JP2005098555A - Neural network type air conditioner - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a temperature sensing characteristic corresponding to an individual difference without extremely raising memory capacity and processing ability when controlling an air conditioner 300 with a neural network type air conditioner 400. <P>SOLUTION: The neural network type air conditioner 400 is provided with a control device 200 for controlling the air conditioner 300 on the basis of a result of computing performed by inputting a group of detection values An to a neural network 230. The control device 200 is provided with a re-writing device 220, which takes any one connection load value Cn2 from a data server 600 and re-writes a present connection load value Cn1 to the connection load value Cn2, to take the desirable connection load value Cn2 from outside and re-writes the present connection load value Cn1. With this structure, temperature sensing characteristic inside a cabin when controlling air conditioning can be changed to a temperature sensing characteristic corresponding to an individual difference without extremely raising memory capacity and processing ability of the neural network type air conditioner 400. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a neural network type air conditioner that controls an air conditioning state in a passenger compartment based on a calculation result of a neural network.

  Conventionally, as a neural network type air conditioner, there are those described in Patent Literature 1 and Patent Literature 2.

  The neural network type air conditioners described in Patent Documents 1 and 2 have a feature that it is possible to finely tune the warmth characteristics that the average user likes on the manufacturer side.

  However, in the neural network type air conditioners described in Patent Documents 1 and 2, the temperature characteristic is determined by a constant (coupling load value) used for the temperature characteristic calculation. It was impossible to change the constants and set warmth characteristics corresponding to individual differences.

  Moreover, as a neural network type air conditioner capable of realizing a warm feeling characteristic corresponding to individual differences, there are those described in Patent Document 3 and Patent Document 4.

  The neural network type air conditioner described in Patent Documents 3 and 4 sequentially learns the setting operation of the temperature characteristic by the user after shipment, and performs the next air conditioning operation based on the accumulated learning data. Has the feature of being possible.

However, the neural network type air conditioners described in Patent Documents 3 and 4 require complicated control and calculation, so that the processing of an 8-bit microcomputer or a 16-bit microcomputer usually used for a vehicle air conditioner is usually performed. In terms of capacity, it may not be able to be processed completely, and it is necessary to write and store the learned warmth characteristic information at the time of use after shipment, which requires an external storage device connected to the microcomputer. There was a problem that the cost of the neural network type air conditioner would increase.
JP-A-11-301248 JP-A-11-310023 JP 2000-293204 A JP 2001-273475 A

  In order to solve the above-described problems, the present invention makes the temperature characteristics of the passenger compartment temperature characteristic corresponding to individual differences without significantly increasing the storage capacity and processing capacity of the control device of the neural network type air conditioner. An object is to provide a neural network type air conditioner capable of conversion.

  In order to achieve the above object, the present invention employs the following technical means.

  The neural network type air conditioner according to claim 1 inputs and calculates a product-sum value composed of a detected value group detected by the sensor group and a current combined load value stored in the storage means by the neural network, and the calculation A neural network type air conditioner comprising a control means for controlling air conditioning equipment according to a result, wherein the control means accumulates a plurality of combined load values for realizing various air conditioning controls, and is a data storage means provided outside the control means From the above, there is provided a rewrite means for taking in an arbitrary connection load value selected by the user and rewriting the current connection load value to an arbitrary connection load value.

  According to the present invention, a desired combined load value is fetched from the outside, and the combined load value currently stored in the storage device is rewritten, so that the storage capacity and processing capacity of the neural network type air conditioner are not increased so much. In addition, it is possible to provide a neural network type air conditioner that can change the temperature sensation characteristic in the passenger compartment during air conditioning control to a temperature sensation characteristic corresponding to individual differences.

