Building energy intelligent management scheduling method based on digital twinning

Abstract

The invention relates to the technical field of energy management, in particular to an intelligent building energy management scheduling method based on digital twinning, and the method comprises the steps: constructing a digital twinning model corresponding to a real target building region, and dividing an energy region; constructing a regional data set, and storing the regional data set to a historical database; establishing an energy consumption prediction model, and predicting the energy consumption of each energy region; collecting energy consumption data in each energy area in real time; predicting the energy consumption data in each area in the next time period by using the currently collected energy consumption data in each energy area; and comparing the predicted energy consumption data of each energy region with preset energy consumption, and scheduling the energy of the region smaller than the preset energy consumption to the region exceeding the preset energy consumption. According to the method, the energy management scheduling efficiency is improved, and the purposes of energy conservation and emission reduction are achieved.

Description

Technical Field

[0001] The present invention relates to the field of energy management technology, and in particular to a building energy intelligent management and scheduling method based on digital twins. Background Art

[0002] In recent years, the frequent occurrence of global extreme climate has sounded the alarm for mankind. Excessive emissions of greenhouse gases are causing irreparable negative impacts on the environment, climate, and health. A self-rescue of the human living environment centered on low-carbon and environmental protection is imminent. With the acceleration of urbanization, building energy consumption has become an important part of global energy consumption. Traditional building energy management systems often rely on manual experience for scheduling, and after the building area is demarcated, it is often based on how to schedule between the demarcated building area and the superior energy management system, and does not involve how to schedule and allocate between the same-level building areas within the building area. As a result, although energy is scheduled, the energy distribution between users within the demarcated building area is unreasonable, resulting in energy waste. Therefore, how to efficiently and reasonably manage building energy has become an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide a building energy intelligent management and scheduling method based on digital twins to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A building energy intelligent management and scheduling method based on digital twins includes the following steps:

[0006] S1: Use the model building module to build a digital twin model corresponding to the real target building area and divide it into energy areas;

[0007] S2: The data analysis module obtains historical energy consumption and constructs regional data sets according to energy regions, and saves the regional data sets to the historical database module;

[0008] S3: The prediction module establishes an energy consumption prediction model and calls the regional data set corresponding to the energy region in the historical database module to perform prediction sample training on the future energy consumption of each energy region in each interval time period t;

[0009] S4: The data collection module collects the current energy consumption corresponding to each energy area in real time in each interval time period t;

[0010] S5: The prediction module uses the current energy consumption to predict the future energy consumption in the next interval time period t;

[0011] S6: The data analysis module compares the future energy consumption of the corresponding energy area with the preset energy consumption, and the emergency dispatch module dispatches energy according to the first-come-first-served algorithm.

[0012] Furthermore, the step S1 includes the following steps:

[0013] S11: Acquire architectural design information corresponding to the real target building area, and generate a virtual three-dimensional model in a model building module;

[0014] S12: The data analysis module transmits the three-dimensional model to the digital twin display module for real-time display;

[0015] S13: The data analysis module divides the three-dimensional model into energy areas according to corresponding building design information;

[0016] S14: The data acquisition module collects the actual building information of the real target building area and verifies it with the three-dimensional model displayed by the digital twin display module to determine whether it corresponds. If it does not correspond, the generated virtual three-dimensional model is corrected, and the process returns to S12 to transmit the corrected three-dimensional model to the digital twin display module for real-time display, and enters S13 to re-divide the three-dimensional model into energy areas; if it corresponds, jump to S15;

[0017] S15: Map the divided energy areas to the digital twin display module, and update the three-dimensional model of the digital twin display module.

[0018] Furthermore, the data acquisition module sends the corresponding actual building information to the data analysis module through the RFID electronic tag of the energy collector, and performs corresponding verification with the architectural design information of the corresponding energy area in the three-dimensional model.

[0019] Further, the S3 comprises the following steps:

[0020] S31: dividing the regional data set into a first statistical data set, a second statistical data set, a third statistical data set, and a fourth statistical data set according to seasons, and dividing each statistical data set into a training set, a validation set, and a test set according to a set number of days;

[0021] S32: establishing a corresponding BP neural network prediction model according to each statistical data set, and for each statistical data set, taking the historical energy consumption in each interval time period t as sample data, and inputting it into the corresponding neural network model for sample training;

[0022] S33: Use gradient descent method and back propagation method to adjust the weight and threshold of each BP neural network prediction model;

[0023] S34: using the DE algorithm to find the optimal values ​​of the initial weights and thresholds of each BP neural network prediction model;

[0024] S35: Use the validation set to evaluate the training process and decide whether to continue training;

[0025] S36: If yes, return to S33 to adjust the weight and threshold of each BP neural network prediction model again; if no, use the test set to test the trained model.

[0026] Further, the S5 comprises the following steps:

[0027] The prediction module selects the corresponding BP neural network model according to the current season;

[0028] The prediction module predicts the future energy consumption of the next interval time period t of the real target building area based on the current energy consumption.

