Air conditioner energy saving method and device based on neural network and genetic algorithm

Abstract

The invention discloses an air conditioner energy saving method and device based on a neural network and a genetic algorithm, and the air conditioner energy saving method based on the neural network and the genetic algorithm comprises the steps that S1, data information corresponding to the temperature in a building is obtained; s2, the obtained data information is input into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs predicted indoor temperature and total active power; wherein the recurrent neural network model is obtained by training the historical data information of the building; s3, the predicted indoor temperature and total active power are input into a genetic algorithm module for optimization calculation, and an optimization result is obtained; and S4, whether the obtained optimization result reaches an optimization target or not is judged, and if yes, the optimized indoor set temperature and the optimized air supply temperature are output.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power energy saving, and in particular to an air conditioning energy saving method and device based on a neural network and a genetic algorithm. Background Art

[0002] The intelligent power distribution system for buildings is a community power distribution terminal system developed for intelligent residential buildings that are increasingly widely used in cities. It can cooperate with the distribution network automation master station and power grid automation substation system to realize data collection, load detection, fault detection, automatic switching of power supply, etc. for multiple residences, multiple electrical equipment, and multiple power lines.

[0003] At present, how to ensure that the indoor temperature of numerous electrical equipment in buildings reaches the human comfort standard while achieving energy saving under complex and changeable weather conditions is a technical issue that needs to be solved by the Internet of Things technology of building power equipment.

[0004] In order to solve the above technical problems, the present invention proposes an air conditioning energy-saving method and device based on neural network and genetic algorithm. Summary of the invention

[0005] The purpose of the present invention is to address the defects of the prior art and provide an air conditioning energy-saving method and device based on a neural network and a genetic algorithm. The air supply temperature and the set temperature of the air conditioning system are adjusted in real time through an optimization algorithm. Under the premise of ensuring human comfort, the energy consumption of the air conditioning unit is minimized and electricity costs are saved.

[0006] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0007] An air conditioning energy saving method based on neural network and genetic algorithm, comprising:

[0008] S1. Obtain data information corresponding to the temperature in the building;

[0009] S2. Input the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is trained on the historical data information of the building;

[0010] S3. The predicted indoor temperature and total active power are input into the genetic algorithm module for optimization calculation to obtain the optimization result;

[0011] S4. Determine whether the obtained optimization result reaches the optimization target. If so, output the optimized indoor set temperature and supply air temperature.

[0012] Furthermore, the data information corresponding to the temperature in step S1 includes indoor set temperature, supply air temperature, return air temperature, supply air volume, and outdoor temperature.

[0013] Furthermore, the recurrent neural network model in step S2 adopts a multi-layer structure and is trained by a back propagation algorithm.

[0014] Furthermore, the step S3 is specifically as follows:

[0015] S31. Construct objective functions of human comfort, energy consumption and energy efficiency of indoor temperature;

[0016] S32. randomly generate a set of solutions for indoor set temperature and supply air temperature, and evaluate the fitness of each solution according to the objective function;

[0017] S33. Generate a new generation of population through selection, crossover and mutation operations;

[0018] S34. Repeat steps S32-S33 until the optimal solution is found or the predetermined number of iterations is reached.

[0019] Furthermore, the objective function in step S31 includes minimizing the difference between the indoor temperature and the preset target temperature and minimizing the weighted sum of the total active power.

[0020] Furthermore, after step S4, the following steps are further included:

[0021] Displays the output indoor set temperature and supply air temperature.

[0022] Accordingly, an air conditioning energy-saving device based on a neural network and a genetic algorithm is also provided, which is used to execute the air conditioning energy-saving method based on a neural network and a genetic algorithm. The air conditioning energy-saving device includes a processor, and the processor is provided with:

[0023] An acquisition module is used to acquire data information corresponding to the temperature in the building;

[0024] A processing module, used for inputting the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is obtained by training the historical data information of the building;

[0025] The optimization module is used to input the predicted indoor temperature and total active power into the genetic algorithm module for optimization calculation to obtain the optimization result;

[0026] The output module is used to determine whether the optimization result reaches the optimization target. If so, the optimized indoor set temperature and supply air temperature are output.

