JP2672768B2 - Air conditioner - 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 for automatically controlling the indoor environment so that residents are comfortable.

[0002]

2. Description of the Related Art A conventional air conditioner is controlled so as to maintain a room temperature within a certain temperature range. However, the air conditioner should be designed to keep the sensation of heat of a person resident therein comfortable. In order to realize such comfort, a comfort index called PMV is proposed, and based on this index, an air conditioner is controlled as disclosed in Japanese Patent Application Laid-Open Nos. 2-178555 and 2-2.
No. 42037 is disclosed. PMV (Pred
(Icted MeanVote) is translated as an average expected thermal sensation report and is one index for evaluating the comfort of a thermal environment. It was proposed by Professor Wanger of the Technical University of Denmark, and was internationally standardized as ISO-7730 in 1984. It is a thing.

[0003] This PMV is a function of temperature, humidity, radiation temperature and air flow velocity which are environmental elements and the amount of activity and clothes which are human body elements.
730. It is assumed that the PMV value 0 is neutral and comfortable, 3 is hot, 2 is warm, 1 is slightly warm, -3 is cold, -2.
Is defined as cool and -1 is defined as slightly cool. Also,-
In the range of 0.5 <PMV value <0.5, 90% of the persons are comfortable, and this range is defined as the comfortable range. The description of the formula and the calculation method is omitted.

Japanese Patent Laid-Open Publication No. Hei 2-178555 discloses an air conditioner that calculates an optimal PMV value by expressing the amount of activity and the amount of clothing, which are factors on the human body side of a PMV, in a fuzzy set, and makes the PMV value neutral. To control. In addition, Japanese Unexamined Patent Publication
Japanese Patent No. 242037 discloses a method of installing a detecting means for detecting temperature, humidity and radiation temperature in a living area, calculating a PMV value from a setting means for setting an air flow velocity, an activity amount, and a clothing amount, and
In addition to cooperatively controlling various air conditioners so that the value becomes neutral, the control target value set in the comfortable range by the input of warmth of "hot" or "cold" is corrected.

[0005]

However, in the conventional method, the user does not know the current PMV value,
Since the control target value is corrected by a constant value by the input of the temperature sensation, the PMV value is outside the range of the comfort zone, and it is "cold" or PM when the cooling operation is on the hot side.
On the other hand, when the V value is on the cold side and the warm feeling is input during the heating operation, it is judged that the user has the warm feeling at that time, and the PMV value is changed all at once. The correction must be large and accurate, which is difficult.

In addition, it is inconvenient that a warm feeling input operation is frequently required each time the vehicle is driven. And the like.

The present invention is to solve the above-mentioned conventional problems. When the PMV value is outside the comfortable range and is on the hot side and the cooling operation is performed, conversely, it is "cold" or the PMV value is on the cold side. On the contrary, even if a "hot" temperature sensation is input during heating operation, the PMV value can be corrected in a large and accurate manner, eliminating the inconvenience of frequent temperature sensation input. The purpose is to provide an air conditioner.

[0008]

In order to achieve this object, an air conditioner of the present invention comprises a cooling / heating means for cooling or heating an interior of a room, and a PMV value calculation means for sequentially calculating a PMV value of a living area. Temperature sensation input means for inputting a user's sensation, such as hot and cold, once or in a plurality of degrees, correction amount designating means for determining a correction amount based on the frequency input by the temperature sensation input means and its predetermined width, and the PMV. PMV value correcting means for correcting the PMV value obtained by the value calculating means by the correction amount determined by the correction amount designating means, and "cold" or PMV value when the PMV value is on the hot side outside the comfort region. Change the specified width of the first number entered in the "hot" direction when it is on the cold side to be larger than the specified width of the first number entered when the PMV value is within the comfort zone. And a predetermined width changing means that, and a cooling and heating control means for controlling said heating and cooling means as PMV value is stable within the comfort region.

Further, the correction amount designating means designates the correction amount corresponding to the input amount of the temperature sensation input means based on the basic predetermined width and the input frequency of the temperature sensation input means. And when the PMV value is outside the comfort range and is on the hot side, it is "cold" or when the PMV value is on the cold side, it is "hot".
Direction, the predetermined width changing means specifies the correction amount according to the input amount of the temperature sensation input means when the predetermined width changing means changes the first predetermined width to be larger than the basic predetermined width. The second temperature sensation input amount designating means, the correction amount determined by the first temperature sensation input amount designating means, and factors such as the room temperature and the outside temperature at the time when the temperature sensation is input are associated with each other as the teacher signal Neuron generating means for generating,
A correction amount learning neural network which is composed of a plurality of neurons generated by the neuron generating means and outputs a result of arithmetic processing of output values of each neuron as a correction amount; the first and second thermal sensation input designating means; The correction amount learning neural network is configured by adding means for adding the correction amount.

Further, the correction amount designating unit designates a correction amount corresponding to the input amount of the temperature sensation input unit based on the basic predetermined width and the input frequency of the temperature sensation input unit. And when the PMV value is outside the comfort range and is on the hot side, it is "cold" or when the PMV value is on the cold side, it is "hot".
Direction, the predetermined width changing means specifies the correction amount according to the input amount of the temperature sensation input means when the predetermined width changing means changes the first predetermined width to be larger than the basic predetermined width. The second sense of heat input amount designating means, the correction amount determined by the first and second sense of heat input amount designating means, and the factors such as the room temperature and the outside temperature at the time when the sense of heat is inputted are associated with each other. A neuron generating means for generating a neuron, a correction amount learning neural network configured by a plurality of neurons generated by the neuron generating means and outputting as a correction amount the result of arithmetic processing of the output value of each neuron; And a second temperature sensation input designating means and an adding means for adding the correction amount of the correction amount learning neural network.