  The control means in the neural network type air conditioner according to claim 2 includes a receiving means for receiving a detection value group detected by the sensor group, a storage means for storing the neural network and a current combined load value, a neural network, It is characterized by comprising a calculation means for reading the combined load value and calculating the product sum value by a neural network, a rewriting means, and an air conditioning control means for controlling the air conditioning operation of the air conditioning equipment based on the calculation result. And

  According to the present invention, by configuring the rewriting means as a part of the control means having the storage means for storing the current connection load value, any connection load value rewritten by the rewriting means can be directly and simply stored in the storage means. Since it is possible to write, it is possible to easily rewrite the current combined load value.

  The rewriting means in the neural network type air conditioner according to claim 3 is a selection means that allows a user to select an arbitrary combined load value from a plurality of combined load values stored by the data storage means, and a user selects It is characterized by comprising a capturing means for capturing an arbitrary bond load value selected by the means, and a writing means for writing the arbitrary bond load value captured by the capturing means to the storage means.

  According to the present invention, the neural network type air conditioner can be controlled to perform air conditioning control of any desired temperature characteristic, that is, a user's preference without significantly increasing the storage capacity and processing capacity.

  The capturing means in the neural network type air conditioner according to claim 4 is an external storage output device for transmitting an arbitrary combined load value captured to the writing means via a connecting means provided in the control means. To do.

  According to the present invention, since the capturing means can be physically separated from the control means, the neural network type air conditioner is coupled to allow the user to perform air conditioning control of any desired temperature characteristic, that is, the user's preference. It is possible to simplify the load value capturing operation and the writing operation.

  The writing means in the neural network type air conditioner according to claim 5 includes: a first I / O area that receives an arbitrary connection load value from the connection means; and an arbitrary connection load received by the first I / O area. It is characterized by comprising a memory for temporarily storing a value and a second I / O area for writing an arbitrary combined load value temporarily stored in the memory to the storage means.

  According to the present invention, even after the storage means for storing the coupling load value is mounted on the control means, an arbitrary coupling load value can be written to the storage means at any time with a simple structure.

  The capturing means in the neural network type air conditioner according to claim 6 comprises at least one of a network connecting means for connecting to at least one of the Internet and an intranet or a media reading means for reading a recording medium, and a personal computer having various arithmetic functions. It is characterized by being a computer.

  According to the present invention, by configuring the capturing means with a personal computer that is generally widespread, for example, an arbitrary combined load value that can realize a user's favorite thermal sensation characteristics with an air conditioner can be obtained from a network such as the Internet. User's preference from FD, CD, MO disk, smart media, DVD and other recording media that can be obtained by downloading from network connection means or storing multiple binding load values that realize various warmth characteristics Since the media reading means can acquire an arbitrary combined load value that can realize the thermal sensation characteristics suitable for the air conditioner, the current combined load value stored in the storage means can be easily determined arbitrarily. It is possible to rewrite to the combined load value.

  The capturing means in the neural network type air conditioner according to claim 7 is a car navigation device that provides driving guidance of a vehicle including at least one of a network connection means and a media reading means.

  According to the present invention, it is possible to rewrite the combined load value currently stored in the storage means to an arbitrary combined load value by the car navigation device mounted on the vehicle on which the neural network type air conditioner is mounted. This eliminates the need to bring equipment into the vehicle compartment and reduces the effort required for rewriting.

  The connection means in the neural network type air conditioner according to claim 8 is a media reading means.

  According to the present invention, it is not necessary to prepare a separate capturing means, and a recording medium storing an arbitrary combined load value capable of realizing a user's favorite thermal sensation characteristic by the air conditioner is directly stored in the neural network type air conditioner. Therefore, it is possible to rewrite the combined load value currently stored in the storage unit with a simple structure to an arbitrary combined load value that suits the user's preference.

  The storage means and the writing means in the neural network type air conditioner according to claim 9 constitute a FLASH microcomputer as one package.

  According to the present invention, since the storage means and the writing means can be handled as one package, the manufacturing process of the control means can be simplified.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to FIGS.

  First, the outline of the neural network type air conditioning control will be briefly described with reference to FIG.

  FIG. 1 is a model diagram showing the contents of neural network type air conditioning control in a vehicle.

