Intelligent building energy consumption monitoring system

Abstract

The invention relates to the technical field of building energy consumption monitoring, and discloses an intelligent building energy consumption monitoring system, which comprises a data acquisition module for acquiring indoor environment data and outdoor environment data of M sub-regions; the space-time heterogeneous graph construction module is used for constructing a space-time heterogeneous graph according to the indoor environment data and the outdoor environment data of the M sub-regions; the energy consumption prediction module is used for analyzing the space-time heterogeneous graph through an energy consumption prediction model to obtain the power consumption of the M sub-regions in N time periods in the future; the energy consumption optimization scheme generation module is used for generating an energy consumption optimization scheme through an optimization algorithm according to the power consumption of the M sub-regions in the future N time periods; according to the method, the space-time heterogeneous graph is formed by the correlation influence among the areas in the building, the space-time analysis is performed on the space-time heterogeneous graph through the energy consumption prediction model to realize the multi-period energy consumption prediction of all the sub-areas, so that the prediction precision is improved, the intelligent active energy consumption optimization is realized in combination with the optimization algorithm, and the operation cost is minimized.

Description

Technical Field

[0001] The invention relates to the technical field of building energy consumption monitoring, and more specifically, to an intelligent building energy consumption monitoring system. Background Art

[0002] Building energy consumption refers to the total amount of various energy consumed by a building during its operation, including electricity, gas, etc., which are mainly used for heating, cooling, lighting, ventilation and other scenarios. Building energy consumption monitoring can not only improve energy utilization, but also reduce energy waste and reduce operating costs.

[0003] Existing building energy consumption monitoring usually collects indoor energy consumption data and outdoor environmental data first. Indoor energy consumption data includes: power consumption, water consumption and gas consumption, etc. Outdoor environmental data includes: temperature, rainfall, wind speed and lighting conditions, etc. Then a nonlinear mapping relationship between it and the actual building energy consumption is established through a mathematical model, so as to realize the monitoring and prediction of building energy consumption. The mathematical model can be a convolutional neural network model or a time series analysis model, etc. However, the above schemes usually only focus on the overall building energy consumption data, ignoring the correlation between the various areas inside the building, such as: changes in population density, temperature distribution and air flow, which will affect the overall building energy consumption, resulting in deviations in the prediction results. In addition, the existing building energy consumption management system is usually limited to monitoring, prediction and early warning, lacking intelligent active optimization control strategies, resulting in unnecessary energy waste. Summary of the invention

[0004] The present invention provides an intelligent building energy consumption monitoring system to solve the technical problems in the above-mentioned background technology.

[0005] The present invention provides an intelligent building energy consumption monitoring system, comprising: A data collection module is used to divide the building to be monitored into M sub-areas, and collect indoor environment data and outdoor environment data of the M sub-areas at equal time intervals t within a preset time period T; Indoor environmental data include: occupancy density, temperature and carbon dioxide concentration; Outdoor environmental data includes: temperature, rainfall, wind speed and lighting conditions; A spatiotemporal heterogeneous graph construction module, which is used to construct a spatiotemporal heterogeneous graph according to the indoor environment data and outdoor environment data of M sub-areas; The spatiotemporal heterogeneous graph includes G subgraphs, each of which consists of M nodes, the initial vector of the node and the undirected edges between the nodes, where G = T / t; One node corresponds to one sub-area, the initial vector of the node is generated according to the indoor environment data and outdoor environment data of the corresponding sub-area, and the undirected edges between nodes are generated according to the indoor environment data of the corresponding sub-area; An energy consumption prediction module is used to analyze the spatiotemporal heterogeneous graph through an energy consumption prediction model to obtain the power consumption of M sub-areas in the next N time periods; An energy consumption optimization scheme generation module is used to generate an energy consumption optimization scheme through an optimization algorithm according to the power consumption of M sub-areas in the future N time periods; The energy consumption optimization scheme is represented by the output power of M sub-areas in the future N time periods.

[0006] Furthermore, the number of sub-areas M, the preset time period T, the time interval t, the number of future time periods N and the duration of each time period are all custom parameters.