[0027] Furthermore, the processor is connected to a memory for storing data or instructions.

[0028] Furthermore, the processor is connected to a transmission device for transmitting data with an external device.

[0029] Furthermore, the processor is connected to an input / output device for inputting or outputting information.

[0030] Compared with the prior art, the present invention, under the premise of ensuring the temperature stability of the computer room or the high comfort level of people in the office, adjusts the air supply temperature and set temperature of the air-conditioning system in real time through an optimization algorithm, so that the air-conditioning unit has the lowest power consumption and the highest energy efficiency, thus saving electricity costs; and the present invention can more effectively achieve stable control of indoor temperature and reduction of energy consumption, and has a higher energy efficiency ratio. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flow chart of an air conditioning energy saving method based on a neural network and a genetic algorithm provided in Example 1;

[0032] Figure 2 It is a schematic diagram of the local optimal solution provided by the second embodiment. DETAILED DESCRIPTION

[0033] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0034] The purpose of the present invention is to provide an air conditioning energy-saving method and device based on a neural network and a genetic algorithm in view of the defects of the prior art.

[0035] Embodiment 1

[0036] This embodiment provides an air conditioning energy saving method based on neural network and genetic algorithm. Figure 1 As shown, including:

[0037] S1. Obtain data information corresponding to the temperature in the building;

[0038] S2. Input the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is trained on the historical data information of the building;

[0039] S3. The predicted indoor temperature and total active power are input into the genetic algorithm module for optimization calculation to obtain the optimization result;

[0040] S4. Determine whether the obtained optimization result reaches the optimization target. If so, output the optimized indoor set temperature and supply air temperature.

[0041] In step S1, data information corresponding to the temperature in the building is obtained.

[0042] This embodiment designs an energy-saving control method for an air-conditioning system in a building to achieve intelligent regulation of indoor temperature and minimization of energy consumption.

[0043] The acquired data information will be used as the input of the neural network model, and the data information includes outdoor temperature (℃), return air temperature (℃), supply air volume (L / s), indoor set temperature (℃), and supply air temperature (℃).

[0044] In step S2, the acquired data information is input into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power.

[0045] The recurrent neural network RGU model of this embodiment adopts a multi-layer structure, and the constructed multi-layer RGU model includes the following layers:

[0046] Input layer: The corresponding number of neurons is set according to the parameters of the historical data information obtained in the data collection phase; that is, the input layer includes multiple neurons to process the corresponding data information.

[0047] Hidden layer: contains at least one hidden layer, each hidden layer contains a certain number of neurons for processing nonlinear relationships; that is, the hidden layer includes multiple neurons to process the corresponding data information.

[0048] Output layer: predict the indoor temperature (℃) and total active power (Kw) at the next moment.

[0049] This embodiment uses the historical energy consumption data information of the building to train the multi-layer RGU model through the back propagation algorithm, and selects appropriate activation functions and optimization algorithms, such as ReLU and Adam, to minimize the mean square error cost function.

[0050] For example, if the average hourly air supply volume, outdoor temperature, and return air temperature are known and calculated by the trained model, a series of indoor set temperatures and supply air temperatures can be assumed, as shown in Table 1.

[0051] Table 1 Data information

[0052]

[0053]

[0054] In this embodiment, the training method of the multi-layer RGU model by the back propagation algorithm is specifically as follows:

[0055] 1. Obtain historical data information and divide the acquired data information into training samples and test samples;

[0056] 2. Input the training samples into the input layer for processing and output the processed data information;

[0057] 3. Perform weighted processing on the data information output by the input layer, and transmit the weighted data information to the hidden layer for processing, and output the processed data information;

[0058] 4. The data information output by the hidden layer is weighted and sent to the output layer for processing. The output layer diffusion network obtains the prediction result of the sample;

[0059] 5. The recurrent neural network RGU is trained in the above manner to obtain the RUGURGU model;

[0060] 6. Test the RUGURGU model with the test sample to determine whether the test result meets the requirements. If not, re-execute steps 1-4. If yes, output the final RUGURGU model, that is, the neural network prediction model.

