Abnormal energy consumption diagnosis method and system based on short-term building energy consumption prediction model

Through the method based on the short-term building energy consumption prediction model, Pearson correlation analysis and multi-input and multi-output strategy optimize long and short-term memory networks, the problem of difficulty in diagnosis of abnormal energy consumption on a small time scale in the existing technology is solved, and the rapid development of the rapid development of the existing technology is achieved. , accurate diagnosis, and support building operation optimization.

CN114169254BActive Publication Date: 2024-11-08ZHONGYIFENG CONSTR GRP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111648640.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-29
Publication Date
2024-11-08
Estimated Expiration
2041-12-29

AI Technical Summary

Technical Problem

The existing methods for diagnosing energy consumption of building abnormalities are difficult to effectively carry out on a small time scale and cannot meet the needs of actual operational optimization.

Method used

Using a method based on short-term building energy consumption prediction model, features are screened through Pearson correlation analysis, energy consumption sample set is constructed and divided into training set, verification set and test set, and long-term short-term memory network is optimized using multi-input and multi-output strategies to establish supervision Multi-step advance short-term energy consumption prediction model, and abnormal energy consumption diagnosis is carried out on an hourly time scale.

Benefits of technology

It realizes fast and accurate abnormal energy consumption diagnosis on a smaller time scale, improves diagnosis speed and accuracy, has good versatility and practicality, and can effectively support the operational optimization of the building.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114169254B_ABST
    Figure CN114169254B_ABST
Patent Text Reader

Abstract

The invention discloses an abnormal energy consumption diagnosis method and system based on a short-term building energy consumption prediction model. After Pearson correlation analysis and data preprocessing, an energy consumption sample set is constructed and divided into a training set, a verification set and a test set. An LSTM network multi-step advance short-term energy consumption prediction model based on 10 input parameters such as historical energy consumption data, occupant behavior, meteorological factors and time factors is proposed. The MIMO strategy is adopted to predict the hourly energy consumption of the building under supervised learning mode. According to the prediction model, a diagnosis method based on the mathematical relationship between the predicted value and the actual value is proposed on a smaller time scale. The invention has the advantages of fast diagnosis speed, high diagnosis accuracy and strong practicality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of building energy consumption prediction, and in particular relates to an abnormal energy consumption diagnosis method and system based on a short-term building energy consumption prediction model. Background Art

[0002] Currently, the construction industry accounts for 39% and 38% of the total global energy consumption and greenhouse gas emissions, respectively. Compared with the transportation and industrial sectors, the energy-saving potential of buildings is much greater. At the same time, with the gradual increase in building area and energy consumption in recent years, energy-saving research has become an important research direction. Fast and accurate energy consumption forecasting can provide data for optimizing operations and help China achieve the peak of carbon dioxide emissions in 2030.

[0003] The role of building energy consumption forecasting varies depending on the forecast period. Long-term forecasts (e.g., more than one year) are usually used for energy maintenance planning and power distribution; medium-term forecasts (e.g., months, years) are mainly used for predictive maintenance and component operation mode determination; short-term forecasts (e.g., sub-hourly, hourly, or daily) are usually used for predictive model control, fault detection, and control optimization. Short-term building energy consumption forecasting has aroused great interest among building professionals because it is closely related to the daily operation of various service systems.

[0004] Existing abnormal energy consumption diagnosis of buildings is usually based on a daily or monthly basis, which is challenging on a smaller time scale and cannot well meet the needs of actual optimized operations. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an abnormal energy consumption diagnosis method and system based on a short-term building energy consumption prediction model in view of the deficiencies in the above-mentioned prior art, so as to solve the problem that abnormal energy consumption diagnosis cannot be performed on a smaller time scale.

