JPH0694292A - Air-conditioning system accommodating feeling of warmth - Google Patents

Abstract

PURPOSE:To make air-conditioning equipment controllable in a manner of accommodating the operation to changes in the occupant's feeling of warmth so as to ensure the amenity of the indoor environment. CONSTITUTION:A VAV unit 5 capable of increasing or decreasing the amount of conditioned air in proportion to the load of indoor air-conditioning is connected to an air conditioner 3. By exercising a learning function a PMV equation reflecting individuals' feeling of warmth with the declared value of the warmth in an occupant's feeling as a tutor's signal is constructed. To the air conditioner 3 and the VAV unit 5 a neural network device 9 is connected, which controls the air conditioner 3 and the VAV unit 5 on the basis of a PMV value calculated by the PMV equation.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a thermal sensation air conditioning system that reflects the thermal sensation of a resident.

[0002]

2. Description of the Related Art In a conventional air conditioning system, a temperature / humidity sensor is attached to a position representative of indoor conditions (hereinafter referred to as "indoor representative point"), and air conditioners are controlled by detection information from the temperature / humidity sensor. Is common. The operation of the air conditioners is controlled based on the detection value of the temperature / humidity sensor attached to the indoor representative point so that the temperature / humidity of the indoor representative point falls within the set range. This is because in the conventional control, it is considered that the indoor environment is almost equal to the temperature and humidity at the representative point.

On the other hand, in recent years, an air conditioning system has been proposed which controls air conditioners based on a PMV (Predic-ted Mean Vote) value which is a comprehensive temperature and heat index. The human thermal sensation is affected not only by factors such as temperature and humidity, air flow velocity, and radiation level, but also by clothing and work volume. The PMV value is calculated by the PMV equation, which is a thermal equilibrium equation between the human body and the environment, in which the temperature environment factors that influence the comfort of the occupants as described above are variables. Therefore, by controlling the air conditioners based on the PMV value, the air conditioners can be controlled with a value close to the actual human sense, and the indoor environment can be made closer to the comfortable condition.

[0004]

In recent years, improvement of intellectual productivity has been taken up as an important issue in offices and the like, and it has been demanded to secure high amenities in the indoor environment.
In response to this, the importance of an air conditioning system capable of satisfying a person's thermal sensation tends to increase more and more. However, the comfort of thermal sensation is
It is a complex subject that is greatly influenced by its physical condition and mood. On the other hand, with the simple control that keeps the indoor representative point temperature constant in conventional air-conditioning systems, it is impossible to achieve a comfortable indoor environment because the thermal sensation of the occupants is significantly different. It was In addition, PM that is an index of warmth
There are air conditioning systems that control air conditioners based on the V value, but since the PMV value is an averaged index that was originally obtained from experiments, it is subject to individual differences (preferences) and changes due to physical condition. It was impossible to satisfy the residents sufficiently.

The present invention has been made in view of the above circumstances, and provides a thermal sensation air-conditioning system capable of controlling air-conditioning equipment by following changes in the thermal sensation of the occupant. To secure high amenities.

[0006]

To achieve the above object, a thermal sensation-compatible air conditioning system according to the present invention has an air conditioner and a conditioned air amount, which is a heat medium connected to the air outlet side of the air conditioner. VAV that increases or decreases in proportion to the air conditioning load in the room
The unit is connected to the air conditioner and VAV unit, and is connected to the man / machine interface that detects the occupant's thermal declaration value. The thermal declaration value and temperature / humidity, air flow velocity, and radiation amount at the time of declaration are used as teacher signals. It is characterized in that a PMV equation reflecting a thermal sensation is constructed by a learning function, and is composed of a neural network section for controlling an air conditioner and a VAV unit with a PMV value calculated from the PMV equation.