The correction amount designating means designates a correction amount corresponding to the input amount of the temperature sensation input means based on the basic predetermined width and the input frequency of the temperature sensation input means. And when the PMV value is outside the comfort range and is on the hot side, it is "cold" or when the PMV value is on the cold side, it is "hot".
Direction, the predetermined width changing means specifies the correction amount according to the input amount of the temperature sensation input means when the predetermined width changing means changes the first predetermined width to be larger than the basic predetermined width. Second temperature sensation input amount specifying means, a correction amount specifying means for outputting a correction amount smaller than the changed correction amount to a neuron generating means, which will be described later, when the predetermined width is changed, and generating a neuron; Neuron generation means for generating a neuron as a teacher signal by associating the correction amount determined by the one-sense input amount designating means and the teacher correction amount designating means with factors such as room temperature and outside temperature at the time of inputting the sense of temperature , A correction amount learning neural network which is composed of a plurality of neurons generated by the neuron generating means and outputs the result of arithmetic processing of the output value of each neuron as a correction amount When, and characterized in that it is constituted by an adder means for adding the correction amount of the said first and second temperature sensitive input specifying means correction amount learning neural network.

The neuron generation order storage means for recording the generation order of the plurality of neurons generated by the neuron generation means, and the oldest neuron stored by the neuron generation order storage means when the number of neurons exceeds a predetermined value. Neuron replacement means for replacing the latest neuron generated by the generation means.

[0013]

According to the present invention, when the PMV value is on the hot side outside the comfortable range, the present invention is "cold" or P-valued.
When the MV value is on the cold side, the specified width of the first number entered in the "hot" direction is greater than the specified width of the first number entered when the PMV value is within the comfort zone. Therefore, it is possible to correct the PMV value all at once and accurately.

Further, every time the temperature sensation input operation is performed, a neuron is generated which correlates the correction amount with factors such as the room temperature and the outside temperature at that time. Since the neuron determines the optimum correction amount according to the state of factors such as the outside temperature, it is not necessary to frequently perform the temperature sensation input operation.

Further, since the oldest neuron is replaced with the latest neuron when the number of neurons generated by the neuron generating means exceeds a predetermined value, the inconvenience of frequent thermal sensation input operation is eliminated with a small storage capacity. be able to.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of the air conditioner of this embodiment. In FIG. 1, reference numeral 1 denotes a PMV value calculating means for calculating and calculating a PMV value according to a formula described in ISO-7730. 2 is a room temperature detecting means, 3 is a humidity detecting means, and 4 is a radiant temperature detecting means, each of which is installed at an arbitrary position in the room to detect a state in the air-conditioned room. Reference numeral 5 is an air flow velocity setting means, 6 is a user's activity amount setting means, and 7 is a user's clothing amount setting means, which is provided on a remote controller 16 (to be described later) installed indoors for the user to operate. The room temperature detection means 2, the humidity detection means 3, the radiation temperature detection means 4, the air flow velocity setting means 5, the activity amount setting means 6, and the clothing amount setting means 7 are PMVs.
It is input to the PMV value calculation means 1 to calculate the value.

A reference numeral 8 is provided in a remote controller 16 which will be described later, and a warm sensation input means for inputting a warm sensation of the user such as hot and cold, 9
Is a correction amount specifying means for determining a correction amount corresponding to the input amount of the thermal sensation input means 8;
The frequency is set so as to correspond to the correction of 0.25 (basic predetermined width) in V value. Reference numeral 10 denotes a PMV value correction unit that corrects the PMV value calculated by the PMV value calculation unit 1 by the correction amount specified by the correction amount designation unit 9. Reference numeral 11 denotes comfort display means for displaying whether or not the corrected PMV value has entered the comfort region (-0.5 to 0.5), which is referred to as a comfort sign in this embodiment. The air conditioner of this embodiment has a PMV value
(If there is a correction by the PMV value correction means 10, the corrected P
(MV value) controls the heating and cooling operation so that it enters the comfortable range.
And it is comfortable with a PMV value of 0 being neutral.
3 hot, 2 warm, 1 slightly warm, -3 cold
I define -2 as cool and -1 as slightly cool.
Therefore, if PMV value <0, heating operation, PMV value> 0
In case of, cooling operation is performed. 12 is the current PMV value (PM
PMV after correction, if there is correction by the V value correction means 10.
Value) is on the hot side outside the comfort range, that is, “cold” during cooling operation, or when the current PMV value is on the cold side, that is, the first time input in the “hot” direction during heating operation. The predetermined width changing means changes the predetermined width of the number to the opposite sign of the integer multiple of 0.25 that is the closest to the current PMV value in the direction away from the comfort area. However, the predetermined width per number after the second frequency is the basic predetermined width of 0.
25, and the correction amount is 0.25 from the correction amount for the first time.
They are added and subtracted one by one . For example, the PMV value is −
In the condition of 0.3 (during heating operation), there is a feeling of "hot"
When input twice, 0.25 is close to -0.3.
Of -0.25 and -0.5 which are multiples, it is a comfortable area
The numerical value in the direction away from -0.5 is selected and vice versa.
The code value of 0.5 is the correction amount for the first several times. Further
In addition to this, the correction amount of the second basic predetermined width (0.25)
Is added, it becomes 0.75. Therefore, the PMV value is-
In the condition of 0.3 (during heating operation), there is a feeling of "hot"
If input twice, 0.75 will be corrected.
The PMV value after correction is -0.3 which is the PMV value before correction.
The correction amount becomes 0.75 by adding 0.75, and the heating operation is immediately stopped. Furthermore, (in cooling operation) in the state of PMV value of 0.6, if the warming of "cold" is one power input, 0.5 to be an integral multiple value of 0.25 is close to 0.6
Of 0.75, it is the numerical value in the direction away from the comfort zone
0.75 is selected and -0.75 of its opposite sign value is corrected
It becomes the amount. Therefore, when the PMV value is 0.6 (cooling operation
When the warm feeling of "cold" is entered once.
Is corrected by -0.75, and the corrected PMV value is
The correction amount of -0.75 is added to the PMV value of 0.6 before correction.
The added value becomes -0.15, and the cooling operation is immediately stopped.