[0007] Furthermore, the population density is represented by the ratio of the number of people to the area of ​​the sub-region, and the lighting conditions are represented by positive integer codes, where 1, 2, and 3 represent rainy days, cloudy days, and sunny days, respectively.

[0008] Furthermore, constructing a spatiotemporal heterogeneous graph includes the following steps: Step S201, concatenating and normalizing the indoor environment data and the outdoor environment data of the sub-region to obtain the initial vector of the node corresponding to the sub-region; The calculation formula for normalization is as follows:

[0009] in and represent the indoor environment data and outdoor environment data of the sub-region corresponding to the m-th node of the g-th sub-graph after normalization and before normalization, respectively. and Respectively represent the maximum and minimum values ​​of the indoor environment data and outdoor environment data of the sub-area corresponding to all nodes of the g-th subgraph before normalization; Step S202, constructing undirected edges between the same nodes of adjacent subgraphs; Step S203: if the difference in the personnel density of the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the personnel density threshold, an undirected edge is constructed between the adjacent nodes; The population density threshold is a custom parameter; Step S204: if the difference in temperature between the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the temperature threshold, an undirected edge is constructed between the adjacent nodes; The temperature threshold is a custom parameter; Step S205: if the difference in carbon dioxide concentration between sub-regions corresponding to adjacent nodes of the same subgraph is greater than or equal to a carbon dioxide concentration threshold, an undirected edge is constructed between the adjacent nodes; The CO2 concentration threshold is a custom parameter.

[0010] Furthermore, the energy consumption prediction model consists of a time series extraction layer, a time series analysis layer, a splicing layer, a spatial aggregation layer, and N classifiers; The time series extraction layer is used to extract the initial vector of each node in the G subgraphs of the spatiotemporal heterogeneous graph and construct the corresponding time series for each node; The time series analysis layer is used to extract features from the time series of each node to obtain a first update vector, where the number of dimensions of the first update vector is a custom parameter; The concatenation layer is used to concatenate the initial vector of each node in the G subgraphs with the corresponding first update vector to obtain a combined vector; The spatial aggregation layer is used to aggregate the information of the combined vectors of all nodes in the spatiotemporal heterogeneous graph in G subgraphs to obtain the second update vector, where the number of dimensions of the second update vector is a custom parameter; The second update vectors of all nodes are sequentially input into N classifiers, and the power consumption of the sub-area corresponding to each node in the next N time periods is output.

[0011] Furthermore, the time series analysis layer is built based on the long short-term memory neural network model.

[0012] Furthermore, the calculation formula of the spatial aggregation layer includes:

[0013] in represents the set of nodes in the g-th subgraph that have undirected edges with the m-th node, represents the second update vector of the mth node in the gth subgraph, and They represent the combined vectors of the mth node and the nth node in the gth subgraph, respectively. and They represent the combination vector and gate vector of the mth node in the g-1th subgraph, respectively. and All are assigned a value of 0. represents the comprehensive influence coefficient between the mth node and the nth node in the gth subgraph, W represents the weight matrix, Offset represents the bias vector, Swish represents the Swish activation function, represents the Hadamard product;

[0014] Where MLP represents multi-layer perceptron, and Concat represents concatenation function;

[0015] in , and They represent the influence coefficient of personnel density, temperature and carbon dioxide concentration between the mth node and the nth node in the gth subgraph respectively. , and are user-defined first, second, and third weight coefficients, and , and The sum of is 1;

[0016] in and They represent the personnel density of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively;

[0017] in and They represent the temperatures of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively;

[0018] in and They respectively represent the carbon dioxide concentrations in the sub-areas corresponding to the m-th node and the n-th node in the g-th sub-graph.

