Public building indoor air conditioner temperature set value optimization and control method

Abstract

In order to solve the problems of high energy consumption and unbalanced comfort of a traditional static control strategy of the air conditioning system during environment change, a building three-dimensional model is constructed, and sensor data and energy consumption simulation software are fused to construct a joint simulation environment. The improved extreme learning machine neural network is used for predicting the air conditioner energy consumption and the personnel thermal comfort degree, the network structure and parameters are optimized through the rotary particle swarm optimization, and the prediction precision and generalization ability are improved. And then, in combination with a Newton-Raphson multi-objective optimization algorithm, solving an air conditioner temperature set value optimal dynamic sequence in a rolling manner, and balancing energy consumption and comfort through a Pareto frontier solution set. Through dynamic simulation and data driving optimization, precise regulation and control of the air conditioner are achieved, and the thermal comfort degree is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of modeling and optimal control of building thermal environment, and particularly relates to a method for optimizing and controlling the set value of indoor air-conditioning temperature in public buildings. Background Technique

[0002] According to the statistical data of the Ministry of Housing and Urban-Rural Development, the total energy consumption of the current construction industry accounts for more than 35% of the total social energy consumption. Among them, the energy consumption intensity per unit area of public buildings is 2.09 times that of residential buildings. In particular, it is worth noting that in public buildings such as large commercial complexes, transportation hubs, and super high-rise office buildings, the proportion of air-conditioning system consumption has remained in the range of 40%-60% all year round, posing a severe challenge to the realization of the "dual carbon" goal.

[0003] At present, the air-conditioning temperature control mostly adopts a static set value strategy, which is difficult to adapt to the dynamic changes of the indoor and outdoor environment (such as fluctuations in personnel density and time-varying characteristics of temperature and humidity) and individual differences in the thermal comfort needs of personnel, resulting in the air-conditioning system being in a contradictory state of "over-cooling / heating" or "insufficient cooling and heating" for a long time. This not only causes a large amount of energy waste, but also leads to potential health hazards and a decline in work efficiency due to the imbalance of thermal comfort, further exacerbating the conflict between building energy consumption and user experience. The limitations of traditional control methods make it difficult for building energy management to balance energy-saving goals and personnel comfort needs. Therefore, the present invention proposes a method for setting the indoor air-conditioning temperature that integrates a data-driven prediction model and model predictive control technology to balance the thermal comfort of indoor personnel and the energy consumption of the air-conditioning system. Summary of the Invention

[0004] The present invention starts from the prediction of indoor thermal comfort in public buildings and the optimization of air-conditioning temperature set values:

[0005] In order to solve the above problems, the present invention provides a data-driven model for predicting indoor comfort and air-conditioning energy consumption in public buildings and a control method for air-conditioning temperature set values based on the prediction model. In the constructed building energy consumption simulation environment, the present invention imports weather files and sets the capacity and working schedule of personnel, equipment (lighting, electrical appliances, air conditioners, etc.). By setting the working mode of the air-conditioning system, data such as indoor environment and equipment energy consumption are obtained. Then, the data is normalized, and an extreme learning machine neural network model is used to establish a prediction model to predict the energy consumption of the indoor air-conditioning system and the thermal comfort of personnel. The trained model is embedded in a co-simulation environment, and a multi-objective optimization algorithm is used to roll-optimize the air-conditioning temperature set value to balance indoor thermal comfort and energy consumption simultaneously.

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

[0007] A method for optimizing and controlling the set value of indoor air-conditioning temperature in public buildings includes the following steps:

[0008] Draw a 3D building model, and set the number of people, lamp power, electrical equipment power, air conditioning system of each heat zone, and parameters of personnel and air conditioning equipment for the drawn model;

[0009] Use sensors and electricity meters to collect indoor and outdoor environmental data, and embed the data collected by sensors into the building energy consumption simulation model through the external interface of building simulation software to simulate the real indoor thermal environment and energy consumption of equipment;

[0010] Through data collection, storage and visualization functions, realize the real-time display of simulation results, and construct a training sample set and a test sample set;

[0011] Construct an extreme learning machine neural network model, and use a swarm intelligence optimization algorithm to optimize the structure and parameters of the network to determine the structure and optimal parameters of the network, and then predict the energy consumption of the indoor air conditioning system and the thermal comfort of personnel;

[0012] Embed the established prediction model into the co-simulation module to realize the dynamic coupling calculation with the building energy consumption simulation model, and solve the optimal dynamic sequence of the air conditioning temperature setting value within the prediction interval based on the multi-objective optimization algorithm;

[0013] For the obtained optimal dynamic sequence, select the first element in it as the air conditioning temperature setting value for the next moment to avoid the deviation of the control effect caused by environmental changes in the future.

