A collaborative diagnosis method, device and medium for abnormal building energy consumption

Abstract

The embodiment of the present application discloses a collaborative diagnosis method, device and medium for abnormal energy consumption in buildings. It belongs to the technical field of building energy consumption diagnosis and solves the problem of low accuracy in building energy consumption diagnosis. It includes obtaining the historical uncertainties corresponding to different historical energy consumption data in the building to be tested and establishing an uncertainty model; generating a Bayesian network sub-unit corresponding to the energy-consuming equipment based on a preset building energy consumption anomaly detection library; constructing a topological structure diagram of the energy-consuming equipment based on the distribution and influence of each energy-consuming equipment in the building to be tested, and connecting the Bayesian network sub-units based on the energy consumption equipment topological structure diagram to generate a Bayesian network unit of the building to be tested; outputting the predicted energy consumption information corresponding to the building to be tested based on the uncertainty model and the preset GRU model; inputting the predicted energy consumption information into the Bayesian network unit of the building to be tested to output the energy consumption anomaly prediction information corresponding to the building to be tested.

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 collaborative diagnosis of abnormal building energy consumption. Background Art

[0002] Faced with the severe challenge of global climate change, energy efficiency and sustainable development have become a global consensus, driving all industries to transition towards greener, low-carbon approaches. As a major energy consumer, modern buildings require energy efficiency management and energy-saving optimization, not only for efficient resource utilization but also as a key component in achieving global sustainable development goals.

[0003] Modern building energy consumption data comes from a wide range of devices and systems, including power systems, HVAC systems, and lighting systems. These data sources vary significantly in terms of collection methods, data formats, and update frequencies, creating a complex, multi-source, and heterogeneous data environment. Current building energy consumption diagnostic technologies often rely on statistical regression based on historical data, but these methods often lack sufficient flexibility and adaptability to address the complex and ever-changing building energy consumption environment. In particular, when building usage patterns, equipment configurations, or the external environment undergo significant changes, the diagnostic accuracy of the model often decreases significantly. Summary of the Invention

[0004] The embodiments of the present application provide a collaborative diagnosis method, device and medium for abnormal building energy consumption, which are used to solve the following technical problems: Current building energy consumption diagnosis technology mostly uses statistical regression based on historical data. When the building usage pattern, equipment configuration or external environment changes significantly, the diagnostic accuracy is low.

[0005] The embodiments of this application adopt the following technical solutions:

[0006] The embodiment of the present application provides a collaborative diagnosis method for abnormal energy consumption in a building. The method comprises: obtaining historical uncertainties corresponding to different historical energy consumption data in the building to be tested, and establishing an uncertainty model based on the historical uncertainties; obtaining various energy-consuming devices in the building to be tested, and generating Bayesian network subunits corresponding to the energy-consuming devices based on a preset building energy consumption abnormality detection library; constructing a topological structure diagram of the energy-consuming devices corresponding to the building to be tested based on the distribution and influence of various energy-consuming devices in the building to be tested, and connecting the Bayesian network subunits based on the energy consumption device topological structure diagram to generate a Bayesian network unit for the building to be tested; adjusting the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model, and inputting the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; and inputting the predicted energy consumption information into the Bayesian network unit of the building to be tested to output predicted energy consumption abnormality information corresponding to the building to be tested.

[0007] By establishing an uncertainty model, the embodiment of the present application can take the volatility and uncertainty factors in the energy consumption data into consideration in the prediction, thereby improving the accuracy and reliability of the prediction results. The embodiment of the present application regards each energy-consuming device as an independent subunit, simplifies the complexity of the system, and improves the accurate detection of energy consumption anomalies of the device by establishing a Bayesian network subunit for each energy-consuming device. The Bayesian network subunits are connected by a topological structure diagram to achieve collaborative analysis and prediction of energy consumption data between devices, thereby improving the accuracy of the overall prediction. In addition, the uncertainty model and the GRU model can improve the data quality and the efficiency of energy consumption prediction. Inputting the predicted energy consumption information into the Bayesian network unit for anomaly detection can improve the efficiency of discovering energy consumption anomalies and provide a basis for energy-saving optimization and troubleshooting.

