Air conditioner temperature control method based on multiple temperature probes

Abstract

The invention provides an air conditioner temperature control method based on multiple temperature probes. The method comprises the following steps: (1) deploying the multiple temperature probes; (2) the number of air conditioners needing to be installed is calculated; and (3) automatically adjusting the temperature of the air conditioner based on the comfort of the human body. The problems that the heating effect is poor and slow are solved, and the problems of air conditioner energy consumption and cost are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of refrigeration, and particularly relates to an air conditioner temperature control method based on multiple temperature probes. Background Art

[0002] Air conditioner control systems are common systems in modern life.

[0003] Air conditioners are divided into two categories: industrial air conditioners and domestic air conditioners. Industrial air conditioners focus on meeting control precision indicators, and the types of parameters to be adjusted depend on the requirements of the production process. For example, in integrated circuit production, the key is to adjust the temperature and control the cleanliness to ensure constant temperature and purification. In the textile industry, the key is to ensure a constant relative humidity.

[0004] Artificial climate chambers require the temperature and humidity to change according to a predetermined program. Domestic air conditioners focus on meeting the requirements of human comfort. Temperature is the main parameter affecting comfort. The value of the comfortable temperature varies with different regions and seasons. Air conditioner equipment is characterized by high power, long operating time, and wide usage range. The energy consumption of air conditioners accounts for a quite large proportion in the total energy consumption of developed countries. Energy conservation is a major indicator in the design of air conditioner control systems.

[0005] Air conditioner control belongs to process control (see process control systems). Most air conditioner control systems are feedback control systems. With the increasing requirements of humans for the air environment, a technology that comprehensively studies and processes air conditioners, heating, and ventilation - artificial climate environment engineering is developing rapidly. Since the 1970s, due to the popularization of microcomputers, electronic computers have begun to be used as the core components of air conditioner control. Direct digital control technology has been widely applied.

[0006] The integration of air conditioner equipment and control systems has become an important direction for the update of air conditioner control technology. Hierarchical distributed air conditioner control systems composed of multiple computers have begun to be used in large - scale multi - functional buildings or building complexes. In the 1980s, with the increasing prominence of the energy - saving problem, the development of air conditioner control optimization software with the goal of minimizing the consumption of cooling capacity, heating capacity, and electricity consumption has received extensive attention on the premise of meeting usage requirements.

[0007] With the improvement of building levels, the internal space of buildings is getting larger and larger. At this time, it is not easy for a single air conditioner to well solve the problem of the comfort of indoor personnel, and when there are multiple air conditioners, the relationship between their quantity and economy needs to be considered. Currently, the control of air conditioners mainly uses the PID control algorithm without modeling the object to be controlled. Although this method is simple and general, it is essentially a black - box control, that is, it does not understand the temperature relationship of the control object and controls it. Its main disadvantages are low precision, no predictive effect on future temperature changes, and a lag characteristic. Summary of the Invention

[0008] In view of the above technical problems, the present invention provides an air conditioner temperature control method based on multiple temperature probes, aiming to overcome the problems that occur in the above prior art.

[0009] To solve the above technical problems, the technical solution of the present invention is as follows:

[0010] An air conditioner temperature control method based on multiple temperature probes, the method comprising the following steps:

[0011] (1) Deploy multiple temperature probes;

[0012] (2) Calculate the number of air conditioners to be installed;

[0013] (3) Automatically adjust the temperature of the air conditioner based on the comfort of the human body.

[0014] A further optimization scheme is that

[0015] Deploy the temperature probes at the four corners of the indoor floor to obtain multiple temperature data indoors;

[0016] When calculating the number of indoor air conditioners, assume the room as a standard cuboid, and all air conditioners adopt the top and side air outlet method. Set the distance between the air conditioner and one side wall as distance one d 1 , and set the length of the air conditioner itself as distance two d 2 , set the distance between air conditioners as distance three d 3 , and set the distance between the air conditioner and the other side wall as distance four d 4 , and calculate the number of air conditioners according to the reasonable arrangement of the distance lengths on the wall where the air conditioners are installed based on the above distance data.