[0029] Further, the S6 comprises the following steps:

[0030] The data analysis module sets preset energy consumption for each corresponding energy area in each quarter;

[0031] The future energy consumption of each energy area is compared with the preset energy consumption, the consumption difference is calculated, and it is determined whether the consumption difference is within the error range. If it is within the error range, the digital twin display module displays the current energy consumption and future energy consumption; when the consumption difference is outside the error range, the emergency dispatch module obtains the information of the corresponding energy area and generates a dispatch task, creates corresponding dispatch task queues according to the positive and negative conditions of the consumption difference, and performs energy dispatch between the two corresponding dispatch task queues according to the first-come, first-served algorithm.

[0032] Furthermore, the emergency dispatch module creates corresponding dispatch task queues according to the positive and negative conditions of the consumption difference, and performs energy dispatch between two corresponding dispatch task queues according to the first-come-first-served algorithm. The specific method is:

[0033] The emergency dispatch module creates two dispatch task queues, namely, a positive demand dispatch task queue and a negative demand dispatch task queue, according to whether the received energy area consumption difference is a positive energy demand or a negative energy demand, and arranges the priority of each dispatch task queue according to a first-come, first-served algorithm;

[0034] Update the positive demand scheduling task queue and the negative demand scheduling task queue;

[0035] Find the scheduling tasks corresponding to the adjacent energy areas in the positive demand scheduling task queue and the negative demand scheduling task queue respectively and pair them according to the first-come-first-served algorithm;

[0036] The two paired energy areas are dispatched according to the minimum value of positive energy demand and negative energy demand;

[0037] The scheduling tasks corresponding to the energy areas whose consumption difference after scheduling satisfies the condition that it is within the error range are eliminated from the corresponding queue;

[0038] Determine whether the number of remaining scheduling tasks in the two scheduling task queues is greater than or equal to 1 at the same time. If both are greater than or equal to 1, return to update the corresponding scheduling task queue and reschedule; if both are 0, end the scheduling task, and the digital twin display module displays the current energy consumption and future energy consumption as well as the difference in scheduled consumption; if the number of scheduling tasks in the positive demand scheduling task queue is 0, apply for scheduling energy to the superior energy allocator, or if the number of scheduling tasks in the negative demand scheduling task queue is 0, save the negative energy demand to cope with the sudden energy demand in the next interval time period t.

[0039] Furthermore, the method of adjusting the weights and thresholds of each BP neural network prediction model using the gradient descent method and the back propagation method includes the following steps:

[0040] First, the weights and thresholds of the neural network are randomly initialized. Then, using forward propagation, the historical energy consumption is input from the input layer through the hidden layer to the output layer, and the output of each layer is calculated. During the forward propagation process, the output of each layer is the input of the next layer. The calculation formula for forward propagation is:

[0041] Z (l) =W (l) a (l-1) +b l

[0042] a (l) =f(z (l) );

[0043] Among them, W (l) is the weight matrix of the lth layer, b (l) is the threshold vector of the lth layer, a (l-1) is the output of the l-1th layer and the input of the lth layer, Z (l) is the weighted sum of the lth layer, a (l) is the output of layer l, and f() is the activation function;

[0044] In the output layer, the loss between the predicted historical energy consumption and the actual historical energy consumption is calculated. The loss function is calculated as:

[0045]

[0046] Among them, y iis the historical energy consumption of the real output, y i is the historical energy consumption of the predicted output, and N is the number of samples;

[0047] Through the back propagation algorithm, starting from the output layer, the gradient of the loss function with respect to each weight and threshold is calculated layer by layer, and the weight and threshold are updated using the gradient descent method. The update formula is:

[0048]

[0049]

[0050] Among them, W (l+1) is the updated weight, η is the learning rate, is the loss function J with respect to w (l) The partial derivative is the gradient, b (l+1) is the updated threshold, b (l) represents the bias of the lth layer, is the loss function with respect to the bias b (l) The offset number.

[0051] Furthermore, the method of using the DE algorithm to find the optimal values ​​of the weights and thresholds of each BP neural network prediction model includes the following steps:

[0052] Initialize the historical energy consumption of the predicted output, input evolution parameters: sample number N, t is the number of iterations, the initial iteration number t = 0, and randomly generate initial sample data

[0053]

[0054] Evaluate all historical energy consumption and calculate

[0055]

[0056] Among them, f(X i (t)) is the target value of historical energy consumption, h is the difference in the number of selected samples;

[0057] Reproduction and recombination, according to the adaptive selection, crossover and mutation operation methods of DE, for each energy consumption data X in the sample i (t), randomly generate three different random energy data where r1, r2, r3∈{1, 2, …N}, j rand ∈{1, 2, …, n};

[0058]

[0059] in, is the new value of the jth gene of individual i at iteration number t+1, F is the crossover factor F∈(0,1), The new value of gene j remains unchanged, that is, it is inherited from the old value at iteration number t, P c is the crossover probability;

[0060] Select the optimal value of the initial weight and threshold of the output BP neural network model

[0061]

[0062] Among them, if the t+1 generation sample X i (t+1) satisfies the termination condition, then output X i The historical energy consumption of the predicted output with the minimum target in (t+1) is taken as the optimal individual, and its corresponding weight and threshold are the optimal values. If they are not satisfied, return to the second step to continue execution.

