Building air conditioning system optimization control method and device based on spatio-temporal data

Abstract

The invention provides a building air conditioning system optimization control method and device based on spatio-temporal data, and the method comprises the steps: executing temperature simulation, and obtaining N temperature simulation distribution conditions at different moments in a year; establishing a room temperature distribution prediction model based on a graph attention neural network and a gating circulation unit; executing energy consumption simulation to obtain N pieces of energy consumption simulation data at different moments in a year; setting parameters of the air conditioner are used as input, energy consumption at the next moment is used as output, and an energy consumption prediction model is established based on a BP neural network; based on the room temperature distribution prediction model and the energy consumption prediction model, a multi-target particle swarm algorithm is adopted for optimization, an optimal scheme is determined through a weighted sum method, optimization targets are room thermal comfort and air conditioner energy consumption, and decision variables are setting parameters of an air conditioner; and the air conditioning system is controlled according to the optimal scheme, so that the control reliability of the building air conditioning system is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of air conditioning system control, and in particular to a method and device for optimizing and controlling a building air conditioning system based on spatiotemporal data. Background Art

[0002] With the rapid development of society, people's living standards are constantly improving, and people have higher requirements for thermal comfort in their living and working environments. At the same time, the world's energy shortage is becoming increasingly serious. The construction industry accounts for a large proportion of global energy consumption, and air-conditioning systems are the main part of building energy consumption. Therefore, optimizing and controlling the operation of air-conditioning systems is an important means to achieve building energy conservation and alleviate the energy crisis.

[0003] In current research, the optimization of indoor thermal comfort is often based on the average room temperature. However, in reality, the temperature distribution in the room is often uneven, especially when the air-conditioning system is just turned on. The temperature distribution in the room is more different. Moreover, improving thermal comfort often leads to an increase in energy consumption. These two goals are contradictory and need to be considered in a coordinated manner, resulting in low reliability of air-conditioning system operation control.

[0004] In view of this problem, the present invention provides a method and device for optimizing and controlling a building air conditioning system based on spatiotemporal data to solve the above problem. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention innovatively proposes a building air-conditioning system optimization control method and device based on spatiotemporal data, which effectively solves the problem of low reliability of building air-conditioning system control caused by the prior art and effectively improves the reliability of building air-conditioning system control.

[0006] A first aspect of the present invention provides a building air conditioning system optimization control method based on spatiotemporal data, comprising:

[0007] Establish a 3D BIM model of the building room, mesh the 3D BIM model, and establish a CFD model;

[0008] Formulate N simulation conditions of air conditioning setting parameters in the CFD model, perform temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in a year, and obtain N temperature simulation distributions at different times in a year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle;

[0009] Based on graph attention neural network and gated recurrent unit, a room temperature distribution prediction model is established;

[0010] Input the 3D BIM model into the energy consumption calculation software, perform energy consumption simulation for N simulated working conditions of air conditioning setting parameters according to weather conditions in a year, and obtain N energy consumption simulation data at different times in a year;

[0011] Taking the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, an energy consumption prediction model is established based on the BP neural network.

[0012] Based on the room temperature distribution prediction model and energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and the weighted sum method is used to determine the optimal solution. The optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the setting parameters of the air conditioner.

[0013] Control the air conditioning system according to the optimal solution.

[0014] A second aspect of the present invention provides a building air conditioning system optimization control device based on spatiotemporal data, comprising:

[0015] The first module is to build a 3D BIM model of the building room, mesh the 3D BIM model, and build a CFD model;

[0016] The first simulation module formulates N simulation conditions of air conditioning setting parameters in the CFD model, performs temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in a year, and obtains N temperature simulation distributions at different times in a year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle;

[0017] The second building module builds a room temperature distribution prediction model based on graph attention neural network and gated recurrent unit;

[0018] The second simulation module inputs the 3D BIM model into the energy consumption calculation software, performs energy consumption simulation for N simulated working conditions of air conditioning setting parameters according to weather conditions within a year, and obtains N energy consumption simulation data at different times within a year;

[0019] The third building module takes the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, and builds an energy consumption prediction model based on the BP neural network;

[0020] The optimization module uses a multi-objective particle swarm algorithm to optimize based on the room temperature distribution prediction model and energy consumption prediction model, and uses the weighted sum method to determine the optimal solution. The optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the air conditioning setting parameters.

[0021] The control module controls the air conditioning system according to the optimal solution.

[0022] The technical solution adopted by the present invention includes the following technical effects:

[0023] 1. The technical solution of the present invention obtains N temperature simulation distributions and energy consumption simulation data at different times of the year; based on the room temperature distribution prediction model and the energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and a weighted sum method is used to determine the optimal solution, wherein the optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the setting parameters of the air conditioner, which effectively solves the problem of low control reliability of the building air conditioning system caused by the prior art, and effectively improves the reliability of the building air conditioning system control.