[0019] Furthermore, an energy consumption optimization scheme is generated by an optimization algorithm, including the following steps: Step S301, randomly generate an initialization population that meets the constraint conditions, and define the current number of iterations as 0; The number of individuals U in the initialized population is a custom parameter; Individuals are represented by matrix encoding Matrix, as follows: ,in Respectively represent the output power of the 1st to the Mth sub-areas in the first period of the future, They represent the output power of the first sub-area in the first to Nth time periods in the future, represents the output power of the Mth sub-area in the Nth period in the future; The constraints include: Constraint 1: , where 1≤i≤M, 1≤j≤N, and They represent the power consumption and output power of the ith sub-area in the jth period in the future, L represents the duration of the period, Indicates the conversion efficiency, with a value of 0.9. Indicates the deviation, which is a custom parameter; Constraint 2: ,in represents the maximum available power of the building to be monitored in the jth period in the future, is a custom parameter; Step S302, calculating the fitness values ​​of all individuals in the initialized population through the objective function; Step S303, generating an evolution coefficient according to the current number of iterations and the maximum number of iterations, and judging that the evolution coefficient is greater than or equal to a preset threshold, then proceeding to step S304, otherwise proceeding to step S305; The calculation formula of the evolution coefficient Evolve is as follows:

[0020] in represents the maximum number of iterations, and Rand represents a random number ranging from 0 to 1; The maximum number of iterations and the preset threshold are both custom parameters; Step S304, updating the matrix codes of random U / 2 individuals in the initialized population through the exploration strategy; Step S305, updating the matrix codes of all individuals in the initialized population through the development strategy; Step S306, the current number of iterations is added by 1, and if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is terminated, and the matrix encoding of the individual with the largest fitness value is used as the energy consumption optimization solution, otherwise, the process returns to step S302 to continue execution.

[0021] Furthermore, the objective function is calculated as follows:

[0022] Where Fit represents the fitness value, and represents the electricity price and carbon emission factor per unit power of the ith sub-region in the jth period in the future, Indicates the carbon emission weight coefficient with a value range of 0.1 to 0.5.

[0023] Furthermore, the calculation formula of the exploration strategy includes:

[0024]

[0025] Among them, 1≤u≤U, 1≤v1≤U, 1≤v2≤U, u≠v1≠2, and Respectively represent the matrix code of the u-th individual with the current iteration number t+1 and t, and represents the matrix encoding of the v1th and v2th individuals with the current iteration number t, Indicates the first change coefficient for the current iteration number t, and Respectively represent the first random number and the second random number whose values ​​range from 0 to 1; The calculation formula for the development strategy is as follows:

[0026]

[0027] in Represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, Indicates the second coefficient of change for the current iteration number t, and They respectively represent a third random number and a fourth random number ranging from 0 to 1.

[0028] The beneficial effects of the present invention are as follows: the present invention constructs a spatiotemporal heterogeneous graph of the correlation influences between various areas inside the building, and performs spatiotemporal analysis on it through an energy consumption prediction model to realize multi-period energy consumption prediction for all sub-areas, thereby improving the prediction accuracy, and combines with the optimization algorithm to realize intelligent active energy consumption optimization, thereby minimizing operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a schematic diagram of an intelligent building energy consumption monitoring system of the present invention; Figure 2 is a flow chart of constructing a spatiotemporal heterogeneous graph of the present invention; Figure 3 is a flow chart of generating an energy consumption optimization scheme through an optimization algorithm of the present invention; Figure 4 It is the energy consumption prediction comparison chart of the present invention.

[0030] In the figure: data collection module 101, spatiotemporal heterogeneous graph construction module 102, energy consumption prediction module 103, energy consumption optimization plan generation module 104. DETAILED DESCRIPTION