[0008] In one implementation of the present application, historical uncertainties corresponding to different historical energy consumption data in the building to be tested are obtained, and an uncertainty model is established based on the historical uncertainties, specifically including: determining the data sources corresponding to different historical energy consumption data in the building to be tested, and determining a first uncertainty based on different data sources; determining the type of energy consumption measuring equipment, and determining a first reference degree based on the type; and determining the service life and maintenance records corresponding to the energy consumption measuring equipment, and determining a second reference degree based on the service life and maintenance records; and determining the placement environment information corresponding to the energy consumption measuring equipment, and determining a third reference degree based on the placement environment information; based on the interaction between the first reference degree, the second reference degree and the third reference degree, superimposing the influence degrees to determine the second uncertainty corresponding to the energy consumption measuring equipment; and establishing an uncertainty model based on the first uncertainty and the second uncertainty.

[0009] In one implementation of the present application, each energy-consuming device in the building to be tested is obtained, and based on a preset building energy consumption anomaly detection library, a Bayesian network sub-unit corresponding to the energy-consuming device is generated, specifically including: obtaining each energy-consuming device corresponding to the building to be tested, and comparing each energy-consuming device with the preset building energy consumption anomaly detection library, and deleting the energy-consuming device that does not exist in the preset building energy consumption anomaly detection library; and obtaining node data corresponding to each energy-consuming device, and comparing the node data with the preset building energy consumption anomaly detection library, and deleting the energy-consuming device corresponding to the node data that does not exist in the preset building energy consumption anomaly detection library; and obtaining edge data corresponding to each energy-consuming device, and comparing the edge data with the preset building energy consumption anomaly detection library, and deleting the energy-consuming device corresponding to the edge data that does not exist in the preset building energy consumption anomaly detection library; based on the remaining energy-consuming devices, generating a Bayesian network sub-unit corresponding to the energy-consuming device.

[0010] In one implementation of the present application, based on the distribution and influence of each energy-consuming equipment in the building to be tested, a topological structure diagram of the energy-consuming equipment corresponding to the building to be tested is constructed, specifically including: dividing the energy-consuming equipment based on the type of each energy-consuming equipment in the building to be tested, and determining the distribution information corresponding to the divided energy-consuming equipment; based on the connection relationship between each energy-consuming equipment, the energy-consuming equipment is initially connected; wherein the connection relationship includes Internet of Things connection and logical association; based on a preset energy consumption association table, the energy consumption influence relationship between different energy-consuming equipment is determined, and based on the energy consumption influence relationship, the energy-consuming equipment is secondary connected; wherein, the preset energy consumption association table includes energy consumption influence relationships between multiple different energy-consuming equipment, and also includes energy consumption influence values ​​corresponding to each energy consumption influence relationship; based on the initial connection and the secondary connection, the connection relationship corresponding to each energy-consuming equipment is determined to construct a topological structure diagram of the energy-consuming equipment.

[0011] In one implementation of the present application, based on the energy-consuming equipment topology diagram, the Bayesian network subunits are connected to generate a Bayesian network unit of the building to be tested, specifically including: matching the Bayesian network subunits corresponding to the energy-consuming equipment with the equipment nodes in the energy-consuming equipment topology diagram based on the distribution of the energy-consuming equipment in the building to be tested; connecting the Bayesian network subunits corresponding to the energy-consuming equipment based on the connection relationship corresponding to the energy-consuming equipment topology diagram to generate a reference Bayesian network unit of the building to be tested; determining the first probability value corresponding to each node based on the energy consumption impact value and historical data in the connection relationship; and determining the error data corresponding to the Bayesian network unit of the reference building to be tested based on the historical data and the connection relationship, and generating a second probability value based on the error data; detecting the reference Bayesian network unit of the building to be tested based on the first probability value and the second probability value, and generating a Bayesian network unit of the building to be tested if the detection passes.