[0063] Furthermore, after the S4 data acquisition module collects the current energy consumption in real time within each interval time period t, the data analysis module matches the collected current energy consumption to the corresponding energy area through the RFID electronic tag; then saves it to the corresponding area data set and stores it as historical energy consumption.

[0064] Furthermore, a building energy intelligent management and scheduling system based on digital twin is applied to the building energy intelligent management and scheduling method based on digital twin, and the system includes the following modules:

[0065] The data collection module includes a perception module, a communication module, and an energy collector. The energy collector is connected to the perception module, and the perception module is connected to the data analysis module through the communication module. RFID electronic tags are set in the energy collectors. Multiple energy collectors are set in the real target building area according to the actual spatial position parameters of the building. The RFID electronic tags in the energy collectors are set with corresponding actual size parameters of the building. The data collection module is used to collect the current energy consumption of each area within the interval time period t in real time;

[0066] Data analysis module, the data analysis module includes a receiving module, a processing module, and a transmission module. The receiving module is used to receive relevant information from the data acquisition module, the model building module, the emergency dispatch module, the digital twin display module, the prediction module, and the historical database module. The transmission module is used to transmit the processed information. The processing module is used to analyze and process the received relevant information, such as saving the regional data set to the historical database module, setting a preset energy consumption for each energy area according to each quarter, and comparing the future energy consumption with the preset energy consumption;

[0067] A model building module, the model building module is connected with the data analysis module, and the model building module is used to build a digital twin model corresponding to the real target building;

[0068] The emergency dispatch module is connected with the data analysis module. The emergency dispatch module is used to create a positive demand dispatch task queue and a negative demand dispatch task queue, arrange the priorities of related dispatch tasks, and dispatch positive energy demand and negative energy demand in corresponding energy areas;

[0069] A digital twin display module, the digital twin display module is connected to the data analysis module, and the digital twin display platform is used to display the three-dimensional model, the divided energy areas, the current energy consumption in the current interval time period t, the predicted future energy consumption in the next interval time period t, and the difference in consumption scheduled in the next interval time period t in real time;

[0070] A prediction module is connected to the data analysis module. The prediction module is used to establish a quarterly energy consumption prediction model and predict the future energy consumption quantity in the next interval time period t.

[0071] The historical database module is connected with the data analysis module. The historical database module includes a regional data set. The content of the regional data set includes a historical energy consumption in water supply, heating, electricity and gas supply. The historical database module is used to save the historical energy consumption in each time period t.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] The present invention provides a building energy intelligent management and scheduling method based on digital twins, which obtains the building design information corresponding to the real target building area to construct a three-dimensional model and uses RFID technology to verify the actual building information and the three-dimensional model. The digital twin technology is used, which can facilitate rapid response to energy collection and energy distribution in the later stage. In addition, this method only uses key data for modeling and verification, instead of mapping all detailed building modules one by one, which simplifies the construction data and construction process of the three-dimensional model, can greatly shorten the modeling time, verification time and subsequent analysis and display time, and is conducive to the intelligent management of building energy.

[0074] The present invention provides a building energy intelligent management and scheduling method based on digital twins. It predicts future energy consumption in real time according to the prediction model, and realizes real-time energy scheduling with priority in the energy areas of the same level, avoiding direct energy scheduling from the superior energy distributor to the areas with high energy consumption demand. It can reasonably allocate energy and greatly reduce energy loss, so as to achieve the purpose of energy saving. The first-come-first-served algorithm is used to save queue processing time, so that the emergency scheduling module can respond quickly.

[0075] The present invention provides a method for intelligent management and scheduling of building energy based on digital twins, which can finely manage the use of building energy through real-time collection of energy consumption, energy consumption forecast, visual display and emergency scheduling in the building, improve energy efficiency and reduce energy consumption. Managers can intuitively see the fluctuations of various data, discover potential problems in time and take measures to deal with them. Digital twin technology uses virtual models to represent and reproduce physical spaces in the real world, which not only improves work efficiency, but also deepens insight into data and provides strong support for decision makers. This real-time and accuracy provides managers with a fast and accurate basis for decision-making, which helps to respond quickly in complex and changing situations. Through simulation and prediction, digital twin technology can find the best energy scheduling plan, optimize energy scheduling, reduce scheduling costs, and find the optimal scheduling method for energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 It is a flow chart of a method for intelligent management and scheduling of building energy based on digital twins of the present invention;

[0077] Figure 2 This is a flow chart of step S1 of a method for intelligent management and scheduling of building energy based on digital twins of the present invention;

[0078] Figure 3 This is a flow chart of step S3 of a method for intelligent management and scheduling of building energy based on digital twins of the present invention;

[0079] Figure 4 This is a schematic diagram of the energy dispatching process of the emergency dispatch module in step S6 of a building energy intelligent management and dispatching method based on digital twins of the present invention;

[0080] Figure 5 This is a structural schematic diagram of a building energy intelligent management and scheduling system based on digital twins of the present invention;

[0081] Figure 6 This is a structural schematic diagram of a data acquisition module of a building energy intelligent management and scheduling system based on digital twins of the present invention;

[0082] Figure 7 This is a structural schematic diagram of a data analysis module of a building energy intelligent management and scheduling system based on digital twins in the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] In the following description of the invention, it should be noted that the terms "upper", "lower", "left", "right", "inner", "outer", etc., indicating directions or positional relationships, are based on directions or positional relationships shown in the drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific direction or be constructed and operated in a specific direction. The term "connection" only indicates the connection between devices and has no special meaning.