[0017] A further optimization scheme is that

[0018] The aforementioned reasonable arrangement is specifically:

[0019] Calculate the maximum value d of distance three according to the installation height of the air conditioner and the size of the indoor space 3,max , according to d 3,max and the heat conduction relationship between the left and right wall surfaces, calculate the maximum values d 1,max , d 4,max , according to d 1,max , d 3,max and d 4,max Select an air conditioner model that meets the length;

[0020] When the length of the wall on the side where the air conditioner is installed indoors minus d 1,max +d 4,max is less than d 2 +d 3,maxWhen it indicates, only one air conditioner needs to be installed; otherwise, two air conditioners need to be installed. And so on to judge whether three air conditioners need to be installed.

[0021] A further optimization solution is that

[0022] temperature control is carried out by using the calculation method;

[0023] An air division model is established, and the indoor air is divided into many small squares in space. Each small square carries its own temperature and humidity, and heat transfer and humidity transfer can occur between the squares. The transfer speed is different according to the air supply speed and angle of the air conditioner;

[0024] A temperature and humidity transfer model is established. The air at the air outlet of the air conditioner is ejected in a straight line. Taking the air outlet as an endpoint, a ray is made in the direction of the air outlet. For the squares close to the ray direction, the temperature and humidity will be transferred faster than those far from the ray;

[0025] A controller is used to adjust the air conditioner. The user can input the length, width, and height of the room into the controller. According to the N - S equation of the fluid dynamics model and Fourier's law of heat conduction, combined with the size of the indoor space and local weather data, an indoor heat model is established by integration;

[0026] Through this indoor heat model, the air flow in the room is predicted, and the jet path that makes the heat transfer fastest and the jet path that keeps the maximum human comfort are calculated. The jet path is corrected through the temperature probes at the four corners on the ground, and relevant parameters are optimized.

[0027] A further optimization solution is that

[0028] When the wind speed near the human body is obtained, combined with Fourier's law of heat conduction, given the human body boundary conditions, human body temperature, and the temperature of the surrounding air volume, the heat Φ conducted by the air volume to the human body is calculated 1 and the heat radiation amount Φ 2 ;

[0029] The heat Φ conducted by the air volume to the human body is calculated 1 and the difference ΔΦ between the heat radiation amount Φ 2 meets the set requirements for the surrounding air volume temperature T;

[0030] According to the N - S equation, combined with the given boundary conditions, including the room size and air conditioner position given by the user, the wind speed v that meets the national standard when the speed at the air conditioner outlet reaches the human body after a certain distance is calculated;

[0031] The wind speed v and temperature T around the human body will jointly affect the heat transfer coefficient on the human body surface, and the heat Φ radiated outward by the human body is calculated 3 ;

[0032] By calculating the heat φ brought to the human body by the air around the human body 1 , and the heat Φ radiated by the human body outward 3 , and finally feedback it to the air conditioner, so that the air conditioner controls the temperature and wind speed at the air outlet, achieving a comfortable feeling for the human body, that is, Φ 1 and Φ 3 The difference ΔΦ′ meets the set requirements.

[0033] A further optimization scheme is as follows

[0034] Use the reinforcement learning method for temperature control; the process of the reinforcement learning method is as follows

[0035] First, establish a CFD simulation space for the indoor scene where the temperature needs to be controlled, that is, use a computer to build a simulation scene, determine the current position of the air conditioner and the size of the room; then, set the initial conditions and operating conditions to generate a variety of virtual data; conduct training on the reinforcement learning model strategy therein, set relevant parameters, including the action space, state space and reward, start the training process, and the method used is to update the Q-table. After waiting for it to converge, it can be put into actual use; finally, deploy it to the actual controller.

[0036] A further optimization scheme is as follows

[0037] The action space: The air conditioner can output a fixed temperature, a fixed angle, and a wind speed

[0038] The state space: The temperature values measured at the four corners of the floor described above plus the PMV value of the human body form the state space s t =(T1, T2, T3, T4, PMV);

[0039] Among them, (T1, T2, T3, T4) respectively represent the temperatures of the four corners, and PMV represents the predicted comfort of the human body

[0040] The reward: The reward function is designed according to the gap between the current temperature and the target temperature and the human body comfort

[0041] Target temperature gap: Set the reward by calculating the difference between the temperatures of the four corners in the current state and the set target temperature. If the temperatures of the four corners are close to the target value, a higher reward can be given

