Accurate prediction and distribution analysis method and device for building energy and medium

By using the historical behavior of building energy use objects and external trigger mechanism information to generate user portraits and energy use object groups, the problems of low prediction accuracy and insufficient flexibility in the existing technology are solved, and precise management of building energy consumption and energy are achieved. Improve efficiency.

CN119090669APending Publication Date: 2024-12-06山东浪潮智慧建筑科技有限公司
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411051345.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing technology has low prediction accuracy and lacks flexibility in building energy prediction and distribution analysis, making it difficult to meet the personalized needs of different users, resulting in serious energy waste.

Method used

By determining the energy consumption characteristic data set based on the historical behavior of the target energy consumption object and external trigger mechanism information, a user portrait is generated, and the energy consumption difference object group is divided according to the user's similarity and trigger event sequence, and accurate energy consumption prediction and distribution analysis are carried out. .

Benefits of technology

It has achieved all-round and precise management of building energy consumption, improved energy utilization efficiency, reduced energy waste, and met the personalized needs of different users.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090669A_ABST
    Figure CN119090669A_ABST
Patent Text Reader

Abstract

The invention provides a building energy consumption accurate prediction and distribution analysis method and device, and a medium, and belongs to the technical field of building energy consumption prediction. The method comprises the following steps: determining corresponding energy consumption characteristic data sets based on historical behaviors of target energy consumption objects and external trigger mechanism information; and according to the energy consumption feature data set and a preset portrait construction model, generating a user portrait of the target energy consumption object. Determining user similarities among the user portraits based on a preset matching rule, and taking the user similarities as first grouping parameters; and according to the energy consumption characteristic data set and a preset report event list, generating an energy consumption trigger event sequence, and determining a second grouping parameter based on the energy consumption trigger event sequence and an event sequence generation model. And according to the first grouping parameter and the second grouping parameter, dividing each target energy consumption object into an energy consumption difference object group. And based on each energy consumption difference object group, determining energy consumption data and energy consumption distribution information in a preset time, and sending the energy consumption data and the energy consumption distribution information to a corresponding management terminal.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of building energy consumption prediction, and in particular to a method, device and medium for accurate prediction and distribution analysis of building energy consumption. Background Art

[0002] With the acceleration of urbanization and the increase in commercial activities, the energy consumption problem of commercial buildings has become increasingly prominent. In addition, the energy consumption data of commercial buildings in terms of electricity, water, gas, heat, etc. is huge and complex, and there is a lack of effective prediction and analysis methods, which leads to serious energy waste and difficulty in improving energy efficiency.

[0003] Accurate prediction and distribution analysis of building energy consumption is an important part of building energy management. It helps building managers to formulate scientific energy use plans, optimize energy allocation, and improve energy efficiency, thereby achieving the goal of energy conservation and emission reduction. Most existing technologies rely on historical energy consumption data to predict and manage energy consumption and formulate usage plans for a period of time. This method has problems such as low prediction accuracy and lack of flexibility. It is too direct and lacks humanization. It is difficult to meet the personalized needs of different users. It cannot provide good constraints and services for different users such as enterprises and institutions in commercial buildings.

[0004] Therefore, there is an urgent need for a technical solution that can accurately predict and analyze the distribution of building energy consumption in an efficient and humane way. Summary of the invention

[0005] The embodiments of the present application provide a method, device and medium for accurate prediction and distribution analysis of building energy consumption, which are used to solve the problems of low prediction accuracy, lack of flexibility and difficulty in meeting the personalized needs of different users in the current methods of building energy consumption prediction and distribution analysis, so as to achieve all-round and accurate management of building energy consumption, improve energy utilization efficiency and reduce energy waste.

[0006] On the one hand, an embodiment of the present application provides a method for accurate prediction and distribution analysis of building energy consumption, which is applied to a system for accurate prediction and distribution analysis of building energy consumption, and the method includes:

[0007] Based on the historical behavior and external trigger mechanism information of each target energy-consuming object, determine the corresponding energy consumption characteristic data sets; wherein the historical behavior and external trigger mechanism information at least include the historical energy consumption data of the target energy-consuming object and the data of external influencing factors of energy consumption; the energy consumption characteristic data set at least includes the key indicators characterizing the energy consumption characteristics of the target energy-consuming object;

[0008] A model is constructed based on the energy usage characteristic data set and the preset portrait to generate a user portrait of the target energy user; the user portrait includes a number of energy usage dimension information associated with the target energy user; the energy usage dimension information includes at least energy usage habits, energy consumption sensitivity, and potential energy saving potential;

[0009] Based on a preset matching rule, determine the user similarity between the user portraits, so as to use the user similarity as a first grouping parameter;

[0010] Generate a corresponding energy usage triggering event sequence according to the energy usage feature data set and the preset reporting event list, and determine a second grouping parameter based on the energy usage triggering event sequence and a pre-trained event sequence generation model;

[0011] According to the first grouping parameter and the second grouping parameter, each of the target energy-consuming objects is divided into a corresponding energy-consuming difference object group;

[0012] Based on each of the energy usage difference object groups, corresponding energy usage data and energy consumption distribution information within a preset time are determined, and the energy usage data and the energy consumption distribution information are sent to a corresponding management terminal.