[0006] The present invention adopts the following technical solutions:

[0007] The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model includes the following steps:

[0008] S1. Collect historical building energy consumption data and data on factors affecting energy consumption, and select features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model through Pearson correlation analysis;

[0009] S2, preprocessing the input feature data obtained in step S1, constructing an energy consumption sample set, and dividing the energy consumption sample set into a training set, a verification set, and a test set;

[0010] S3, using multi-input multi-output strategy to optimize long short-term memory network and establish a supervised multi-step ahead short-term energy consumption prediction model;

[0011] S4, using the training set data obtained in step S2 to train the supervised multi-step-ahead short-term energy consumption prediction model established in step S3, using the validation set obtained in step S2 to optimize the hyperparameters of the trained supervised multi-step-ahead short-term energy consumption prediction model, to obtain an optimized short-term building energy consumption prediction model, and using a portion of the test set obtained in step S2 to evaluate the short-term building energy consumption prediction model;

[0012] S5. Based on the short-term building energy consumption prediction model evaluated in step S4, determine an abnormal energy consumption diagnosis method on an hourly time scale;

[0013] S6. Perform energy consumption diagnosis on the remaining test sets obtained in step S2 using the abnormal energy consumption diagnosis method determined in step S5, and perform abnormal energy consumption analysis based on the diagnosis results.

[0014] Specifically, in step S1, the data affecting the energy consumption factors include occupant behavior, meteorological factors and time factors.

[0015] Specifically, in step S1, the Pearson correlation coefficient is calculated as follows:

[0016]

[0017] Among them, ρ X,Y is the correlation coefficient between variables X and Y, and E is the mathematical expectation of the sample.

[0018] Specifically, in step S2, the energy consumption sample set is divided into a training set, a validation set and a test set according to a ratio of 6:2:2.

[0019] Furthermore, the test set is divided into two parts at a ratio of 1:1, which are used for evaluating the short-term building energy consumption prediction model in step S4 and for energy consumption diagnosis in step S6.

[0020] Specifically, in step S4, the short-term building energy consumption prediction model when the coefficient of variation of the root mean square error at an hourly time scale is less than 30% is used for energy consumption diagnosis.

[0021] Furthermore, the coefficient of variation of the root mean square error CV-RMSE is specifically:

[0022]

[0023] Among them, y i and are the actual value and predicted value at the i-th moment, respectively, and n is the number of samples.

[0024] Specifically, in step S5, the abnormal energy consumption diagnosis method is specifically as follows:

[0025] S501, considering the data obtained by the prediction model as energy-saving data, and the actual data as non-energy-saving data;

[0026] S502, using the residual ε to represent the difference between the actual energy consumption value and the predicted value, and using the mean absolute error MAE to represent the mean absolute error when the actual value is greater than the predicted value in the test set used to evaluate the performance of the prediction model;

[0027] S503: According to the evaluation criteria, the energy usage is divided into an energy-saving state, a cautionary state and a non-energy-saving state.

[0028] Furthermore, in step S503, the evaluation criteria are:

[0029] When ε≤0, it is an energy-saving state; when 0<ε≤MAE, it is an attention state; and when ε>MAE, it is a non-energy-saving state.

[0030] Another technical solution of the present invention is an abnormal energy consumption diagnosis system based on a short-term building energy consumption prediction model, comprising:

[0031] The screening module collects historical building energy consumption data and data on factors affecting energy consumption, and screens out features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model through Pearson correlation analysis;

[0032] A processing module pre-processes the input feature data obtained by the screening module, constructs an energy consumption sample set, and divides the energy consumption sample set into a training set, a validation set, and a test set;

[0033] The output module uses a multi-input multi-output strategy to optimize the long short-term memory network and establish a supervised multi-step advance short-term energy consumption prediction model;

[0034] The optimization module uses the training set data obtained by the processing module to train the supervised multi-step-ahead short-term energy consumption prediction model established by the output module, uses the validation set obtained by the processing module to optimize the hyperparameters of the supervised multi-step-ahead short-term energy consumption prediction model after training, obtains the optimized short-term building energy consumption prediction model, and uses a part of the test set obtained by the processing module to evaluate the short-term building energy consumption prediction model;

[0035] The diagnosis module determines the abnormal energy consumption diagnosis method on an hourly time scale based on the short-term building energy consumption prediction model evaluated by the optimization module;

[0036] The analysis module uses the abnormal energy consumption diagnosis method determined by the diagnosis module to perform energy consumption diagnosis on the remaining test sets obtained by the processing module, and implements abnormal energy consumption analysis based on the diagnosis results.