[0042] PMV (Predicted Mean Vote) index: PMV is used to measure the human body's perception of environmental comfort and can calculate the comfort score of the current environment for people; if PMV is within ±0.5, it means that the human body feels comfortable, and it can also be used as an index of the reward; if PMV is far from the target value, a negative reward can be given

[0043] The reward function is designed as follows:

[0044] R(s t ,a t ) = -|T avg -T target | - λ·|PMV|

[0045] where:

[0046] T avg is the average temperature of the four corners,

[0047] T target is the target temperature,

[0048] PMV is the comfort score of the current environment,

[0049] λ is the weight that adjusts the impact of comfort on the reward,

[0050] A negative reward represents the temperature difference and the uncomfortable state, while a positive reward represents that both the temperature and comfort meet the expectations.

[0051] A further optimization scheme is as follows:

[0052] The training process includes:

[0053] State update: At each time step, the agent observes the current state s t , and selects an action a t according to the policy;

[0054] Reward and next state: After executing the action, the environment will feedback a reward R(s t ,a t ) and update to the next state s t+1 ;

[0055] Q-value update: Use the Q-learning update rule to update the Q-value:

[0056]

[0057] where:

[0058] α is the learning rate;

[0059] γ is the discount factor, representing the importance of future rewards;

[0060] is the maximum Q-value in the next state, used to estimate the maximum possible future return;

[0061] Termination event: Set that the difference between the set temperature and the collected temperature is 2 degrees, and the human body PMV is within ±0.5. At this time, it is considered that the reinforcement model has been trained well;

[0062] PMV = [0.303exp(-0.036M) + 0.0275]TL, where TL is the human body heat load, representing the difference between the heat production and heat dissipation of the human body; M is the human body metabolic rate.

[0063] A further optimization solution is as follows.

[0064] Deploy the trained Q-table to the controller for control, and maintain an exploration rate of 0.2 and a conservatism rate of 0.8. Then, during the actual operation of the air conditioner, the Q-table is still slightly updated according to the actual environment.

[0065] 1. Solve the problem of poor heating effect. The present invention changes the position of the temperature probe, avoiding the problem that the air conditioner misjudges the heating effect by receiving an ambient temperature that is too different from the temperature around the human body.

[0066] 2. Solve the problem of slow heating effect. The present invention establishes an air model based on the positions of four temperature probes and the air outlet of the air conditioner. Through this model, the air outlet angle and strength that can reach the heating target fastest during the heating of the air conditioner can be solved.

[0067] 3. Solve the trade-off problem of multiple factors such as the spacing and height of the air conditioners, energy consumption and cost, and human body comfort. Through the algorithm, finally, inputting the cost and room layout and outputting data solutions such as the number, spacing, height, and model of the air conditioners. This solution can provide certain reference for designers. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is an example of the placement positions of the sensors and the air conditioner in the present invention;

[0069] Figure 2 It is a schematic diagram of the placement distance of the air conditioner in the room in the present invention;

[0070] Figure 3 It is the air division model of the present invention;

[0071] Figure 4 It is the temperature and humidity transfer rate of the present invention (the arrow indicates the air jet direction of the air outlet, and the oval circle represents the isotherm and isohumidity lines of the temperature and humidity transfer);

[0072] Figure 5 It is a schematic diagram of the control method of the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0073] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0074] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "center", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the present invention.

[0075] In the present invention, unless otherwise clearly specified and defined, the terms "arrange", "install", "connect", "join", "fix", etc. shall be understood in a broad sense. For example, it may be a fixed connection or a detachable connection; it may be a mechanical connection; it may be a direct connection or an indirect connection through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0076] In addition, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first", "second" may explicitly or implicitly include one or more of such features.

[0077] The present invention discloses a multi-probe air-conditioning temperature control system. Aiming at the problem of uneven air temperature during air-conditioning heating, a set of air-conditioning temperature sensor deployment solutions are proposed for air-conditioning manufacturers to choose from.