[0061] In this embodiment, a suitable activation function and optimization algorithm are selected to minimize the mean square error cost function as follows:

[0062] GRU has two gates, a reset gate r and an update gate. Intuitively, the reset gate determines how to combine the new input with the previous memory, and the update gate determines how much previous memory works.

[0063] 1. GRU forgets old information: The model decides how much of the previous memory (hidden state) to forget, which is controlled by a mechanism called "update gating". The closer the value is to 0, the more it is forgotten; the closer the value is to 1, the more it is retained.

[0064] 2. GRU remembers new information: At the same time, the model also decides how much new input information to remember. This is also controlled by the update gating. The larger the value, the more new information is remembered.

[0065] 3. GRU combines old and new information: The model combines the previous forgetting and memory, retaining some old information while adding new information to obtain updated memory. This ensures that the model does not completely lose past information, but can also flexibly adapt to new inputs.

[0066] This embodiment uses historical data to train the GRU timing model, constructs a mapping relationship between input and output, and obtains a GRU timing model prediction model. Finally, the test data (i.e., the acquired data information) is input into the trained GRU timing model prediction model, and the GRU timing model prediction model outputs the predicted indoor temperature and total active power.

[0067] In this embodiment, a suitable activation function and optimization algorithm are selected to minimize the mean square error cost function as follows:

[0068] 1. Cost function:

[0069] In RGU, the cost function plays an important role. Classification data points are hardly affected by small adjustments in weights and biases. Therefore, a smooth cost function can be used to determine the way to adjust weights and biases to improve performance.

[0070] The following is the mean squared error cost function:

[0071]

[0072] Among them, w represents the weight collected in the network; b represents the bias; n represents the number of training inputs; a represents the output vector; x represents the input; ||v|| represents the normal length of the vector v.

[0073] 2. Loss function:

[0074] The loss function of a neural network is used to determine if adjustments need to be made during the learning process. The number of neurons in the output layer is equal to the number of classes. The difference between the predicted and actual probability distributions is displayed. Below is the cross entropy loss for binary classification:

[0075]

[0076] The significance of RGU is that it can train the neural network by fitting the training data with a custom curve, and optimize the parameter kernel and bias with the minimization of loss as a guide.

[0077] This embodiment uses historical data to train the RGU model, constructs a mapping relationship between input and output, and obtains a neural network prediction model. Finally, the test data (i.e., the acquired data information) is input into the trained neural network prediction model, and the neural network prediction model outputs the predicted indoor temperature and total active power.

[0078] In step S3, the predicted indoor temperature and total active power are input into the genetic algorithm module for optimization calculation to obtain an optimization result.

[0079] Genetic Algorithm (GA): originated from computer simulation research on biological systems. It is a random global search and optimization method developed by imitating the biological evolution mechanism in nature, drawing on Darwin's theory of evolution and Mendel's genetics. Its essence is an efficient, parallel, global search method that can automatically acquire and accumulate knowledge about the search space during the search process, and adaptively control the search process to obtain the best solution.

[0080] In this embodiment, the output of the recurrent neural network RGU model is used as the input of the genetic algorithm module, and the output of the neural network model is optimized through the following steps, specifically:

[0081] S31. Construct objective functions of human comfort, energy consumption and energy efficiency of indoor temperature;

[0082] About human comfort:

[0083] In a certain area, the environmental level and human comfort are usually related to indoor temperature, humidity, air flow rate, number of people indoors, human activity, sunshine, human metabolism level, etc. Considering the decisive influence of indoor temperature on human comfort, the indoor temperature of 21-25 degrees is taken as the optimal range of human comfort. The closer the indoor temperature of the office is to 23 degrees, the higher the human comfort. For the computer room, in order to ensure the normal and efficient operation of the server, the indoor temperature in the computer room needs to be maintained at 18-22 degrees. It is assumed here that the closer the temperature is to 20 degrees, the higher the operating efficiency of the computer room. Therefore, the first goal needs to be optimized: in the environmental level in the area, the value range of the objective function x and y values, such as Figure 2 The values ​​of the x-axis and y-axis, the size of the F(x,y) value: for offices, the indoor temperature is as close to 23 degrees as possible, and cannot be lower than 21 degrees or higher than 25 degrees; for computer rooms, the indoor temperature is as close to 20 degrees as possible, and cannot be lower than 18 degrees or higher than 22 degrees.

[0084] About energy consumption:

[0085] The total active power (P) of the air-conditioning unit represents the power loss per unit time, so the second goal needs to be optimized: minimizing the total active power of the unit to achieve the goal of energy saving and consumption reduction.

[0086] About energy efficiency:

[0087] The greater the temperature change that can be achieved under the same power, the higher the energy efficiency. Therefore, the third goal needs to be optimized: maximize ∑ΔT / total active power, where ∑ΔT represents the total heat transfer drive.

[0088] S32. Randomly generate a set of indoor set temperature and supply air temperature solutions, and evaluate the fitness of each solution according to the objective function. The objective function needs to include minimizing the difference between the indoor temperature and the preset target temperature and minimizing the weighted sum of the total active power; specifically:

[0089] Initialize the population: randomly generate a set of solutions for indoor set temperature and supply air temperature.

[0090] In a given area, a population with a moderate number of individuals existing in different positions in the environment and capable of reproduction (a group that continuously iterates to seek the optimal solution) is the population that remains after continuous reproduction and survival of the fittest (a group of solutions after the iteration, each of which should be very close to the optimal solution). In this embodiment, the initial group is randomly distributed on F(x, y). In addition, the generated group of solutions should not be too large (complex in calculation) nor too small.

[0091] Fitness evaluation: The fitness of each solution is evaluated according to the objective function, which includes the difference between the indoor temperature and the target temperature and the total active power.

[0092] Fitness refers to the measure of the degree of advantage of an individual in the survival of a population, and is used to distinguish between "good" and "bad" individuals. Fitness is calculated using a fitness function. The fitness function is also called an evaluation function, which mainly judges the fitness of an individual through individual characteristics. The fitness function is transformed from the objective function.

[0093] In this embodiment, the fitness of each individual solution can be evaluated as follows:

[0094] 1. After decoding the individual coding string, the individual phenotype can be obtained.

[0095] 2. The objective function value of the corresponding individual can be calculated based on the individual's phenotype.

[0096] 3. According to the type of optimization problem, the fitness of the individual is calculated from the objective function value according to certain transformation rules.

[0097] The design of the fitness function should be single-valued, continuous, non-negative, maximized, reasonable, consistent, with the smallest possible computational complexity and the strongest possible generality.

[0098] The ratio of the fitness function value of each solution to the sum of the fitness function values ​​of a group of solutions represents the probability that the solution is retained in the iteration. This embodiment introduces a fitness function to evaluate a solution; wherein the fitness function value is non-negative.

[0099] The fitness function is expressed as Fit(f(x)), where f(x) is a solution. Note that the fitness function value is non-negative.

[0100] Cmin : The minimum value in this set of solutions; G max : The maximum value in this set of solutions.

[0101] Maximum value problem:

[0102] Minimum problem:

[0103] The above two piecewise functions both represent probabilities and are therefore non-negative, and the larger the fitness function value of the solution, the easier it is to retain.

[0104] S33. Generate a new generation of population through selection, crossover and mutation operations;

[0105] Crossover, variation - reproduction: Chromosome crossover, gene mutation:

[0106] When producing offspring, the chromosomes of the parents cross over, and gene mutations may occur during reproduction, resulting in changes in the genotype of the offspring (creating part of the next set of solutions based on the existing solutions). Note: not all solutions need to be crossed over or mutated. In genetic algorithms, the binary code of the solution is not halved and crossed over, but a random section of indefinite length is exchanged. Crossover is to ensure that the general direction is towards the optimal solution, and mutation is to avoid generating local optimal solutions.