[0024] 2. The technical solution of the present invention is based on BIM, CFD and EnergyPlus modeling, which effectively reduces the data set collection cycle and improves the efficiency of model training; and based on spatiotemporal data, a room temperature distribution prediction model is established through graph attention neural network and gated recurrent unit, which improves the accuracy of room temperature distribution prediction.

[0025] 3. In the technical solution of the present invention, the nodes in the room temperature field topology diagram can be either sitting type nodes and standing type nodes of people whose main stay time is longer than the preset time, or nodes can be determined by calculating the temperature gradient of the temperature distribution in the room area, thereby improving the flexibility of node selection.

[0026] 4. The feature vector h' of the target node i output by the aggregated graph attention neural network at time t in the technical solution of the present invention i Preprocessing is performed, wherein the preprocessing is multi-head attention processing to improve the learning ability of the room temperature distribution prediction model.

[0027] 5. The technical solution of the present invention controls the air-conditioning system according to the optimal solution, including: air-conditioning pre-start optimization control and air-conditioning optimization control during the room use period; different optimization controls can be performed according to the use period of the building air-conditioning, further improving the reliability and flexibility of the building air-conditioning system optimization control.

[0028] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1It is a schematic diagram of a process in a method of Example 1 in the scheme of the present invention;

[0031] Figure 2 Another schematic diagram of a process in the method of embodiment 1 of the present invention

[0032] Figure 3 Schematic diagram of the process of step S3 in the method of embodiment 1 of the present invention;

[0033] Figure 4 A schematic diagram of determining the location of nodes in a building room according to the length of stay of a person and the location of the person in the method of Embodiment 1 of the solution of the present invention;

[0034] Figure 5 Schematic diagram of the structure of the gated cycle unit in the method of Example 1 of the solution of the present invention;

[0035] Figure 6 It is a flow chart of a multi-objective particle swarm algorithm in the method of Example 1 in the scheme of the present invention;

[0036] Figure 7 Schematic diagram of the flow of the weighted sum method in the method of Example 1 in the scheme of the present invention;

[0037] Figure 8 It is a schematic diagram of the structure of the device of Example 2 in the scheme of the present invention. DETAILED DESCRIPTION

[0038] In order to clearly illustrate the technical features of the present solution, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings. The disclosure below provides many different embodiments or examples for realizing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and settings of specific examples are described below. In addition, the present invention may repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. It should be noted that the components illustrated in the accompanying drawings are not necessarily drawn to scale. The present invention omits the description of known components and processing techniques and processes to avoid unnecessary limitations on the present invention.

[0039] Embodiment 1

[0040] like Figure 1-Figure 2 As shown, the present invention provides a building air conditioning system optimization control method based on spatiotemporal data, comprising:

[0041] Step S1, establishing a three-dimensional BIM model of the building room, meshing the three-dimensional BIM model, and establishing a CFD model;

[0042] Step S2, formulating N simulation conditions of air conditioning setting parameters in the CFD model, performing temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in a year, and obtaining N temperature simulation distributions at different times in a year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle;

[0043] Step S3, building a room temperature distribution prediction model based on graph attention neural networks (GAT) and gated recurrent units (GRU);

[0044] Step S4, inputting the three-dimensional BIM model into the energy consumption calculation software, performing energy consumption simulation for N simulated working conditions of air conditioning setting parameters according to weather conditions within a year, and obtaining N energy consumption simulation data at different times within a year;

[0045] Step S5, taking the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, establishing an energy consumption prediction model based on the BP neural network;

[0046] Step S6, based on the room temperature distribution prediction model and the energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and a weighted sum method is used to determine the optimal solution, wherein the optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the setting parameters of the air conditioner;

[0047] Step S7, controlling the air conditioning system according to the optimal solution.

[0048] Among them, in step S1, firstly, a three-dimensional solid model of the building room is established using BIM (Building Information Modeling) software based on drawings and field measurement information. The model includes inner and outer walls, doors and windows, floor slabs, roofs, structural columns, etc.

[0049] Import the BIM model into the meshing software for meshing and establish a mesh model; import the mesh model into the CFD (Computational Fluid Dynamics) software, and then select and set the model, set the material, set the boundary, set the initial condition, set the solution, etc., and finally establish the CFD model.

[0050] In step S2, N simulation conditions of air conditioning setting parameters are formulated in the CFD model (air conditioning setting parameters include air conditioning pre-start time, temperature set point, air supply speed and air supply angle). One of the conditions is selected to perform simulation, and multiple temperature measuring instruments are evenly placed in the room to measure the room temperature. The measured room temperature is compared with the simulated temperature to verify the accuracy of the CFD model.

[0051] For the N working conditions that set the air conditioning setting parameters in the CFD model, temperature simulation is performed for the weather conditions within a year (taking 365 days as an example), and the temperature distribution output step is set to R minutes. Therefore, N temperature simulation distributions at different times in a year are obtained, that is, 365×(24×60) / R×N temperature distributions.