[0078] First, equip the air-conditioning with multiple temperature probes and deploy them reasonably. Considering that the hot air at the outlet floats upward during air-conditioning heating and the people on the ground cannot obtain a comfortable temperature, the present invention proposes to replace the single temperature probe above the air-conditioning with multiple temperature probes and deploy them at the four corners of the ground. As Figure 1 shown. The purpose of doing this is to enable the air-conditioning controller to obtain more temperature data in the room.

[0079] Then, in the design, assume that the room is a standard cuboid and all air-conditionings adopt the air outlet mode of the top side. As Figure 2As shown in the figure, the distances between the air conditioner and the wall ①, the length of the air conditioner itself ②, the distance between air conditioners ③, and the distance between the air conditioner and another wall ④ can be used as control variables in the design. The current idea is: based on the height and the size of the room space, calculate the maximum distance ③, which will minimize the number of air conditioners and achieve the purpose of energy conservation. Assuming that the distances ① and ④ are related to the wall surfaces, the maximum distances of ① and ④ can be deduced from the heat conduction relationship between ③ and the left and right wall surfaces. Then, select the air conditioner model that meets the length according to the distances of ①, ③, and ④. When the length of the room minus the distance of (① + ④) is less than the distance of (② + ③), it means that only one air conditioner needs to be installed. Otherwise, two air conditioners need to be installed. The choice between two and three air conditioners is also the same.

[0080] Scheme 1: Calculation method.

[0081] Assume that the initial temperature in the room is uniform and there is no air flow. In the heat and humidity transfer model, it is assumed that the indoor air is divided into many small squares in space, and each small square carries its own temperature and humidity. Heat transfer and humidity transfer can occur between the squares. The transfer rate varies according to the air supply speed and angle of the air conditioner, as Figure 3 shown.

[0082] If the air from the air conditioner outlet is ejected in a straight line. Then, with the outlet as the endpoint and the outlet direction as a ray, this model is established. For the squares close to the ray direction, the temperature and humidity will be transferred faster than those far from the ray, as Figure 4 shown. The heat and humidity transfer rate of the squares (inside the yellow circle (elliptical circle)) around the red ray (the ray with an arrow) will be faster than that of the squares outside the yellow circle.

[0083] Under this assumption, the user can input the length, width, and height of the room into the controller. The controller generates a heat model for the room. This model is established by combining the N - S equation in the fluid dynamics model and Fourier's law of heat conduction, considering the size of the indoor space and local weather data. This model uses mathematical methods to theoretically predict various situations of air flow in the actual space, and calculates the jet path that makes heat transfer the fastest and the jet path that maintains the maximum human comfort in theory. During the actual operation, the jet path is corrected through the temperature sensors at the four corners of the ground, and relevant parameters are optimized.

[0084] Detailed steps:

[0085] After obtaining the wind speed near the human body, combined with Fourier's law of heat conduction. Given the human body boundary conditions, such as a human body temperature of 37°C and the surrounding air volume temperature T (for the final solution), calculate the heat conducted by the air volume to the human body.

[0086] Fourier's law of heat conduction:

[0087]

[0088] where the system emissivity is

[0089]

[0090] Finally, calculate the heat Φ of the heat conduction law and the heat radiation amount Φ 1，2 The ambient air volume temperature T that is close to or slightly smaller.

[0091] Then list the N - S equations. First, simplify the gas in three - dimensional space to two - dimensional gas, list the N - S equations, and combine specific boundary conditions (the size of the room given by the user, the position of the air conditioner, etc.) to calculate the speed at the air conditioner outlet. After a certain distance, the wind speed reaches the human body, and this wind speed should be less than the national standard of 0.3 m / s.

[0092]

[0093] u is the fluid velocity, p is the fluid pressure, ρ is the fluid density, and μ is the fluid dynamic viscosity. In the formula, term 1 corresponds to the inertial force, term 2 corresponds to the pressure, term 3 corresponds to the viscous force, and term 4 corresponds to the external force acting on the fluid.

[0094] The wind speed and temperature around the human body will jointly affect the heat transfer coefficient on the human body surface. According to the formula, the relationship between the wind speed and the heat carried away can be obtained.

[0095]

[0096] In short, through the heat φ brought by the air around the human body calculated and the heat φ radiated by the human body outward 1，2 , finally feedback to the air conditioner, so that the air conditioner controls the temperature and wind speed at the air outlet to achieve a comfortable feeling for the human body (φ and φ 1，2 are approximately equal or slightly smaller).