[0013] In one implementation of the present application, based on the historical behavior of each target energy-consuming object and the external trigger mechanism information, the corresponding energy-consuming characteristic data sets are determined, specifically including:

[0014] Standardizing the historical behavior and external trigger mechanism information corresponding to the target energy-consuming object;

[0015] Extracting features from the standardized historical energy consumption data and the energy consumption external influencing factor data to obtain corresponding historical behavior features and external trigger mechanism features;

[0016] A correlation analysis is performed on each of the historical behavior characteristics, each of the external trigger mechanism characteristics, and the historical energy consumption data, and a feature association network corresponding to each of the historical behavior characteristics and each of the external trigger mechanism characteristics is established based on the correlation analysis result and the correlation coefficient screening threshold; the feature association network includes a plurality of feature nodes and corresponding node edges; the node edges correspond to at least one of the following node relationships: direct relationship, indirect relationship, and interaction;

[0017] Determine the undetermined key indicator value corresponding to each of the feature nodes according to the feature association network and the preset feature weight list;

[0018] When the value of the undetermined key indicator is greater than the first preset threshold, the corresponding feature node is used as a key indicator and added to the corresponding energy consumption feature data set.

[0019] In one implementation of the present application, a model is constructed based on the energy usage feature data set and the preset portrait to generate a user portrait of the target energy user, specifically including:

[0020] Inputting the energy usage feature data set into the preset portrait construction model, so as to perform clustering processing on each of the key indicators through the preset portrait construction model to obtain one or more energy usage dimension information corresponding to each of the key indicators;

[0021] According to the classification standards corresponding to each of the energy consumption dimension information, each of the key indicators is mapped to a corresponding portrait label, so as to determine the user portrait according to each of the portrait labels.

[0022] In one implementation of the present application, based on a preset matching rule, determining the user similarity between the user portraits so as to use the user similarity as the first grouping parameter specifically includes:

[0023] By using the preset matching rule and according to the energy consumption dimension, calculating the portrait similarity between each pair of the user portraits;

[0024] Determining the user similarity according to the preset portrait weight corresponding to the energy consumption dimension and the similarity of each portrait;

[0025] The user similarity is used as the first grouping parameter corresponding to the corresponding two user portraits.

[0026] In one implementation of the present application, a corresponding energy usage triggering event sequence is generated according to the energy usage characteristic data set and the preset reporting event list, specifically including:

[0027] Comparing the energy usage characteristic data set with the specific conditions corresponding to the preset reporting event list to determine the triggering event corresponding to the energy usage characteristic data set according to the comparison result; the preset reporting event list comes from the user terminal corresponding to each of the target energy users;

[0028] Determine the energy consumption change data and duration corresponding to the trigger event according to the historical energy consumption data corresponding to each target energy-consuming object;

[0029] Determining an energy consumption impact value corresponding to the triggering event according to the energy consumption change data and the duration;

[0030] In the order of the energy usage impact degree values ​​from large to small, each of the trigger events is sequentially added to the energy usage trigger event sequence.

[0031] In one implementation of the present application, determining the second grouping parameter based on the energy triggering event sequence and a pre-trained event sequence generation model specifically includes:

[0032] Inputting the energy trigger event sequence into the pre-trained event sequence generation model; wherein the event sequence generation model is a GRU model trained by a number of energy trigger event sequence samples and grouping parameter labels;

[0033] Determine a predicted event sequence corresponding to a preset time period output by the event sequence generation model;

[0034] Matching the predicted event sequences corresponding to the target energy users to determine the similarity of the predicted events corresponding to the predicted event sequences;

[0035] The similarities of each predicted event are respectively used as the second grouping parameters corresponding to the two target energy users.

[0036] In an implementation of the present application, according to the first grouping parameter and the second grouping parameter, each of the target energy-consuming objects is divided into corresponding energy-consuming difference object groups, specifically including:

[0037] Calculating a sequence of energy consumption difference weighted values ​​corresponding to the target energy consumption object according to each of the first grouping parameters, each of the second grouping parameters and a preset weight tuple corresponding to the target energy consumption object;

[0038] According to the energy usage difference weighted value sequence and the second preset threshold, one or more target energy usage objects that match the energy usage characteristics corresponding to each target energy usage object are screened respectively, and each of the screened target energy usage objects that match the energy usage characteristics are added to the same energy usage difference object group.