[0014] As an alternative implementation, use AutoCAD, BIM, SketchUp, etc. to draw a 3D building model, and use energy consumption simulation software such as TRNSYS and EnergyPlus to set the internal parameters of the building; the 3D building model and its enclosure structure parameters are constructed based on the actual building construction drawings, and the air conditioning equipment capacity (including COP, cooling capacity, heating capacity, etc.) is determined with reference to the actually deployed air conditioning system or HVAC construction drawings, and the indoor electrical equipment and lighting electrical power are set according to the actually deployed equipment power or electrical construction drawings.

[0015] As an alternative implementation, the number of indoor people and their hourly occupancy rate can be detected by passive infrared sensors, the operating status of electrical equipment and air conditioning systems can be detected by electricity meters, and the indoor and outdoor temperature and humidity can be detected by temperature and humidity sensors. The detected data is transmitted into the building energy consumption simulation model through the external interface to set the work schedules of personnel and various equipment.

[0016] As an alternative implementation, the data sets for training and testing the model need to be collected by setting different air conditioning temperatures at different times; after data collection, the data should be stored in an Excel table or a database.

[0017] As an alternative embodiment, the neural network model is an extreme learning machine model. This model incorporates a particle swarm optimization algorithm that fuses the characteristics of particle rotational motion to determine the model's structure and optimal parameters. This model mainly includes an input layer, a hidden layer, an output layer, and a network structure solution layer.

[0018] As a further limitation, the implementation process of the extreme learning machine prediction model is as follows:

[0019] Step 101: Select the outdoor temperature , indoor temperature , indoor relative humidity , indoor radiant temperature , indoor air-conditioning energy consumption , indoor human comfort , air-conditioning temperature setpoint as the input data of the prediction model, represented by the vector ; Select the indoor air-conditioning system energy consumption and human thermal comfort output by the building energy consumption simulation model as the output data of the prediction model, represented by the vector ; The described dataset is ;

[0020] Step 102: Split the dataset into a training dataset and a test dataset. To balance the feature dimensions, improve numerical stability, optimize the activation function efficiency, and significantly improve the prediction accuracy, convergence speed, and generalization ability of the model, preprocess the data using standardization or normalization methods.

[0021] As an alternative embodiment, use maximum-minimum normalization to preprocess the dataset:

[0022]

[0023] where is the original data; is the normalized input data; is the column vector composed of the minimum values of each feature of the original data; is the column vector composed of the maximum values of each feature of the original data; The ranges of the normalized input data and output data are within ; represents the Hadamard product;

[0024] Step 103: For the normalized dataset, train the extreme learning machine model using different activation functions and optimize it with a specific objective function.

[0025] As an alternative implementation, the specific objective function is the least squares method with an L2 regularization term, as shown in the following formula:

[0026]

[0027] In the formula, is the number of hidden layers of the extreme learning machine prediction model; is the weight matrix connecting the input layer and the hidden layer, where: ; is the bias matrix of the hidden layer, where: ; is the weight matrix connecting the hidden layer and the output layer, where: ; is the input data of the model; is the output data of the model; is the normalized output data; is the activation function of the neural network;

[0028] As an alternative implementation, the optimal activation function for the hidden layer can be sigmoid, tansig, tanh, relu, leaky_relu, elu, etc.

[0029] Step 104: Initialize the population, set the hyperparameters of the rotating particle swarm algorithm, and define the motion state equation of the particles to iteratively update each particle;

[0030] As an alternative implementation, the number of hidden layers, the weight matrix connecting the input layer and the hidden layer, and the bias matrix of the hidden layer are used as particles in the particle swarm algorithm, represented by the vector ; Define the population size , the maximum number of iterations , the charge carried by each particle , the magnetic field strength and the particle mass , and initialize each population ;

[0031] As an alternative implementation, the population initialization method is uniform random initialization:

[0032]

[0033] In the formula, is the component of the k-th particle in the q-th dimension of the search space; is a generated random number, uniformly distributed in the interval [0,1]; and are the upper and lower bounds of the search space respectively;