[0097] Scheme 2: Reinforcement learning method (neural network method).

[0098] First, establish a CFD simulation space for the indoor scene where the temperature needs to be controlled, that is, use a computer to build a simulation scene and determine the current position of the air conditioner and the size of the room.

[0099] Then, set the initial conditions and operating conditions to generate a variety of virtual data; perform training on the reinforcement learning model strategy in it. The method used is to update the Q - table. After waiting for it to converge, it can be put into actual use. When using, maintain an exploratory rate of 0.2 (randomly select strategies) and a conservative rate of 0.8 (use the strategies in the Q - table).

[0100] Training process:

[0101] 1. Set relevant parameters, including the action space, state space, and reward.

[0102] Action space: The air conditioner can output a fixed temperature (integers, such as 27, 28, 29), a fixed angle (swing, up, middle, down), and wind speed (high, medium, low). Therefore, the action space is a discrete action space.

[0103] Assuming only the above controllable outputs, there will be 27 actions, that is, 3 (temperatures) × 3 (wind speeds) × 3 (swing angles) = 27 discrete actions.

[0104] State space: The state space is composed of the temperature values measured at the four corners of the floor described above plus the PMV value of the human body (s t = (T1, T2, T3, T4, PMV)),

[0105] where (T1, T2, T3, T4) represent the temperatures at the four corners respectively, and PMV represents the predicted comfort level of the human body.

[0106] Reward: The reward function is designed based on the gap between the current temperature and the target temperature and the human comfort level.

[0107] Target temperature gap: The reward is set by calculating the difference between the temperatures at the four corners in the current state and the set target temperature. If the temperatures at the four corners are close to the target value, a higher reward can be given.

[0108] PMV (Predicted Mean Vote) index: PMV is used to measure the human perception of environmental comfort and can calculate the comfort score of the current environment for people. If PMV is within ±0.5, it means that the human body feels comfortable, and it can also be used as an indicator for rewards. If PMV is far from the target value, a negative reward can be given.

[0109] The reward function is designed to be:

[0110] R(s t ,a t ) = -|T avg -T target | - λ·|PMV|

[0111] where,

[0112] T avg is the average temperature of the four corners,

[0113] T target is the target temperature (assumed to be the central value of a certain temperature range),

[0114] PMV is the comfort score of the current environment.

[0115] λ is the weight that adjusts the impact of comfort on rewards.

[0116] Negative rewards represent temperature differences and discomfort states, while positive rewards represent both the temperature and comfort meeting expectations.

[0117] II. Start the training process

[0118] State update: At each time step, the agent observes the current state (s t ) and selects an action (a t ) according to the policy.

[0119] Reward and next state: After executing the action, the environment will feedback a reward (R(s t , a t )) and update to the next state (s t+1 ).

[0120] Q-value update: Use the Q-learning update rule to update the Q-value:

[0121]

[0122] Where:

[0123] α is the learning rate;

[0124] γ is the discount factor, representing the importance of future rewards;

[0125] is the maximum Q-value in the next state, used to estimate the maximum possible future return.

[0126] Termination event: Set the temperature difference between the set temperature and the collected temperature to be 2 degrees, and the human PMV to be within ±0.5. At this time, it is considered that the reinforcement model has been trained well.

[0127] PMV = [0.303exp(-0.036M) + 0.0275]TL, where TL is the human thermal load, representing the difference between the human heat production and heat dissipation; M is the human metabolic rate.

[0128] III. Deploy to the actual controller, refer to the appendix Figure 5 .

[0129] Deploy the trained Q-table to the controller for control and maintain an exploration rate of 0.2. After that, during the actual operation of the air conditioner, still update the Q-table slightly according to the actual environment.

[0130] The present invention first sets the temperature and blows air normally; then it starts to calculate the PMV of the human body based on the measured values of the temperature sensors at the four corners; then it uses reinforcement learning to update the strategy of the controller. The calculated PMV is fed back to the controller. When the PMV is outside ±1, a penalty is given to the controller, and when the PMV is within ±0.5, a reward is given to the controller. The controller changes the value of PMV by adjusting three variables: the air outlet speed, angle, and temperature. Finally, the controller will learn the control strategy that can make the human body comfortable. Then this strategy is deployed to the controller for actual control.