[0052] like Figure 3 Said step S3 specifically comprises:

[0053] S31, establishing a room temperature field topology diagram, wherein the room temperature field topology diagram is composed of nodes and edges, wherein the nodes are used to represent different selected positions in the building room, and the edges are used to represent the correlation between different nodes;

[0054] In step S31, different points are selected from the rooms of the building to establish a room temperature field topology diagram. The room temperature field topology diagram is composed of nodes and edges, which are represented by a graph G = (Q, O), where Q is a set of nodes of the room temperature field topology diagram, representing different selected locations in the room, and O is a set of edges of the room temperature field topology diagram, representing the correlation between different nodes.

[0055] There are two ways to select nodes in the room temperature field topology diagram: one is subjective selection, which is mainly to select from the areas where people stay for a long time in the room, such as office buildings. Figure 4 As shown, if a person sits at a workstation for a long time, three nodes can be selected from the room space positions corresponding to the human head, chest and legs respectively; a node can be selected from the room space position corresponding to the head of a person standing at the workstation; and nodes can be selected from the room space positions corresponding to the head, chest and legs of a standing person in the aisle; specifically, they can be a person sitting posture type node and a person standing posture type node whose stay time is greater than a preset time, wherein the person sitting posture type node includes the room position node corresponding to the head when the person stands up from a sitting position, the room position node corresponding to the head when the person is sitting, the room position node corresponding to the chest when the person is sitting, and the room position node corresponding to the legs when the person is sitting; the person standing posture type node includes the room position node corresponding to the head when the person stands up, the room position node corresponding to the chest when the person is standing, and the room position node corresponding to the legs when the person is standing.

[0056] Another way to determine the nodes in the room temperature field topology diagram can be objectively selected. The temperature distribution of the room is measured by a temperature acquisition instrument, and the temperature gradient of the temperature distribution in the room area is calculated. When the gradient value of the temperature corresponding to a certain position in the room is greater than the preset gradient threshold and there are no other nodes within the preset distance, then the position is a node; the specific process is: first, use thermal imagers, thermocouples and other temperature acquisition instruments and equipment, and ultrasonic travel time tomography technology can also be used as needed (divided into two steps: first, the signal propagation time is measured by an ultrasonic sensor to calculate the sound speed on the sound propagation path, and then the spatial temperature distribution of the entire test area is reconstructed through tomography. Commonly used reconstruction methods in hierarchical imaging include iterative methods and non-iterative methods. Iterative methods include Landweber iteration method, synchronous iterative reconstruction technology, synchronous algebraic reconstruction technology, etc. Non-iterative methods include truncated singular value decomposition method, Tikhonov regularization method, etc.) to measure the temperature distribution of the room, and then calculate the temperature gradient of the temperature distribution in the area. When If there are no other nodes within the surrounding distance r, a node is determined, where ΔT is the temperature difference, Δh is the spatial distance change, and ε is the node selection threshold. Under specific spatial distance conditions, the smaller the threshold, the higher the room temperature requirement. The selection of the threshold ε is generally first determined by experts scoring different levels of thresholds, and then the corresponding level of thresholds is selected according to the room temperature requirements;

[0057] Common empirical values ​​include, for example, the vertical temperature difference at 0.1m and 1.1m above the floor (considering sitting posture) is divided into three levels, with temperature differences less than 2°C, 3°C, and 4°C respectively; the vertical temperature difference between 1.8m and 0.1m (considering standing posture) is no more than 3°C.

[0058] The edges of the room temperature field topology graph are the connections between the selected nodes.

[0059] S32, constructing a time series feature matrix for each moment within a preset time length, and normalizing the time series feature matrix; wherein the time series feature matrix includes outdoor meteorological parameters, node temperatures output after temperature simulation, and relevant setting parameters of air conditioners; the time series feature matrix at a certain moment is e rows and b columns, wherein e represents the number of nodes, and b represents the number of feature data;

[0060] The input features of the room temperature distribution prediction model include b features, namely, outdoor meteorological parameters, node temperature (obtained from the temperature distribution output by the CFD model, that is, the N working conditions for setting air conditioning setting parameters in the CFD model are respectively subjected to temperature simulation for the weather conditions within a year, and the temperature distribution output step is set to R minutes, and N temperature simulation distributions at different times within a year are obtained), and the relevant setting parameters of the air conditioner (i.e., the air conditioner pre-start time, temperature setting point, air supply speed and air supply angle). Among them, the outdoor meteorological parameters and the relevant setting parameters of the air conditioner correspond to each node, and together with the node temperature of each node, they are used as the characteristics of the node. Therefore, a feature matrix is ​​established for each moment with a time interval of R minutes within the first n hours, where the feature matrix at a certain moment is e rows and b columns, e represents the number of nodes, and b represents the feature data.