[0034] As an alternative implementation, the particle motion equation is mainly determined by the velocity of the particle's rotational motion and the velocity during the basic motion, and is calculated by the following formula:

[0035]

[0036] In the formula, is the current iteration number; is the velocity of the particle's rotational motion; is the velocity of the particle's basic motion; is the integration constant; is the charge carried by each particle; is the magnetic field strength; is the particle mass; is the inertia weight of the particle's rotational motion; is the inertia weight of the particle's basic motion; , are learning factors; , , is a random number between [0, 1]; is the individual optimal value at the current iteration number; is the global optimal value at the current iteration number;

[0037] Step 105: During the algorithm iteration process, adopt the population adaptive adjustment strategy and the inertia weight dynamic optimization strategy to suppress the excessive accumulation of local optimal solutions and ensure the full exploration of potential solutions at the same time.

[0038] As an alternative implementation, the population adaptive adjustment strategy is the particle chaotic collision strategy, and the specific mathematical model is:

[0039]

[0040] In the formula, is the new particle generated by the particle collision strategy; is the particle randomly selected from the current iteration number; is a random number between [0, 1] and follows a uniform distribution;

[0041] As an alternative implementation, the inertia weight dynamic optimization strategy is the particle chaotic collision strategy, and the specific mathematical model is:

[0042]

[0043] Step 106: When all particles participate in the update, if the maximum number of optimization processes is reached, terminate the optimization;

[0044] Step 107: After the prediction model training is completed, denormalize the data output by the model to test the model accuracy, as shown in the following formula:

[0045]

[0046] Step 108: Embed the tested model into the joint simulation module to achieve the prediction of air-conditioning energy consumption and thermal comfort during the simulation process;

[0047] As an alternative implementation, the multi-objective optimization algorithm for solving the optimal dynamic sequence of air-conditioning temperature set values during the prediction period is the Newton-Raphson multi-objective optimization algorithm. The model trained through the neural network is applied to the joint simulation module, and the building energy consumption simulation software is used to calculate the indoor average radiant temperature, thermal comfort, and air-conditioning system energy consumption in real time. Based on the calculation results of the energy consumption simulation model and the prediction model, the Newton-Raphson multi-objective optimization algorithm is used to determine the optimal dynamic sequence of air-conditioning temperature set values within the prediction interval.

[0048] As a further limitation, within the prediction step of steps, substitute the collected data into the trained neural network for prediction, and solve the temperature set value within the prediction step through the rolling optimization method , and the optimized objective function is shown in the following formula:

[0049]

[0050] As an alternative implementation, the specific process of the optimal dynamic sequence optimization method of the air-conditioning temperature set value within the prediction interval is as follows:

[0051] Step 201: According to the data collected at the current moment and the established prediction model, update the model output within the prediction step to make the objective function of the rolling optimization reach the Pareto front:

[0052]

[0053] In the formula, is the input data of the model within the prediction step; is the output data of the model within the prediction step; is other feasible solutions in the solution set; is the Pareto front solution;

[0054] Step 202: Determine the air-conditioning temperature set value through the defined objective function and constraint conditions, based on the multi-objective optimization algorithm of the Newton-Raphson search rule, define the population size , the maximum number of iterations , and use Latin hypercube initialization to randomly generate N p populations:

[0055]

[0056] Wherein, is the j-th dimensional component of the i-th element of the population in the search space; is a generated random number, uniformly distributed in the interval [0, 1]; and are the upper and lower bounds of the search space respectively; is a random permutation of the element in the j-th dimension;

[0057] Step 203: After the parameters and the population are initialized, substitute each element into the objective function, calculate the fitness value of each element, perform non-dominated sorting, and screen out the non-inferior solutions at the current iteration; then, use the ideal point method to calculate the distance between the fitness value of each element at the current iteration and the current optimal fitness value, as shown in the following formula:

[0058]

[0059] Wherein, is the Euclidean distance between the fitness value corresponding to each element and the optimal fitness value at the current iteration; is the fitness value corresponding to the i-th optimization objective of each element at the current iteration; is the optimal fitness value of the i-th optimization objective of each element at the current iteration; is the worst fitness value of the i-th optimization objective of each element at the current iteration;