[0131] Comfort optimization: The collected data set corresponds to the relationship between people, PMV, and control strategies, enabling the use of a better strategy than the original strategy at the beginning.

[0132] The innovations of the present invention are as follows: 1. The temperature sensors are arranged at the four corners of the ground; 2. A model of the air in the space is established; 3. The method of using reinforcement learning + CFD (computer fluid simulation) to train the agent in advance and then deploy it to the physical object.

[0133] The beneficial effects of this embodiment are as follows: This method can achieve the two goals of rapid heating of the air conditioner and maintaining the ground comfort. By adjusting the position of the sensors and combining with the mathematical model algorithm, the air outlet angle and strength of the air conditioner are controlled to make the ground environment temperature reach the set temperature fastest. At the same time, by comparing the temperatures at the four points on the ground, the air outlet angle and strength of the air conditioner are controlled to avoid the problem of uneven room temperature in the room. Compared with the current air conditioner control technology, the temperature control is more refined, taking into account factors such as the upward floating of hot air. The reinforcement learning algorithm is adopted, and more output variables are used than in other papers, which can improve the heating speed and efficiency of the air conditioner and the comfort of the people on the ground during air conditioner heating.

[0134] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An air conditioning temperature control method based on multiple temperature probes, characterized in that: The method comprises the following steps: (1) Deploy multiple temperature probes; (2) Calculate the number of air conditioners that need to be installed; (3) Automatically adjust the temperature of the air conditioner based on human comfort.

2. The air conditioning temperature control method based on multiple temperature probes according to claim 1, characterized in that: The temperature probes are deployed at the four corners of the indoor floor to obtain multiple temperature data indoors; When calculating the number of indoor air conditioners, the room is assumed to be a standard rectangular parallelepiped. All air conditioners use the top and side air outlet method. The distance between the air conditioner and one side wall is set as distance one d1, the length of the air conditioner itself is set as distance two d2, the distance between air conditioners is set as distance three d3, and the distance between the air conditioner and the other side wall is set as distance four d4. According to the above distance data, the reasonable arrangement of the distance length on the wall on the side where the air conditioner is installed is used to calculate the number of air conditioners.

3. The air conditioning temperature control method based on multiple temperature probes according to claim 2 is characterized in that: The above reasonable arrangement is specifically as follows: Calculate the maximum value of distance 3 d according to the height of the air conditioner installation and the size of the indoor space 3,max , according to d 3,max The maximum value d of distance 1 and distance 4 is calculated based on the thermal conductivity relationship between the left and right walls. 1,max ,d 4,max , according to d 1,max ,d 3,max and d 4,max Choose an air conditioner model that matches your length; When the length of the wall on the indoor side where the air conditioner is installed is reduced by d 1,max +d 4,max Less than d2+d 3,max , it means that only one air conditioner needs to be installed, otherwise two air conditioners need to be installed; and so on to determine whether three air conditioners need to be installed.

4. The air conditioning temperature control method based on multiple temperature probes according to claim 3 is characterized in that: Temperature control is carried out by calculation method; Establish an air partitioning model to divide the indoor air into many small cubes in space. Each small cube carries its own temperature and humidity. Heat and humidity can be transferred between cubes. The transfer speed varies according to the air supply speed and angle of the air conditioner. Establish a temperature and humidity transfer model. The air from the air conditioner outlet is ejected in a straight line. With the outlet as the endpoint, a ray is drawn in the direction of the outlet. The blocks close to the ray will have their temperature and humidity transferred faster than those far from the ray. The air conditioner is adjusted by a controller. The user can input the length, width and height of the room into the controller. The indoor heat model is established based on the NS equation of the fluid dynamics model and Fourier's heat conduction law, combined with the size of the indoor space and local weather data. The indoor thermal model is used to predict the indoor air flow, and the jet path that enables the fastest heat transfer and the jet path that maintains the greatest human comfort are calculated. The jet path is corrected through temperature probes at the four corners of the ground, and related parameters are optimized.