[0107] Coding - individual, genotype:

[0108] Individuals in the population (a solution, i.e. Figure 2 A black dot in the image) and its genotype (usually stored as a binary code in genetic algorithms).

[0109] Decoding - Phenotype:

[0110] According to the length of binary code of the phenotype caused by the genotype (convert the binary array of the solution into the corresponding x, y values), a line segment is divided into countless small line segments, and the distance of each small line segment (the reason for using precision is that we cannot get every point on the line segment, so we can only change it to 2 on the line segment l points. The 2 l -1 small line segment length for precision).

[0111] Selection - Survival of the fittest:

[0112] When generating a new generation of population, parents and children are bound to be eliminated (the next group of solutions of the same size is selected with bias based on fitness).

[0113] S34. Repeat steps S32-S33 until the optimal solution is found or the predetermined number of iterations is reached.

[0114] The fitness evaluation, selection, crossover, and mutation steps are repeated until the optimal solution is found or a predetermined number of iterations is reached.

[0115] In step S4, it is determined whether the obtained optimization result reaches the optimization target. If so, the optimized indoor set temperature and supply air temperature are output.

[0116] The RGU model outputs predicted indoor temperature and total active power. In this embodiment, a set of indoor set temperatures and supply air temperatures that minimize the objective function need to be selected for recommendation.

[0117] This embodiment adjusts the air supply temperature and indoor set temperature of the air conditioning system in real time according to the optimal solution obtained by genetic algorithm optimization to achieve the lowest energy consumption. The performance of the energy-saving control system can also be evaluated by comparing the energy consumption data before and after implementation. It ensures that the system achieves significant reduction in energy consumption while meeting the indoor comfort requirements.

[0118] Embodiment 2

[0119] The air conditioning energy saving method based on neural network and genetic algorithm provided in this embodiment is different from that in the first embodiment in that:

[0120] This embodiment is implemented through the software of the energy-saving control system:

[0121] 1. Software architecture design

[0122] Design a software system to realize the training of RGU model, optimization of genetic algorithm and implementation of control strategy.

[0123] 2. Database establishment

[0124] Establish a database to store historical data of parameters such as indoor and outdoor temperature, supply air temperature, return air temperature, and supply air volume.

[0125] 3. User Interface Design

[0126] Design a user-friendly interface that allows users to input parameters, view prediction results and output indoor set temperature, supply air temperature, and manually adjust control parameters.

[0127] 4. Model training and optimization

[0128] Implements a module for training neural network models and performing genetic algorithm optimization.

[0129] 5. Automation of control strategies

[0130] Develop an automation module to automatically adjust the control parameters of the air conditioning system based on the optimization results.

[0131] 6. System testing and deployment

[0132] Deploy the software system in an actual building, conduct testing and debugging, and ensure that the system runs stably and achieves the expected energy-saving effect.

[0133] Embodiment 3

[0134] This embodiment further provides an air conditioning energy-saving device based on a neural network and a genetic algorithm, which is used to execute an air conditioning energy-saving method based on a neural network and a genetic algorithm in Embodiment 1. The air conditioning energy-saving device includes a processor, wherein the processor is provided with:

[0135] An acquisition module is used to acquire data information corresponding to the temperature in the building;

[0136] A processing module, used for inputting the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is obtained by training the historical data information of the building;

[0137] The optimization module is used to input the predicted indoor temperature and total active power into the genetic algorithm module for optimization calculation to obtain the optimization result;

[0138] The output module is used to determine whether the optimization result reaches the optimization target. If so, the optimized indoor set temperature and supply air temperature are output.