[0039] In one implementation of the present application, based on each of the energy usage difference object groups, determining the corresponding energy usage data and energy consumption distribution information within a preset time, and sending the energy usage data and the energy consumption distribution information to the corresponding management terminal, specifically includes:

[0040] According to the energy usage difference object groups, determine the energy usage statistical indicators of all the target energy usage objects in each energy usage difference object group within a preset time, so as to add the energy usage statistical indicators to the energy usage chart to obtain the energy usage data; wherein the energy usage statistical indicators include: total energy usage, average energy usage, maximum energy usage and minimum energy usage;

[0041] Determine the energy consumption distribution information within the preset time according to the historical energy consumption data corresponding to any target energy-consuming object in the energy-consuming difference object group; the energy consumption distribution information includes energy consumption conditions in different peak time periods;

[0042] The energy consumption data and the energy consumption distribution information are sent to a corresponding management terminal.

[0043] On the other hand, the embodiment of the present application further provides a device for accurately predicting and analyzing the energy consumption of a building, the device comprising:

[0044] At least one processor; and a memory in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for accurate prediction and distribution analysis of building energy consumption.

[0045] On the other hand, an embodiment of the present application further provides a non-volatile computer storage medium storing computer executable instructions, characterized in that the computer executable instructions can execute the above-mentioned method for accurate prediction and distribution analysis of building energy consumption.

[0046] Compared with the prior art, the present invention has the following significant effects:

[0047] Through the above scheme, this application can make full use of the historical behavior of the target energy-consuming object and the external trigger mechanism information for portrait analysis, and analyze the triggering events that affect energy consumption according to the reported events, so as to divide each target energy-consuming object into a group of energy-consuming difference objects with close energy consumption differences. In this way, the personalized energy consumption characteristics of the target energy-consuming object can be fully utilized to conduct building energy consumption prediction and distribution analysis. It can meet the personalized needs of different users, realize all-round and precise management of building energy consumption, improve energy utilization efficiency, and reduce energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0049] Figure 1 A schematic diagram of a flow chart of a method for accurate prediction and distribution analysis of building energy consumption in an embodiment of the present application;

[0050] Figure 2 A schematic diagram of a feature association network in a method for accurate prediction and distribution analysis of building energy consumption in an embodiment of the present application;

[0051] Figure 3 This is a schematic diagram of the structure of a device for accurate prediction and distribution analysis of building energy consumption in an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0053] The embodiments of the present application provide a method, device and medium for accurate prediction and distribution analysis of building energy consumption, which are used to solve the problems of low prediction accuracy, lack of flexibility and difficulty in meeting the personalized needs of different users in the current methods of prediction and distribution analysis of building energy consumption, so as to achieve all-round and accurate management of building energy consumption, improve energy utilization efficiency and reduce energy waste.

[0054] The following describes in detail various embodiments of the present application in conjunction with the accompanying drawings.

[0055] The present application embodiment provides a method for accurate prediction and distribution analysis of building energy consumption, which is applied to a system for accurate prediction and distribution analysis of building energy consumption. Figure 1 As shown, the method may include steps S101-S106:

[0056] S101, the server determines corresponding energy usage characteristic data sets based on the historical behavior of each target energy-consuming object and external trigger mechanism information.

[0057] The historical behavior and external trigger mechanism information at least includes the historical energy consumption data of the target energy-consuming object and the data of external factors affecting energy consumption. The energy consumption characteristic data set at least includes key indicators that characterize the energy consumption characteristics of the target energy-consuming object.

[0058] It should be noted that the server, as the executor of the building energy accurate prediction and distribution analysis method, is only an example, and the executor is not limited to the server, and this application does not make specific restrictions on this. When the executor is a server, the building energy accurate prediction and distribution analysis system can be placed on the server for operation.

[0059] The target energy-using object can be understood as various shops, enterprises and other POI points of interest in commercial buildings. The historical behavior and external trigger mechanism information of the target energy-using object can be understood as a series of regular and identifiable behaviors and energy consumption habits of the target energy-using object (such as enterprise employees) in daily work, such as fixed work and rest, work habits, lighting habits, air conditioning usage habits, office equipment usage habits, etc. Among them, the historical behavior mainly includes the main energy-using activities and behaviors of users in the target energy-using object in the building, and the external trigger mechanism mainly includes the external factors that affect the user's energy use, such as weather, holidays, social and economic activities, and traffic conditions around the building.

[0060] In the embodiment of the present application, based on the historical behavior of each target energy-consuming object and the external trigger mechanism information, the corresponding energy-consuming feature data sets are determined, specifically including:

[0061] The historical behavior and external trigger mechanism information corresponding to the target energy-consuming object are standardized. The standardized historical energy consumption data and energy consumption external influencing factor data are subjected to feature extraction to obtain the corresponding historical behavior features and external trigger mechanism features. Correlation analysis is performed on each historical behavior feature, each external trigger mechanism feature, and historical energy consumption data, and a feature association network corresponding to each historical behavior feature and each external trigger mechanism feature is established based on the correlation analysis results and the correlation coefficient screening threshold. The feature association network includes multiple feature nodes and corresponding node edges. The node edge corresponds to at least one of the following node relationships: direct relationship, indirect relationship, and interaction. According to the feature association network and the preset feature weight list, the pending key indicator values ​​corresponding to each feature node are determined. When the pending key indicator value is greater than the first preset threshold, the corresponding feature node is used as a key indicator and added to the corresponding energy consumption feature data set.