[0031] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0032] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0033] like Figure 1 to Figure 4 As shown, an intelligent building energy consumption monitoring system includes: The data collection module 101 is used to divide the building to be monitored into M sub-areas, and collect indoor environment data and outdoor environment data of the M sub-areas at equal time intervals t within a preset time period T; Indoor environmental data include: occupancy density, temperature and carbon dioxide concentration; Outdoor environmental data includes: temperature, rainfall, wind speed and lighting conditions; A spatiotemporal heterogeneous graph construction module 102, which is used to construct a spatiotemporal heterogeneous graph according to the indoor environment data and outdoor environment data of the M sub-areas; The spatiotemporal heterogeneous graph includes G subgraphs, each of which consists of M nodes, the initial vector of the node and the undirected edges between the nodes, where G = T / t; One node corresponds to one sub-area, the initial vector of the node is generated according to the indoor environment data and outdoor environment data of the corresponding sub-area, and the undirected edges between nodes are generated according to the indoor environment data of the corresponding sub-area; An energy consumption prediction module 103 is used to analyze the spatiotemporal heterogeneous graph through an energy consumption prediction model to obtain the power consumption of M sub-areas in the next N time periods; An energy consumption optimization solution generation module 104 is used to generate an energy consumption optimization solution through an optimization algorithm according to the power consumption of the M sub-areas in the future N time periods; The energy consumption optimization scheme is represented by the output power of M sub-areas in the future N time periods.

[0034] It should be noted that the present invention is not limited to realizing the prediction and optimization of power consumption of multiple sub-areas of the monitored building through the energy consumption prediction model and optimization algorithm. The concept of the present invention can also be used to realize the prediction and optimization of energy consumption such as water consumption and gas consumption. The prediction and optimization of a single type of energy consumption can improve the prediction accuracy and optimization granularity. The high-precision power consumption optimization scheme can provide more targeted guidance for the building energy management system, thereby achieving more efficient energy utilization and lower operating costs.

[0035] In one embodiment of the present invention, the number of sub-areas M, the preset time period T, the time interval t, the number of future time periods N and the duration of each time period are all custom parameters, wherein the number of sub-areas M is determined according to the actual energy consumption monitoring needs, for example, it can be divided according to the number of rooms of the building to be monitored or the functional area division, the preset time period T, the time interval t and the number of future time periods N are determined according to the predicted time span and granularity, for example, the long-term prediction can appropriately increase the values ​​of T, t and N, and the short-term prediction can appropriately reduce the values ​​of T, t and N. Preferably, the preset time period T, the time interval t and the number of future time periods N are set to 1 day, 1 hour and 12 respectively, then the number of sub-graphs G of the spatiotemporal heterogeneous graph is 24, and the duration of each time period is set to 2 hours.

[0036] In one embodiment of the present invention, the population density is represented by the ratio of the number of people to the area of ​​the sub-area, and the lighting conditions are represented by positive integer codes, where 1, 2, and 3 represent rainy days, cloudy days, and sunny days, respectively. In addition, the lighting conditions can also be directly represented by collecting the light intensity through a light intensity sensor, which will not be elaborated here.

[0037] In one embodiment of the present invention, Figure 2 As shown, constructing a spatiotemporal heterogeneous map based on the indoor environment data and outdoor environment data of M sub-areas includes the following steps: Step S201, concatenating and normalizing the indoor environment data and the outdoor environment data of the sub-region to obtain the initial vector of the node corresponding to the sub-region; The calculation formula for normalization is as follows:

[0038] in and represent the indoor environment data and outdoor environment data of the sub-region corresponding to the m-th node of the g-th sub-graph after normalization and before normalization, respectively. and Respectively represent the maximum and minimum values ​​of the indoor environment data and outdoor environment data of the sub-area corresponding to all nodes of the g-th subgraph before normalization; Step S202, constructing undirected edges between the same nodes of adjacent subgraphs; Step S203: if the difference in the personnel density of the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the personnel density threshold, an undirected edge is constructed between the adjacent nodes; The population density threshold is a custom parameter. Preferably, the population density threshold is set to 20% of the maximum historical population density. Step S204: if the difference in temperature between the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the temperature threshold, an undirected edge is constructed between the adjacent nodes; The temperature threshold is a custom parameter. Preferably, the temperature threshold is set to 10% of the historical maximum temperature. Step S205: if the difference in carbon dioxide concentration between sub-regions corresponding to adjacent nodes of the same subgraph is greater than or equal to a carbon dioxide concentration threshold, an undirected edge is constructed between the adjacent nodes; The carbon dioxide concentration threshold is a custom parameter. Preferably, the carbon dioxide concentration threshold is set to 15% of the maximum historical carbon dioxide concentration.