[0139] The processor described in this embodiment may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0140] The processor described in the present embodiment is connected with a memory for storing data or instructions. The memory may include a large capacity memory for data or instructions. For example, but not limitation, the memory may include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory may be inside or outside the data processing device. In a specific embodiment, the memory is a non-volatile memory. In a specific embodiment, the memory includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable Programmable Read-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. Under appropriate circumstances, the RAM can be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM can be a fast page mode dynamic random access memory (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0141] The memory can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor. The processor reads and executes the computer program instructions stored in the memory.

[0142] The processor described in this embodiment is connected to a transmission device for transmitting data with an external device. The transmission device can be used to receive or send data via a network. The above-mentioned specific examples of the network may include a wired or wireless network provided by a communication provider of the electronic device. In one example, the transmission device includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0143] The processor of this embodiment is connected with an input / output device for inputting or outputting information. The input / output device is used for inputting or outputting information. In this embodiment, the input information is data information corresponding to the temperature, and the output information is the indoor set temperature and the air supply temperature.

[0144] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. An air conditioning energy saving method based on neural network and genetic algorithm, characterized in that: include: S1. Obtain data information corresponding to the temperature in the building; S2. Input the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is trained on the historical data information of the building; S3. The predicted indoor temperature and total active power are input into the genetic algorithm module for optimization calculation to obtain the optimization result; S4. Determine whether the obtained optimization result reaches the optimization target. If so, output the optimized indoor set temperature and supply air temperature.

2. The air conditioning energy saving method based on neural network and genetic algorithm according to claim 1, characterized in that: The data information corresponding to the temperature in step S1 includes the indoor set temperature, the supply air temperature, the return air temperature, the supply air volume, and the outdoor temperature.

3. The air conditioning energy saving method based on neural network and genetic algorithm according to claim 2, characterized in that: The recurrent neural network model in step S2 adopts a multi-layer structure and is trained by a back propagation algorithm.

4. The air conditioning energy saving method based on neural network and genetic algorithm according to claim 3, characterized in that: The step S3 is specifically as follows: S31. Construct objective functions of human comfort, energy consumption and energy efficiency of indoor temperature; S32. randomly generate a set of solutions for indoor set temperature and supply air temperature, and evaluate the fitness of each solution according to the objective function; S33. Generate a new generation of population through selection, crossover and mutation operations; S34. Repeat steps S32-S33 until the optimal solution is found or the predetermined number of iterations is reached.

5. The air conditioning energy saving method based on neural network and genetic algorithm according to claim 4, characterized in that: The objective function in step S31 includes minimizing the difference between the indoor temperature and the preset target temperature and minimizing the weighted sum of the total active power.

6. The air conditioning energy saving method based on neural network and genetic algorithm according to claim 5, characterized in that: After step S4, the following steps are also included: Displays the output indoor set temperature and supply air temperature.

7. An air conditioning energy-saving device based on a neural network and a genetic algorithm, used to execute an air conditioning energy-saving method based on a neural network and a genetic algorithm as claimed in any one of claims 1 to 6, the air conditioning energy-saving device comprising a processor, characterized in that: The processor is provided with: An acquisition module is used to acquire data information corresponding to the temperature in the building; A processing module, used for inputting the acquired data information into a recurrent neural network model for prediction processing, and the recurrent neural network model outputs the predicted indoor temperature and total active power; wherein the recurrent neural network model is obtained by training the historical data information of the building; The optimization module is used to input the predicted indoor temperature and total active power into the genetic algorithm module for optimization calculation to obtain the optimization result; The output module is used to determine whether the optimization result reaches the optimization target. If so, the optimized indoor set temperature and supply air temperature are output.

8. The air conditioning energy saving device based on neural network and genetic algorithm according to claim 7, characterized in that: The processor is connected to a memory for storing data or instructions.

9. The air conditioning energy saving device based on neural network and genetic algorithm according to claim 7, characterized in that: The processor is connected with a transmission device for transmitting data with an external device.

10. The air conditioning energy saving device based on neural network and genetic algorithm according to claim 7, characterized in that: The processor is connected to an input / output device for inputting or outputting information.

Images