[0061] S33, establishing an adjacency matrix of the room temperature field topology structure graph, and determining whether any two nodes in the room temperature field topology structure graph are adjacent according to the adjacency matrix;

[0062] The adjacency matrix is ​​used to represent the relationship between nodes. The matrix element U of the adjacency matrix is IJ Indicates the relationship between node I and node J, U IJ The calculation method is:

[0063]

[0064] Among them, P(vx I ,vx J )、P(vx I ,vx J ) and P(vx I ,vx J ) are the Pearson correlation coefficients of the velocity components of node I and node J on the x, y and z axes, respectively. The velocity component is the velocity component of the node on the three coordinate axes, representing the magnitude of the velocity in different directions, describing the operation of the airflow, and thus affecting the distribution of the temperature field; ε is the threshold of the correlation coefficient; U IJ The value of U is 1, indicating that there is an edge between node I and node J, the two nodes are strongly correlated, and they are neighbors of each other; IJ If the value of is 0, it means that the two nodes I and J are not adjacent and there is no correlation between the two nodes;

[0065] S34, extracting spatial features of each node based on a graph attention neural network, wherein the spatial features of the node include an attention coefficient between a target node and an adjacent node;

[0066] S35, calculating the attention coefficient of all neighbor nodes to the target node. The attention coefficient of the neighbor node to the target node indicates the importance of the neighbor node to the target node. The specific calculation method of the attention coefficient is:

[0067]

[0068] Among them, α ij is the normalized attention coefficient of the target node i and the neighbor node j; i is the target node; j is the neighbor node j; k is the neighbor node k; σ1 is the LeakyReLU activation function, LeakyReLU is a nonlinear activation function; N i are all neighbor nodes of the target node i; || is the series operation between feature vectors; a is the learnable parameter vector; W1 is the linear transformation matrix used for node feature transformation, the matrix size is the same as the feature matrix, and is used for feature transformation; h j is the feature vector of target node i, i.e., the i-th row of the feature matrix; h j is the feature vector of neighbor node j; h k is the feature vector of neighbor node k;

[0069] S36, according to the attention coefficients of all neighbor nodes to the target node, the spatial features of the neighbor nodes adjacent to the target node are aggregated to obtain the feature vector of the target node output by the aggregated graph attention neural network; wherein the expression of the feature vector of the target node output by the aggregated graph attention neural network is:

[0070]

[0071] Among them, h' i is the feature vector of the target node output by the graph attention neural network after aggregation, σ2 is the Sigmoid activation function;

[0072] S37, based on the gated recurrent unit, extracts the corresponding time feature of each target node i output by the aggregated graph attention neural network to obtain a room temperature distribution prediction model; wherein, Figure 5 As shown, the gated recurrent unit includes a reset gate and an update gate. The expression of the gated recurrent unit specifically includes:

[0073] r t =σ2(W r x t +U r h t-1 +b r ),

[0074] z t =σ2(W z x t +U z h t-1 +b z ),

[0075]

[0076] Among them, r t is the reset gate output, z t is the update gate output; x t is the feature vector h' of each target node i output by the aggregated graph attention neural network at time t i ; W r is the weight matrix input to the reset gate; W z is the weight matrix of the update gate; W h is the weight matrix of the candidate hidden state; U r is the implicit state h at time t-1 t-1 To the weight matrix of the reset gate; U z is the weight matrix of the update gate at time t-1; U h is the weight matrix of the candidate hidden state at time t-1; b r , b z and b h are the reset gate, update gate and bias of candidate hidden state respectively; h t-1 is the hidden state at time t-1; is the product of matrix elements; tanh is the hyperbolic tangent function; is the candidate hidden state at time t; h t is the hidden state at time t, the last h in the time series t It is input into the fully connected layer to predict the temperature distribution at the next moment;

[0077] S38, setting basic parameters of the room temperature distribution prediction model, dividing the data set into a training set and a test set according to a preset ratio, and using the training set to train the room temperature distribution prediction model to obtain a trained room temperature distribution prediction model.

[0078] The preset ratio can be to divide the temperature distribution simulation data set into a training set and a test set in a ratio of 7:3, and use the training set to train the room temperature distribution prediction model based on S31~S37. After the room temperature distribution prediction model is trained, the test set is used to test the performance of the room temperature distribution prediction model. If the performance of the room temperature distribution prediction model meets the requirements, the temperature distribution prediction result for the next moment is output.

[0079] Model evaluation indicators can be selected from root mean square error RMSE, mean absolute error MAE, mean absolute percentage error MAPE, correlation coefficient R 2 One or more of the above, the present invention is not limited here.

[0080]

[0081]

[0082] In the formula, is the predicted value; f is the actual value; is the average value of the actual value; n is the number of samples.

[0083] Preferably, the feature vector h' of the target node i output by the aggregated graph attention neural network at time t i Preprocessing is performed, and the preprocessing is multi-head attention processing. The feature vector of the target node i output by the graph attention neural network after preprocessing at time t is specifically:

[0084]

[0085] Among them, h' i ' is the feature vector of the target node i output by the preprocessed graph attention neural network at time t; M represents the number of multi-head attention heads; is the attention coefficient of the mth head; is the weight matrix of the mth head.