[0037] Compared with the prior art, the present invention has at least the following beneficial effects:

[0038] The invention discloses an abnormal energy consumption diagnosis method based on a short-term building energy consumption prediction model. After Pearson correlation analysis and data preprocessing, an energy consumption sample set is constructed and divided into a training set, a verification set and a test set. An LSTM network multi-step advance short-term energy consumption prediction model based on 10 input parameters such as historical energy consumption data, occupant behavior, meteorological factors and time factors is proposed. The MIMO strategy is adopted to predict the hourly energy consumption of the building under supervised learning mode. According to the prediction model, a diagnosis method based on the mathematical relationship between the predicted value and the actual value is proposed on a smaller time scale. The general method steps of machine learning are adopted to solve the problems of long physical modeling cycle and large engineering workload.

[0039] Furthermore, the factors affecting energy consumption are divided into three categories, which is conducive to selecting the factors with greater influence in each category.

[0040] Furthermore, using 0.3 as the threshold of the Pearson correlation coefficient can correctly select factors with greater influence, eliminate unimportant factors, and eliminate the impact on the performance of the prediction model.

[0041] Furthermore, the energy consumption sample set is divided into training set, validation set and test set according to the ratio of 6:2:2. Compared with the traditional method of dividing the energy consumption sample set into training set and test set according to the ratio of 7:3, the validation set can be fully utilized to select the optimal hyperparameters of the model.

[0042] Furthermore, the test set is divided into two parts at a ratio of 1:1, which can not only meet the needs of evaluating model performance but also be used for energy consumption diagnosis.

[0043] Furthermore, the coefficient of variation of the root mean square error is less than 30% as a standard to measure whether the prediction model on an hourly time scale can be used for energy consumption diagnosis. At this time, the model prediction accuracy is high and meets the needs of actual engineering.

[0044] Furthermore, the coefficient of variation of the root mean square error (CV-RMSE) uses the ratio of the data standard deviation to the mean, overcoming the influence of the traditional statistical measurement scale and dimension.

[0045] Furthermore, the abnormal energy consumption diagnosis method is based on the mathematical relationship between the predicted value and the actual value, and therefore has good versatility and practicality in practical applications.

[0046] Furthermore, energy usage is divided into three levels according to the evaluation criteria, which helps managers to understand abnormal energy consumption in a timely manner.

[0047] In summary, the present invention has the characteristics of fast diagnosis speed, high diagnosis accuracy, good versatility and practicality in practical applications.

[0048] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a Pearson correlation diagram between other factors affecting energy consumption and energy consumption of the present invention;

[0050] Figure 2 It is the LSTM model structure diagram of the present invention;

[0051] Figure 3 It is a structural diagram of the MIMO strategy of the present invention;

[0052] Figure 4 is a prediction flow chart of the present invention;

[0053] Figure 5 is a diagnostic flow chart of the present invention;

[0054] Figure 6 This is a diagram of the diagnosis results of the present invention using this embodiment. DETAILED DESCRIPTION

[0055] 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 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.

[0056] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0057] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0058] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0059] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0060] The present invention provides an abnormal energy consumption diagnosis method based on a short-term building energy consumption prediction model. After Pearson correlation analysis and data preprocessing, an energy consumption sample set is constructed and divided into a training set, a verification set and a test set. An LSTM network multi-step advance short-term energy consumption prediction model based on 10 input parameters such as historical energy consumption data, occupant behavior, meteorological factors and time factors is proposed. The MIMO strategy is adopted to predict the hourly energy consumption of the building under supervised learning mode. According to the prediction model, a diagnosis method based on the mathematical relationship between the predicted value and the actual value is proposed on a smaller time scale. The present invention has the advantages of fast diagnosis speed, high diagnosis accuracy and strong practicality.

[0061] As shown in the figure, the present invention provides an abnormal energy consumption diagnosis method based on a short-term building energy consumption prediction model, comprising the following steps:

[0062] S1. Collect historical building energy consumption data and other data affecting energy consumption, and through Pearson correlation analysis, select features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model;

[0063] Other factors that affect energy consumption include: occupant behavior, meteorological factors and time of day.