[0039] It should be noted that the dimensions of the initial vectors of all nodes in the spatiotemporal heterogeneous graph are 7. By constructing undirected edges between the same nodes of adjacent subgraphs, the continuity of the data in the time dimension can be effectively captured. In order to consider the correlation between the various areas inside the building, the adjacent nodes of the same subgraph are constructed into undirected edges based on the population density, temperature and carbon dioxide concentration, which can effectively realize the correlation of the data in the spatial dimension, thereby improving the accuracy of the prediction.

[0040] In one embodiment of the present invention, the energy consumption prediction model is composed of a time series extraction layer, a time series analysis layer, a splicing layer, a spatial aggregation layer, and N classifiers; The time series extraction layer is used to extract the initial vector of each node in the G subgraphs of the spatiotemporal heterogeneous graph and construct the corresponding time series for each node; The time series analysis layer is used to extract features from the time series of each node to obtain a first update vector. The number of dimensions of the first update vector is a custom parameter. Preferably, the number of dimensions of the first update vector is set to 32. The concatenation layer is used to concatenate the initial vector of each node in the G subgraphs with the corresponding first update vector to obtain a combined vector; The spatial aggregation layer is used to aggregate information of the combined vectors of all nodes in the spatiotemporal heterogeneous graph in G subgraphs to obtain a second update vector. The number of dimensions of the second update vector is a custom parameter. Preferably, the number of dimensions of the second update vector is set to 64. The second update vectors of all nodes are sequentially input into N classifiers, and the power consumption of the sub-area corresponding to each node in the next N time periods is output.

[0041] It should be noted that the time series of a node includes G sequence units, each sequence unit is represented by an initial vector. Assuming that the number of dimensions of the first update vector is set to 32, the number of dimensions of the combined vector is 32+7=39. Each node has a global timing feature in the G subgraphs, thereby improving the accuracy of model prediction. In addition, it is also helpful to pre-train the energy consumption prediction model. The first update vector of each node output by the timing analysis layer can be input into an additional classifier. The classification space of these classifiers can be the indoor environment data of the next time interval. In this way, it is easier to obtain pre-trained training samples, which can reduce the training time of the actual model training to a certain extent and accelerate the convergence speed of the model.

[0042] In one embodiment of the present invention, the time series analysis layer is constructed based on a long short-term memory neural network model (LSTM), and can also be constructed based on a gated recurrent neural network model (GRU), which will not be described in detail here.

[0043] In one embodiment of the present invention, the calculation formula of the spatial aggregation layer includes:

[0044] in represents the set of nodes in the g-th subgraph that have undirected edges with the m-th node, represents the second update vector of the mth node in the gth subgraph, and They represent the combined vectors of the mth node and the nth node in the gth subgraph, respectively. and They represent the combination vector and gate vector of the mth node in the g-1th subgraph, respectively. and All are assigned a value of 0. represents the comprehensive influence coefficient between the mth node and the nth node in the gth subgraph, W represents the weight matrix, Offset represents the bias vector, Swish represents the Swish activation function, represents the Hadamard product;

[0045] Where MLP represents multi-layer perceptron, and Concat represents concatenation function;

[0046] in , and They represent the influence coefficient of personnel density, temperature and carbon dioxide concentration between the mth node and the nth node in the gth subgraph respectively. , and are user-defined first, second, and third weight coefficients, and , and The sum of is 1;

[0047] in and They represent the personnel density of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively;

[0048] in and They represent the temperatures of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively;

[0049] in and They respectively represent the carbon dioxide concentrations in the sub-areas corresponding to the m-th node and the n-th node in the g-th sub-graph.

[0050] It should be noted that, assuming that the number of dimensions of the combination vector is 39 and the number of dimensions of the second update vector is set to 64, the gating vector needs to be designed to be 1×39 in size, the weight matrix needs to be designed to be 39×64 in size, and the bias vector needs to be designed to be 1×64 in size. Both the weight matrix and the bias vector are learnable hyperparameters, which are reversely updated using the chain rule combined with the gradient descent algorithm, which will not be elaborated here.