[0086] That is, in step S37, based on the feature vector of each target node i output by the gated recurrent unit after aggregation or the feature vector of the target node i output by the preprocessed graph attention neural network at time t, the corresponding time feature is extracted to obtain the room temperature distribution prediction model; wherein, in the expression of the gated recurrent unit, x t is the feature vector h' of each target node i output by the aggregated graph attention neural network at time t i Or the feature vector h' of the target node i output by the preprocessed graph attention neural network at time t i '.

[0087] In step S4, the BIM model is input into the EnergyPlus energy consumption calculation software (energy consumption model), and the measured energy consumption data of a period is selected to compare with the simulated energy consumption data to verify the accuracy of the energy consumption model; then, the energy consumption simulation is performed for the N working conditions of the air-conditioning setting parameters for 365×N weather conditions in a year, and the output step length of the energy consumption is R minutes, so 365×(24×60) / R×N energy consumption data are obtained.

[0088] In step S5, the setting parameters of the air conditioner are taken as input, and the energy consumption at the next moment is taken as output (the 365×(24×60) / R×N energy consumption data obtained in step S4), and deep training is performed based on the BP neural network to establish an energy consumption prediction model; that is, the energy consumption data set is divided into a training set and a test set at a ratio of 7:3, and the training set is used to train the energy consumption prediction model. After the energy consumption prediction model is trained, the test set is used to test the performance of the energy consumption prediction model. If the performance of the energy consumption prediction model meets the requirements, the energy consumption prediction result for the next moment is output.

[0089] In step S6, Figure 6 As shown in Figure 2, the optimization using multi-objective particle swarm algorithm specifically includes:

[0090] Initialize the particle population and external archive, and set the termination condition of the multi-objective particle swarm algorithm;

[0091] Calculating the fitness of each particle, wherein the fitness of the particle includes a thermal comfort objective function F1 and an energy consumption objective function F2;

[0092] Update the historical optimal position pBest and the global optimal position gBest of each particle;

[0093] Update the speed and position of each particle. The speed and position of the particle are updated as follows:

[0094]

[0095] in, are the particle velocities of particle h at the e+1th iteration and the eth iteration respectively; c1 and c2 are acceleration coefficients respectively; γ1 and γ2 are random numbers between [0,1] respectively; are the historical optimal position and global optimal position of particle h in the e-th iteration respectively; ω is the inertia weight; are the positions of particle h at the e+1th iteration and the eth iteration respectively;

[0096] Update the external archive with the updated velocity and position of each particle;

[0097] Until the current number of iterations or the fitness function value meets the termination condition, the Pareto knot is output, and the termination condition includes that the number of iterations meets the preset maximum number of iterations or the fitness function value is less than the preset fitness threshold.

[0098] The thermal comfort objective function is expressed as the cumulative sum of the differences between the predicted temperature of all target nodes and the set most comfortable temperature, specifically:

[0099]

[0100] The energy consumption objective function is as follows:

[0101] min F2=E,

[0102] Among them: F1 is the thermal comfort objective function, which is expressed by the absolute value of the difference between the predicted room temperature at the next moment and the set most comfortable temperature; F2 is the energy consumption objective function; Q is the total number of target nodes; T q is the predicted temperature of the qth target node; Set the most comfortable temperature for the qth target node; E is the energy consumption of the air-conditioning system between the current moment and the next moment (one time step).

[0103] like Figure 7 As shown in the figure, the optimal solution determined by the weighted sum method specifically includes:

[0104] Assign weights to the optimization target objective functions respectively;

[0105] According to the weight of each optimization objective function, the weighted sum value of each solution is calculated, where the weighted sum value of each solution is calculated as follows:

[0106] S z =ζ1F z1 +ζ2F z2 ,

[0107] Among them, S z is the weighted sum value of the zth scheme; ζ1 and ζ2 are the weights of the thermal comfort objective function and the energy consumption objective function respectively; F z1 、F z2 are the thermal comfort objective function value and energy consumption objective function value of the z-th scheme respectively;

[0108] Sort all the options according to the weighted sum value (total value) of each option;

[0109] The solution with the smallest weighted sum value is selected as the optimal solution.

[0110] In step S7, controlling the air conditioning system according to the optimal solution includes: optimizing the air conditioning pre-startup control and optimizing the air conditioning during the room usage period;

[0111] The optimization variables in the air conditioning pre-start optimization control include the air conditioning pre-start time, temperature set point, air supply speed and air supply angle. The optimization process is performed during the non-use period (early morning) of the building room to improve the thermal comfort and energy consumption at the initial use time of the room.

[0112] The optimization variables in the optimal control of air conditioning during the room usage period include temperature set point, air supply speed and air supply angle, and the optimization process is performed during the usage period of the building room.