  As is well known, the neural network type air conditioning control is one of the control methods for controlling the air conditioner based on the calculation result of the neuro software that mathematically realizes the information processing of the biological nervous system.

  Prior to the creation of the neuro software, the teacher data that combines the detection value group of various sensors mounted on the vehicle and the expected output value is created, and the correlation defined by this teacher data is established. A constant Cn called a load value is calculated by a workstation or the like using a generally known method such as backpropagation.

  Specifically, the neuro-soft integrates the detection value group of various sensors mounted on the vehicle with the combined load value Cn as the input value An, and each input value An and the combined load value (constant Cn, 1 is equivalent to w1, w2,..., And is calculated using a neural network function (such as a sigmoid function), and the air conditioner is controlled based on the output result En. It is.

  In this program, one arithmetic unit corresponding to an input value is arranged in parallel, and arithmetic unit groups composed of the plural arithmetic units are arranged in several stages in a hierarchical manner.

The content of the calculation by the program having the above-described configuration is shown in the following formula 1.
(Equation 1) Output result En = f (input 1 × coupled load value 1+... Input n × coupled load value n)
= F (A1W1 + A2W2 +... AnWn)
n
= F (Σ AnWn)
n = 1
Below, the neural network type air-conditioning control by the above-mentioned program does not increase the storage capacity and processing capacity so much, and is not limited to the currently stored combined load value. An example that makes it possible to match the user's preference will be described with reference to FIGS.

  FIG. 2 is an overall configuration diagram showing a neural network type air conditioner 400 including a sensor group 100 and air conditioners 300 arranged in various places of a vehicle (not shown).

  The sensor group 100 includes an inside air sensor 1, an outside air sensor 2, a solar radiation sensor 3, a post-evaporation sensor 4, and a water temperature sensor 5.

  The sensors 1 to 5 described above are connected in parallel to a receiving device 201 described later, and transmit each detected value to the receiving device 201.

  The neural network type air conditioner 400 includes a control device 200 and an air conditioner 300.

  The control device 200 includes a receiving device 201, a storage device 202, a computing device 203, an air conditioning control device 204, and a rewriting device 220.

  Among these, the receiving device 201, the storage device 202, the arithmetic device 203, and the air conditioning control device 204 are mounted on one ECU as the control ECU 240 and are one of the components of the rewriting device 220. Only a personal computer 501 described later is located outside the control ECU 240.

  As described above, the receiving device 201 is connected in parallel with each of the sensors 1 to 5 of the sensor group 100, and converts the analog signals (A1 to A5) detected by the sensors 1 to 5 into digital signals. At the same time, it is a well-known A / D conversion / interface device that inputs to the arithmetic unit 203 described later.

  The storage device 202 corresponds to a FLASH ROM area of a FLASH microcomputer 210 (described later) mounted on the control ECU 240. This storage device 202 is a part of the FLASH microcomputer 210.

  In general, a FLASH microcomputer is a kind of storage device in which a FLASHROM area and a FLASHRAM area are packaged into one, and data stored in the FLASHRAM area can be rewritten to data stored in the FLASHROM area at any time.

  FIG. 3 shows an internal configuration diagram of the FLASH microcomputer 210. The interior is mainly divided into four layers of a first I / O area 222a, a FLASHRAM area 222b, a second I / O area 222c, and a FLASHROM area (that is, the storage device 202).

  The first I / O area 222a receives an arbitrary coupling load value (constant Cn2) transmitted from the personal computer 501 described later, and also stores the arbitrary coupling load value (constant Cn2) in the FLASHRAM area 222b described later. Is an input / output device for writing.

  The FLASHRAM area 222b is a memory that temporarily stores an arbitrary coupling load value (constant Cn2) received by the first I / O area 222a.

  The second I / O area 222c is an input / output device that reads an arbitrary coupling load value (constant Cn2) temporarily stored in the FLASHRAM area 222b and writes it in the FLASHROM area (that is, the storage device 202).