[0060] Step 204: Select the element corresponding to the farthest distance at the current iteration as the worst position , and its j-th dimensional component is ; select the element corresponding to the nearest distance at the current iteration as the worst position , and its j-th dimensional component is ; use the Newton-Raphson search rule to calculate:

[0061]

[0062] Wherein, is a guiding parameter for guiding the population at the current iteration into the correct direction; , are random numbers between (0, 1); , are the j-th dimensional components of randomly selected elements in the population, and ; is the result of Newton-Raphson solution;

[0063] Step 205: After calculating the values of each element in each dimension using the Newton-Raphson search rule, calculate the updated element position using the following formula:

[0064]

[0065] In the formula, is the j-th dimensional component of the i-th element after update; is a random number between [0, 1]; , , are the update parameters of ; is the adaptive coefficient, which is used to ensure the diversity of the population;

[0066] Step 206: When the algorithm reaches the maximum number of iterations, terminate the algorithm and output the Pareto surface; then, use the ideal point method to calculate the distance of the solution set on the Pareto surface and select the optimal solution .

[0067] As an alternative implementation, in order to minimize the impact of future environmental changes on the air-conditioning control effect, for the obtained optimal solution, take its first element as the air-conditioning temperature setting value.

[0068] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0069] The present invention constructs a prediction model based on an extreme learning machine neural network, combines a rotating particle swarm optimization algorithm to optimize network parameters and a Newton-Raphson multi-objective optimization algorithm to iteratively solve the dynamic temperature setting value, so as to realize the collaborative optimization of the energy consumption and indoor thermal comfort of the air-conditioning system. The present invention uses a building energy consumption simulation model and sensor data to dynamically and coupledly establish a joint simulation model. At the same time, the present invention uses a neural network model for real-time prediction and uses a multi-objective optimization algorithm to dynamically adjust the air-conditioning temperature setting value, which not only solves the problem of "over-cooling / heating" caused by the dynamic change of the environment in the traditional static control strategy, significantly reduces energy consumption, but also dynamically balances the thermal comfort index and energy efficiency through the Pareto front solution set, effectively improving the adaptability to personalized comfort requirements. In addition, the combination of the rotating particle swarm algorithm and the multi-objective optimization mechanism enhances the generalization ability and calculation efficiency of the model, providing high-precision and real-time response technical support for the intelligentization and energy-saving regulation of the building air-conditioning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0071] Figure 1 Schematic diagram of building joint simulation for at least one embodiment of the present invention;

[0072] Figure 2 Neural network architecture diagram for at least one embodiment of the present invention;

[0073] Figure 3 Flowchart of prediction model modeling for at least one embodiment of the present invention;

[0074] Figure 4 Flowchart for solving the set value of air conditioner temperature for at least one embodiment of the present invention. Detailed implementation manners

[0075] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0076] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0077] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0078] An optimization and control method for the set value of indoor air conditioner temperature in public buildings includes the following steps:

[0079] S1: Draw a three-dimensional building model, and set the number of people, lamp power, electrical equipment power, air conditioning systems of each room, and parameters of personnel and air conditioning equipment for the drawn model;

[0080] S2: Collect indoor and outdoor environmental data using sensors and electricity meters, and embed the data collected by the sensors into the energy consumption simulation software through the external interface of the building simulation software to simulate the real indoor thermal environment of the building and the energy consumption of the equipment;

[0081] S3: Through data collection, storage and visualization functions, realize the real-time display of simulation results, and construct a training sample set and a test sample set;

[0082] S4: Construct a neural network prediction model, and use a swarm intelligence optimization algorithm to optimize the structure and parameters of the network to determine the structure and optimal parameters of the network, so as to realize the prediction of the energy consumption of the indoor air conditioning system and the thermal comfort of personnel;

[0083] S5: Embed the established prediction model into the co-simulation module to achieve its dynamic coupling calculation with the building energy consumption simulation model, and solve the optimal dynamic sequence of the air-conditioning temperature set value during the prediction period based on the multi-objective optimization algorithm;

[0084] S6: For the obtained optimal dynamic sequence, select the first element therein as the air-conditioning temperature set value for the next moment.

[0085] Specifically, the building three-dimensional model described in step S1 is drawn by AutoCAD, BIM or SketchUp, and the internal parameters of the building are set using TRNSYS or EnergyPlus energy consumption simulation software; the building three-dimensional model and its enclosure structure parameters are constructed based on the actual building construction drawings, and the air-conditioning equipment capacity (including COP, cooling capacity, heating capacity, etc.) is determined according to the actually deployed air-conditioning system or HVAC construction drawings, and the indoor electrical equipment and lighting electric power are set according to the actually deployed equipment power or electrical construction drawings.

