JP3128156B2 - Air conditioner controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an air conditioner control device for controlling indoor temperature, air volume and direction to enhance the comfort of humans in a room. BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an air conditioner control device that automatically performs a comfortable air conditioning operation in an air conditioner.

[0002]

2. Description of the Related Art An air conditioner controller is disclosed in, for example,
The example described in 3-108145 is generally known. Hereinafter, a conventional air conditioner control device will be described with reference to FIG.

FIG. 7 is a block diagram showing the configuration of inference control software in a conventional air conditioner control device.
In FIG. 7, reference numeral 61 denotes an environment detection program for sensing the air-conditioning environment such as temperature, humidity, and external light from respective sensors (not shown) and accumulating data, and 62 for deriving a specific fact from the data. Is a condition setting program for setting the above conditions, 63 is a processing program for performing element syntactic semantic symbolization processing of the above-mentioned conditions and the accumulated knowledge database for air conditioning, and 64 is a predicate of the semantic symbolized knowledge. An inference program 65 for inferring an optimal solution by inference using a logic type programming language. Reference numeral 65 denotes a drive interface control program for distributing and outputting the digital signal of the optimal solution to a control output display, an actuator, a motor, and the like. Is an air conditioner.

Next, the operation of the above conventional device will be described with reference to FIG. First, in step 71, facts from the environment, which are inputs from the sensor group, are stored as data. Next, in step 72, knowledge having a new meaning is created from the data by using a control knowledge rule based on conditions or the like, for example, IF to THEN format. For example, if a 24-hour clock device is electronically provided, and a day exceeding 25 degrees at a specific time continues for 5 days, a knowledge rule of a determination condition such as "summer" is digitally encoded, and a microcomputer is used. In the memory.

Next, in step 73, the cold air is blown out in a short time in a low direction, for example, in summer, since the person is lightly dressed, based on the knowledge generated by the control knowledge rule based on the above conditions and the like. It is inferred that a pattern that changes the direction upward so as not to hit directly is good, and is controlled by the wind direction switching means. In addition, it outputs control outputs such as operation start time, operation mode switching, wind speed and wind direction switching, and temperature setting.

Next, in step 74, when the user performs an operation input to change the wind direction, for example, on the inferred output, the information is stored. Then, in the next step 75, the accumulated data is analyzed, the habits and preferences of the user are determined, and the result is fed back to the step 72,
It is designed to change the premise of inference.

[0007]

However, in the above-mentioned conventional device, knowledge is detected from the input of the sensor group,
In order to generate and infer new knowledge from detected knowledge, control knowledge rules must be generated in advance based on experience already obtained by a skilled person, and it takes much time and effort to create control knowledge rules. Had the problem that

Further, since the sensor group also detects only the air-conditioning environment, it cannot directly detect indoor human information or indoor heat distribution, and lacks information necessary for comfortable air-conditioning. Had.

The present invention solves such a conventional problem, and provides air conditioning that can automatically and easily perform a comfortable air-conditioning operation in consideration of a room environment and human information of a user. It is an object to provide a machine control device.

[0010]

In order to achieve the above object, the present invention provides a plurality of sensors for detecting indoor and outdoor environmental conditions.
Storage means for storing the previous state of the sensor means
And image input means for detecting two-dimensional image information in a room
From the image data obtained from the image input means.
Feature extraction means for extracting the amount of features of a user, and indication of user's intention
Operating means for inputting the
Operating time detecting means for detecting overtime, and the sensor means
And indoor and outdoor environmental conditions which are output values from the storage means.
And the feature extraction means, operation means, and operation time
Guess the user's operation content based on information from the detection means
Refer to the operation content estimating means and the results obtained from each of the above means.
Learning means for learning operation contents by referring to the operation contents;
Control the air conditioner based on the signal output from the measurement means.
Control signal generating means .