  In the FLASHROM area (that is, the storage device 202), the above-described neural network 230, the current connection load value (constant Cn1) for the neural network operation by the neural network 230, and constants used for other than the neural network operation are stored. Has been.

  Then, the arithmetic device 203 reads out the joint load value (constant Cn1) currently stored in the storage device 202 as the joint load value for neural network computation.

  As described above, the FLASH microcomputer 210 includes two storage areas, the FLASHRAM area 222b and the FLASHROM area (that is, the storage device 202). Therefore, the neural network 230 and the current combined load value (constant Cn1) are stored in the FLASHROM area That is, the storage device is provided with the convenience that the combined load value (constant Cn2) of the neural network 230 can be rewritten to an arbitrary value even after being mounted on the control ECU 240 in the state stored in the storage device 202).

  Therefore, when the FLASH microcomputer 210 is used, the specification as a product of the neural network type air conditioner is determined, and even after actually entering the production system or after shipment, the combined load of the current neural network 230 that has already been stored. The value (constant Cn1) can be rewritten to a new combined load value (constant Cn2).

  For reference, even in a general control ECU that does not use a neural network, the use of a FLASH microcomputer is common because it can be commercialized with a short delivery time and the specification can be easily changed after the product flows. It is becoming.

  Therefore, an increase in cost by using the FLASH microcomputer 210 is small as compared with a period when MASKROM is frequently used.

  Further, the components of the FLASH microcomputer 210 other than the FLASH ROM area (that is, the storage device 202), that is, the three components of the first I / O area 222a, the FLASHRAM area 222b, and the second I / O area 222c will be described later. The writing device 222 is configured, and this writing device 222 corresponds to the writing means described in the claims.

  The arithmetic unit 203 is a well-known 8-bit or 16-bit CPU. The combined load value Cn1 currently stored in the storage device 202 (that is, the FLASHROM area) is read out, and the detected values An by the sensors 1 to 5 transmitted from the receiving device 201 and the combination read out from the current storage device 202 are read. This is a known arithmetic device that calculates a product sum value from a load value (constant Cn1) and performs a neural network operation based on the product sum value.

  In the case of the present embodiment, the neural network calculation performed by the above-described calculation device 203 is a calculation using a sigmoid function, and the calculation device 203 uses a calculation result En based on this sigmoid function as an air conditioning control device 204 described later. To communicate.

  The air-conditioning control device 204 is a well-known control ECU that controls the air-conditioning operation of the air-conditioning device 300 described later based on the calculation result En of the sigmoid function of the arithmetic device 203 described above.

  Specifically, an internal / external air switching door motor 301, an A / M motor 302, and a blowing mode switching door motor 303 constituting an air conditioner 300 to be described later are connected, and the output result of the sigmoid function by the arithmetic unit 203 By controlling the three motors 301, 302, and 303 based on En, the temperature sensation characteristic in the air conditioner 300 is variably controlled.

  The rewriting device 220 is a kind of interface system including a personal computer 501, a connection unit 221, and a writing device 222.

  The personal computer 501 is a well-known personal computer. Among the constituent elements constituting the rewriting device 220, the only constituent element separate from the control ECU 240 is equivalent to the capturing means shown in the claims.

  The personal computer 501 includes a serial connector port such as RS-232C or a port that communicates with the outside, such as a USB port, and a modem 510 that can be connected to the Internet. The modem 510 corresponds to the network connection means described in the claims.

  Further, a drive device 520 for reading a recording medium such as FD, CD, smart media, or DVD is also installed. The drive device 520 corresponds to a medium reading unit described in the claims.

  The connection unit 221 is a well-known connector unit that is electrically connected to a personal computer 501 to be described later, and usually corresponds to a serial connector or a USB connector.

  The writing device 222 is a recording device that includes a first I / O area 222a, a FLASHRAM area 222b, and a second I / O area 222c that constitute the FLASH microcomputer 210 described above.