[0086] Specifically, the environmental data described in step S2 are measured by various sensors and electricity meters. The main environmental data to be detected are that the indoor number of people and their hourly occupancy rate can be detected by passive infrared sensors, the operating status of electrical equipment and air-conditioning systems can be detected by electricity meters, and the indoor and outdoor temperature and humidity can be detected by temperature and humidity sensors. The detected data are transmitted into the building energy consumption simulation model through an external interface to set the work schedules of personnel and various equipment.

[0087] Specifically, for the data acquisition, storage and visualization functions described in step S3, the data sets for training and testing the model are acquired by setting different air-conditioning temperatures at different times. After the data are acquired, they are stored in an Excel table or a database.

[0088] Specifically, the neural network prediction model described in step S4 is an extreme learning machine model integrated with the rotating particle swarm optimization algorithm. This model introduces the rotating particle swarm optimization algorithm to determine the structure and optimal parameters of the model. This model mainly includes an input layer, a hidden layer, an output layer and a network structure solving layer.

[0089] Specifically, the implementation process of the neural network prediction model described in step S4 is as follows:

[0090] Step 101: Select the outdoor temperature , indoor temperature , indoor relative humidity , indoor radiant temperature and the air-conditioning temperature set value as the input data of the prediction model, and use the vector Indicate; select the indoor room air conditioning system energy consumption output by the building energy consumption simulation model and human thermal comfort As the output data of the prediction model, use a vector Indicate; the described data set is ;

[0091] Step 102: Split the data set into a training data set and a test data set. In order to balance the feature dimensions, improve the numerical stability, optimize the activation function efficiency, and significantly improve the prediction accuracy, convergence speed, and generalization ability of the model, preprocess the data by standardization or normalization;

[0092] As an alternative implementation, use max-min normalization to preprocess the data set:

[0093]

[0094] In the formula, is the original data; is the input data after normalization; is the column vector composed of the minimum values of each feature of the original data; is the column vector composed of the maximum values of each feature of the original data; the ranges of the input data and output data after normalization are within the range; represents the Hadamard product;

[0095] Step 103: For the normalized data set, train the extreme learning machine model with different activation functions and optimize it with a specific objective function;

[0096] As an alternative implementation, the specific objective function is the least squares method with an L2 regularization term, as shown in the following formula:

[0097]

[0098] In the formula, is the number of hidden layers of the extreme learning machine prediction model; is the weight matrix connecting the input layer and the hidden layer, where: ; is the bias matrix of the hidden layer, where: ; is the weight matrix connecting the hidden layer and the output layer, where: ; is the input data of the model; is the output data of the model; is the output data after normalization; is the activation function of the neural network;

[0099] As an alternative embodiment, the optimal activation function of the hidden layer can be selected from sigmoid, tansig, tanh, relu, leaky_relu, elu, etc.

[0100] Step 104: Initialize the population, set the hyperparameters of the rotating particle swarm algorithm, and define the motion state equation of the particles to iteratively update each particle;

[0101] As an alternative embodiment, the number of hidden layers, the weight matrix of the input layer and the hidden layer, and the bias matrix of the hidden layer are used as particles in the particle swarm algorithm, represented by the vector ; Define the particle population size , the maximum number of iterations , the charge carried by each particle , the magnetic field strength and the particle mass , and initialize each population ;

[0102] As an alternative embodiment, the population initialization method is uniform random initialization:

[0103]

[0104] where is the component of the k-th particle in the q-th dimension of the search space; is a generated random number, uniformly distributed in the interval [0,1]; and are the upper and lower bounds of the search space respectively;

[0105] As an alternative embodiment, the motion equation of the particles is mainly determined by the velocity of the particles' rotating motion and the velocity during basic motion, and is calculated by the following formula:

[0106]

[0107] where is the current iteration number; is the velocity of the particles' rotating motion; is the velocity of the particles' basic motion; is the integral constant; is the charge carried by each particle; is the magnetic field strength; is the particle mass; is the inertia weight of the particles' rotating motion; is the inertia weight of the particles' basic motion; , is the learning factor; , , is a random number between [0, 1]; is the individual optimal value at the current iteration; is the global optimal value at the current iteration;

[0108] Step 105: During the algorithm iteration process, adopt the population adaptive adjustment strategy and the inertia weight dynamic optimization strategy to suppress the excessive accumulation of local optimal solutions and ensure the full exploration of the potential solution set.