The present invention also provides a method for detecting indoor and outdoor environmental conditions.
The sensor means that detects the indoor and outdoor temperatures, the air volume of the air conditioner,
And means for detecting the direction and humidity. The present invention
The storage means for holding the previous state of the sensor is N seconds (N is
Positive real value) Record the air temperature gradient of the air conditioner at intervals.
It is something to remember. The present invention also provides a
The operating means to be input is used when the user
I will express my feelings in two categories of hot and cold
It is something that has been done. The present invention also provides an operation time detecting means.
But a calendar that outputs signals about the month, day, and time
It has a function. The present invention also proposes operation contents.
Means to measure and learn using neural networks
It is a thing. In the present invention, the image input means may be an indoor
This is a means for detecting the distribution of thermal energy. The present invention
In addition, the feature extraction means includes: the number of persons in the room, the position and posture of a person,
Extract the number of moving objects, degree of movement, floor and wall temperatures
It is a means to issue. According to the present invention, the feature extracting means may include
For extracting information indicating the life scenes taking place inside
It is a step.

[0012]

According to the present invention, a rule for associating inputs and outputs from a group of sensors is not required by using a neural network as a means for estimating and learning the contents of an operation. hot,
The operation from the operation means for displaying two intentions of cold is learned, and the anticipation of the operation is predicted. In addition, the air conditioning of the automatic mode operation according to the room environment, personal preference and time, time and season from the start of operation is performed, and the user is released from the trouble of remote control setting by learning,
By using the indoor human information and the indoor heat distribution, the comfort of the user can be improved.

[0013]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic block diagram of an air conditioner control device according to one embodiment of the present invention. In FIG.
1 is a sensor means, 101 is a sensor signal, 2 is a storage means for storing a sensor value from the sensor means 1, 102 is a stored sensor value, 3 is an image input means, 103 is image data, 4 is a feature extraction means, Reference numeral 104 denotes an information signal from which a feature has been extracted, 5 denotes an operation means which is a remote control or an operation panel, 1
Reference numeral 05 denotes a signal indicating the operation content of the user, reference numeral 105a denotes a signal indicating the start of operation of the air conditioner 10, reference numeral 6 denotes an operation time detecting means for detecting an elapsed time from the start of operation, reference numeral 106 denotes a time signal from the start of operation, and reference numeral 7 Is a learning means, 107 is a learning result signal, 8 is an operation content estimating means, 108 is an operation content estimating signal, 9 is a control signal generating means, 109 is a control signal, and 10 is an air conditioner.

The operation of the above configuration will be described below. A large number of sensor means 1, for example, an outside air temperature sensor, a suction temperature (room temperature) sensor, a humidity sensor and the like are provided in the air conditioner 10. When the sensor signal 1 is output from the sensor unit 1, the contents such as the outdoor temperature, the suction temperature, and the humidity are output to the learning unit 7, the operation content estimation unit 8, and the storage unit 2. Storage means 2
Stores the history of the input sensor signal 101 for the past N seconds (N is a positive real number) and learns the state before the stored sensor value, for example, the inclination sensor value 102 of the room temperature at intervals of N seconds. Output to the means 7 and the operation content estimating means 8.

The image input means 3 outputs the image data 103 to the characteristic extracting means 4, and the characteristic extracting means 4
The feature amount 104 obtained from the step S03 is output to the learning means 7 and the operation content estimating means 8. Here, the image input means 3 is an indoor two-dimensional image input device, and the image data 103 is indoor two-dimensional image data. This image data 103
When the image input unit 3 is a visible image input device, a luminance value is output, and when the image input unit 3 is a pyroelectric infrared sensor, a heat temperature value is output.

The operation means 5, which is a remote controller or an operation panel, outputs two types of signals, for example, "hot" and "cold", which are indications of the user's current intention in air conditioning, as well as temperature and air volume. , A signal 105 indicating the setting or change of the wind direction or the like is output to the learning means 7 and the operation content estimating means 8. The operation means 5 outputs an operation start signal 105 a to the operation time detection means 6. The operating time detecting means 6 calculates a time t1 from the start of the operation of the air conditioner 10 to the present time, and outputs the calculated time t1 to the learning means 7 and the operation content estimating means 8.