[0085] In addition, the technical fields and installation methods involved in the embodiments of the present invention described below can be combined with each other as long as there is no conflict between them.

[0086] Specific implementation: Please refer to Figure 1-Figure 7 , a building energy intelligent management and scheduling method based on digital twins, comprising the following steps:

[0087] S1: Use the model building module to build a digital twin model corresponding to the real target building area and divide it into energy areas;

[0088] S2: The data analysis module obtains historical energy consumption and constructs regional data sets according to energy regions, and saves the regional data sets to the historical database module;

[0089] S3: The prediction module establishes an energy consumption prediction model and calls the regional data set corresponding to the energy region in the historical database module to perform prediction sample training on the future energy consumption of each energy region in each interval time period t;

[0090] S4: The data collection module collects the current energy consumption corresponding to each energy area in real time in each interval time period t;

[0091] S5: The prediction module uses the current energy consumption to predict the future energy consumption in the next interval time period t;

[0092] S6: The data analysis module compares the future energy consumption of the corresponding energy area with the preset energy consumption, and the emergency dispatch module dispatches energy according to the first-come-first-served dispatch algorithm.

[0093] Furthermore, the step S1 includes the following steps:

[0094] S11: Acquire architectural design information corresponding to the real target building area, and generate a virtual three-dimensional model in the model building module, wherein the architectural design information includes architectural setting space position parameters and building setting size parameters, wherein the architectural setting space position parameters include building setting building distribution and building setting floor distribution; and the building setting size parameters include building setting outline size, building setting spacing, and building setting main body height;

[0095] S12: The data analysis module transmits the three-dimensional model to the digital twin display module for real-time display;

[0096] S13: The data analysis module divides the three-dimensional model into energy areas according to corresponding building design information;

[0097] In an embodiment, buildings are not limited to industrial buildings, residential buildings, commercial buildings, agricultural buildings, etc. The data analysis module divides the energy areas and divides and numbers them according to the distribution of spatial position parameters of buildings in the real target building area; the energy is one of electricity, heat, water, and gas.

[0098] S14: The data acquisition module collects the actual building information of the real target building area and verifies it with the three-dimensional model displayed by the digital twin display module to determine whether it corresponds. If it does not correspond, the generated virtual three-dimensional model is corrected, and the process returns to S12 to transmit the corrected three-dimensional model to the digital twin display module for real-time display, and enters S13 to re-divide the three-dimensional model into energy areas; if it corresponds, jump to S15;

[0099] The actual building information includes the actual spatial location parameters of the building and the actual size parameters of the building. The actual spatial location parameters of the building include the actual building distribution and the actual floor distribution of the building; the actual size parameters of the building include the actual outline size of the building, the actual spacing between the buildings, and the actual main height of the building;

[0100] S15: Map the divided energy areas to the digital twin display module, and update the three-dimensional model of the digital twin display module.

[0101] The data collection module includes an energy collector, each of which is provided with an RFID electronic tag. A plurality of energy collectors are set in the real target building area according to the actual spatial position parameters of the building, and the RFID electronic tags in the energy collectors are provided with corresponding actual size parameters of the building;

[0102] The data acquisition module sends the corresponding actual building information to the data analysis module through the RFID electronic tag of the energy collector, and verifies the correspondence with the architectural design information of the corresponding energy area in the three-dimensional model. If it corresponds, the number of the corresponding energy area will be displayed in the digital twin display module. If it does not correspond, the generated virtual three-dimensional model will be checked and corrected to make the architectural design information consistent with the actual building information.

[0103] The architectural design information corresponding to the real target building area is obtained to construct a three-dimensional model, and RFID technology is used to verify the actual building information with the three-dimensional model. The digital twin technology is used, which can facilitate rapid response to energy collection and energy distribution in the later stage. This method only uses key data for modeling and verification, rather than mapping all detailed building modules one by one, which simplifies the construction data and construction process of the three-dimensional model, and can greatly shorten the modeling time, verification time and subsequent analysis and display time, which is conducive to the intelligent management of building energy.