[0109] As an alternative implementation, the population adaptive adjustment strategy is the particle chaos collision strategy, and the specific mathematical model is:

[0110]

[0111] In the formula, is the new particle generated by the particle collision strategy; is the particle randomly selected from the current iteration; is a random number between [0, 1] and follows a uniform distribution;

[0112] As an alternative implementation, the inertia weight dynamic optimization strategy is the particle chaos collision strategy, and the specific mathematical model is:

[0113]

[0114] Step 106: When all particles participate in the update, if the maximum number of optimization processes is reached or the objective function meets the requirements, terminate the optimization;

[0115] Step 107: After the prediction model training is completed, denormalize the data output by the model to test the model accuracy, as shown in the following formula:

[0116]

[0117] Step 108: Embed the tested model into the joint simulation module to realize the prediction of air-conditioning energy consumption and thermal comfort during the simulation process;

[0118] Specifically, the multi-objective optimization algorithm for solving the optimal dynamic sequence of the air-conditioning temperature setting value during the prediction period in step S5 is the Newton-Raphson multi-objective optimization algorithm. The model trained by the neural network is applied to the joint simulation module, and the building energy consumption simulation software is used to calculate the indoor average radiant temperature, thermal comfort, and air-conditioning system energy consumption in real time. Based on the calculated results and the prediction model, the Newton-Raphson multi-objective optimization algorithm is used to determine the optimal dynamic sequence of the air-conditioning temperature setting value within the prediction interval.

[0119] Specifically, the specific process of the optimal dynamic sequence optimization method for the air conditioner temperature setting value within the prediction interval described in step S5 is as follows:

[0120] Step 201: According to the data collected at the current moment and the established prediction model, update the model output within the prediction step length to make the objective function of the rolling optimization reach the Pareto front:

[0121]

[0122] In the formula, is the input data of the model within the prediction step length; is the output data of the model within the prediction step length; is other feasible solutions in the solution set; is the Pareto front solution;

[0123] Step 202: Determine the air conditioner temperature setting value according to the defined objective function and constraint conditions, using a multi-objective optimization algorithm based on the Newton-Raphson search rule. Define the population size , the maximum number of iterations , and use Latin hypercube initialization to randomly generate populations:

[0124]

[0125] In the formula, is the component of the i-th element of the population in the j-th dimension of the search space; is the generated random number, uniformly distributed in the interval [0,1]; and are the upper and lower bounds of the search space respectively; is the random permutation of the element in the j-th dimension;

[0126] After initializing the parameters and the population, substitute each element into the objective function, calculate the fitness value of each element, perform non-dominated sorting, and screen out the non-inferior solutions at the current iteration; then, use the ideal point method to calculate the distance between the fitness value of each element at the current iteration and the current optimal fitness value, as shown in the following formula:

[0127]

[0128] In the formula, is the Euclidean distance between the fitness value corresponding to each element and the optimal fitness value at the current iteration; is the fitness value corresponding to the i-th optimization objective of each element at the current iteration; is the optimal fitness value of the i-th optimization objective of each element at the current iteration; is the worst fitness value of the i-th optimization objective for an element at the current iteration count;

[0129] Step 204: Select the farthest distance at the current iteration count The corresponding element is the worst position , and its j-th dimension component is ; Select the closest distance at the current iteration count The corresponding element is the worst position , and its j-th dimension component is ; Calculate using the Newton-Raphson search rule:

[0130]

[0131] In the formula, is a guiding parameter used to guide the population at the current iteration count in the correct direction; , are random numbers between (0, 1); , are the j-th dimension components of randomly selected elements in the population, and ; is the Newton-Raphson solution result;

[0132] Step 205: When the values of each element in each dimension are calculated using the Newton-Raphson search rule, calculate the updated element position using the following formula:

[0133]

[0134] In the formula, is the j-th dimension component of the i-th updated element; is a random number between (0, 1); , , are update parameters; is an adaptive coefficient used to ensure the diversity of the population;

[0135] Step 206: When the algorithm reaches the maximum iteration count, terminate the algorithm and output the Pareto surface; then, use the ideal point method to calculate the distances of the solution set on the Pareto surface and select the optimal solution .