  The air conditioner 300 includes a blower fan (not shown), an evaporator, a heater core, a refrigerant circuit, an inside / outside air switching door motor 301, an A / M switching door motor 302, a blowing mode switching door motor 303, and the like. It is a well-known air conditioner.

  Among these, the inside / outside air switching door motor 301, the A / M switching door motor 302, and the blowing mode switching door motor 303 are controlled by the air conditioning control device 204 described above.

  The data server 600 is connected to the personal computer 501 via the well-known Internet, and stores a plurality of combined load values (Cnx) for enabling the air conditioning operation of the air conditioner 300 with various temperature characteristics. This is a well-known file server that transmits a coupling load value (Cn2) selected from the personal computer 501 side by a predetermined protocol in response to a request from the personal computer 501.

  When accessing via the Internet using the modem 510 of the personal computer 501, a menu screen (not shown) is displayed on the personal computer 501.

  With the writing device 220 included in the neural network type air conditioner 400 having the above-described configuration, it is possible to easily rewrite the combined load value used when the neural network type air conditioner 400 performs the neural network calculation.

  Next, the operation of the rewriting device 220 for rewriting the connection load value (Cn1) currently stored in the storage device 202 to an arbitrary connection load value (Cn2) desired by the user will be described with reference to the flowchart shown in FIG. I will explain.

  In step S1, the personal computer 501 accesses the data server 600 via the Internet, and the user desires (that is, himself / herself to the air conditioner 300) from a plurality of combined load values (Cnx) stored in the data server 600. The combined load value (Cn2) capable of realizing the air conditioning operation to be operated is selected from the menu screen. If selected, the process proceeds to step S2.

  Incidentally, the processing content of this step S1 corresponds to the selection means shown in the claims.

  In step S2, the desired combined load value Cn2 selected in step S1 is downloaded from the data server 600 to the personal computer 501 via the Internet. If it downloads, it will progress to step S2.

  In step S3, a connector unit (not shown) of the personal computer 501 is connected to the connection unit 221 of the neural network type air conditioner 400. When the connection is completed, the process proceeds to step S4.

  In step S4, transmission of a write command from the personal computer 501 to the connection unit 221 is started. When transmission is started, the process proceeds to step S5.

  In step S5, the downloaded combined load value Cn2 is transferred from the connection unit 221 to the first I / O area 222a of the FLASH microcomputer 210. When the transfer is completed, the process proceeds to step S6.

  In step S6, an arbitrary combined load value Cn2 is once written from the first I / O area 222a to the FLASHRAM area 222b. When the writing is completed, the process proceeds to step S7.

  In step S7, an arbitrary combined load value Cn2 is written from the FLASHRAM area 222b to the storage device 202 (FLASHROM area) via the second I / O area 222c. When writing is started, the process proceeds to step S8.

  In step S8, it is determined whether or not writing has been completed. When the writing is completed, the process proceeds to step S9. When the writing is not completed, step S8 is circulated in a closed loop.

  In step S9, the personal computer 501 displays that the writing of the arbitrary combined load value Cn2 has been completed normally. When the display is completed, this flow ends.

  With the configuration and operation described above, the neural network type air conditioner 400 is not limited to the characteristics of the current combined load value Cn1 stored in the storage device 202, and the control of the warmth characteristic in the vehicle interior corresponds to individual differences. It becomes possible to make it a warm feeling characteristic.

  In this embodiment, the case where the FLASH microcomputer 210 is mounted is shown, but the storage device 202 may be a storage device other than the FLASH ROM area, for example, a well-known memory. Similarly, the writing device 222 may be a memory writing device when the storage device 202 is a memory.

  Further, in the present embodiment, a desired arbitrary combined load value (Cn2) is downloaded from the Internet using the modem 510. However, the drive device 520 uses a medium (not shown) in which various patterns of combined load values are stored in advance. A desired arbitrary bond load value (Cn2) may be acquired by the personal computer 501 and the bond load value (Cn1) in the storage device 202 may be rewritten.