[0062] In other words, the application can extract key indicators related to historical energy consumption data from historical behavior and external trigger mechanism information, so as to carry out subsequent accurate energy consumption prediction and distribution analysis. Specifically, the server can standardize the historical behavior and external trigger mechanism information. The standardization process can be to subtract the original data in the historical behavior and external trigger mechanism information from the corresponding data mean, and divide the difference by the data standard deviation corresponding to the original data to obtain the standardized data. This eliminates the influence of different dimensions and magnitudes on data analysis.

[0063] Furthermore, the server can extract features from the standardized data. Feature extraction of historical energy consumption data can include trend analysis, periodic analysis, and peak-valley period identification. Trend analysis includes calculating the moving average or exponential smoothing of historical energy consumption data. Periodic analysis includes using Fourier transform or seasonal decomposition (such as STL decomposition) to identify periodic patterns. Peak-valley period identification includes determining the peak and trough periods of energy consumption through clustering or threshold methods, thereby extracting historical behavior characteristics. Feature extraction of external factors affecting energy consumption includes extracting environmental factors, socio-economic factors, holiday effects, etc., such as temperature and humidity in environmental factors, socio-economic factors such as GDP growth rate, electricity price changes, etc. Holiday effects, such as analyzing the designed holiday virtual variables to obtain the special impact of holidays on energy consumption (such as a significant increase or decrease in energy consumption), thereby obtaining external trigger mechanism characteristics.

[0064] The server will use historical behavior features, external trigger mechanism features and historical energy consumption data to perform correlation analysis. The correlation analysis can be completed by calculating the Pearson correlation coefficient. The server can compare the correlation analysis results, that is, the correlation coefficients between historical behavior features, external trigger mechanism features and historical energy consumption data, with the pre-set correlation coefficient screening threshold, screen historical behavior features and external trigger mechanism features with correlation coefficients greater than the correlation coefficient screening threshold, and use historical behavior features and external trigger mechanism features as feature nodes to determine the relationship between them and historical energy consumption data, including direct relationship, indirect relationship, interaction, etc. Direct relationship can be understood as some features may be directly related to energy consumption, such as the positive correlation between temperature and air conditioning energy consumption. Indirect relationship means that some features may indirectly affect energy consumption by affecting other features. For example, holidays may affect energy consumption by changing the activity pattern of personnel. Interaction can be understood as there may be interactions between features, that is, the impact of two or more features changing at the same time on energy consumption is not the simple sum of their respective impacts.

[0065] Feature association network Figure 2 As shown, if the feature node A1 of the historical behavior feature is directly related to the power consumption B, and the feature node C1 of the external trigger feature can affect the historical behavior feature A1, then C1 is an indirect relationship (the dotted line indicates an indirect relationship), the feature node A2 of the historical behavior feature and the feature node C2 of the external trigger feature interact with the power consumption B, and both affect the power consumption at the same time.

[0066] A preset feature weight list is pre-set in the server. Different feature nodes in the feature association network can be associated with different weights in the preset feature weight list. The value of the node edge of each feature node in the feature association network is the value of the correlation coefficient. The server can calculate the value of the pending key indicator by multiplying the correlation coefficient and the feature weight corresponding to the feature node. When the server determines that the value of the pending key indicator is greater than the first preset threshold, the characteristics of one or more feature nodes corresponding to the pending key indicator value are used as key indicators and added to the energy consumption feature data set. This application generates an energy consumption feature data set for each target energy-consuming object. The above-mentioned first preset threshold is set by the user according to actual needs, and this application does not make specific limitations on this.

[0067] Through the above scheme, the present application can screen out key indicators related to energy consumption from historical behavior characteristics and external trigger mechanism characteristics, reduce irrelevant calculations, and improve prediction and analysis efficiency.

[0068] S102: The server builds a model based on the energy usage feature data set and the preset portrait to generate a user portrait of the target energy user.

[0069] The user profile includes several energy consumption dimension information associated with the target energy user, which at least includes energy consumption habits, energy consumption sensitivity, and potential energy saving potential.

[0070] In the embodiment of the present application, a model is constructed based on an energy usage feature data set and a preset portrait to generate a user portrait of a target energy user, specifically including:

[0071] The energy consumption feature data set is input into the preset portrait construction model to cluster each key indicator through the preset portrait construction model to obtain one or more energy consumption dimension information corresponding to each key indicator. According to the classification standards corresponding to each energy consumption dimension information, each key indicator is mapped to a corresponding portrait label to determine the user portrait according to each portrait label.

[0072] The preset portrait construction model can be a k-means clustering algorithm, which can cluster the key indicators in the energy consumption feature data set to obtain the energy consumption dimension information contained in each cluster after clustering. The energy consumption dimension information can be preset by the user, and this application does not make specific restrictions on this. The user can pre-set a classification standard for each energy consumption dimension information, and the server maps the key indicators to the portrait label through the classification standard to obtain the user portrait. The classification standard can include different portrait label pools associated with different energy consumption dimensions.