Reference numeral 13 denotes a temperature sensation input reception prohibiting means for prohibiting the reception of the temperature sensation input, which is transmitted from the PMV value correcting means 10 to the PMV.
When a value is input and the temperature is outside the range of the comfortable area (0.5 or more during cooling, −0.5 or less during heating), a warm feeling other than the condition under which the predetermined width changing unit 12 operates is input. In this case, acceptance of the warm sensation input by the correction amount designating means 9 is prohibited. Reference numeral 19 denotes a warmth sensation monitor, which will be described later, and displays a warmth sensation input amount. Reference numeral 14 is a cooling / heating means for cooling or heating the inside of the room by a refrigeration cycle, and 15 is a cooling / heating control means for controlling the cooling / heating means 14 so that the corrected PMV value is stable in the comfort region.

FIG. 2 is a front view of the remote controller, and 16 is the remote controller main body. Reference numeral 17 is a cold key that the user inputs when it is cold, 18 is a hot key that is input when the user is hot, and the temperature sensation input means 8 is composed of a cold key 17 and a hot key 18. Each time the cold key 17 and the hot key 18 are pressed, the PMV value is corrected by 0.25 each time (however, when the predetermined width changing means 12 is activated, the PMV value at that time is used as a reference for the first one-time operation). The predetermined width of 0.25 has been changed
It becomes larger, and 0.25 per number once after the second frequency
If you keep pressing the button, the correction will be continued by 0.25.

Reference numeral 19 is a temperature sensation monitor for displaying the sensation of temperature input by the temperature sensation input means 8. The segment in the neutral position is always displayed, and when the key 17 is pressed in the cold, the left segment is one. When the key 18 is pressed when it is hot, the right segments are displayed one by one.
The PMV correction amount per segment is 0.25 (when the predetermined width changing means 12 is not operating). Reference numeral 11 denotes a comfort sign of the comfort display means.

Numeral 20 designates the air volume of an indoor blower 32 described later as "strong", "medium", and "weak" (air flow velocities of 0.3, 0.
(Equivalent to 2, 0.1 m / s) With the air volume setting key set in 3 stages, each time the key is pressed, "strong" → "medium" → "weak" →
It changes to "strong". Reference numerals 21a, 21b, and 21c are air volume display lamps for displaying air volume settings of "weak", "medium", and "strong", respectively. The air flow velocity setting means 5 comprises an air volume setting key 20 and air volume display lamps 21a, 21b, 21c.

22 indicates the activity amount as "light", "medium", "heavy"
(Equivalent to 1.2, 1.4 and 1.6 met, respectively.) With the activity amount setting keys set in 3 levels, each time you press the key, "light" → "medium" → "heavy" → "light" And changes. 2
3a, 23b and 23c are "light", "medium",
It is an activity amount display lamp that displays the activity amount setting of “heavy”. The activity amount setting means 6 comprises an activity amount setting key 22 and activity amount display lamps 23a, 23b, 23c.

Numeral 24 is a clothing amount setting key for setting the amount of clothing in three stages of "summer clothing", "combined clothing" and "winter clothing" (corresponding to activity amounts of 0.5, 0.7 and 1.0 clo, respectively). , Every time you press the key, it changes from "summer clothes" to "combined clothes" to "winter clothes" to "summer clothes". 25a, 25b, 25c are "summer clothes",
It is a clothing amount display lamp that displays the clothing amount setting of “combined clothing” and “winter clothing”. The clothing amount setting means 7 is a clothing amount setting key 24.
And clothing amount display lamps 25a, 25b, 25c. Reference numeral 26 denotes an operation / stop key for instructing the operation and stop of the cooling / heating means 14.

FIG. 3 is a schematic diagram of the cooling / heating means 14. In FIG. 3, 27 is a compressor, 28 is a four-way valve, and 29.
Is an indoor heat exchanger that is a heat exchange means for heating or cooling the intake air in the room, 30 is a decompressor, and 31 is an outdoor heat exchanger, and these are connected in a ring to form a refrigeration cycle. 32 sucks indoor air, and indoor heat exchanger 2
An indoor blower that blows out the air heated or cooled by 9 and 33 is an outdoor blower. The indoor blower 32 can set the air volume in three stages of "strong", "medium", and "weak". Reference numeral 34 is an indoor unit installed indoors, and 35 is an outdoor unit installed outdoors. Switching between the cooling operation and the heating operation is performed by switching the four-way valve 28 to switch the flow of the refrigerant in the refrigeration cycle.

An operation example of the air conditioner having the above configuration will be briefly described with reference to the drawings. In the example of FIG. 2, the air volume is “strong”,
The activity amount is set to "light" and the clothing amount is set to "winter clothes". In this state, the PMV value calculated by the PMV value calculation means 1 is -1.
In the case of 5, the PMV value deviates from the comfort area to the negative side in this state, so the cooling / heating means 14 performs heating operation, and the PMV value gradually increases and approaches the comfort area.
At this time, the comfort sign 11 is off.

When the PMV value is -0.9 (during heating operation), the user feels that it is hot for some reason and presses the key 18 once when it is hot. The predetermined width of the number is an integer of 0.25 close to -0.9
Of the double value -0.75 and -1.0, from the comfort zone
-1.0, which is the numerical value in the direction of leaving, is selected and its opposite
It becomes 1.0 of the issue number. In other words, since the value is changed from 0.25 to 1.0, the corrected PMV value is the uncorrected PMV value.
Is -0.9 which is the correction amount of 1.0 plus 0.1.
Then, the comfort zone is entered, and the comfort sign 11 lights up. After that, the cooling / heating means 14 repeats the operation / stop of the heating operation within the range of the comfortable region, and the PMV value becomes neutral and stable.

On the other hand, when the key 17 is pressed when the PMV value is -0.9 when it is cold, the temperature sensation input prohibiting means 13 prohibits the temperature sensation input from being accepted because the heating operation is being performed. The PMV value will not be corrected at all.
However, if the PMV value falls within the range of the comfort region, the prohibition of acceptance of the temperature sensation input is released, and the user can correct the PMV value by the temperature sensation input. In this case, the predetermined width per number is 0.25.