[0051] In one embodiment of the present invention, Figure 3 As shown, an energy consumption optimization plan is generated by an optimization algorithm according to the power consumption of M sub-areas in the next N time periods, including the following steps: Step S301, randomly generate an initialization population that meets the constraint conditions, and define the current number of iterations as 0; The number of individuals U of the initial population is a custom parameter. Preferably, U is set to 20; Individuals are represented by matrix encoding Matrix, as follows: ,in Respectively represent the output power of the 1st to the Mth sub-areas in the first period of the future, They represent the output power of the first sub-area in the first to Nth time periods in the future, represents the output power of the Mth sub-area in the Nth period in the future; The constraints include: Constraint 1: , where 1≤i≤M, 1≤j≤N, and They represent the power consumption and output power of the ith sub-area in the jth period in the future, L represents the duration of the period, Indicates the conversion efficiency, with a value of 0.9. Indicates the deviation, which is a custom parameter; Constraint 2: ,in represents the maximum available power of the building to be monitored in the jth period in the future, is a custom parameter; Step S302, calculating the fitness values ​​of all individuals in the initialized population through the objective function; Step S303, generating an evolution coefficient according to the current number of iterations and the maximum number of iterations, and judging that the evolution coefficient is greater than or equal to a preset threshold, then proceeding to step S304, otherwise proceeding to step S305; The calculation formula of the evolution coefficient Evolve is as follows:

[0052] in represents the maximum number of iterations, and Rand represents a random number ranging from 0 to 1; The maximum number of iterations and the preset threshold are both custom parameters. Preferably, the maximum number of iterations is set to 20, and the preset threshold is set to 0.5; Step S304, updating the matrix codes of random U / 2 individuals in the initialized population through the exploration strategy; Step S305, updating the matrix codes of all individuals in the initialized population through the development strategy; Step S306, the current number of iterations is added by 1, and if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is terminated, and the matrix encoding of the individual with the largest fitness value is used as the energy consumption optimization solution, otherwise, the process returns to step S302 to continue execution.

[0053] In one embodiment of the present invention, the calculation formula of the objective function is as follows:

[0054] Where Fit represents the fitness value, and represents the electricity price and carbon emission factor per unit power of the ith sub-region in the jth period in the future, Indicates the carbon emission weight coefficient with a value range of 0.1 to 0.5.

[0055] In one embodiment of the present invention, the calculation formula of the exploration strategy includes:

[0056]

[0057] Among them, 1≤u≤U, 1≤v1≤U, 1≤v2≤U, u≠v1≠2, and Respectively represent the matrix code of the u-th individual with the current iteration number t+1 and t, and represents the matrix encoding of the v1th and v2th individuals with the current iteration number t, Indicates the first change coefficient for the current iteration number t, and Respectively represent the first random number and the second random number whose values ​​range from 0 to 1; The calculation formula for the development strategy is as follows:

[0058]

[0059] in Represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, Indicates the second coefficient of change for the current iteration number t, and They respectively represent a third random number and a fourth random number ranging from 0 to 1.

[0060] It should be noted that the power supply priority of the electrical appliances in each sub-area can be set at different time periods. When the output power of the sub-area is insufficient to meet the needs of all electrical appliances, the electrical appliances with higher power supply priorities can be powered first, and the output power of electrical appliances with lower priority can be reduced, thereby minimizing the operating costs and carbon emissions of the monitored buildings.

[0061] like Figure 4 As shown, the energy consumption prediction model provided by the present invention and a single time series analysis model (LSTM) are used to predict the power consumption of the monitored building in the next 2 hours (in kWh). It is found through observation that the energy consumption prediction model provided by the present invention has higher prediction accuracy.