T1 = tn−t0 (1) tn: current time t0: operation start time However, when absolute time tn, month m, and day d are used, tn set in operation time detecting means 6 , M, and d to the learning means 7 and the operation content estimating means 8.

Signals 101, 102, 104 and 105 output from the means 1, 2, 4, and 5 and a signal 106 (absolute) output from the operation time detecting means 6 and indicating the time t1 based on the above equation (1) The operation content estimating means 8 having input the time tn, the month m, and the day d) uses the neural network to determine the operation content of the operating means 5 such as "hot" or "cold". Infer. The operation content estimating signal 108 output from the operation content estimating means 8 is input to the control signal generating means 9 to generate a control signal 109 and generate the control signal 109.
Control 0.

On the other hand, similarly, the signals 101, 102, 104, 105, 1 output from the respective means 1, 2, 4, 5, 6
06 is also input to the learning means 7. The learning means 7 learns the neural network using the input values of each means and the actual operation contents of the operation means 5 as a teacher signal. After the learning is completed, the learning result signal (reference information) 107 in the neural network is output to the operation content estimating means 8, and the operation content estimating means 8 updates the internal reference information.

Next, the operation of the feature extracting means 4 will be described with reference to FIG. The image data 101 output from the image input unit 3 is passed to the feature extraction unit 4. The feature extracting means 4 is a memory 1 for storing image data 101
1, an image processing unit 12, and a feature amount extracting unit 13. Image data 1 temporarily stored in memory 11
01 is called by the image processing unit 12, and after performing processing such as noise removal and edge detection, the feature quantity extraction unit 13 sets the number of humans 14, the position of humans 15, the posture of humans 16, and the number of moving objects 17. , Degree of movement 18, floor and wall temperature 19
And other information. 14 to 18 can be collectively used as information indicating the indoor living scene as the indoor human information 20. Information indicating a life scene includes, for example, gathering, going to bed, housework, eating, studying, going home,
Leaving the room. As for extracting life scenes,
Japanese Patent Application No. 3-285106, filed earlier by the present applicant
The techniques disclosed in these specifications are applied to the practice of the present invention as it is or with some modifications.

Next, referring to FIG. 3, the operation content estimating means 8 will be described.
Will be described in detail. FIG. 3 shows a configuration of a neural network.
Although various models of the pattern classification type can be used, in this embodiment, an LVQ (Learning Vector Quantum: Reference,
Kohonen, "Self-Organization and Associative Memor
y ", 2nd, Springer-Verlag, 1988). In FIG. 3, reference numeral 21 denotes an input signal normalizing unit, and reference numeral 22 denotes a reference information unit, which is" hot "or" cold "according to human intentions. "
It has three categories (events) of “satisfied” and m pieces of reference information for each category. 23 is a distance calculation unit,
4 is a category calculation unit.

The operation of the above configuration will be described below. The input signal 210 is normalized in the input signal normalization unit 21. That is, all input signals x1 are processed as in the following equation.

Sx1 = (xmax−x1) / (xmax−xmin) (2) Sx1: normalized signal value xmax: maximum value of input signal xmin: minimum value of input signal

The input signal 210 includes the outside air temperature, the suction temperature, the slope of the suction temperature (the suction temperature before N seconds), the air volume, the set temperature, the operation time, the number of people, the position of the person, the position of the person, the number of moving objects, and the movement. Degree, wall temperature, and the like.
The signal normalized by the input signal normalization unit 21 is output to the distance calculation unit 23 as an input vector 211. The distance calculation unit 23 calculates the distance between the input vector and the reference information in each of the categories “hot”, “cold”, and “satisfied” representing the human intention of the reference information unit 22. That is, for example, the distance dAj from the reference information of the "hot" category
Is calculated from the following equation (3), and then the number of categories 3 and the number of pieces of reference information of each category are calculated by referring to the directory 221 of the reference information section 22. The contents of the directory 221 of the reference information section 22 are shown in FIG.