  In this embodiment, the network connection means is the modem 510, but it may be a network card (LAN board).

(Second embodiment)
In the above-described first embodiment, the personal computer 501 is taken as an example of the capturing unit, but the navigation device 502 mounted on a vehicle on which the air conditioner 300 is mounted may be used.

  In recent years, a navigation apparatus 502 including a modem 510 corresponding to a network connection means for connecting to the Internet and the like and a drive device 520 corresponding to a medium reading means for reading and writing a recording medium has been commercialized.

  FIG. 5 is an image diagram showing an image in which an arbitrary combined load value taken in by the navigation device 502 having the above-described features is transmitted to the control ECU 240.

  The desired combined load value Cn2 is downloaded from the Internet by the navigation device 502 having the above-described features, temporarily stored in the drive device 520 provided in the navigation device 502, and from this step in the flowchart shown in FIG. You may write in the memory | storage device 202 through the processing content shown to S1-step S7.

  With the above-described configuration and operation, it is possible to save the trouble of bringing the personal computer 501 into the vehicle interior, and to easily control the air-conditioning apparatus 300 without depending on the constant range after shipment. It is possible to make the control contents that realize the characteristics.

(Third embodiment)
In the first and second embodiments described above, the desired arbitrary combined load values Cn2 are all downloaded from the Internet.

  However, as shown in FIG. 6, the manufacturer distributes a medium 530 storing the maximum number of possible combined load values Cnx, and from the medium 530, the drive unit 520 included in the navigation apparatus 502 is used to store the medium 530. Any desired combined load value Cn2 may be written.

  With the configuration and operation described above, even if the navigation device 502 without the modem 510 is installed, the control of the air conditioner 300 can be easily changed to the control content that realizes the thermal characteristics corresponding to individual differences. It becomes.

(Fourth embodiment)
In each of the first to third embodiments described above, a desired combined load value (Cn2) is temporarily stored in the personal computer 501 or the external storage output device 500 (capture means) such as the navigation device 502, From here, the combined load value (Cn1) in the storage device 202 is rewritten.

  However, the drive unit 520 may be directly provided in the control ECU 240 as the connection unit 221.

  In this case, in the medium 530 read by the drive device 520, as shown in the third embodiment, not all possible combined load values (Cnx) but combined load values (Cn2) classified according to individual discrimination. It is necessary to separately prepare a medium 531 storing itself.

  FIG. 7 shows a configuration diagram in the case where the drive device 520 is provided in the control ECU 240 and the medium 531 is inserted into the drive device 520.

  As described above, the drive device 520 corresponds to a medium reading unit that reads various recording media such as FD, CD, smart media, and DVD.

  The drive device 520 is set to automatically start a reading operation when the medium 531 described above is inserted.

  Therefore, the coupling load value (Cn2) desired by the user is automatically written in the storage device 202.

  With the configuration and operation described above, it is possible to more easily rewrite the current combined load value Cn1 in the storage device 202 to a desired combined load value Cn2.

(Other embodiments)
In each of the first to third embodiments described above, the personal computer 501 or the navigation device 502 and the control ECU 240 are connected by wire. However, the wireless computer such as the Blue tooth communication method can be used arbitrarily. The combined load value Cn2 may be transmitted.

It is a model diagram of neural network type air conditioning control. It is the block diagram which showed the whole sensor group 100 and neural network type air conditioner 400 which concern on 1st embodiment of this invention. A FLASH microcomputer comprising a storage device 202, a first I / O area 222a, a FLASHRAM area 222b, and a second I / O area 222c constituting a part of the writing device 222 according to the first embodiment of the present invention 2 is a configuration diagram showing the configuration of 210. FIG. In the neural network type air conditioner 400 according to the first embodiment of the present invention, the temperature in the passenger compartment can be reduced at low cost without depending on the upper and lower limits of the set value after shipment from the manufacturer (that is, the initial combined load value Cn1). It is a flowchart which shows the operation | movement which makes the control of a feeling characteristic the warm feeling characteristic corresponding to an individual difference. It is an image figure which shows the connection aspect of the navigation apparatus 502 and the neural network type | formula air conditioner 400 which store the downloaded new combined load Cn2 based on 2nd embodiment of this invention. It is an image figure which shows the transmission aspect of the joint load Cn between the media 530 which stores and distributes several possible joint loads Cn, and the neural network type air conditioner 400 based on 3rd embodiment of this invention. . It is the block diagram which showed the structure at the time of providing the drive apparatus 520 directly in control ECU240 based on 4th embodiment of this invention.