[0062] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. An intelligent building energy consumption monitoring system, characterized in that: include: A data collection module is used to divide the building to be monitored into M sub-areas, and collect indoor environment data and outdoor environment data of the M sub-areas at equal time intervals t within a preset time period T; Indoor environmental data include: occupancy density, temperature and carbon dioxide concentration; Outdoor environmental data includes: temperature, rainfall, wind speed and lighting conditions; A spatiotemporal heterogeneous graph construction module, which is used to construct a spatiotemporal heterogeneous graph according to the indoor environment data and outdoor environment data of M sub-areas; The spatiotemporal heterogeneous graph includes G subgraphs, each of which consists of M nodes, the initial vector of the node and the undirected edges between the nodes, where G = T / t; One node corresponds to one sub-area, the initial vector of the node is generated according to the indoor environment data and outdoor environment data of the corresponding sub-area, and the undirected edges between nodes are generated according to the indoor environment data of the corresponding sub-area; An energy consumption prediction module is used to analyze the spatiotemporal heterogeneous graph through an energy consumption prediction model to obtain the power consumption of M sub-areas in the next N time periods; An energy consumption optimization plan generation module is used to generate an energy consumption optimization plan through an optimization algorithm according to the power consumption of M sub-areas in the next N time periods; The energy consumption optimization scheme is represented by the output power of M sub-areas in the future N time periods.

2. The intelligent building energy consumption monitoring system according to claim 1 is characterized in that: The number of sub-areas M, the preset time period T, the time interval t, the number of future time periods N, and the duration of each time period are all custom parameters.

3. The intelligent building energy consumption monitoring system according to claim 1 is characterized in that: The population density is represented by the ratio of the number of people to the area of ​​the sub-region, and the lighting conditions are represented by positive integer codes, where 1, 2, and 3 represent rainy days, cloudy days, and sunny days, respectively.

4. The intelligent building energy consumption monitoring system according to claim 1 is characterized in that: Constructing a spatiotemporal heterogeneous graph includes the following steps: Step S201, concatenating and normalizing the indoor environment data and the outdoor environment data of the sub-region to obtain the initial vector of the node corresponding to the sub-region; The calculation formula for normalization is as follows: in and represents the indoor environment data and outdoor environment data of the sub-region corresponding to the m-th node of the g-th sub-graph after normalization and before normalization, respectively. and Respectively represent the maximum and minimum values ​​of the indoor environment data and outdoor environment data of the sub-area corresponding to all nodes of the g-th subgraph before normalization; Step S202, constructing undirected edges between the same nodes of adjacent subgraphs; Step S203: if the difference in the personnel density of the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the personnel density threshold, an undirected edge is constructed between the adjacent nodes; The population density threshold is a custom parameter; Step S204: if the difference in temperature between the sub-regions corresponding to the adjacent nodes of the same sub-graph is greater than or equal to the temperature threshold, an undirected edge is constructed between the adjacent nodes; The temperature threshold is a custom parameter; Step S205: if the difference in carbon dioxide concentration between sub-regions corresponding to adjacent nodes of the same subgraph is greater than or equal to a carbon dioxide concentration threshold, an undirected edge is constructed between the adjacent nodes; The CO2 concentration threshold is a custom parameter.

5. The intelligent building energy consumption monitoring system according to claim 1 is characterized in that: The energy consumption prediction model consists of a time series extraction layer, a time series analysis layer, a splicing layer, a spatial aggregation layer, and N classifiers; The time series extraction layer is used to extract the initial vector of each node in the G subgraphs of the spatiotemporal heterogeneous graph and construct the corresponding time series for each node; The time series analysis layer is used to extract features from the time series of each node to obtain a first update vector, where the number of dimensions of the first update vector is a custom parameter; The concatenation layer is used to concatenate the initial vector of each node in the G subgraphs with the corresponding first update vector to obtain a combined vector; The spatial aggregation layer is used to aggregate the information of the combined vectors of all nodes in the spatiotemporal heterogeneous graph in G subgraphs to obtain the second update vector, where the number of dimensions of the second update vector is a custom parameter; The second update vectors of all nodes are sequentially input into N classifiers, and the power consumption of the sub-area corresponding to each node in the next N time periods is output.

6. The intelligent building energy consumption monitoring system according to claim 5 is characterized in that: The time series analysis layer is built based on the long short-term memory neural network model.

