However, other types of sigmoid functions such as a tanh sigmoid function as shown in the following equation (2) Of course, it may be used. y _j = (tanh (x _j ) +1) / 2 ‥‥‥ (2)

[0023]

As described above, the vehicle air conditioner according to the first aspect of the present invention provides a vehicle air conditioner which detects the temperatures detected by at least two vehicle interior temperature detecting means for detecting the temperatures of the front and rear portions of the vehicle interior. Two temperature estimators configured by a neural network using the environmental factor of the vehicle and the state of the air conditioner as input signals, and using the estimated value of the temperature at the front and the estimated value of the temperature at the rear in the vehicle cabin as output values. With
Based on the above two estimated values, the temperature and air volume of the conditioned air blown to the front and rear portions of the vehicle interior are adjusted to control the temperature in the vehicle interior. Even when the temperature and the amount of air are different, control can be performed in consideration of the influence of each other, and the temperatures of the front part and the rear part in the vehicle compartment can be quickly set to the respective target temperatures.

According to a second aspect of the present invention, the temperature of the vehicle interior is adjusted by adjusting the temperature and air volume of the conditioned air blown to the front and rear portions of the vehicle interior using the two estimated values as feedback values. Since the control is performed, the temperatures at the front and rear portions of the vehicle interior can be accurately and quickly set to the respective target temperatures.

In the vehicle air conditioner according to the third aspect, the environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is determined by the blow mode, the mix door opening, the blow air flow at the front and rear of the vehicle compartment. Or all of the above information, it is possible to reliably control the front seat and the rear seat in consideration of each other's setting situation.

In the vehicle air conditioner according to the fourth aspect, since the number of inputs and the structure of the two neural networks are the same, the calculation of the room temperature can be performed using one program, and the development efficiency can be improved. Can be significantly improved.

According to a fifth aspect of the present invention, there is provided a vehicle air conditioner, wherein a teacher signal used for learning a neural network in a front part of a vehicle cabin is an average of positions corresponding to heads and feet of a driver seat and a passenger seat. Since the temperature is used, the difference between the estimated value of the vehicle interior temperature and the actual vehicle interior temperature can be made extremely small.

According to the vehicle air conditioner of the present invention, the teacher signal used when learning the neural network in the front and rear portions of the vehicle cabin corresponds to the head, foot, and foot of the right, center, and left rear seats, respectively. Since the average temperature at the position is set, the difference between the estimated value of the vehicle interior temperature and the actual vehicle interior temperature can be made extremely small.

In the vehicle air conditioner according to the present invention, the input state of the teacher signal used at the time of learning the neural network is normalized from 0.02 to 0.98. Efficiency and safety can be improved.

In the air conditioner for a vehicle according to the present invention, the sigmoid function used in learning of the neural network is determined by setting an absolute value of an error between an output value in each input range of a linear function and an output value of the sigmoid function to 3; %, The number of required memories can be reduced and the calculation time can be greatly reduced by approximating the input range and the function of the coefficient of the linear function. Target temperature.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a configuration of a vehicle air conditioner according to a first embodiment of the present invention.

FIG. 2 is a diagram illustrating Inc. according to the first embodiment of the present invention. It is a figure showing composition of an s model.

FIG. 3 is a diagram showing a sigmoid function used in a neural network.

FIG. 4 is a diagram illustrating Inc. according to the first embodiment of the present invention. It is a flowchart figure which shows the calculation method of an s model.

FIG. 5 is a diagram illustrating a relationship between an estimated value of a vehicle interior temperature and an actually measured value by a temperature estimator according to the first embodiment of the present invention.

FIG. 6 is a diagram showing a control flow of the vehicle air conditioner according to the first embodiment of the present invention.

FIG. 7 is a diagram illustrating an approximation method of a sigmoid function according to the second embodiment of the present invention.

FIG. 8 is a diagram showing an error between a sigmoid function and a linear function.

FIG. 9 is a diagram showing a control flow of a conventional vehicle air conditioner.

FIG. 10 is a diagram showing a configuration of a neural network of a conventional vehicle air conditioner.

[Explanation of symbols]

1 control device, 2A, 2B air duct, 3 setting panel, 4, 4a, 4b temperature setting device, 5a, 5b driving device, 6, 6a, 6b inside temperature sensor, 7 outside temperature sensor, 8 sunlight sensor, 9 front seat , 10 rear seats,
11a Front seat control logic, 11b Rear seat control logic, 12a (front) temperature estimator, 12b (rear) temperature estimator.

Claims (8)

[Claims] 1. At least two vehicle interior temperature detecting means for detecting temperatures of a front part and a rear part of a vehicle interior, respective detected temperatures from the respective vehicle interior temperature detecting means, environmental factors of the vehicle, and states of air conditioners. Are input signals, and an estimated value of the temperature in the front part of the vehicle compartment and an estimated value of the temperature in the rear part are output values, respectively.
An air conditioner for a vehicle, comprising: two temperature estimators, wherein the temperature in the vehicle compartment is controlled based on the two estimated values. 2. The vehicle air conditioner according to claim 1, wherein the temperature in the passenger compartment is controlled using the two estimated values as feedback values. 3. The environmental factors are the outside air temperature and the amount of solar radiation, and the state of the air conditioner is determined by the front and rear air outlet modes of the vehicle compartment,
2. The vehicle air conditioner according to claim 1, wherein any one or all of the information on the mix door opening and the amount of blown air is combined. 4. The vehicle air conditioner according to claim 1, wherein the number of inputs and the structure of the two neural networks are the same. 5. In a neural network for obtaining an estimated value of the temperature in the front part of the vehicle compartment, a teacher signal used at the time of learning is an average temperature at a position corresponding to a head and a foot of a driver seat and a passenger seat. The vehicle air conditioner according to claim 1, wherein: 6. A neural network for obtaining an estimated value of the temperature at the rear of the vehicle compartment, wherein a teacher signal used for learning is obtained by calculating the average temperature at the positions corresponding to the right, center, and left heads and feet of the rear seat. The vehicle air conditioner according to claim 1, wherein: 7. The vehicle air conditioner according to claim 1, wherein an input state of a teacher signal used in learning of the neural network is normalized from 0.02 to 0.98. 8. A sigmoid function used for a neural network is constituted by a linear function having coefficients different depending on an input range, and an error between an output value in each input range of the linear function and an output value of the sigmoid function. 2. The vehicle air conditioner according to claim 1, wherein the input range and the coefficient of the linear function are approximated by a function in which an absolute value is within 3%.