[0073] Through the above scheme, the present application can further analyze the energy consumption characteristic data set and cluster the key indicators so as to obtain energy consumption dimension information related to the target energy-consuming object, which is more in line with the properties of the target energy-consuming object itself, so as to obtain personalized energy consumption information of the target energy-consuming object.

[0074] S103: The server determines the user similarity between the user portraits based on a preset matching rule, and uses the user similarity as the first grouping parameter.

[0075] In the embodiment of the present application, based on the preset matching rule, the user similarity between the user portraits is determined, so as to use the user similarity as the first grouping parameter, which specifically includes:

[0076] By using the preset matching rules and according to the energy consumption dimension, the portrait similarity between each user portrait is calculated. The user similarity is determined according to the preset portrait weight corresponding to the energy consumption dimension and the similarity of each portrait. The user similarity is used as the first grouping parameter corresponding to the corresponding two user portraits.

[0077] That is to say, the server can calculate the similarity of user portraits in multiple energy consumption dimensions by presetting matching rules. This application can encode each user portrait to obtain a portrait set corresponding to the user portrait, and then calculate the portrait similarity between the portrait vectors. The portrait similarity can be the Jaccard coefficient, which measures the similarity of two portrait vectors. The formula is as follows:

[0078]

[0079] Similarity is the portrait similarity, A is the portrait set of the first user portrait, and B is the portrait set of the second user portrait.

[0080] At the same time, the application can pre-set different preset portrait weights for different energy consumption dimensions, for example, the energy consumption habit weight is 0.4, the energy consumption sensitivity is 0.3, the potential energy saving potential is 0.3... Calculate the product value of the preset portrait weight and the corresponding portrait similarity to obtain the user similarity corresponding to the two target energy consumption objects. The server uses the user similarity as the first grouping parameter.

[0081] S104, the server generates a corresponding energy usage triggering event sequence according to the energy usage feature data set and the preset reporting event list, and determines a second grouping parameter based on the energy usage triggering event sequence and a pre-trained event sequence generation model.

[0082] In the embodiment of the present application, a corresponding energy usage triggering event sequence is generated according to the energy usage characteristic data set and the preset reporting event list, specifically including:

[0083] Compare the energy consumption characteristic data set with the specific conditions corresponding to the preset reporting event list to determine the trigger event corresponding to the energy consumption characteristic data set based on the comparison result. The preset reporting event list comes from the user terminal corresponding to each target energy consumption object. According to the historical energy consumption data corresponding to each target energy consumption object, determine the energy consumption change data and duration corresponding to the trigger event. According to the energy consumption change data and duration, determine the energy consumption impact value corresponding to the trigger event. In the order of energy consumption impact value from large to small, add each trigger event to the energy consumption trigger event sequence.

[0084] That is to say, the server can collect preset reporting events from user terminals and generate a preset reporting event list. The reporting events in the preset reporting event list can be understood as a series of events that trigger energy consumption behaviors. The events may cause changes in the original historical behaviors and external trigger mechanism information of the target energy-consuming objects, such as policy changes, energy-saving publicity, etc. The specific content of the preset reporting event list can be set by the user according to the actual scenario, and this application does not make specific restrictions on this. The preset reporting event list may include specific conditions for different key indicators of the energy consumption feature data set. The specific condition may be a threshold. If the value corresponding to the key indicator is greater than the threshold, the reporting event corresponding to the key indicator is determined to be triggered. Furthermore, the server determines the energy consumption change data and duration of the trigger event history from the historical energy consumption data. The energy consumption change data can be understood as the difference change data between the energy consumption in the historical time and the energy consumption data when the trigger event did not occur, and the duration of the energy consumption difference change caused by it. The server may compare the energy consumption change data with the preset difference data samples, determine the first impact of the corresponding preset difference data samples, and determine the second impact corresponding to the duration of the corresponding trigger event from the preset impact list, and obtain the energy consumption impact value of the trigger event according to the product value of the first impact and the second impact. Subsequently, the server will add the above trigger events to the energy consumption trigger event sequence in the order of the energy consumption impact values.

[0085] Therefore, various triggering events having different degrees of influence on the target energy-consuming object can be arranged in order, so as to obtain the energy consumption data of the target energy-consuming object when it is affected by different triggering events.

[0086] In the embodiment of the present application, the second grouping parameter is determined based on the energy triggering event sequence and the pre-trained event sequence generation model, specifically including:

[0087] The server inputs the energy trigger event sequence into a pre-trained event sequence generation model. The event sequence generation model is a GRU model trained by a number of energy trigger event sequence samples and grouping parameter labels. Determine the predicted event sequence corresponding to the preset time period output by the event sequence generation model. Match the predicted event sequences corresponding to each target energy user object to determine the similarity of the predicted events corresponding to each predicted event sequence. Use each predicted event similarity as the second grouping parameter corresponding to the two target energy users.