[0007]

[Function] Within the range of standard air conditioning conditions, temperature and humidity, air flow velocity,
Environmental factors such as radiation amount, clothing amount, and work amount are set, and a standard PMV equation is learned by a neural network using these variables as teacher signals. If there is no thermal declaration from the resident, the air conditioners are controlled by this standard PMV value. If the occupant is not satisfied with the air conditioning condition based on the standard PMV value, the degree of discomfort is transmitted to the neural network unit via the man / machine interface, and the reported value of heat and temperature / humidity at the time of declaration, air flow velocity, A neural network reconstructs a PMV equation reflecting the thermal sensation of an individual by using the detected value of the radiation amount as a teacher signal, and the air-conditioning equipment is controlled by the PMV value calculated from this which reflects the thermal sensation of the occupant.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of a thermal sensation air conditioning system according to the present invention will be described in detail below with reference to the drawings. Figure 1
FIG. 2 is an explanatory diagram showing the basic configuration of a thermal sensation-compatible air conditioning system of the present invention. The thermal sensation air conditioning system 1 includes an air conditioner 3, a variable air volume (VAV) unit 5,
The control output calculation unit 7, the neural network unit 9, the sensor 11, and the man / machine interface 13 are main components. The hot and cold water coil 3 is provided in the air conditioner 3.
a, a humidifier 3b, a blower 3c, etc. are provided, and the flow rate of cold / hot water supplied to the cold / hot water coil 3a is controlled by a cold / hot water valve 15. A VAV unit 5 is connected to the air conditioner 3 via a supply duct, and the VAV unit 5 can increase or decrease the conditioned air amount which is a heat medium in proportion to the air conditioning load in the room. A control output calculation unit 7 is connected to the cold / hot water valve 15 of the air conditioner 3 and the VAV unit 5, and the control output calculation unit 7 is based on a thermal sensation index calculated by a neural network unit 9 described later.
It is designed to efficiently control these devices.

Next, the principle of the neural network unit 9 will be described. The neural network is
It consists of neurons that process signals and synaptic connections that connect each neuron. A neuron is an element that weights signals from output connections and adds them together, and outputs a signal to the output connections when a certain threshold is exceeded. Given the inputs and outputs, the network can adjust the connection weights accordingly and store the pattern. This is the learning function of neural networks, which is attracting attention today. Neural networks can be set by appropriately setting the transfer function and network structure of neurons according to the problem, and by repeatedly learning (re-learning) the example,
The weights of the connections can be adjusted to handle the example correctly. That is, it has a self-organizing function that determines an internal processing method according to an input and an output. According to a mathematical model, a neuron gives weights to the inputs from other connected neurons, takes them in, applies a transfer function to the sum of these inputs, and sends them to the output. The output may be an input to another neuron.
Although there are various network structures, the present invention uses a hierarchical network shown in FIG. 2 as a basic structure.

FIG. 2 is a structural diagram of the neural network used in the present invention. The basic structure is a hierarchical network consisting of three layers of an input layer, an intermediate layer, and an output layer. The weighted sum is used as the summation method of signals from each neuron, and the sigmoid function is used as the signal output function. The learning algorithm is backpropagation (error backpropagation learning: a learning method that gradually changes the coupling between neurons so that the error between the actual reaction of the final output layer and the correct answer given as the teacher signal is minimized). . In the present invention, the seven thermal elements (temperature / humidity / air velocity / radiation amount / clothing amount / metabolic amount / work amount) constituting the PMV are associated with each neuron in the input layer, and the PMV is converted into one neuron in the output layer. Correspond. On the other hand, the PMV described in the related art is the temperature / humidity / air velocity / radiation amount /
It is a comprehensive thermal index that considers the amount of clothes, the amount of metabolism, and the amount of work, and is close to the actual human sense. However, PM
The V index is made assuming the average person. FIG. 3 is a control block diagram of the neural network. In the present invention, the above neural network is made to learn this PMV equation, the thermal sensation index is dynamically acquired from the thermal sensation report of the individual, and control is performed based on this acquired index.