As described above, according to the present embodiment, the cooling / heating means 14 for cooling or heating the room, the PMV value calculating means 1 for sequentially calculating the PMV value of the living area, and the feeling of the user such as hot and cold. Is input in a plurality of degrees, a correction amount designating means 9 that determines a correction amount based on the input frequency of the temperature sensation input means 8 and its predetermined width, and a PMV value obtained by the PMV value calculating means 1. PMV value correcting means 10 for correcting only the correction amount determined by the correction amount designating means 9.
And when the PMV value is outside the comfort range and is on the hot side, it is "cold" or when the PMV value is on the cold side, it is "hot".
The predetermined width of the first number of times input in the direction is set to 0. 0 which is the closest to the current PMV value in the direction away from the comfort area.
Specified width changing means 12 for changing to the opposite sign of the multiple of 25
And the cooling / heating control means 15 for controlling the cooling / heating means 14 so that the PMV value is stable within the range of the comfort region, the PMV value is made large and accurate according to the warmth of the user. Can be corrected to.

A second embodiment of the present invention will be described below with reference to the drawings. FIG. 4 is a block diagram of the air conditioner of the second embodiment. In FIG. 4, reference numeral 36 denotes a correction amount designating means for learning the correction amount of the thermal sensation input by means of a neural network. The first thermal sensation input amount designating means 37, the second thermal sensation input quantity designating means 38, and the neuron generating means 39. When,
It is composed of a neuron generation order storage means 40, a neuron replacement means 41, a correction amount learning neural network 42, and an addition means 43.

The first thermal sensation input amount designating means 37 designates a correction amount corresponding to the input amount of the thermal sensation inputting means 8, and the second thermal sensation input amount designating means 38 operates when the predetermined width changing means 12 operates. A correction amount corresponding to the input amount of the temperature sensation input means 8 is designated.
The method of changing the predetermined width in this case is the same as that of the first embodiment, and a description thereof will be omitted.

The neuron generating means 39 learns the correction amount of one neuron that correlates the correction amount by the first temperature input amount designating means 37 with the room temperature, the outside air temperature and the PMV value at the time of inputting the temperature sensation. It is generated in the neural network 42. The neuron generation order storage unit 40 stores the generation order of the neurons generated by the neuron generation unit 39, and the neuron replacement unit 41 replaces the latest neuron with the oldest neuron when the number of generated neurons exceeds a predetermined number. To do.

The correction amount learning neural network 42 is composed of a plurality of neurons generated by the neuron generating means 39 (a predetermined number of intermediate layer neurons to be described later).
The input value of the room temperature, the ambient temperature, and the PMV value are sequentially input, and the average value of the outputs of the fired neurons is output to the adding means 43 as a correction amount . In addition, "ignition" means new
Terminology commonly used in the Ral network field
Then, the state in which the neuron generated an action potential pulse (excitation
I was excited about it.
The non-existent state is called the stationary state. And the neuro that fired
Signals to another synaptic neuron
Because The adding means 43 is the first thermal sensation input amount designating means 3
7, the second thermal sensation input amount designating means 38 and the correction amount learning neural network 42 are added, and the correction amounts are added to obtain PM.
The value is output to the V value correction means 10.

Thus, when the temperature sensation is input, the correction amount of the correction amount learning neural network 42 is the correction amount of the first temperature sensation input amount designating means 37 or the second temperature sensation input amount designating means 38. In that case, it will be added and output. Reference numeral 44 is an outside air temperature detecting means for detecting the outside air temperature. Hereinafter, the same components as those in FIG. 3 are designated by the same reference numerals and the description thereof will be omitted.

An example of the operation of the air conditioner having the above structure will be briefly described with reference to the drawings. FIGS. 5A and 5B are conceptual diagrams showing the relationship between one neuron generated in the correction amount learning neural network 42 by the neuron generating means 39 and the input / output conditions. FIG. Neural network 42 for learning correction amount when temperature sensation is input
Represents the structure of. 45, 46, and 47 are the first and second inputting room temperature X1, outside air temperature X2, and PMV value X3, respectively.
And the third input layer neuron. Reference numeral 48 denotes a first intermediate-layer neuron generated by the neuron generating means 39 at the time of the first input of the temperature sensation, which outputs a temperature sensation input correction amount Y1 which will be described later when the firing state occurs. Reference numeral 49 denotes an output layer neuron, which is the amount of the temperature-sensing input correction Y1 of the first intermediate layer neuron
And outputs the average value (in this case, Y1 is output). There is one output layer neuron 49. FIG.
(B) shows the input / output relationship of the first intermediate layer neuron 48, and the output Y1 is expressed by the following equation.

[0036]

(Equation 1)

[0037]

(Equation 2)

F (X) is a threshold function (f (X) = 1 (X
≧ 0), f (X) = 0 (X <0)), A is the temperature sensation input amount designating means 3 when the first intermediate layer neuron 48 is generated.
It is stored in the first intermediate layer neuron 48 with the amount of correction of the temperature sensation input determined by 7. W1, W2 and W3 are coupling weights with respect to the input signal, T is a threshold value,
It is determined by the neuron generating means 39 when the neuron is generated.

When X ≧ 0, the neuron is in a firing state. It should be noted that the intermediate layer neurons are generated one by one by one thermal sensation input operation, and the maximum number of intermediate layer neurons are configured by a predetermined number of operations. The neuron generated by the warm sensation input operation after the predetermined number of times is replaced by the neuron replacement unit 41 with the oldest neuron in the generation order of the intermediate layer neurons stored by the neuron generation order storage unit 40. On the other hand, in the initial state in which the temperature sensation input is not operated even once, the correction amount learning neural network 42
There are no neurons in the hidden layer and no connection from the input layer to the output layer. For this reason, even if an input signal is input from the input layer, the output layer outputs 0. That is, the correction amount is zero.

FIG. 6 is a structural diagram of the correction amount learning neural network 42 learned by the temperature sensation input operation five times. 48, 50, 51, 52, and 53 are the first, second, third, fourth, and fifth generated by the temperature sensation input operation five times.
Is a middle layer neuron. The output layer neuron 49 outputs, as a correction amount, the average value of the outputs (Y1 to Y5) of the intermediate layer neurons in the firing state according to the condition of the input signal.
For example, when the first and second intermediate layer neurons 48 and 50 are in the firing state, the output layer neuron 49 is (Y1 + Y).
2) / 2 is output, and (Y1 + Y2 + Y3 + Y4 + Y5) / 5 when all the hidden neurons are in the firing state.
Is output.