DAj = Σ (xi−Raji) ² (3) dAj: distance from j-th reference information xi: i-th input vector value Raji: i-th vector value of j-th reference information

Next, the distance between the reference information and the input vector is calculated for each category, and the same processing is performed for all the categories to calculate the distance. The total number dnum of distances obtained by the distance calculation unit 23 is as shown in the following equation (4).

Dnum = m1 + m2... + M3 (4)

Next, the distance dnum obtained by the distance calculator 23
Is output to the category calculation unit 24. The category calculator 24 calculates the shortest distance dmin among the dnum distances, and further calculates the category to which the corresponding reference information belongs. FIG. 5 shows a general algorithm for outputting the shortest distance. The output of the category calculation unit 24 that has calculated the category to which the reference information located at the shortest distance from the input vector belongs is the output of the operation content estimation unit 8 in FIG. For example, "Increase the set temperature by 1 degree" for the "Cold" category
For example, "set temperature is lowered by one degree" for the "hot" category, and "do not change the set temperature" for the "satisfied" category.

Next, the details of the learning means 7 will be described with reference to FIG. 6, reference numeral 51 normalizes an input signal common to the input signal normalization unit 21 in FIG. 2 and converts it into an input vector. Reference numeral 52 is also a reference information section common to FIG. 5
3 is a reference information search unit, 54 is a reference information generation unit, and 55 is a reference information learning unit.

The operation of the above configuration will be described below. The input signal normalized by the input signal normalization section 51 is used as an input vector as a reference information search section 53.
Is input to Furthermore, the correct category T of the input vector
C (teacher data) is also input to the reference information search unit 53 at the same time. The reference information search unit 53 refers to the directory of the reference information unit 52 and determines the correct category TC of the input vector.
Is the "hot" category, for example, when the number of reference information items in the "hot" category has the following relationship, the following processing is performed.

If ARnum <ARnummax, the processing of the reference information generator 54 is performed. If ARnum ≧ ARnummax, the reference information learning unit 55
Is performed. ARnum: the number of reference information in the "hot" category ARnummax: the maximum number of reference information in the "hot" category

The reference information generator 54 generates the input vector as reference information of the correct category TC. Here, reference information of the “hot” category is generated, and ARnum is incremented by one.

The reference information learning unit 55 compares the output value of the operation means 5 with the correct category TC of the input vector,
If they match, the shortest reference information is brought closer. If the output value of the operation means 5 does not match the correct category TC of the input vector, the reference information having the shortest distance is kept away. That is, if OC = TC, then RV = RV + α (RV−Sx) If OC ≠ TC, then RV = RV−α (RV−Sx) RV: reference information having the shortest distance from the input vector OC: output category (RV TC: correct category of input vector Sx: input vector α: learning rate

Therefore, the reference information learning unit 55 generates the input vector as the reference information of the correct category TC of the input vector until the number of reference information exceeds the maximum value of the reference information. After exceeding the maximum value, if the output category and the input category match, the reference information is slightly moved in the direction of the input vector, and if they do not match, the reference information is slightly moved away from the input vector. The distance to approach or move away is set by α.

As described above, according to the above-described embodiment, the use of the neural network as a means for estimating and learning the operation content does not require a rule for associating the input and the output from the sensor group. In addition to learning the operation from the operating means that displays the two intentions, hot and cold, in the environment of the environment, predicting the operation ahead of time, the room environment, personal preference, and the time, time, and season from the start of driving Air conditioning for automatic mode operation can be performed, and learning can relieve the user from the hassle of remote control setting, and improve user comfort by using indoor human information and indoor heat distribution. be able to.

[0036]

According to the present invention, as is apparent from the above embodiment, the indoor and outdoor environmental conditions obtained from the sensor means and the indoor characteristics obtained from the two-dimensional image information of the room output from the image input means. It is equipped with a means for estimating and learning the contents of the user's operation from the amount and the user's intention information obtained from the operation means, so that the room environment, indoor living scene and personal preference, time, season The air-conditioning operation suitable for can be performed automatically and easily.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of an air conditioner control device according to an embodiment of the present invention.