The neural network unit 9 includes a sensor 1
1 is connected, and the sensor 11 detects indoor air temperature / humidity, average radiation temperature, and indoor air velocity. Further, a man / machine interface 13 is connected to the neural network unit 9, and the man / machine interface 13 transmits a resident's thermal declaration. The man / machine interface 13 is a data communication device in which a sensor is incorporated in a telephone,
It may be of a touch panel type. In the case of a sensor-embedded data communication device, it is of a desktop type, equipped with a solar cell as an auxiliary power source, and provided with a suction fan for guiding indoor air to an internal sensor. Also,
A connection terminal to a personal computer (neural network unit 9), a telephone line connection terminal, etc. are provided on the back surface. A monitor screen display unit is provided on the front surface of the sensor-embedded data communication device, and the monitor screen display unit can display various data of the indoor environment. On the touch panel, the degree of discomfort is, for example, "very cold; -3", "cold;-2",
"Slightly cold; -1", "Neither;0","Slightlyhot;+1","Hot;+2","Veryhot;+3"
It is conceivable that a signal can be directly detected as a discrete value from -3 to +3.

The operation of the thermal sensation air-conditioning system configured as described above will be described. FIG. 3 is a control block diagram of the thermal sensation air conditioning system of the present invention. First, according to the conventionally proposed PMV equation, considering the combination of temperature and humidity, air flow velocity, and radiation amount within the range of standard air conditioning conditions,
Set the standard amount of clothing, metabolism and work amount for each season
Calculate the MV value. Using these results as a teacher signal, a standard PMV equation is learned by a neural network. If there is no thermal declaration from the resident, the PMV value calculated by this standard PMV equation is fed back from the detected values of temperature and humidity, air flow velocity, and radiation amount, and based on the deviation between this and the set value of PMV, Control output calculator 7
Change the set value of the room temperature / humidity thermostat, control the air conditioner cold / hot water coil valve 15 and the humidifier 3b based on the set value, change the blown air temperature / humidity, and use the damper or the inverter or the VAV unit 5. Control the amount of blown air.

When the occupant is hot or cold and is not satisfied with the current air-conditioning state based on the standard PMV value, the degree of discomfort is transmitted to the neural network unit via the man / machine interface 13. Next, let the neural network re-learn as the teacher signal the detected temperature value and the detected temperature / humidity / air flow velocity / radiation amount at the time of declaration,
Reconstruct the PMV equation that reflects the personal thermal sensation. According to the above control algorithm, the blown air temperature and the blown air amount are controlled so that the PMV value calculated from this is within the target range.

[0014]

As described in detail above, in the thermal sensation air conditioning system according to the present invention, a thermal sensation model is constructed using a neural network, and the thermal sensation of the occupant is dynamically reflected by additional learning. As a result, the air conditioners can be controlled by following changes in the sense of the occupants, and an ideal indoor environment that suits the taste can be realized.
As a result, it is possible to secure high amenities in the indoor environment.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing the basic configuration of a thermal sensation air conditioning system of the present invention.

FIG. 2 is a structural diagram of a neural network used in the present invention.

FIG. 3 is a control block diagram of a neural network.

[Explanation of symbols]

 1 Air conditioning system for thermal sensation 3 Air conditioner 5 VAV unit 9 Neural network 13 Man / machine interface

Front page continuation (72) Inventor Hideo Tanaka 1-2-7 Moto-Akasaka, Minato-ku, Tokyo Kashima Construction Co., Ltd. (72) Inventor Masahiko Tokunaga 1-2-7 Moto-Akasaka, Minato-ku, Tokyo Kashima Ken Inside the corporation

Claims (1)

[Claims] 1. An air conditioner, a VAV unit connected to a blow-out side of the air conditioner for increasing / decreasing a conditioned air amount as a heat medium in proportion to an indoor air conditioning load, and connected to the air conditioner and the VAV unit. In addition, a man / machine interface for detecting the occupant's thermal declaration value and a PMV equation reflecting the thermal sensation of an individual, which is connected to the man / machine interface and uses the thermal declaration value as a teacher signal, is constructed by a learning function. A thermal sensation-compatible air conditioning system comprising: the air conditioner and a neural network unit that controls the VAV unit with a PMV value calculated from a PMV equation.