[0088] The event sequence generation model may be a GRU (Gated Recurrent Unit) model, and the server may pre-train the model with a number of energy-triggering event sequence samples labeled with grouping parameter labels until the model output accuracy is greater than a preset third preset threshold, thereby obtaining a trained event sequence generation model. The energy-triggering event sequence is processed by the event sequence generation model, and the predicted event sequence within a subsequent preset time period is predicted, and the predicted event sequences of each target energy-using object are matched, thereby obtaining the predicted event similarity between each target energy-using object based on the triggering event, and obtaining the second grouping parameter.

[0089] Through the above scheme, the similarity between the target energy-consuming objects in the triggering event dimension can be obtained, and the target energy-consuming objects can be divided more accurately.

[0090] S105: The server divides each target energy-consuming object into a corresponding energy-consuming difference object group according to the first grouping parameter and the second grouping parameter.

[0091] In the embodiment of the present application, each target energy-consuming object is divided into a corresponding energy-consuming difference object group according to the first grouping parameter and the second grouping parameter, specifically including:

[0092] According to each first grouping parameter, each second grouping parameter and preset weight tuple corresponding to the target energy-using object, calculate the energy usage difference weighted value sequence corresponding to the target energy-using object. According to the energy usage difference weighted value sequence and the second preset threshold, select one or more target energy-using objects that match the energy usage characteristics corresponding to each target energy-using object, and add the selected target energy-using objects that match the energy usage characteristics to the same energy usage difference object group.

[0093] That is to say, the server can calculate the energy difference weighted value through the first grouping parameter, the second grouping parameter and the preset weight tuple. For example, the first grouping parameter of the target energy user a includes {x1, x2, x3, ..., x n}, x n represents the first grouping parameter between the target energy-consuming object a and the nth target energy-consuming object, and the second grouping parameter includes {y1, y2, y3, ..., y n}, the preset weight tuple is {w1, w2}, where w1 is the parameter weight of the first grouping parameter, w2 is the parameter weight of the second grouping parameter, and w1+w2=1. The energy difference weighted value calculation formula is as follows:

[0094] CY i =w1*x i +w2*y i

[0095] Among them, CY iis the weighted value of the energy consumption difference between target energy-consuming object a and the i-th target energy-consuming object, 0 <i<n+1。

[0096] Subsequently, the server adds the energy usage difference weighted values ​​calculated by the calculation to the energy usage difference weighted value sequence in order of magnitude, and uses the preset second preset threshold to filter the energy usage difference weighted values, and can filter the energy usage difference weighted values ​​greater than the second preset threshold. And based on the filtered energy usage difference weighted values, one or more target energy usage objects matching the energy usage characteristics of the target energy usage objects are obtained, and the energy usage difference object groups are divided.

[0097] S106: The server determines the corresponding energy usage data and energy consumption distribution information within a preset time based on each energy usage difference object group, and sends the energy usage data and energy consumption distribution information to the corresponding management terminal.

[0098] In the embodiment of the present application, based on each energy difference object group, determining the corresponding energy consumption data and energy consumption distribution information within a preset time, and sending the energy consumption data and energy consumption distribution information to the corresponding management terminal specifically includes:

[0099] According to the energy consumption difference object group, determine the energy consumption statistical indicators of all target energy consumption objects in each energy consumption difference object group within the preset time, so as to add the energy consumption statistical indicators to the energy consumption chart to obtain energy consumption data. Among them, the energy consumption statistical indicators include: total energy consumption, average energy consumption, maximum energy consumption and minimum energy consumption. According to the historical energy consumption data corresponding to any target energy consumption object in the energy consumption difference object group, determine the energy consumption distribution information within the preset time. The energy consumption distribution information includes the energy consumption in different peak time periods. Send the energy consumption data and energy consumption distribution information to the corresponding management terminal.

[0100] In other words, the present application can count the energy consumption statistical indicators in a preset time according to the energy consumption difference object group, and display the energy consumption data in the form of an energy consumption chart. At the same time, the historical energy consumption data of any target energy consumption object in the energy consumption difference object group is used as a standard to analyze the energy consumption distribution information in the preset time, and the energy consumption distribution information is used as the energy consumption distribution information of the energy consumption difference object group. Subsequently, the server sends the energy consumption data and energy consumption distribution information to the management terminal, so that the management terminal performs energy management and energy consumption strategy optimization according to the energy consumption difference object group. The management terminal can be a mobile phone, computer or other device of the electricity department or building management personnel, and the present application does not make specific restrictions on this.

[0101] In addition, the present application can obtain the first grouping parameters and the second grouping parameters of each target energy-consuming object and each other target energy-consuming object in the energy-consuming difference object group in real time, thereby updating the energy-consuming difference object group.