FIG. 2 is a schematic block diagram of a feature extracting unit in the apparatus.

FIG. 3 is a schematic block diagram of operation content estimating means in the apparatus.

FIG. 4 is a schematic diagram showing the contents of a directory of a reference information section in the operation content estimating means.

FIG. 5 is a flowchart showing an example of an algorithm for outputting the shortest distance in the operation content estimating means.

FIG. 6 is a schematic block diagram of a learning unit according to an embodiment of the present invention.

FIG. 7 is a schematic block diagram illustrating a configuration of an inference control software in the conventional air conditioner control device.

FIG. 8 is a schematic block diagram showing an operation of inference control software in the conventional device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Sensor means 2 Storage means 3 Image input means 4 Feature extraction means 5 Operation means 6 Operating time detection means 7 Learning means 8 Operation content estimation means 9 Control signal generation means 10 Air conditioner

──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor ▲ Yoshi ▼ Kunio Tada 3-10-1, Higashi Mita, Tama-ku, Kawasaki City, Kanagawa Prefecture Inside Matsushita Giken Co., Ltd. (72) Inventor Ikuo Akamine Kadoma City, Osaka Prefecture 1006 Oji Kadoma Matsushita Electric Industrial Co., Ltd. (72) Inventor Makoto Shimizu 1006 Oji Kadoma, Kadoma City, Osaka Pref. Matsushita Electric Industrial Co., Ltd. (72) Inventor Katsuhiko Fujiwara 1006 Okadoma Kadoma, Osaka Prefecture Address Matsushita Electric Industrial Co., Ltd. (56) References JP-A-3-177742 (JP, A) JP-A-62-147391 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) ) F24F 11/02

Claims (9)

(57) [Claims] A plurality of sensors for detecting indoor and outdoor environmental conditions.
Storage means for storing the previous state of the sensor means
And image input means for detecting two-dimensional image information in a room
From the image data obtained from the image input means.
Feature extraction means for extracting the amount of features of a user, and indication of user's intention
Operating means for inputting the
Operating time detecting means for detecting overtime, and the sensor means
And indoor and outdoor environmental conditions which are output values from the storage means.
And the feature extraction means, operation means, and operation time
Guess the user's operation content based on information from the detection means
Refer to the operation content estimating means and the results obtained from each of the above means.
Learning means for learning operation contents by referring to the operation contents;
Control the air conditioner based on the signal output from the measurement means.
An air conditioner control device comprising: a control signal generating unit . 2. A sensor means for sensing indoor and outdoor environmental conditions, indoor and outdoor temperatures, the air volume of the air conditioner, the wind direction, the air conditioner control apparatus of claim 1 further comprising a means for detecting the humidity. 3. A storage means for holding a previous state of the sensor,
3. The air conditioner control device according to claim 1, wherein an intake air temperature gradient of the air conditioner at intervals of N seconds (N is a positive real number) is stored. 4. The operation means for inputting a user's intention display, wherein the user indicates his or her bodily sensation in an indoor environment in two classifications of hot and cold during learning.
The air conditioner control device according to any one of 1 to 3 . 5. The operation time detection means, month, day, claim 1 having a calendar function for outputting a signal for time
5. The air conditioner control device according to any one of items 1 to 4 . 6. The air conditioner control device according to claim 1, wherein the means for estimating and learning the operation content uses a neural network. 7. The air conditioner control device according to claim 1, wherein the image input means is means for detecting a distribution of indoor thermal energy. 8. feature extracting means, room number, human position, orientation, number of moving objects, according to one degree of movement, claim 1 is a means for extracting the temperature of the floor and the wall 7 of the Air conditioner control device. 9. The method according to claim 1, wherein the feature extracting means is means for extracting information indicating a life scene performed indoors .
9. The air conditioner control device according to any one of claims 8 to 8 .