Claims

1. A method for optimizing and controlling the temperature setting value of indoor air conditioning in a public building, characterized in that: include: Use building energy consumption simulation software to build a building energy consumption simulation model, which outputs key parameters such as indoor occupancy, indoor occupant thermal comfort, indoor and outdoor environment, and air conditioning system energy consumption; Through the joint simulation function of the building energy consumption simulation software, the prediction model and optimization algorithm are embedded in the building energy consumption simulation model to achieve data acquisition, data preprocessing and air conditioning temperature setting value optimization; Through the data visualization function of the joint simulation, the collected data is visualized (data collection charts, optimization result charts, etc.), and the training sample set and test sample set for training the prediction model are obtained; Construct a neural network prediction model and use optimization algorithms to optimize the network structure and parameters to predict indoor occupant thermal comfort and air conditioning system energy consumption; Based on the established prediction model, the prediction interval and objective function are defined, and the optimal air conditioning temperature setting value is solved using a multi-objective optimization algorithm; After the solution process is completed, the obtained results are input into the external interface of the building energy consumption simulation model for joint calculation until the simulation process is completed.

2. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: The building energy consumption simulation model is characterized by comprising: Use AutoCAD, BIM or SketchUp to draw a 3D building model, set the number of people in each thermal zone, lamp power, electrical equipment power, and air conditioning system in each room for the drawn model; use sensors and electricity meters to collect indoor and outdoor environmental data, and embed the data collected by sensors into the building energy consumption simulation model through the external interface of the building simulation model to set the work schedule of personnel and equipment; use data collection and visualization functions to realize real-time display of simulation results, and build training sample sets and test sample sets.

3. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: include: The neural network prediction model is an extreme learning machine neural network; this neural network integrates the extreme learning machine network structure and the convolutional particle swarm algorithm, including an input layer, a hidden layer, an output layer and a network structure solution layer.

4. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: include: The air conditioning temperature set point optimization and control method is the Newton-Raphson multi-objective optimization algorithm; The method is based on the Newton-Raphson search rule to solve the defined objective function.

5. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: The implementation process of the neural network prediction model includes: Step 101: Select the outdoor temperature output by the building energy consumption simulation model , Indoor temperature , Indoor relative humidity , Indoor radiant temperature , Indoor air conditioning energy consumption , Indoor occupant comfort , Air conditioning temperature setting value As the input data of the neural network prediction model, the vector Indicates; Select the indoor room air conditioning system energy consumption output by the building energy consumption simulation model Thermal comfort As the output data of the neural network prediction model, we use vector express; Step 102: Perform maximum-minimum normalization on the data set: ； in, is the original data; is the normalized input data; is the matrix composed of the minimum values ​​of each feature of the original data; is a matrix composed of the maximum values ​​of each feature of the original data; the range of the normalized input data and output data is within the scope; represents the Hadamard product; Step 103: Input the data into the extreme learning machine neural network for training. The objective function is as follows: ； in, is the number of hidden layers of the extreme learning machine neural network; is the weight matrix connecting the input and hidden layers, where: ; is the bias matrix of the hidden layer, where: ; is the weight matrix connecting the hidden layer and the output layer, where: ; is the input data of the model; is the output data of the model; The activation function of the neural network can be sigmoid, tansig, tanh, relu, leaky_relu, elu, etc. Step 104: The number of hidden layers, the weights of the hidden layers connected to the input layer, and the bias of the hidden layers are used as particles in the particle swarm algorithm. Represents; defines population size , maximum number of iterations , the charge carried by each particle , magnetic field strength and particle mass , initialize each population , as shown below: ； in, is the component of the kth particle in the qth dimension of the search space; r is a generated random number, uniformly distributed in the interval [0,1]; and are the upper and lower bounds of the search space respectively; ； in, is the number of iterations; is the speed of the particle when it rotates; The speed of the particle when doing basic motion; is the integration constant; is the charge carried by each particle; is the magnetic field strength; is the particle mass; The inertia weight of the particle when it rotates; The inertia weight of the particle when doing basic motion; , is the learning factor; , , is a random number between [0,1]; is the individual optimal value under the current number of iterations; is the global optimal value under the current number of iterations; Step 105: To prevent the particle swarm algorithm from falling into the local optimum, the particle chaos collision strategy and the inertia weight adaptive adjustment strategy are used to prevent the population from falling into the local optimum, as shown in the following formula: ； in, are new particles generated by the particle collision strategy; is a particle randomly selected from the current number of iterations; is a random number between [0,1] and is uniformly distributed; if the fitness value calculated by substituting the updated particle into the objective function is less than the fitness value of the particle before the update, then a new particle is created; otherwise, the original particle is retained and the next iteration is entered; Step 106: After all particles have participated in the update, if the maximum number of iterations of the optimization process is reached, the optimization is terminated; Step 107: After the prediction model training is completed, the data output by the model is denormalized and the model accuracy is tested, as shown in the following formula: ； Step 108: Embed the tested model into the joint simulation module to realize the prediction of air conditioning energy consumption and thermal comfort during the simulation process.

6. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that The prediction step length is Within the step, the collected data is substituted into the trained neural network for prediction, and the temperature setting value within the prediction step is solved by the rolling optimization method. , the objective function of rolling optimization includes: 。 7. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: The set value optimization method is a Newton-Raphson multi-objective optimization algorithm, which specifically includes: Apply the trained extreme learning machine neural network to the building energy consumption simulation model; The data is acquired through the building energy consumption simulation model, and the extreme learning machine neural network is used to predict the air conditioning energy consumption and personnel comfort; Newton-Raphson multi-objective optimization algorithm is used for optimization; Based on the optimization results, update the air conditioning temperature setting value at the next moment.

8. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: The process of optimizing the air conditioning temperature setting value by the multi-objective optimization algorithm based on the Newton-Raphson search rule includes: Step 201: Based on the data collected at the current moment and the established prediction model, the model output within the prediction step is updated so that the objective function of the rolling optimization reaches the Pareto frontier: ； in, The input data of the model within the prediction step; The output data of the model within the prediction step; are other feasible solutions in the solution set; is the Pareto front solution; Step 202: Determine the air conditioning temperature setting value by using the defined objective function and constraints, and define the population size according to the multi-objective optimization algorithm based on the Newton-Raphson search rule , maximum number of iterations , using Latin hypercube to initialize the random generation Populations: ； in, is the component of the i-th element of the population in the j-th dimension of the search space; is a generated random number, uniformly distributed in the interval [0,1]; and are the upper and lower bounds of the search space respectively; is the random arrangement of elements on the jth; Step 203: After initializing the parameters and population, substitute each element into the objective function, calculate the fitness value of each element, perform non-dominated sorting, and screen out the non-inferior solution under the current number of iterations; then, use the ideal point method to calculate the distance between the fitness value of each element under the current number of iterations and the current optimal fitness value, as shown in the following formula: ； in, It is the Euclidean distance between the fitness value corresponding to each element and the optimal fitness value under the current number of iterations; is the number of elements in the current iteration. i The fitness value corresponding to the optimization goal; is the optimal fitness value of the i-th optimization target for each element under the current number of iterations; is the worst fitness value of the element under the current number of iterations for the i-th optimization objective; Step 204: Select the maximum distance under the current number of iterations The corresponding element is the worst position , whose j-th dimension component is ; Select the shortest distance under the current number of iterations The corresponding element is the worst position , whose j-th dimension component is ; Calculate using the Newton-Raphson search rule: ； in, is the bootstrap parameter, which is used to guide the population at the current iteration number into the correct direction; , is a random number between (0,1); , is the j-th dimension component of randomly selected elements in the population, and ; Solve for Newton-Raphson; Step 205: After calculating the value of each element in each dimension using the Newton-Raphson search rule, the updated element position is calculated using the following formula: ； in, is the j-th dimension component of the i-th element after update; is a random number between (0,1); , , for Update parameters of is the adaptive coefficient, which is used to ensure the diversity of the population; Step 206: When the algorithm reaches the maximum number of iterations, the algorithm is terminated and the Pareto surface is output; then, the ideal point method is used to calculate the distance between the solution set on the Pareto surface and the optimal solution is selected. .

9. A method for optimizing and controlling indoor air conditioning temperature setting values ​​in a public building as claimed in claim 1, characterized in that: In order to make the air conditioning control effect less affected by future environmental changes as possible, the first element of the optimal solution obtained is taken as the air conditioning temperature setting value.

Images