7. The intelligent building energy consumption monitoring system according to claim 5, characterized in that: Calculation formula of spatial aggregation layer include: in represents the set of nodes in the g-th subgraph that have undirected edges with the m-th node, represents the second update vector of the mth node in the gth subgraph, and They represent the combined vectors of the mth node and the nth node in the gth subgraph, respectively. and They represent the combination vector and gate vector of the mth node in the g-1th subgraph, respectively. and All are assigned a value of 0. represents the comprehensive influence coefficient between the mth node and the nth node in the gth subgraph, W represents the weight matrix, Offset represents the bias vector, Swish represents the Swish activation function, represents the Hadamard product; Where MLP represents multi-layer perceptron, and Concat represents concatenation function; in , and They represent the influence coefficient of personnel density, temperature and carbon dioxide concentration between the mth node and the nth node in the gth subgraph respectively. , and are user-defined first, second, and third weight coefficients, and , and The sum of is 1; in and They represent the personnel density of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively; in and They represent the temperatures of the sub-regions corresponding to the m-th node and the n-th node in the g-th sub-graph respectively; in and They respectively represent the carbon dioxide concentrations in the sub-areas corresponding to the m-th node and the n-th node in the g-th sub-graph.

8. The intelligent building energy consumption monitoring system according to claim 1 is characterized in that: Generate an energy consumption optimization plan through the optimization algorithm, including the following steps: Step S301, randomly generate an initialization population that meets the constraint conditions, and define the current number of iterations as 0; The number of individuals U in the initialized population is a custom parameter; Individuals are represented by matrix encoding Matrix, as follows: ,in Respectively represent the output power of the 1st to the Mth sub-areas in the first period of the future, They represent the output power of the first sub-area in the first to Nth time periods in the future, represents the output power of the Mth sub-area in the Nth period in the future; The constraints include: Constraint 1: , where 1≤i≤M, 1≤j≤N, and They represent the power consumption and output power of the ith sub-area in the jth period in the future, L represents the duration of the period, Indicates the conversion efficiency, with a value of 0.

9. Indicates the deviation, which is a custom parameter; Constraint 2: ,in represents the maximum available power of the building to be monitored in the jth period in the future, is a custom parameter; Step S302, calculating the fitness values ​​of all individuals in the initialized population through the objective function; Step S303, generating an evolution coefficient according to the current number of iterations and the maximum number of iterations, and judging that the evolution coefficient is greater than or equal to a preset threshold, then proceeding to step S304, otherwise proceeding to step S305; The calculation formula of the evolution coefficient Evolve is as follows: in represents the maximum number of iterations, and Rand represents a random number ranging from 0 to 1; The maximum number of iterations and the preset threshold are both custom parameters; Step S304, updating the matrix codes of random U / 2 individuals in the initialized population through the exploration strategy; Step S305, updating the matrix codes of all individuals in the initialized population through the development strategy; Step S306, the current number of iterations is added by 1, and if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is terminated, and the matrix encoding of the individual with the largest fitness value is used as the energy consumption optimization solution, otherwise, the process returns to step S302 to continue execution.

9. The intelligent building energy consumption monitoring system according to claim 8, characterized in that: The objective function is calculated as follows: Where Fit represents the fitness value, and represents the electricity price and carbon emission factor per unit power of the ith sub-region in the jth period in the future, Indicates the carbon emission weight coefficient with a value range of 0.1 to 0.

5.

10. The intelligent building energy consumption monitoring system according to claim 8, characterized in that: The calculation formula of the exploration strategy includes: Among them, 1≤u≤U, 1≤v1≤U, 1≤v2≤U, u≠v1≠2, and Respectively represent the matrix code of the u-th individual with the current iteration number t+1 and t, and represents the matrix encoding of the v1th and v2th individuals with the current iteration number t, Indicates the first change coefficient for the current iteration number t, and Respectively represent the first random number and the second random number whose values ​​range from 0 to 1; The calculation formula for the development strategy is as follows: in Represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, Indicates the second change coefficient for the current iteration number t, and They respectively represent a third random number and a fourth random number ranging from 0 to 1.

Images