[0102] Through the above scheme, this application can make full use of the historical behavior of the target energy-consuming object and the external trigger mechanism information for portrait analysis, and analyze the triggering events that affect energy consumption according to the reported events, so as to divide each target energy-consuming object into a group of energy-consuming difference objects with close energy consumption differences. In this way, the personalized energy consumption characteristics of the target energy-consuming object can be fully utilized to conduct building energy consumption prediction and distribution analysis. It can meet the personalized needs of different users, realize all-round and precise management of building energy consumption, improve energy utilization efficiency, and reduce energy waste.

[0103] Figure 3 A schematic diagram of a device for accurately predicting and analyzing building energy consumption provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the device includes:

[0104] At least one processor; and a memory in communication with the at least one processor. The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0105] Based on the historical behavior and external trigger mechanism information of each target energy-consuming object, the corresponding energy-using characteristic data sets are determined. Among them, the historical behavior and external trigger mechanism information at least include the historical energy consumption data of the target energy-consuming object and the data of external influencing factors of energy consumption. The energy-using characteristic data set at least includes key indicators that characterize the energy consumption characteristics of the target energy-using object. A model is constructed based on the energy-using characteristic data set and the preset portrait to generate a user portrait of the target energy-using object. The user portrait includes several energy-using dimension information associated with the target energy-using object. The energy-using dimension information at least includes energy-using habits, energy-using sensitivity, and potential energy-saving potential. Based on the preset matching rules, the user similarity between each user portrait is determined, and the user similarity is used as the first grouping parameter. According to the energy-using characteristic data set and the preset reporting event list, a corresponding energy-using trigger event sequence is generated, and a second grouping parameter is determined based on the energy-using trigger event sequence and the pre-trained event sequence generation model. According to the first grouping parameter and the second grouping parameter, each target energy-using object is divided into a corresponding energy-using difference object group. Based on each energy-using difference object group, the corresponding energy-using data and energy-using distribution information within a preset time are determined, and the energy-using data and energy-using distribution information are sent to the corresponding management terminal.

[0106] The embodiment of the present application further provides a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as follows:

[0107] Based on the historical behavior and external trigger mechanism information of each target energy-consuming object, the corresponding energy-using characteristic data sets are determined. Among them, the historical behavior and external trigger mechanism information at least include the historical energy consumption data of the target energy-consuming object and the data of external influencing factors of energy consumption. The energy-using characteristic data set at least includes key indicators that characterize the energy consumption characteristics of the target energy-using object. A model is constructed based on the energy-using characteristic data set and the preset portrait to generate a user portrait of the target energy-using object. The user portrait includes several energy-using dimension information associated with the target energy-using object. The energy-using dimension information at least includes energy-using habits, energy-using sensitivity, and potential energy-saving potential. Based on the preset matching rules, the user similarity between each user portrait is determined, and the user similarity is used as the first grouping parameter. According to the energy-using characteristic data set and the preset reporting event list, a corresponding energy-using trigger event sequence is generated, and a second grouping parameter is determined based on the energy-using trigger event sequence and the pre-trained event sequence generation model. According to the first grouping parameter and the second grouping parameter, each target energy-using object is divided into a corresponding energy-using difference object group. Based on each energy-using difference object group, the corresponding energy-using data and energy-using distribution information within a preset time are determined, and the energy-using data and energy-using distribution information are sent to the corresponding management terminal.

[0108] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0109] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0110] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0111] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for accurate prediction and distribution analysis of building energy consumption, characterized in that: The method is applied to a building energy consumption accurate prediction and distribution analysis system; the method comprises: Based on the historical behavior and external trigger mechanism information of each target energy-consuming object, determine the corresponding energy consumption characteristic data sets; wherein the historical behavior and external trigger mechanism information at least include the historical energy consumption data of the target energy-consuming object and the data of external influencing factors of energy consumption; the energy consumption characteristic data set at least includes the key indicators characterizing the energy consumption characteristics of the target energy-consuming object; A model is constructed based on the energy usage characteristic data set and the preset portrait to generate a user portrait of the target energy user; the user portrait includes a number of energy usage dimension information associated with the target energy user; the energy usage dimension information includes at least energy usage habits, energy consumption sensitivity, and potential energy saving potential; Based on a preset matching rule, determine the user similarity between the user portraits, so as to use the user similarity as a first grouping parameter; Generate a corresponding energy usage triggering event sequence according to the energy usage feature data set and the preset reporting event list, and determine a second grouping parameter based on the energy usage triggering event sequence and a pre-trained event sequence generation model; According to the first grouping parameter and the second grouping parameter, each of the target energy-consuming objects is divided into a corresponding energy-consuming difference object group; Based on each of the energy usage difference object groups, corresponding energy usage data and energy consumption distribution information within a preset time are determined, and the energy usage data and the energy consumption distribution information are sent to a corresponding management terminal.