As described above, according to the second embodiment, the correction amount designating means 36 is similar to the first embodiment, and the first temperature sensation input amount designating means 37 for defining the correction amount of the predetermined width unchanged. The second temperature sensation input amount designating means 38 that determines the correction amount by changing the predetermined width by the method described above, and the correction amount determined by the first temperature sensation input amount designating means 37 and the room temperature and the outside temperature at the time when the temperature sensation is input. And a neuron generating means 39 for generating a neuron as a teacher signal by associating with the above factor and a plurality of neurons generated by the neuron generating means, and outputs the result of arithmetic processing of the output value of each neuron as a correction amount. A correction amount learning neural network 42, a neuron generation order storage unit 40 that stores the generation order of generated neurons, and the latest neuron when the number of generated neurons exceeds a predetermined number. Is composed of a neuron replacing means 41 for replacing old neurons, an adding means 43 for adding the correction amounts of the first and second temperature sensation input designating means 37, 38 and the correction amount learning neural network 42. Therefore, every time the temperature sensation input operation is performed, a neuron is generated that correlates the correction amount with the room temperature, the outside temperature, and the PMV value at that time, and even if the temperature sensation input operation is not performed, the room temperature and the outside temperature are generated. Since the neuron can determine the optimum correction amount according to the state of the PMV value and can preempt the warm sensation input operation, it is not necessary to frequently perform the warm sensation input operation.

Further, since the oldest neuron is replaced with the latest neuron when the number of neurons exceeds a predetermined number, the correction amount learning neural network 42 can be constructed with a limited storage capacity. Further, since the correction amount of the second temperature sensation input designating means 38 is not used for learning of the neural network, when the PMV value is on the hot side outside the range of the comfort region, that is, when the PMV value is “cold” or the PMV value is When the temperature is on the cold side, that is, the temperature sensation input in the "hot" direction during heating operation is in a transitional period due to heat and it cannot be said that it is an accurate temperature sensation, so correct learning of the neural network is required. It can be carried out. However, the PMV value at the time when the temperature sensation is input is corrected.

The third embodiment of the present invention will be described below with reference to the drawings. FIG. 7 is a block diagram of the air conditioner of the third embodiment. In FIG. 7, reference numeral 54 is a correction amount designating means for learning the correction amount of the thermal sensation input by a neural network, which is a first thermal sensation input amount designating means 55, a second thermal sensation input amount designating means 56, and a neuron generating means 57. When,
It comprises a neuron generation order storage means 58, a neuron replacement means 59, a correction amount learning neural network 60, and an addition means 61.

The first thermal sensation input amount designating means 55 designates a correction amount corresponding to the input amount of the thermal sensation input means 8, and the second thermal sensation input amount designating means 56 operates when the predetermined width changing means 12 operates. A correction amount corresponding to the input amount of the temperature sensation input means 8 is designated.
The method of changing the predetermined width in this case is the same as that of the first embodiment, and a description thereof will be omitted.

The neuron generating means 58 correlates the correction amount by the first and second temperature sensation input quantity designating means 55, 56 with the room temperature, the outside air temperature and the PMV value at the time of the temperature sensation input. A neuron is generated in the correction amount learning neural network 57. Neuron generation order storage means 58
Stores the generation order of the neurons generated by the neuron generation means 57, and the neuron replacement means 59 performs an operation of replacing the latest neuron with the oldest neuron when the number of generated neurons exceeds a predetermined number.

Then, the correction amount learning neural network 60 is composed of a plurality of neurons generated by the neuron generating means 57, and sequentially, room temperature, outside temperature and PM.
The V value is input and the average value of the outputs of the fired neurons is output to the adding means 61 as a correction amount. The adding unit 61 adds the correction amounts output by the first temperature sensation input amount designating unit 55, the second temperature sensation input amount designating unit 56, and the correction amount learning neural network 60, and outputs the addition result to the PMV value correcting unit 10.

As a result, when the temperature sensation is input, the correction amount of the first temperature sensation input amount designating unit 55 or the second temperature sensation input amount designating unit 56 includes the correction amount of the correction amount learning neural network 60. In that case, it will be added and output. Hereinafter, the same components as those in FIG. 4 are designated by the same reference numerals and the description thereof will be omitted.

An operation example of the air conditioner having the above configuration will be briefly described. When a temperature sensation is input by the temperature sensation input means 8, it is designated by the first and second temperature sensation input amount designating means 55, 56. The correction amount is input to the neuron generating means 57, and the intermediate layer neuron that stores the correction amount is generated in the correction amount learning neural network 60.

As described above, according to the third embodiment, the correction amount designating means 54 is similar to the first embodiment, and the first temperature sensation input quantity designating means 55 for defining the correction amount of the predetermined width unchanged. Second thermal sensation input amount designating means 56 for determining the correction amount by changing the predetermined width by the method of (1) and first and second thermal sensation input amount designating means 55, 5
6, a neuron generating unit 57 that correlates a correction amount determined by 6 with factors such as room temperature and outside temperature at the time of inputting a temperature sensation to generate neurons as a teacher signal, and a plurality of neurons generated by the neuron generating unit 57. A correction amount learning neural network 60 configured as described above and outputting the result of arithmetic processing of the output value of each neuron as a correction amount.
A neuron generation order storage means 58 for storing the generation order of the generated neurons; a neuron replacement means 59 for performing an operation of replacing the latest neuron with the oldest neuron when the number of generated neurons exceeds a predetermined number; Since the first and second temperature sensation input designating means 55, 56 and the adding means 61 for adding the correction amount of the correction amount learning neural network 60 are added, the correction amount is set each time the temperature sensation input operation is performed. A neuron associated with the room temperature, the ambient temperature, and the PMV value at that time is generated,
Room temperature, outside temperature and PM
Since it is possible for the neuron to determine the optimum correction amount according to the state of the V value and to preempt the warm sensation input operation, it is not necessary to frequently perform the warm sensation input operation.