[0104] Furthermore, the data analysis module obtains the historical energy consumption data within one year and constructs a regional data set according to the corresponding energy area. For example, if there are n energy areas in the real target building area, then the regional data set Q = {A1, A2, A3...A n}, A1, A2, A3…A n The energy consumption data set corresponding to n superior energy regions in one year is used as the sub-regional data set, and the energy consumption data of the sub-regional data set in each day is used as the historical energy consumption;

[0105] Further, the S3 comprises the following steps:

[0106] S31: Divide the regional data set into a first statistical data set Q1, a second statistical data set Q2, a third statistical data set Q3, and a fourth statistical data set Q4 according to season, where Q = Q1∪Q2∪Q3∪Q4, and divide each statistical data set into a training set, a validation set, and a test set according to a set number of days;

[0107] Each statistical data set is divided into a training set, a validation set and a test set by day, and the historical energy consumption of 60% of the days in each month is used as a training set, 20% as a validation set, and 20% as a test set;

[0108] S32: establishing a corresponding BP neural network prediction model according to each statistical data set, and for each statistical data set, taking the historical energy consumption in each interval time period t as sample data, and inputting it into the corresponding neural network model for sample training;

[0109] S33: Use gradient descent method and back propagation method to adjust the weight and threshold of each BP neural network prediction model;

[0110] S34: using the DE algorithm to find the optimal values ​​of the weights and thresholds of each BP neural network prediction model;

[0111] S35: Use the validation set to evaluate the training process and decide whether to continue training;

[0112] S36: If yes, return to S33 to adjust the weight and threshold of each BP neural network prediction model again; if no, use the test set to test the trained model.

[0113] Specifically, in this embodiment, the interval time period t is set to 1 day, and the historical energy consumption within the corresponding number of days is selected as sample data.

[0114] Furthermore, the method of adjusting the weights and thresholds of each BP neural network prediction model using the gradient descent method and the back propagation method includes the following steps:

[0115] First, the weights and thresholds of the neural network are randomly initialized. Then, the historical energy consumption is input through forward propagation, and propagated from the input layer through the hidden layer to the output layer. The output of each layer is calculated. In the forward propagation process, the output of each layer is the input of the next layer. The calculation formula for forward propagation is:

[0116] Z (l) =W (l) a (l-1) +b l

[0117] a (l) =f(z (l) );

[0118] Among them, W (l) is the weight matrix of the lth layer, b (l) is the threshold vector of the lth layer, a (l-1) is the output of the l-1th layer and the input of the lth layer, Z (l) is the weighted sum of the lth layer, a (l) is the output of layer l, and f() is the activation function;

[0119] In the output layer, the loss between the predicted historical energy consumption and the actual historical energy consumption is calculated. The loss function is calculated as:

[0120]

[0121] Among them, y i is the historical energy consumption of the real output, y i is the historical energy consumption of the predicted output, and N is the number of samples;

[0122] Through the back propagation algorithm, starting from the output layer, the gradient of the loss function with respect to each weight and threshold is calculated layer by layer, and the weight and threshold are updated using the gradient descent method. The update formula is:

[0123]

[0124]

[0125] Among them, W (l+1) is the updated weight, η is the learning rate, is the loss function J with respect to w (l) The partial derivative is the gradient, b (l+1) is the updated threshold, b (l) represents the bias of the lth layer, is the loss function with respect to the bias b (l) The offset number.

[0126] Furthermore, the method of using the DE algorithm to find the optimal values ​​of the weights and thresholds of each BP neural network prediction model includes the following steps:

[0127] Initialize the historical energy consumption of the predicted output, input evolution parameters: sample number N, t is the number of iterations, the initial number of iterations t = 0, and randomly generate initial sample data

[0128]

[0129] Evaluate all historical energy consumption and calculate

[0130]

[0131] Among them, f(X i (t)) is the target value of historical energy consumption, h is the difference in the number of selected samples;

[0132] Reproduction and recombination, according to the adaptive selection, crossover and mutation operation methods of DE, for each energy consumption data X in the sample i (t), randomly generate three different random energy data where r1, r2, r3∈{1, 2, …, N}, j rand ∈{1, 2, …, n};

[0133]

[0134] in, is the new value of the jth gene of individual i at iteration number t+1, F is the crossover factor F∈(0,1), The new value of gene j remains unchanged, that is, it is inherited from the old value at iteration number t, P c is the crossover probability;

[0135] Select the optimal value of the initial weight and threshold of the output BP neural network model

[0136]

[0137] Among them, if the t+1 generation sample X i (t+1) satisfies the termination condition, then output X i The historical energy consumption of the predicted output with the minimum target in (t+1) is taken as the optimal individual, and its corresponding weight and threshold are the optimal values. If they are not satisfied, return to the second step to continue execution.

[0138] Furthermore, after the S4 data collection module collects the current energy consumption in real time in each interval time period t, the data analysis module maps the collected current energy consumption to the corresponding energy area through the RFID electronic tag;

[0139] It is then saved to the corresponding regional dataset and stored as historical energy consumption for use in updating the prediction model.

[0140] Further, the S5 comprises the following steps:

[0141] The prediction module selects the corresponding BP neural network model according to the current season;

[0142] The prediction module predicts the future energy consumption of the next interval time period t of the real target building area based on the current energy consumption.