[0113] The technical solution of the present invention obtains N temperature simulation distributions and energy consumption simulation data at different times of the year respectively; based on the room temperature distribution prediction model and the energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and a weighted sum method is used to determine the optimal solution, wherein the optimization objectives are room thermal comfort and air-conditioning energy consumption, and the decision variables are the setting parameters of the air-conditioning, which effectively solves the problem of low control reliability of the building air-conditioning system caused by the prior art, and effectively improves the control reliability of the building air-conditioning system.

[0114] The technical solution of the present invention is based on BIM, CFD and EnergyPlus modeling, which effectively reduces the data set collection cycle and improves the efficiency of model training; and based on spatiotemporal data, a room temperature distribution prediction model is established through a graph attention neural network and a gated recurrent unit, thereby improving the accuracy of room temperature distribution prediction.

[0115] In the technical solution of the present invention, the nodes in the room temperature field topology diagram can be either sitting type nodes and standing type nodes for people whose main stay time is greater than the preset time, or the nodes can be determined by calculating the temperature gradient of the temperature distribution in the room area, thereby improving the flexibility of node selection.

[0116] In the technical solution of the present invention, the feature vector h' of the target node i output by the aggregated graph attention neural network at time t is i Preprocessing is performed, wherein the preprocessing is multi-head attention processing to improve the learning ability of the room temperature distribution prediction model.

[0117] The technical solution of the present invention controls the air-conditioning system according to the optimal solution, including: air-conditioning pre-start optimization control and air-conditioning optimization control during the room use period; different optimization controls can be performed according to the use period of the building air-conditioning, further improving the reliability and flexibility of the building air-conditioning system optimization control.

[0118] Embodiment 2

[0119] like Figure 8 As shown, the technical solution of the present invention also provides a building air conditioning system optimization control device based on spatiotemporal data, comprising:

[0120] The first building module 101 builds a three-dimensional BIM model of a building room, meshes the three-dimensional BIM model, and builds a CFD model;

[0121] The first simulation module 102 formulates N simulation conditions of air conditioning setting parameters in the CFD model, performs temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in one year, and obtains N temperature simulation distributions at different times in one year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle;

[0122] The second building module 103 builds a room temperature distribution prediction model based on a graph attention neural network and a gated recurrent unit;

[0123] The second simulation module 104 inputs the three-dimensional BIM model into the energy consumption calculation software, performs energy consumption simulation on N simulated working conditions of air conditioning setting parameters according to weather working conditions in one year, and obtains N energy consumption simulation data at different times in one year;

[0124] The third building module 105 takes the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, and builds an energy consumption prediction model based on the BP neural network;

[0125] The optimization module 106 uses a multi-objective particle swarm algorithm to optimize based on the room temperature distribution prediction model and the energy consumption prediction model, and uses a weighted sum method to determine the optimal solution, wherein the optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the setting parameters of the air conditioner;

[0126] The control module 107 controls the air conditioning system according to the optimal solution.

[0127] It should be noted that the first establishment module 101, the first simulation module 102, the second establishment module 103, the second simulation module 104, the third establishment module 105, the optimization module 106, and the control module 107 in this embodiment respectively correspond to the method steps in Example 1, and the present invention will not go into details here.

[0128] The technical solution of the present invention obtains N temperature simulation distributions and energy consumption simulation data at different times of the year respectively; based on the room temperature distribution prediction model and the energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and a weighted sum method is used to determine the optimal solution, wherein the optimization objectives are room thermal comfort and air-conditioning energy consumption, and the decision variables are the setting parameters of the air-conditioning, which effectively solves the problem of low control reliability of the building air-conditioning system caused by the prior art, and effectively improves the control reliability of the building air-conditioning system.

[0129] The technical solution of the present invention is based on BIM, CFD and EnergyPlus modeling, which effectively reduces the data set collection cycle and improves the efficiency of model training; and based on spatiotemporal data, a room temperature distribution prediction model is established through a graph attention neural network and a gated recurrent unit, thereby improving the accuracy of room temperature distribution prediction.

[0130] In the technical solution of the present invention, the nodes in the room temperature field topology diagram can be either sitting type nodes and standing type nodes for people whose main stay time is greater than the preset time, or the nodes can be determined by calculating the temperature gradient of the temperature distribution in the room area, thereby improving the flexibility of node selection.

[0131] In the technical solution of the present invention, the feature vector h' of the target node i output by the aggregated graph attention neural network at time t is i Preprocessing is performed, wherein the preprocessing is multi-head attention processing to improve the learning ability of the room temperature distribution prediction model.

[0132] The technical solution of the present invention controls the air-conditioning system according to the optimal solution, including: air-conditioning pre-start optimization control and air-conditioning optimization control during the room use period; different optimization controls can be performed according to the use period of the building air-conditioning, further improving the reliability and flexibility of the building air-conditioning system optimization control.