Further, since the oldest neuron is replaced with the latest neuron when the number of neurons exceeds a predetermined number, the correction amount learning neural network 60 can be constructed with a limited storage capacity. Further, since the correction amount of the second temperature sensation input designating means 56 is also used for learning of the neural network, the temperature sensation input operation in the thermal transitional period can be preempted, and the temperature sensation input operation can be performed. The number of times can be further reduced.

A fourth embodiment of the present invention will be described below with reference to the drawings. FIG. 8 is a block diagram of the air conditioner of the fourth embodiment. In FIG. 8, reference numeral 62 designates a correction amount designating means for learning the correction amount of the thermal sensation input by a neural network. The first thermal sensation input amount designating means 63, the second thermal sensation input amount designating means 64 and the teacher correction amount designation. Means 65,
It is composed of a neuron generation means 66, a neuron generation order storage means 67, a neuron replacement means 68, a correction amount learning neural network 69, and an addition means 70.

The first thermal sensation input amount designating unit 63 designates a correction amount corresponding to the input amount of the thermal sensation input unit 8, and the second thermal sensation input amount designating unit 64 operates when the predetermined width changing unit 12 operates. A correction amount corresponding to the input amount of the temperature sensation input means 8 is designated.
The method of changing the predetermined width in this case is the same as that of the first embodiment, and a description thereof will be omitted.

The teacher correction amount designating means 65 receives the correction amount of the second temperature sensation input amount designating means 64 when the predetermined width changing means 12 is operated, and in this embodiment, the basic predetermined value is smaller than the correction amount after the predetermined width is changed. The width of 0.25 is output to the neuron generating means 66 as a teacher correction amount. The neuron generating means 66 correlates the correction amount designated by the first temperature sensation input amount designating means 63 and the teacher correction amount designating means 65 with the room temperature, the outside air temperature and the PMV value at the time when the temperature sensation is input. The individual neurons are generated in the correction amount learning neural network 69. The neuron generation order storage means 67 stores the generation order of the neurons generated by the neuron generation means 66, and the neuron replacement means 68 replaces the latest neuron with the oldest neuron when the number of generated neurons reaches a predetermined number or more. To do.

The correction amount learning neural network 69 is composed of a plurality of neurons generated by the neuron generating means 66, and sequentially stores room temperature, ambient temperature and PM.
The V value is input and the average value of the outputs of the fired neurons is output to the adding means 70 as a correction amount. The adding unit 70 adds the correction amounts output from the first temperature sensation input amount designating unit 63, the second temperature sensation input amount designating unit 64, and the correction amount learning neural network 69, and outputs them to the PMV value correcting unit 10.

As a result, when the temperature sensation is input, the correction amount of the first temperature sensation input amount designating means 55 or the second temperature sensation input amount designating means 56 includes the correction amount of the correction amount learning neural network 60. In that case, it will be added and output. Hereinafter, the same components as those in FIG. 4 are designated by the same reference numerals and the description thereof will be omitted.

An example of the operation of the air conditioner having the above-mentioned configuration will be briefly described. When a temperature sensation is input by the temperature sensation input unit 8, the first temperature sensation input amount designating unit 63 and the teacher correction amount designating unit 65 are used. The specified correction amount is input to the neuron generating means 66, and the intermediate layer neuron that stores the correction amount is generated in the correction amount learning neural network 69.

As described above, according to the fourth embodiment, the correction amount designating means 62 is similar to the first embodiment in that the first warm sensation input amount designating means 63 determines the correction amount of the predetermined width unchanged. The second temperature sensation input amount designating means 64 that determines the correction amount by changing the predetermined width by the method described above, and the correction amount of the second temperature sensation input amount designating means 64 when the predetermined width changing means 12 operates, and receives the predetermined width. By the teacher correction amount designating means 65 which outputs a basic predetermined width 0.25 smaller than the changed correction amount to the neuron generating means 66 as the teacher correction amount, the first warm sensation input amount designating means 63 and the teacher correction amount designating means 65. A neuron generating unit 66 that generates a neuron using a teacher as a teacher signal by associating the determined correction amount with factors such as room temperature and outside temperature at the time of inputting a warm sensation, and a plurality of neurons generated by the neuron generating unit 66. Each A correction amount learning neural network 69 that outputs the result of arithmetic processing of the output value of the huron as a correction amount, a neuron generation order storage unit 67 that stores the generation order of the generated neurons, and a predetermined number or more of the generated neurons. If so, a neuron replacing means 68 for performing an operation of replacing the latest neuron with the oldest neuron, and an adding means 70 for adding the correction amounts of the first and second temperature sensation input designating means 63 and 64 and the correction amount learning neural network 69. Since it is configured by, the correction amount and the room temperature at that time each time the temperature sensation input operation is performed,
A neuron associated with the outside air temperature and the PMV value is generated, and even if the temperature sensation input operation is not performed, the neuron determines the optimum correction amount according to the state of the room temperature, the outside air temperature and the PMV value, and the temperature sensation input operation is performed. Therefore, it is not necessary to frequently perform the temperature sensation input operation.

Since the oldest neuron is replaced with the latest neuron when the number of neurons exceeds a predetermined number, the correction amount learning neural network 69 can be constructed with a limited storage capacity. Further, since the correction amount smaller than the correction amount of the second temperature sensation input designating means 64 is used for learning of the neural network, the learning of the neural network can be performed relatively accurately and at the same time during the thermal transition period. In this case, the temperature sensation input operation can be pre-empted, and the number of times of temperature sensation input operation can be further reduced.

In the above embodiment, the input conditions were room temperature, the outside air temperature and the PMV value. However, conditions such as the PMV value change and the elapsed time after the start of operation may be added to this, or another input condition may be used. The same effect can be expected. Further, although the cooling / heating means 9 is constituted by the refrigeration cycle, the same effect can be expected by using this as another cooling / heating means such as a radiation cooling / heating panel. Further, the same effect can be expected if the PMV value calculating means 1 inputs related conditions other than the input conditions shown in the embodiment and learns by a neural network to infer the PMV value.