[0064] The Pearson correlation coefficient is calculated as follows:

[0065]

[0066] Among them, ρ X,Y is the correlation coefficient between variables X and Y, and E is the mathematical expectation of the sample.

[0067] S2. Preprocess the input feature data, construct an energy consumption sample set, and divide the energy consumption sample set into a training set, a validation set, and a test set;

[0068] The training set, validation set, and test set are divided into 6:2:2.

[0069] Among them, the test set is divided into 1:1 and is used to evaluate the prediction model performance and energy consumption diagnosis respectively.

[0070] S3, using the multiple-input multiple-output (MIMO) strategy to optimize the long short-term memory (LSTM) network and establish a supervised multi-step-ahead short-term energy consumption prediction model;

[0071] S4. Use the data of the training set to train the model, optimize the hyperparameters of the model on the validation set, obtain the optimized short-term building energy consumption prediction model, and evaluate the prediction model on the test set;

[0072] When the coefficient of variation CV-RMSE of the root mean square error of the short-term building energy consumption prediction model at an hourly time scale is less than 30%, it is considered that the model fully meets the needs of the engineering purpose and can be used for energy consumption diagnosis.

[0073] Specifically, the coefficient of variation of the root mean square error is:

[0074]

[0075] Among them, y i and are the actual value and predicted value at the i-th moment, respectively, and n is the number of samples.

[0076] S5. Develop abnormal energy consumption diagnosis methods on a smaller time scale (hourly) based on the prediction model;

[0077] S501, considering the data obtained by the prediction model as energy-saving data, and the actual data as non-energy-saving data;

[0078] S502, using the residual ε to represent the difference between the actual energy consumption value and the predicted value, and using the mean absolute error MAE to represent the mean absolute error when the actual value is greater than the predicted value in the test set used to evaluate the performance of the prediction model;

[0079] The residual ε and mean absolute error MAE are:

[0080]

[0081]

[0082] Among them, y i and are the actual value and predicted value at the i-th moment, respectively, and n is the number of samples.

[0083] S503. According to the evaluation criteria, the energy usage is divided into an energy-saving state, a cautionary state and a non-energy-saving state.

[0084] The evaluation criteria are:

[0085]

[0086] S6. Perform energy consumption diagnosis on the test set and conduct in-depth analysis of abnormal energy consumption based on the diagnosis results.

[0087] In another embodiment of the present invention, an abnormal energy consumption diagnosis system based on a short-term building energy consumption prediction model is provided. The system can be used to implement the above-mentioned abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model. Specifically, the abnormal energy consumption diagnosis system based on the short-term building energy consumption prediction model includes a screening module, a processing module, an output module, an optimization module, a diagnosis module and an analysis module.

[0088] Among them, the screening module collects historical building energy consumption data and data on factors affecting energy consumption, and screens out features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model through Pearson correlation analysis;

[0089] A processing module pre-processes the input feature data obtained by the screening module, constructs an energy consumption sample set, and divides the energy consumption sample set into a training set, a validation set, and a test set;

[0090] The output module uses a multi-input multi-output strategy to optimize the long short-term memory network and establish a supervised multi-step advance short-term energy consumption prediction model;

[0091] The optimization module uses the training set data obtained by the processing module to train the supervised multi-step-ahead short-term energy consumption prediction model established by the output module, uses the validation set obtained by the processing module to optimize the hyperparameters of the supervised multi-step-ahead short-term energy consumption prediction model after training, obtains the optimized short-term building energy consumption prediction model, and uses a part of the test set obtained by the processing module to evaluate the short-term building energy consumption prediction model;

[0092] The diagnosis module determines the abnormal energy consumption diagnosis method on an hourly time scale based on the short-term building energy consumption prediction model evaluated by the optimization module;

[0093] The analysis module uses the abnormal energy consumption diagnosis method determined by the diagnosis module to perform energy consumption diagnosis on the remaining test sets obtained by the processing module, and implements abnormal energy consumption analysis based on the diagnosis results.

[0094] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0095] As a specific embodiment of the present invention, energy consumption-related data from September 2020 to April 2021 were selected for model training, data from May to July 2021 were used as a validation set, data from August to September 2021 were used for prediction model evaluation, and finally, data from October 2021 were selected to diagnose abnormal energy consumption.