[0143] Further, the S6 comprises the following steps:

[0144] The data analysis module sets preset energy consumption for each corresponding energy area in each quarter;

[0145] The future energy consumption of each energy area is compared with the preset energy consumption, the consumption difference is calculated, and it is determined whether the consumption difference is within the error range. If it is within the error range, the digital twin display module displays the current energy consumption and future energy consumption; when the consumption difference is outside the error range, the emergency dispatch module obtains the information of the corresponding energy area and generates a dispatch task, creates corresponding dispatch task queues according to the positive and negative conditions of the consumption difference, and performs energy dispatch between the two corresponding dispatch task queues according to the first-come, first-served algorithm;

[0146] Specifically, the consumption difference greater than 0 is regarded as positive energy demand, indicating that the future energy consumption in the energy area exceeds the preset energy consumption and needs to be supplemented; the difference consumption less than 0 is regarded as negative energy demand, indicating that the future energy consumption in the energy area is less than the preset energy consumption and can be dispatched to other energy areas; when it is outside the error range, the digital analysis module extracts the actual building information and consumption difference of all energy areas outside the error range and sends them to the emergency dispatch module;

[0147] The emergency dispatch module creates two dispatch task queues, namely, a positive demand dispatch task queue and a negative demand dispatch task queue, according to whether the received energy area consumption difference is a positive energy demand or a negative energy demand, and arranges the priority of each dispatch task queue according to a first-come, first-served algorithm;

[0148] Update the positive demand scheduling task queue and the negative demand scheduling task queue;

[0149] More specifically, in the positive demand scheduling task queue and the negative demand scheduling task queue, the adjacent energy areas are searched according to the actual information of the buildings, and further judged according to the actual distance between the buildings; the energy areas with the smallest distance between the two corresponding energy areas in the two scheduling task queues are regarded as adjacent energy areas;

[0150] Find the scheduling tasks corresponding to the adjacent energy areas in the positive demand scheduling task queue and the negative demand scheduling task queue respectively and pair them according to the first-come-first-served algorithm. If multiple energy areas adjacent to another scheduling task queue are found in one scheduling task queue, for example, an energy area in a positive demand scheduling task queue corresponds to multiple adjacent energy areas in the negative demand scheduling task queue, the scheduling task that comes first in the queue will be paired first;

[0151] The two paired energy areas are dispatched according to the minimum value principle of positive energy demand and negative energy demand. For example, if the positive energy demand corresponding to the energy area is the minimum value, the negative energy demand is dispatched to the other energy area according to the minimum value; if the negative energy demand corresponding to the energy area is the minimum value, the negative energy demand is dispatched to the other energy area as a whole;

[0152] The scheduling tasks corresponding to the energy areas whose consumption difference after scheduling satisfies the condition that it is within the error range are eliminated from the corresponding queue;

[0153] Determine whether the number of remaining scheduling tasks in the two scheduling task queues is greater than or equal to 1 at the same time. If both are greater than or equal to 1, return to update the corresponding scheduling task queue and reschedule; if both are 0, end the scheduling task, and the digital twin display module displays the current energy consumption and future energy consumption as well as the difference in scheduled consumption; if the number of scheduling tasks in the positive demand scheduling task queue is 0, apply for scheduling energy to the superior energy allocator, or if the number of scheduling tasks in the negative demand scheduling task queue is 0, save the negative energy demand to cope with the sudden energy demand in the next interval time period t.

[0154] Specifically in the present embodiment, the preset energy consumption is set for each energy zone in each quarter by calculating the average value of the historical energy consumption in each interval time period t of the energy zone in each quarter as the preset energy consumption, and the error range of the consumption difference is set to ±θ, where θ is a number greater than 0, and its value shall not be greater than the difference between the maximum and minimum values ​​of the historical energy consumption in each interval time period t of the energy zone in this quarter.

[0155] This building energy intelligent management and scheduling method predicts future energy consumption in real time based on the prediction model, and realizes real-time energy scheduling within the same energy area, avoiding direct energy scheduling from the superior energy distributor to the area with high energy demand, greatly reducing energy loss, and achieving the purpose of energy conservation. The first-come-first-served algorithm is used to save queue processing time, so that the emergency scheduling module can respond quickly.

[0156] Furthermore, a building energy intelligent management and scheduling system based on digital twin is applied to a building energy intelligent management and scheduling method based on digital twin, and the system includes the following modules:

[0157] The data collection module includes a perception module, a communication module, and an energy collector. The energy collector is connected to the perception module, and the perception module is connected to the data analysis module through the communication module. RFID electronic tags are set in the energy collectors. Multiple energy collectors are set in the real target building area according to the actual spatial position parameters of the building. The RFID electronic tags in the energy collectors are set with corresponding actual size parameters of the building. The data collection module is used to collect the current energy consumption of each area within the interval time period t in real time;

[0158] Data analysis module, the data analysis module includes a receiving module, a processing module, and a transmission module. The receiving module is used to receive relevant information from the data acquisition module, the model building module, the emergency dispatch module, the digital twin display module, the prediction module, and the historical database module. The transmission module is used to transmit the processed information. The processing module is used to analyze and process the received relevant information, such as saving the regional data set to the historical database module, setting a preset energy consumption for each energy area according to each quarter, and comparing the future energy consumption with the preset energy consumption;