Explanation of symbols

100 sensor group 200 control device (control means)
201 Receiving device (receiving means)
202 Storage device (storage means)
203 Arithmetic unit (calculation means)
204 Air conditioning control device (air conditioning control means)
210 FLASH microcomputer 220 rewriting device (rewriting means)
221 Connecting unit 222 Writing device 240 Control ECU
300 Air conditioner 501 Personal computer (external calculation storage means)
502 navigation device (external calculation storage means)
520 Media reader (media reader)

Claims (9)

A product sum value consisting of the detected value group (An) detected by the sensor group (100) and the current combined load value (Cn1) stored in the storage means (202) is input and calculated by the neural network (230), A neural network type air conditioner comprising control means (200) for controlling the air conditioner (300) according to the calculation result (En),
The control means (200) is selected by the user from the data storage means (600) provided outside the control means (200) for storing a plurality of combined load values (Cnx) for realizing various air conditioning controls. A neural network type air conditioner comprising rewriting means (220) for taking in an arbitrary coupling load value (Cn2) and rewriting the current coupling load value (Cn1) to the arbitrary coupling load value (Cn2) . The control means (200) includes a receiving means (201) for receiving a detection value group (An) detected by the sensor group (100);
Storage means (202) for storing the neural network (230) and the current combined load value (Cn1);
An arithmetic means (203) for reading out the neural network (230) and the combined load value (Cn1), and inputting and calculating the product-sum value in the neural network (230);
The rewriting means (220);
The neural network type air conditioner according to claim 1, comprising air conditioning control means (204) for controlling an air conditioning operation of the air conditioner (300) based on the calculation result (En). The rewriting unit (220) is a selection unit (step) that allows a user to select an arbitrary combined load value (Cn2) from the plurality of combined load values (Cnx) stored in the data storage unit (600). S1)
Capture means (500) for capturing the arbitrary combined load value (Cn2) selected by the user by the selection means (step S1);
The neural network type according to claim 2, characterized by comprising a writing means (222) for writing the arbitrary combined load value (Cn2) taken in by the taking-in means (500) into the storage means (202). Air conditioner. The fetching means (500) transmits the fetched arbitrary coupling load value (Cn2) to the writing means (222) via the connecting means (221) provided in the control means (200). The neural network type air conditioner according to claim 3, wherein the air conditioner is a device. The writing means (222) includes a first I / O area (222a) for receiving the arbitrary coupling load value (Cn2) from the connection means (221),
A memory (222b) for temporarily storing the arbitrary coupling load value (Cn2) received by the first I / O area (222a);
The second I / O area (222c) for writing the arbitrary coupling load value (Cn2) temporarily stored in the memory (222b) to the storage means (202). 5. A neural network type air conditioner according to 4. The capturing means (500) includes at least one of a network connection means (510) connected to at least one of the Internet and an intranet, or a media reading means (520) for reading a recording medium, and a personal computer (provided with various calculation functions). 501), the neural network air conditioner according to claim 3 or 4. The capture means (500) is a car navigation device (502) that provides vehicle travel guidance with at least one of the network connection means (510) or the media reading means (520). The neural network type air conditioner according to 3 or 4. The neural network air conditioner according to claim 3, wherein the connecting means (221) is the media reading means (520). The neural network type air conditioner according to any one of claims 3 to 5, wherein the storage means (202) and the writing means (222) constitute a FLASH microcomputer (210) as one package.

Images