5. The air conditioning temperature control method based on multiple temperature probes according to claim 4 is characterized in that: After obtaining the wind speed near the human body, combined with Fourier's law of heat conduction, given the human body boundary conditions, human body temperature, and surrounding air volume temperature, the heat Φ1 and thermal radiation Φ2 conducted by the air volume to the human body are calculated; Calculate the ambient air temperature T when the difference ΔΦ between the heat Φ1 conducted to the human body by the air volume and the heat radiation Φ2 meets the set requirements; According to the NS equation, combined with the given boundary conditions, including the room size and air conditioner position given by the user, the wind speed v that meets the national standard is calculated from the air conditioner outlet after a certain distance to the human body; The wind speed v and temperature T around the human body will jointly affect the heat transfer coefficient of the human body surface, and the heat radiated outward by the human body Φ3 is calculated; The heat Φ1 brought to the human body by the air around the human body and the heat Φ3 radiated outward by the human body are calculated and finally fed back to the air conditioner, so that the air conditioner controls the temperature and wind speed at the air outlet to achieve a comfortable feeling for the human body, that is, the difference ΔΦ' between Φ1 and Φ3 meets the set requirements.

6. The air conditioning temperature control method based on multiple temperature probes according to claim 3 is characterized in that: The temperature is controlled by using a reinforcement learning method; the process of the reinforcement learning method is: First, a CFD simulation space is established for the indoor scene that needs to control the temperature. That is, a simulation scene is built using a computer to determine the current position of the air conditioner and the size of the room. Then, the initial conditions and operating conditions are set to generate a variety of virtual data. The reinforcement learning model strategy is trained in it, and relevant parameters, including action space, state space and reward, are set. The training process begins by updating the Q table and waiting for it to converge before it can be put into use in reality. Finally, it is deployed to the actual controller.

7. The air conditioning temperature control method based on multiple temperature probes according to claim 6 is characterized in that: The action space: the air conditioner can output a fixed temperature, a fixed angle, and a fixed wind speed; The state space is composed of the temperature values ​​measured at the four corners of the ground and the PMV value of the human body. t =(T1,T2,T3,T4,PMV); Where (T1, T2, T3, T4) represent the temperatures of the four corners, and PMV represents the predicted comfort level of the human body; The reward function is designed according to the difference between the current temperature and the target temperature and the human comfort; Target temperature gap: The reward is set by calculating the difference between the four corner temperatures in the current state and the set target temperature. If the temperatures of the four corners are close to the target values, a higher reward can be given; PMV (Predicted Mean Vote) indicator: PMV is used to measure the human body's perception of environmental comfort and can calculate the current environmental comfort score for people. If PMV is within ±0.5, it means that the human body feels comfortable and can also be used as an indicator for rewards. If PMV is far from the target value, a negative reward can be given. The reward function is designed as: R(s t ,a t )=-|T avg -T target |-λ·|PMV| in, T avg is the average temperature of the four corners, T target is the target temperature, PMV is the comfort rating of the current environment. λ is the weight that adjusts the effect of comfort on rewards, Negative rewards represent temperature gaps and discomfort, while positive rewards represent that both temperature and comfort meet expectations.

8. The air conditioning temperature control method based on multiple temperature probes according to claim 7, characterized in that: The training process includes: State update: At each time step, the agent observes the current state s t , and select an action a according to the strategy t ; Reward and next state: After executing the action, the environment will feedback a reward R(s t ,a t ) and update to the next state s t+1 ; Q-value update: Use Q-learning update rule to update the Q-value: in: α is the learning rate; γ is a discount factor, which indicates the importance of future rewards; It is the maximum Q value in the next state, which is used to estimate the maximum possible reward in the future; Termination event: When the set temperature differs by 2 degrees from the collected temperature and the human PMV is within ±0.5, the reinforcement model is considered to have been trained; PMV=[0.303exp(-0.036M)+0.0275]TL, where TL is the human body heat load, representing the difference between the heat produced and the heat dissipated by the human body; M is the human body metabolic rate.

9. The air conditioning temperature control method based on multiple temperature probes according to claim 8, characterized in that: The trained Q table is deployed to the controller for control, and the exploratory value of 0.2 and the conservative value of 0.8 are maintained. Then, during the actual operation of the air conditioner, the Q table is still updated slightly according to the actual environment.

Images