[0159] A model building module, the model building module is connected with the data analysis module, and the model building module is used to build a digital twin model corresponding to the real target building;

[0160] The emergency dispatch module is connected with the data analysis module. The emergency dispatch module is used to create a positive demand dispatch task queue and a negative demand dispatch task queue, arrange the priorities of related dispatch tasks, and dispatch positive energy demand and negative energy demand in corresponding energy areas;

[0161] A digital twin display module, the digital twin display module is connected to the data analysis module, and the digital twin display platform is used to display the three-dimensional model, the divided energy areas, the current energy consumption in the current interval time period t, the predicted future energy consumption in the next interval time period t, and the difference in consumption scheduled in the next interval time period t in real time;

[0162] A prediction module is connected to the data analysis module. The prediction module is used to establish a quarterly energy consumption prediction model and predict the future energy consumption quantity in the next interval time period t.

[0163] The historical database module is connected with the data analysis module. The historical database module includes a regional data set. The content of the regional data set includes a historical energy consumption in water supply, heating, electricity and gas supply. The historical database module is used to save the historical energy consumption in each time period t.

Claims

1. A building energy intelligent management and scheduling method based on digital twins, characterized in that: The following steps are involved: S1: Use the model building module to build a digital twin model corresponding to the real target building area and divide it into energy areas; S2: The data analysis module obtains historical energy consumption and constructs regional data sets according to energy regions, and saves the regional data sets to the historical database module; S3: The prediction module establishes an energy consumption prediction model and calls the regional data set corresponding to the energy region in the historical database module to perform prediction sample training on the future energy consumption of each energy region in each interval time period t; S4: The data collection module collects the current energy consumption corresponding to each energy area in real time in each interval time period t; S5: The prediction module uses the current energy consumption to predict the future energy consumption in the next interval time period t; S6: The data analysis module compares the future energy consumption of the corresponding energy area with the preset energy consumption, and the emergency dispatch module dispatches energy according to the first-come-first-served dispatch algorithm.

2. According to claim 1, a building energy intelligent management and scheduling method based on digital twins is characterized by: The S1 comprises the following steps: S11: Acquire architectural design information corresponding to the real target building area, and generate a virtual three-dimensional model in a model building module; S12: The data analysis module transmits the three-dimensional model to the digital twin display module for real-time display; S13: The data analysis module divides the three-dimensional model into energy areas according to corresponding building design information; S14: The data acquisition module collects the actual building information of the real target building area and verifies it with the three-dimensional model displayed by the digital twin display module to determine whether it corresponds. If it does not correspond, the generated virtual three-dimensional model is corrected, and the process returns to S12 to transmit the corrected three-dimensional model to the digital twin display module for real-time display, and enters S13 to re-divide the three-dimensional model into energy areas; if it corresponds, jump to S15; S15: Map the divided energy areas to the digital twin display module, and update the three-dimensional model of the digital twin display module.

3. According to claim 2, a building energy intelligent management and scheduling method based on digital twins is characterized by: The data acquisition module sends the corresponding actual building information to the data analysis module through the RFID electronic tag of the energy collector, and performs corresponding verification with the architectural design information of the corresponding energy area in the three-dimensional model.

4. According to claim 1, a building energy intelligent management and scheduling method based on digital twins is characterized by: The S3 comprises the following steps: S31: dividing the regional data set into a first statistical data set, a second statistical data set, a third statistical data set, and a fourth statistical data set according to seasons, and dividing each statistical data set into a training set, a validation set, and a test set according to a set number of days; S32: establishing a corresponding BP neural network prediction model according to each statistical data set, and for each statistical data set, taking the historical energy consumption in each interval time period t as sample data, and inputting it into the corresponding neural network model for sample training; S33: Use gradient descent method and back propagation method to adjust the weight and threshold of each BP neural network prediction model; S34: using the DE algorithm to find the optimal values ​​of the initial weights and thresholds of each BP neural network prediction model; S35: Use the validation set to evaluate the training process and decide whether to continue training; S36: If yes, return to S33 to adjust the weight and threshold of each BP neural network prediction model again; if no, use the test set to test the trained model.

5. According to claim 1, a building energy intelligent management and scheduling method based on digital twins is characterized by: The S5 comprises the following steps: The prediction module selects the corresponding BP neural network model according to the current season; The prediction module predicts the future energy consumption of the next interval time period t of the real target building area based on the current energy consumption.

6. The method for intelligent building energy management and scheduling based on digital twin according to claim 1 is characterized in that: The S6 comprises the following steps: The data analysis module sets preset energy consumption for each corresponding energy area in each quarter; The future energy consumption of each energy area is compared with the preset energy consumption, the consumption difference is calculated, and it is determined whether the consumption difference is within the error range. If it is within the error range, the digital twin display module displays the current energy consumption and future energy consumption; when the consumption difference is outside the error range, the emergency dispatch module obtains the information of the corresponding energy area and generates a dispatch task, creates corresponding dispatch task queues according to the positive and negative conditions of the consumption difference, and performs energy dispatch between the two corresponding dispatch task queues according to the first-come, first-served algorithm.