[0096] The process of pre-screening the factors affecting building energy consumption using Pearson correlation analysis specifically includes the following steps:

[0097] S1. Other factors affecting energy consumption include: occupant behavior, meteorological factors and time factors; Pearson correlation coefficient is calculated. According to the Pearson correlation standard, variables with absolute values ​​of correlation coefficient greater than 0.3 are selected as input variables of the prediction model, including 10 parameters such as energy consumption of historical buildings, indoor flow of people, and flow of people entering the museum.

[0098] S2. Preprocess the input feature data, construct an energy consumption sample set, and divide the energy consumption sample set into a training set, a validation set, and a test set;

[0099] S3. Use the multiple-input multiple-output (MIMO) strategy to optimize the long short-term memory (LSTM) network and establish a supervised multi-step-ahead short-term energy consumption prediction model.

[0100] S4. Use the data of the training set to train the model, optimize the hyperparameters of the model on the validation set, obtain the optimized short-term building energy consumption prediction model, and evaluate the prediction model on the test set;

[0101] S401. The hyper parameters of the model after optimization on the validation set are:

[0102]

[0103]

[0104] S402. When the coefficient of variation of the root mean square error (CV-RMSE) of the short-term building energy consumption prediction model at an hourly time scale is less than 30%, it is considered that the model fully meets the needs of the engineering purpose and can be used for energy consumption diagnosis;

[0105] S403. The CV-RMSE of the optimized model is 0.1434.

[0106] S5. Develop abnormal energy consumption diagnosis methods on a smaller time scale (hourly) based on the prediction model;

[0107] S501, considering the data obtained by the prediction model as energy-saving data, and the actual data as non-energy-saving data;

[0108] S502, using the residual ε to represent the difference between the actual energy consumption value and the predicted value, and using the mean absolute error MAE to represent the mean absolute error when the actual value is greater than the predicted value in the test set used to evaluate the performance of the prediction model;

[0109] S503. According to the evaluation criteria, the energy usage is divided into an energy-saving state, a cautionary state and a non-energy-saving state.

[0110] S6. Perform energy consumption diagnosis on the test set and conduct in-depth analysis of abnormal energy consumption based on the diagnosis results.

[0111] S601, calculating the residual ε between the actual energy consumption value and the predicted value at each time step.

[0112] S602. Calculate and obtain MAE=7.8253.

[0113] S603. When ε is less than or equal to 0, it is considered as an energy-saving state; when ε is greater than 0 and less than or equal to 7.8253, it is considered as an attention state; when ε is greater than 7.8253, it is considered as a non-energy-saving state.

[0114] The experimental results show that among the 744 data in October, 67 data were diagnosed as non-energy-saving status, among which the relative errors of three time points were above 20%, successfully diagnosing the non-energy-saving time in building operation.

[0115] Table 1 Diagnostic results statistics

[0116]

[0117] Table 2 Partial content of non-energy-saving state

[0118]

[0119] In summary, the present invention provides an abnormal energy consumption diagnosis method and system based on a short-term building energy consumption prediction model. Short-term building energy consumption is predicted through LSTM neural network modeling. After training, energy consumption prediction can be quickly provided, which solves the problems of long physical modeling cycle and large engineering workload. The LSTM neural network is optimized through the MIMO strategy and supervised learning is adopted at the same time to improve the model prediction performance, and the energy consumption diagnosis accuracy of the building is significantly improved. The diagnostic method is obtained based on the mathematical relationship between the predicted value and the actual value. Therefore, it has good versatility and practicality in practical applications.