2. A method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: Based on the historical behavior and external trigger mechanism information of each target energy-consuming object, the corresponding energy-consuming characteristic data sets are determined, including: Standardizing the historical behavior and external trigger mechanism information corresponding to the target energy-consuming object; Extracting features from the standardized historical energy consumption data and the energy consumption external influencing factor data to obtain corresponding historical behavior features and external trigger mechanism features; A correlation analysis is performed on each of the historical behavior characteristics, each of the external trigger mechanism characteristics, and the historical energy consumption data, and a feature association network corresponding to each of the historical behavior characteristics and each of the external trigger mechanism characteristics is established based on the correlation analysis result and the correlation coefficient screening threshold; the feature association network includes a plurality of feature nodes and corresponding node edges; the node edges correspond to at least one of the following node relationships: direct relationship, indirect relationship, and interaction; Determine the undetermined key indicator value corresponding to each of the feature nodes according to the feature association network and the preset feature weight list; When the value of the undetermined key indicator is greater than the first preset threshold, the corresponding feature node is used as a key indicator and added to the corresponding energy consumption feature data set.

3. A method for accurate prediction and distribution analysis of building energy consumption according to claim 2, characterized in that: The model is constructed based on the energy usage feature data set and the preset portrait to generate a user portrait of the target energy user, specifically including: Inputting the energy usage feature data set into the preset portrait construction model, so as to perform clustering processing on each of the key indicators through the preset portrait construction model to obtain one or more energy usage dimension information corresponding to each of the key indicators; According to the classification standards corresponding to each of the energy consumption dimension information, each of the key indicators is mapped to a corresponding portrait label, so as to determine the user portrait according to each of the portrait labels.

4. A method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: Based on a preset matching rule, determining the user similarity between the user portraits, so as to use the user similarity as a first grouping parameter, specifically includes: By using the preset matching rule and according to the energy consumption dimension, calculating the portrait similarity between each pair of the user portraits; Determining the user similarity according to the preset portrait weight corresponding to the energy consumption dimension and the similarity of each portrait; The user similarity is used as the first grouping parameter corresponding to the corresponding two user portraits.

5. The method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: Generate a corresponding energy usage triggering event sequence according to the energy usage characteristic data set and the preset reporting event list, specifically including: Comparing the energy usage characteristic data set with the specific conditions corresponding to the preset reporting event list to determine the triggering event corresponding to the energy usage characteristic data set according to the comparison result; the preset reporting event list comes from the user terminal corresponding to each of the target energy users; Determine the energy consumption change data and duration corresponding to the trigger event according to the historical energy consumption data corresponding to each target energy-consuming object; Determining an energy consumption impact value corresponding to the triggering event according to the energy consumption change data and the duration; In the order of the energy usage impact degree values ​​from large to small, each of the trigger events is sequentially added to the energy usage trigger event sequence.

6. The method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: Determining the second grouping parameter based on the energy triggering event sequence and the pre-trained event sequence generation model specifically includes: Inputting the energy trigger event sequence into the pre-trained event sequence generation model; wherein the event sequence generation model is a GRU model trained by a number of energy trigger event sequence samples and grouping parameter labels; Determine a predicted event sequence corresponding to a preset time period output by the event sequence generation model; Matching the predicted event sequences corresponding to the target energy users to determine the similarity of the predicted events corresponding to the predicted event sequences; The similarities of each predicted event are respectively used as the second grouping parameters corresponding to the two target energy users.

7. The method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: According to the first grouping parameter and the second grouping parameter, each of the target energy-consuming objects is divided into corresponding energy-consuming difference object groups, specifically including: Calculating a sequence of energy consumption difference weighted values ​​corresponding to the target energy consumption object according to each of the first grouping parameters, each of the second grouping parameters and a preset weight tuple corresponding to the target energy consumption object; According to the energy usage difference weighted value sequence and the second preset threshold, one or more target energy usage objects that match the energy usage characteristics corresponding to each target energy usage object are screened respectively, and each of the screened target energy usage objects that match the energy usage characteristics are added to the same energy usage difference object group.

8. The method for accurate prediction and distribution analysis of building energy consumption according to claim 1, characterized in that: Based on each of the energy usage difference object groups, determining corresponding energy usage data and energy consumption distribution information within a preset time, and sending the energy usage data and the energy consumption distribution information to a corresponding management terminal, specifically includes: According to the energy usage difference object groups, determine the energy usage statistical indicators of all the target energy usage objects in each energy usage difference object group within a preset time, so as to add the energy usage statistical indicators to the energy usage chart to obtain the energy usage data; wherein the energy usage statistical indicators include: total energy usage, average energy usage, maximum energy usage and minimum energy usage; Determine the energy consumption distribution information within the preset time according to the historical energy consumption data corresponding to any target energy-consuming object in the energy-consuming difference object group; the energy consumption distribution information includes energy consumption conditions in different peak time periods; The energy consumption data and the energy consumption distribution information are sent to a corresponding management terminal.

9. A device for accurate prediction and distribution analysis of building energy consumption, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for accurate prediction and distribution analysis of building energy consumption as described in any one of claims 1 to 8 above.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute a method for accurate prediction and distribution analysis of building energy consumption as described in any one of claims 1 to 8 above.

Citation Information

Cited By

  • AI-based user energy consumption portrait analysis method and system

    CN119295141A