7. The method for intelligent management and scheduling of building energy based on digital twin according to claim 6 is characterized in that: The emergency dispatch module creates corresponding dispatch task queues according to the positive and negative conditions of the consumption difference, and performs energy dispatch between two corresponding dispatch task queues according to the first-come-first-served algorithm. The specific method is as follows: The emergency dispatch module creates two dispatch task queues, namely, a positive demand dispatch task queue and a negative demand dispatch task queue, according to whether the received energy area consumption difference is a positive energy demand or a negative energy demand, and arranges the priority of each dispatch task queue according to a first-come, first-served algorithm; Update the positive demand scheduling task queue and the negative demand scheduling task queue; Find the scheduling tasks corresponding to the adjacent energy areas in the positive demand scheduling task queue and the negative demand scheduling task queue respectively and pair them according to the first-come-first-served algorithm; The two paired energy areas are dispatched according to the minimum value of positive energy demand and negative energy demand; The scheduling tasks corresponding to the energy areas whose consumption difference after scheduling satisfies the condition that it is within the error range are eliminated from the corresponding queue; Determine whether the number of remaining scheduling tasks in the two scheduling task queues is greater than or equal to 1 at the same time. If both are greater than or equal to 1, return to update the corresponding scheduling task queue and reschedule; if both are 0, end the scheduling task, and the digital twin display module displays the current energy consumption and future energy consumption as well as the difference in scheduled consumption; if the number of scheduling tasks in the positive demand scheduling task queue is 0, apply for scheduling energy to the superior energy allocator, or if the number of scheduling tasks in the negative demand scheduling task queue is 0, save the negative energy demand to cope with the sudden energy demand in the next interval time period t.

8. The method for intelligent management and scheduling of building energy based on digital twin according to claim 4 is characterized in that: The method of using the gradient descent method and the back propagation method to adjust the weights and thresholds of each BP neural network prediction model comprises the following steps: First, the weights and thresholds of the neural network are randomly initialized. Then, using forward propagation, the historical energy consumption is input from the input layer through the hidden layer to the output layer, and the output of each layer is calculated. During the forward propagation process, the output of each layer is the input of the next layer. The calculation formula for forward propagation is: Z (l) =W (l) a (l-1) +b l a (l) =f(z (l) ); Among them, W (l) is the weight matrix of the lth layer, b (l) is the threshold vector of the lth layer, a (l-1 ) is the output of the l-1th layer and the input of the lth layer, Z (l) is the weighted sum of the lth layer, a (l) is the output of layer l, and f() is the activation function; In the output layer, the loss between the predicted historical energy consumption and the actual historical energy consumption is calculated. The loss function is calculated as: Among them, y i is the historical energy consumption of the real output, y i is the historical energy consumption of the predicted output, and N is the number of samples; Through the back propagation algorithm, starting from the output layer, the gradient of the loss function with respect to each weight and threshold is calculated layer by layer, and the weight and threshold are updated using the gradient descent method. The update formula is: Among them, W (l+1) is the updated weight, η is the learning rate, is the loss function J with respect to w (l) The partial derivative is the gradient, b (l+1) is the updated threshold, b (l) represents the bias of the lth layer, is the loss function with respect to the bias b (l) The offset number.

9. The method for intelligent building energy management and scheduling based on digital twins according to claim 4 is characterized in that: The method of using the DE algorithm to find the optimal values ​​of the weights and thresholds of each BP neural network prediction model includes the following steps: Initialize the historical energy consumption of the predicted output, input evolution parameters: sample number N, t is the number of iterations, the initial iteration number t = 0, and randomly generate initial sample data Evaluate all historical energy consumption and calculate Among them, f(X i (t)) is the target value of historical energy consumption, h is the difference in the number of selected samples; Reproduction and recombination, according to the adaptive selection, crossover and mutation operation methods of DE, for each energy consumption data X in the sample i (t), randomly generate three different random energy data where r1, r2, r3∈{1, 2, …, N}, j rand ∈{1, 2, …, n}; in, is the new value of the jth gene of individual i at iteration number t+1, F is the crossover factor F∈(0,1), The new value of gene j remains unchanged, that is, it is inherited from the old value at iteration number t, P c is the crossover probability; Select the optimal value of the initial weight and threshold of the output BP neural network model Among them, if the t+1 generation sample X i (t+1) satisfies the termination condition, then output X i The historical energy consumption of the predicted output with the minimum target in (t+1) is taken as the optimal individual, and its corresponding weight and threshold are the optimal values. If they are not satisfied, return to the second step to continue execution.

10. The method for intelligent building energy management and scheduling based on digital twin according to claim 1, characterized in that: After the S4 data acquisition module collects the current energy consumption in real time in each interval time period t, the data analysis module matches the collected current energy consumption to the corresponding energy area through the RFID electronic tag; then saves it to the corresponding area data set and stores it as historical energy consumption.

Images