[0060]

As described above, the air conditioner of the present invention comprises a cooling / heating means for cooling or heating a room, and a PMV in a residential area.
Based on the PMV value calculating means for sequentially calculating the values, the temperature sensation input means for inputting the user's sensation of being hot or cold, etc. once or in a plurality of degrees, and the degree of input by the temperature sensation input means and its predetermined width. A correction amount designating unit that determines a correction amount, and a PMV that corrects the PMV value calculated by the PMV value computing unit by the correction amount defined by the correction amount designating unit.
The V value correction means and the predetermined width of the first number input in the "cold" direction when the PMV value is outside the comfort range and on the hot side or when the PMV value is on the cold side,
A predetermined width changing means for changing the PMV value so as to be larger than a predetermined width of the first input once when the PMV value is within the comfort area, and the cooling and heating so that the PMV value is stabilized within the comfort area. Since it is provided with the cooling / heating control means for controlling the means, there is an effect that the PMV value can be corrected in a large and accurate manner in accordance with the warm feeling of the user.

Further, in the air conditioner of the present invention, the correction amount designating unit designates the correction amount corresponding to the input amount of the temperature sensation input unit based on the basic predetermined width and the input frequency of the temperature sensation input unit. The first temperature sensation input amount designating means and the predetermined width due to the temperature sensation being input in the "cold" direction when the PMV value is on the hot side outside the comfort region or in the "hot" direction when the PMV value is on the cold side. Second thermal sensation input amount designating means for designating a correction amount according to an input amount of the thermal sensation input means when the modifying means modifies the predetermined width of the first one to be larger than the basic predetermined width, A neuron generating means for generating a neuron as a teacher signal by associating a correction amount determined by the first thermal sensation input amount designating means with factors such as room temperature and outside temperature at the time of thermal sensation input, and the neuron generating means. Generated by Correction amount learning neural network which is configured by an arron and outputs the result of calculation processing of the output value of each neuron as a correction amount, and the correction amounts of the first and second temperature sensation input designating means and the correction amount learning neural network are added. It is characterized in that the neuron determines the optimum correction amount according to the room temperature, the outside air temperature, and the PMV value state even if the temperature sensation input operation is not performed. Since the input operation can be pre-empted, it is not necessary to frequently perform the warm sensation input operation. Further, since the correction amount of the second temperature sensation input designating means is not used for learning the neural network, there is an effect that the neural network can be accurately learned.

Further, in the air conditioner of the present invention, the correction amount designating unit designates the correction amount corresponding to the input amount of the temperature sensation input unit based on the basic predetermined width and the input frequency of the temperature sensation input unit. The first temperature sensation input amount designating means and the predetermined width due to the temperature sensation being input in the "cold" direction when the PMV value is on the hot side outside the comfort region or in the "hot" direction when the PMV value is on the cold side. Second thermal sensation input amount designating means for designating a correction amount according to an input amount of the thermal sensation input means when the modifying means modifies the predetermined width of the first one to be larger than the basic predetermined width, A neuron generating means for generating a neuron as a teacher signal by associating a correction amount determined by the first and second thermal sensation input amount designating means with a factor such as room temperature or outside temperature at the time of thermal sensation input; Generated by the neuron generator Number learning neurons for outputting the result of arithmetic processing of the output value of each neuron as a correction amount, a correction amount learning neural network, correction amounts for the first and second temperature sensation input designating means, and the correction amount learning neural network Since it is configured by an adding means for adding, the neuron determines an optimum correction amount according to the room temperature, the outside air temperature, and the PMV value state even if the temperature sensation input operation is not performed. Since the temperature sensation input operation can be pre-empted, it is not necessary to frequently perform the temperature sensation input operation. In addition, since the correction amount of the second temperature sensation input designating means is also used for learning the neural network, the temperature sensation input operation in the thermal transitional period can be preempted, and the number of times of temperature sensation input operation is performed. The effect is that it can be further reduced.

Further, in the air conditioner of the present invention, the correction amount designating unit designates the correction amount corresponding to the input amount of the temperature sensation input unit based on the basic predetermined width and the input frequency of the temperature sensation input unit. The first temperature sensation input amount designating means and the predetermined width due to the temperature sensation being input in the "cold" direction when the PMV value is on the hot side outside the comfort region or in the "hot" direction when the PMV value is on the cold side. Second thermal sensation input amount designating means for designating a correction amount according to an input amount of the thermal sensation input means when the modifying means modifies the predetermined width of the first one to be larger than the basic predetermined width, When the predetermined width is changed, a correction amount smaller than the changed correction amount is output to a neuron generating unit, which will be described later, to generate a neuron correction amount specifying unit,
Neuron generation for generating neurons as teacher signals by correlating the correction amount determined by the first temperature sensation input amount designating unit and the teacher correction amount designating unit with factors such as room temperature and outside temperature at the time of temperature sensation input Means, a correction amount learning neural network composed of a plurality of neurons generated by the neuron generating means and outputting the result of arithmetic processing of the output value of each neuron as a correction amount, and the first and second thermal sensation input designation And a means for adding the correction amount of the correction amount learning neural network, the state of the room temperature, the outside air temperature and the PMV value even if the temperature sensation input operation is not performed. Allows the neuron to determine the optimum correction amount and pre-empt the temperature sensation input operation, so there is no need to perform the temperature sensation input operation frequently. Further, since the correction amount smaller than the correction amount of the second thermal sensation input designating means is used for learning the neural network, the learning of the neural network can be performed relatively accurately, and the temperature is not changed during the transitional period. In some cases, the temperature sensation input operation can be preempted, and the number of times of the temperature sensation input operation can be further reduced.

In the air conditioner of the present invention, the neuron generation order storage means for recording the generation order of the plurality of neurons generated by the neuron generation means, and the neuron generation order storage means when the number of neurons exceeds a predetermined value. Since it has a neuron replacement means for replacing the oldest stored neuron with the latest neuron generated by the neuron generating means, when the number of neurons exceeds a predetermined number, the oldest neuron is replaced with the latest neuron. Therefore, the correction amount learning neural network can be configured with a limited storage capacity.