[0120] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0121] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0122] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0124] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. Abnormal energy consumption diagnosis method based on short-term building energy consumption prediction model, It is characterized in that The following steps are involved: S1. Collect historical building energy consumption data and data on factors affecting energy consumption, and select features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model through Pearson correlation analysis; S2, preprocessing the input feature data obtained in step S1, constructing an energy consumption sample set, and dividing the energy consumption sample set into a training set, a verification set, and a test set; S3, using multi-input multi-output strategy to optimize long short-term memory network and establish a supervised multi-step ahead short-term energy consumption prediction model; S4, using the training set data obtained in step S2 to train the supervised multi-step-ahead short-term energy consumption prediction model established in step S3, using the validation set obtained in step S2 to optimize the hyperparameters of the trained supervised multi-step-ahead short-term energy consumption prediction model, to obtain an optimized short-term building energy consumption prediction model, and using a portion of the test set obtained in step S2 to evaluate the short-term building energy consumption prediction model; S5. Based on the short-term building energy consumption prediction model evaluated in step S4, determine an abnormal energy consumption diagnosis method on an hourly time scale; S6. Perform energy consumption diagnosis on the remaining test sets obtained in step S2 using the abnormal energy consumption diagnosis method determined in step S5, and perform abnormal energy consumption analysis based on the diagnosis results.

2. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 1, It is characterized in that In step S1, the data affecting the energy consumption factors include occupant behavior, meteorological factors and time factors.

3. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 1, It is characterized in that In step S1, the Pearson correlation coefficient is calculated as follows: Among them, ρ X,Y is the correlation coefficient between variables X and Y, and E is the mathematical expectation of the sample.

4. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 1, It is characterized in that In step S2, the energy consumption sample set is divided into a training set, a validation set and a test set according to a ratio of 6:2:

2.

5. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 4, It is characterized in that The test set is divided into two parts at a ratio of 1:1, which are used for evaluating the short-term building energy consumption prediction model in step S4 and for energy consumption diagnosis in step S6.

6. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 1, It is characterized in that In step S4, the short-term building energy consumption prediction model when the coefficient of variation of the root mean square error at an hourly time scale is less than 30% is used for energy consumption diagnosis.

7. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 6, It is characterized in that The coefficient of variation of the root mean square error CV-RMSE is specifically: Among them, y i and are the actual value and predicted value at the i-th moment, respectively, and n is the number of samples.

8. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 1, It is characterized in that In step S5, the abnormal energy consumption diagnosis method is specifically as follows: S501, considering the data obtained by the prediction model as energy-saving data, and the actual data as non-energy-saving data; S502, using the residual ε to represent the difference between the actual energy consumption value and the predicted value, and using the mean absolute error MAE to represent the mean absolute error when the actual value is greater than the predicted value in the test set used to evaluate the performance of the prediction model; S503: According to the evaluation criteria, the energy usage is divided into an energy-saving state, a cautionary state and a non-energy-saving state.

9. The abnormal energy consumption diagnosis method based on the short-term building energy consumption prediction model according to claim 8, It is characterized in that In step S503, the evaluation criteria are: When ε≤0, it is an energy-saving state; when 0<ε≤MAE, it is an attention state; and when ε>MAE, it is a non-energy-saving state.

10. An abnormal energy consumption diagnosis system based on short-term building energy consumption prediction model, It is characterized in that include: The screening module collects historical building energy consumption data and data on factors affecting energy consumption, and screens out features with Pearson correlation coefficients within the threshold as input features of the energy consumption prediction model through Pearson correlation analysis; A processing module pre-processes the input feature data obtained by the screening module, constructs an energy consumption sample set, and divides the energy consumption sample set into a training set, a validation set, and a test set; The output module uses a multi-input multi-output strategy to optimize the long short-term memory network and establish a supervised multi-step advance short-term energy consumption prediction model; The optimization module uses the training set data obtained by the processing module to train the supervised multi-step-ahead short-term energy consumption prediction model established by the output module, uses the validation set obtained by the processing module to optimize the hyperparameters of the supervised multi-step-ahead short-term energy consumption prediction model after training, obtains the optimized short-term building energy consumption prediction model, and uses a part of the test set obtained by the processing module to evaluate the short-term building energy consumption prediction model; The diagnosis module determines the abnormal energy consumption diagnosis method on an hourly time scale based on the short-term building energy consumption prediction model evaluated by the optimization module; The analysis module uses the abnormal energy consumption diagnosis method determined by the diagnosis module to perform energy consumption diagnosis on the remaining test sets obtained by the processing module, and implements abnormal energy consumption analysis based on the diagnosis results.