[0133] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A building air conditioning system optimization control method based on spatiotemporal data, characterized in that it includes: Establish a 3D BIM model of the building room, mesh the 3D BIM model, and establish a CFD model; Formulate N simulation conditions of air conditioning setting parameters in the CFD model, perform temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in a year, and obtain N temperature simulation distributions at different times in a year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle; Based on graph attention neural network and gated recurrent unit, a room temperature distribution prediction model is established; Input the 3D BIM model into the energy consumption calculation software, perform energy consumption simulation for N simulated working conditions of air conditioning setting parameters according to weather conditions in a year, and obtain N energy consumption simulation data at different times in a year; Taking the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, an energy consumption prediction model is established based on the BP neural network. Based on the room temperature distribution prediction model and energy consumption prediction model, a multi-objective particle swarm algorithm is used for optimization, and the weighted sum method is used to determine the optimal solution. The optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the setting parameters of the air conditioner. Control the air conditioning system according to the optimal solution.

2. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 1, characterized in that: The method of establishing a room temperature distribution prediction model based on a graph attention neural network and a gated recurrent unit specifically includes: Establishing a room temperature field topology diagram, wherein the room temperature field topology diagram is composed of nodes and edges, wherein the nodes are used to represent different selected positions in the building room, and the edges are used to represent the correlation between different nodes; Construct a time series feature matrix for each moment within a preset duration, and normalize the time series feature matrix; wherein the time series feature matrix includes outdoor meteorological parameters, node temperatures output after temperature simulation, and relevant setting parameters of air conditioners; the time series feature matrix at a certain moment is e rows and b columns, wherein e represents the number of nodes and b represents the number of feature data; Establish an adjacency matrix of the room temperature field topology structure diagram, and determine whether any two nodes in the room temperature field topology structure diagram are adjacent according to the adjacency matrix; Extracting spatial features of each node based on a graph attention neural network, wherein the spatial features of the node include an attention coefficient between a target node and adjacent nodes; Calculate the attention coefficient of all neighbor nodes to the target node. The calculation method of the attention coefficient is as follows: Among them, α ij is the normalized attention coefficient of target node i and neighbor node j; i is the target node; j is the neighbor node j; k is the neighbor node k; σ1 is the LeakyReLU activation function; N i are all neighbor nodes of target node i; || is a series operation; a is a learnable parameter vector; W1 is a linear transformation matrix for node feature transformation; h i is the feature vector of target node i; h j is the feature vector of neighbor node j; h k is the feature vector of neighbor node k; According to the attention coefficients of all neighboring nodes to the target node, the spatial features of the neighboring nodes adjacent to the target node are aggregated to obtain the feature vector of the target node output by the aggregated graph attention neural network; the expression of the feature vector of the target node output by the aggregated graph attention neural network is: Among them, h' i is the feature vector of the target node output by the graph attention neural network after aggregation, σ2 is the Sigmoid activation function; Based on the gated recurrent unit, the corresponding time feature is extracted from the feature vector of each target node i output by the aggregated graph attention neural network, and the room temperature distribution prediction model is obtained; wherein the gated recurrent unit includes a reset gate and an update gate, and the expression of the gated recurrent unit specifically includes: r t =σ2(W r x t +U r h t-1 +b r ), z t =σ2(W z x t +U z h t-1 +b z ), Among them, r t is the reset gate output, z t is the update gate output; x t is the feature vector h' of each target node i output by the aggregated graph attention neural network at time t i ; W r is the weight matrix input to the reset gate; W z is the weight matrix of the update gate; W h is the weight matrix of the candidate hidden state; U r is the implicit state h at time t-1 t-1 To the weight matrix of the reset gate; U z is the weight matrix of the update gate at time t-1; U h is the weight matrix of the candidate hidden state at time t-1; b r , b z and b h are the reset gate, update gate and bias of candidate hidden state respectively; h t-1 is the hidden state at time t-1; is the product of matrix elements; tanh is the hyperbolic tangent function; is the candidate hidden state at time t; h t is the hidden state at time t, the last h in the time series t It is input into the fully connected layer to predict the temperature distribution at the next moment; The basic parameters of the room temperature distribution prediction model are set, the temperature distribution simulation data set is divided into a training set and a test set according to a preset ratio, and the training set is used to train the room temperature distribution prediction model to obtain the trained room temperature distribution prediction model.

3. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 2 is characterized in that: The nodes in the room temperature field topology diagram are the sitting posture type nodes and the standing posture type nodes of the persons whose stay time is greater than the preset time, wherein the sitting posture type nodes include the room position node corresponding to the head when the person stands up from a sitting position, the room position node corresponding to the head when the person is sitting, the room position node corresponding to the chest when the person is sitting, and the room position node corresponding to the legs when the person is sitting; the standing posture type nodes include the room position node corresponding to the head when the person is standing, the room position node corresponding to the chest when the person is standing, and the room position node corresponding to the legs when the person is standing.

4. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 2 is characterized in that: The nodes in the room temperature field topology diagram are determined as follows: The temperature distribution of the room is measured by a temperature acquisition instrument, and the temperature gradient of the temperature distribution in the room area is calculated. When the gradient value of the temperature corresponding to a certain position in the room is greater than the preset gradient threshold and there are no other nodes within the preset distance, the position is a node.

5. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 2 is characterized in that: The feature vector h' of the target node i output by the attention neural network at time t after aggregation i Preprocessing is performed, and the preprocessing is multi-head attention processing. The feature vector of the target node i output by the graph attention neural network after preprocessing at time t is specifically: Among them, h' i ' is the feature vector of the target node i output by the preprocessed graph attention neural network at time t; M represents the number of multi-head attention heads; is the attention coefficient of the mth head; is the weight matrix of the mth head.

6. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 1 is characterized in that: The optimization using multi-objective particle swarm algorithm specifically includes: Initialize the particle population and external archive, and set the termination condition of the multi-objective particle swarm algorithm; Calculating the fitness of each particle, wherein the fitness of the particle includes a thermal comfort objective function F1 and an energy consumption objective function F2; Update the historical optimal position pBest and the global optimal position gBest of each particle; Update the speed and position of each particle. The speed and position of the particle are updated as follows: in, are the particle velocities of particle h at the e+1th iteration and the eth iteration respectively; c1 and c2 are acceleration coefficients respectively; γ1 and γ2 are random numbers between [0,1] respectively; are the historical optimal position and global optimal position of particle h in the e-th iteration respectively; ω is the inertia weight; are the positions of particle h at the e+1th iteration and the eth iteration respectively; Update the external archive with the updated velocity and position of each particle; Until the current number of iterations or the fitness function value meets the termination condition, the Pareto knot is output, and the termination condition includes that the number of iterations meets the preset maximum number of iterations or the fitness function value is less than the preset fitness threshold.

7. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 1 is characterized in that: The comfort objective function is expressed as the cumulative sum of the differences between the predicted temperature of all target nodes and the set most comfortable temperature, specifically: The energy consumption objective function is as follows: min F2=E, Where: F1 is the thermal comfort objective function; F2 is the energy consumption objective function; Q is the total number of target nodes; T q is the predicted temperature of the qth target node; Set the most comfortable temperature for the qth target node; E is the energy consumption of the air conditioning system between the current moment and the next moment.

8. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 7 is characterized in that: The optimal solution determined by the weighted sum method specifically includes: Assign weights to the optimization objective functions respectively; According to the weight of each optimization objective function, the weighted sum value of each solution is calculated, where the weighted sum value of each solution is calculated as follows: S z =ζ1F z1 +ζ2F z2 , Among them, S z is the weighted sum value of the zth scheme; ζ1 and ζ2 are the weights of the thermal comfort objective function and the energy consumption objective function respectively; F z1 、F z2 are the thermal comfort objective function value and energy consumption objective function value of the z-th scheme respectively; Sort all the options according to the weight and value of each option; The solution with the smallest weighted sum value is selected as the optimal solution.

9. The method for optimizing and controlling a building air conditioning system based on spatiotemporal data according to claim 1, characterized in that: Controlling the air conditioning system according to the optimal solution includes: optimizing the air conditioning pre-start control and optimizing the air conditioning during the room usage period; The optimization variables in the air conditioning pre-start optimization control include the air conditioning pre-start time, temperature set point, air supply speed and air supply angle. The optimization process is performed during the non-use period of the building room to improve the thermal comfort and energy consumption at the initial use time of the room. The optimization variables in the optimal control of air conditioning during the room usage period include temperature set point, air supply speed and air supply angle, and the optimization process is performed during the usage period of the building room.

10. A building air conditioning system optimization control device based on spatiotemporal data, characterized in that: include: The first module is to build a 3D BIM model of the building room, mesh the 3D BIM model, and build a CFD model; The first simulation module formulates N simulation conditions of air conditioning setting parameters in the CFD model, performs temperature simulation on the N simulation conditions of the CFD model according to the weather conditions in a year, and obtains N temperature simulation distributions at different times in a year; wherein the air conditioning setting parameters include the air conditioning pre-start time, the temperature setting point, the air supply speed and the air supply angle; The second building module builds a room temperature distribution prediction model based on graph attention neural network and gated recurrent unit; The second simulation module inputs the 3D BIM model into the energy consumption calculation software, performs energy consumption simulation for N simulated working conditions of air conditioning setting parameters according to weather conditions within a year, and obtains N energy consumption simulation data at different times within a year; The third building module takes the setting parameters of the air conditioner as input and the energy consumption at the next moment as output, and builds an energy consumption prediction model based on the BP neural network; The optimization module uses a multi-objective particle swarm algorithm to optimize based on the room temperature distribution prediction model and energy consumption prediction model, and uses the weighted sum method to determine the optimal solution. The optimization objectives are room thermal comfort and air conditioning energy consumption, and the decision variables are the air conditioning setting parameters. The control module controls the air conditioning system according to the optimal solution.

Images