[Brief description of the drawings]

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

FIG. 2 is a front view of the remote controller according to the embodiment.

FIG. 3 is a schematic configuration diagram of a cooling / heating unit in the embodiment.

FIG. 4 is a block diagram of an air conditioner according to a second embodiment of the present invention.

5A is a structural diagram of a correction amount learning neural network in the same embodiment. FIG. 5B is a neuron input / output relationship diagram in the same embodiment.

FIG. 6 is a structural diagram of a correction amount learning neural network in the embodiment.

FIG. 7 is a block diagram of an air conditioner according to a third embodiment of the present invention.

FIG. 8 is a block diagram of an air conditioner according to a fourth embodiment of the present invention.

[Explanation of symbols]

 1 PMV value calculating means 8 Temperature sensation input means 9, 36, 54, 62 Correction amount designating means 10 PMV value correcting means 12 Specified width changing means 14 Cooling and heating means 15 Cooling and heating control means 37, 55, 63 First temperature sensation input quantity designation Means 38, 56, 64 Second thermal sensation input amount designating means 39, 57, 66 Neuron generating means 40, 58, 67 Neuron generation order storing means 41, 59, 68 Neuron replacing means 44, 60, 69 Correction amount learning neural network 43, 61, 70 adding means 65 teacher correction amount designating means

Front Page Continuation (72) Inventor Yoshitaka Kubota 3-22 Takata Hondori, Higashi-Osaka City, Osaka Prefecture Matsushita Refrigerator Co., Ltd. In the company

Claims (5)

(57) [Claims] 1. A cooling / heating means for cooling or heating an interior of a room, a PMV value calculating means for sequentially calculating a PMV value in a living area, and a temperature sensation for inputting a user's feeling such as hot and cold at once or in a plurality of degrees. An input unit, a correction amount designating unit that determines a correction amount based on the frequency input by the temperature sensation input unit and its predetermined width, and a correction amount that the PMV value obtained by the PMV value computing unit is defined by the correction amount designating unit. A PMV value correction means for correcting only the amount, and a predetermined number of the first number input in the "cold" direction when the PMV value is on the hot side outside the comfort range or when the PMV value is on the cold side. Predetermined width changing means for changing the width so as to be larger than the predetermined width of the first once input when the PMV value is within the comfortable area, and the PMV value is stabilized within the comfortable area. An air conditioner comprising: a cooling / heating control means for controlling the cooling / heating means. 2. The first thermal sensation input amount designating means for designating a correction amount corresponding to the input amount of the thermal sensation input device based on the basic predetermined width and the input frequency of the thermal sensation input device. When,
When the PMV value is outside the comfort range and is on the hot side, the temperature is input in the "cold" direction, or when the PMV value is on the cold side, the temperature is input in the "hot" direction. By the second temperature sensation input amount designating means and the first temperature sensation input amount designating means for designating a correction amount according to the input amount of the temperature sensation input means when the value is changed to be larger than the basic predetermined width. It is composed of a neuron generating means for generating a neuron as a teacher signal by associating a predetermined correction amount with factors such as room temperature and outside temperature at the time of inputting a temperature sensation, and a plurality of neurons generated by the neuron generating means. A correction amount learning neural network for outputting the result of arithmetic processing of the output value of each neuron as a correction amount, the first and second thermal sensation input designating means, and the correction amount learning neural network The air conditioner of claim 1, wherein composed of an adding means for adding the correction amount of Ttowaku. 3. A first thermal sensation input amount designating means for designating a correction amount corresponding to the input amount of the thermal sensation inputting means based on a basic predetermined width and an input frequency of the thermal sensation inputting means. When,
When the PMV value is outside the comfort range and is on the hot side, the temperature is input in the "cold" direction, or when the PMV value is on the cold side, the temperature is input in the "hot" direction. Second thermal sensation input amount designating means for designating a correction amount according to the input amount of the thermal sensation input means in the case where is changed to be larger than the basic predetermined width, and the first and second thermal sensation input amounts. A neuron generating means for generating a neuron as a teacher signal by associating the correction amount determined by the designating means with factors such as room temperature and outside temperature at the time of inputting the temperature sensation, and a plurality of neurons generated by the neuron generating means A correction amount learning neural network for outputting the result of arithmetic processing of the output value of each neuron as a correction amount, the first and second temperature sensation input designating means, and the correction amount learning unit. The air conditioner of claim 1, wherein composed of an adding means for adding the correction amount of over neural network. 4. The first thermal sensation input amount designating unit, wherein the correction amount designating unit designates a correction amount corresponding to the input amount of the thermal sensation input unit based on the basic predetermined width and the input frequency of the thermal sensation input unit. When,
When the PMV value is outside the comfort range and is on the hot side, the temperature is input in the "cold" direction, or when the PMV value is on the cold side, the temperature is input in the "hot" direction. Is set to be larger than the basic predetermined width, the second thermal sensation input amount designating means for designating a correction amount according to the input amount of the thermal sensation input means; A correction amount smaller than the correction amount is output to a later-described neuron generating unit to generate a neuron, a teacher correction amount designating unit, a correction amount and a temperature determined by the first thermal sensation input amount designating unit and the teacher correction amount designating unit. A neuron generating unit that generates a neuron using a teacher signal in association with factors such as room temperature and outside temperature at the time of inputting a feeling, and a plurality of neurons generated by the neuron generating unit A correction amount learning neural network that outputs the result of arithmetic processing of the output value of each neuron as a correction amount, and an addition unit that adds the correction amounts of the first and second temperature sensation input designating means and the correction amount learning neural network The air conditioner according to claim 1, comprising: 5. A neuron generation order storage means for recording the generation order of a plurality of neurons generated by the neuron generation means, and an oldest neuron stored by the neuron generation order storage means when the number of neurons exceeds a predetermined value. The air conditioner according to claim 2, 3 or 4, further comprising: neuron replacement means for replacing the latest neuron generated by the neuron generation means.