[0012] In one implementation of the present application, based on the uncertainty model, the energy consumption data corresponding to the building to be tested obtained are adjusted, and the adjusted data are input into a preset GRU model to output the predicted energy consumption information corresponding to the building to be tested through the GRU model, specifically including: inputting the energy consumption data corresponding to the building to be tested obtained into the uncertainty model, and outputting the adjusted data corresponding to the energy consumption data of the building to be tested based on the uncertainty model; inputting the adjusted data into the preset GRU model to output the energy consumption information corresponding to different time scales through the preset GRU model; weighting the energy consumption information corresponding to different time scales, and performing weighted calculation of multi-scale energy consumption information based on the assigned weights to obtain a weighted sum; based on the weighted sum, determining the predicted energy consumption information corresponding to the building to be tested in a preset time period in the future.

[0013] In one implementation of the present application, the predicted energy consumption information is input into the Bayesian network unit of the building to be tested to output the energy consumption anomaly prediction information corresponding to the building to be tested, specifically including: mapping the predicted energy consumption information to the corresponding nodes in the Bayesian network of the building to be tested; updating the probability distribution of the corresponding nodes according to the input predicted energy consumption information to obtain the posterior probability corresponding to each node; determining the energy consumption anomaly information of the building to be tested based on the posterior probability distribution corresponding to the Bayesian network of the building to be tested; and determining the energy consumption anomaly prediction information corresponding to the building to be tested based on the prediction time period corresponding to the energy consumption anomaly information and the predicted energy consumption data.

[0014] In one implementation of the present application, after determining the energy consumption anomaly prediction information corresponding to the building to be tested based on the predicted time period corresponding to the energy consumption anomaly information and the predicted energy consumption data, the method also includes: determining the number and type of abnormal energy-consuming equipment based on the abnormal node distribution information in the energy consumption anomaly information; inputting the number and type into a preset abnormal cause detection model to output corresponding abnormal information through the preset abnormal cause detection model; sorting the abnormal information based on the predicted time period; and performing abnormal processing in sequence before the abnormality occurs based on the sorted abnormal information.

[0015] An embodiment of the present application provides a collaborative diagnosis device for abnormal energy consumption in a building, comprising: 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 as to enable the at least one processor to: obtain historical uncertainties corresponding to different historical energy consumption data in the building to be tested, and establish an uncertainty model based on the historical uncertainties; obtain each energy-consuming device in the building to be tested, and generate a Bayesian network sub-unit corresponding to the energy-consuming device based on a preset building energy consumption abnormality detection library; construct a topological structure diagram of the energy-consuming device corresponding to the building to be tested based on the distribution and influence of each energy-consuming device in the building to be tested, and connect the Bayesian network sub-units based on the energy consumption device topological structure diagram to generate a Bayesian network unit of the building to be tested; adjust the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model, and input the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; input the predicted energy consumption information into the Bayesian network unit of the building to be tested to output energy consumption abnormality prediction information corresponding to the building to be tested.

[0016] A non-volatile computer storage medium provided by an embodiment of the present application stores computer-executable instructions, wherein the computer-executable instructions are configured to: obtain historical uncertainties corresponding to different historical energy consumption data in a building to be tested, and establish an uncertainty model based on the historical uncertainties; obtain each energy-consuming device in the building to be tested, and generate a Bayesian network sub-unit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library; construct a topological structure diagram of the energy-consuming device corresponding to the building to be tested based on the distribution and influence of each energy-consuming device in the building to be tested, and connect the Bayesian network sub-units based on the energy consumption device topological structure diagram to generate a Bayesian network unit of the building to be tested; adjust the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model, and input the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; and input the predicted energy consumption information into the Bayesian network unit of the building to be tested to output energy consumption anomaly prediction information corresponding to the building to be tested.

[0017] At least one of the above-mentioned technical solutions adopted in the embodiment of the present application can achieve the following beneficial effects: By establishing an uncertainty model, the embodiment of the present application can take the volatility and uncertainty factors in the energy consumption data into consideration in the prediction, thereby improving the accuracy and reliability of the prediction results. The embodiment of the present application regards each energy-consuming device as an independent subunit, simplifies the complexity of the system, and improves the accurate detection of energy consumption anomalies of the device by establishing a Bayesian network subunit for each energy-consuming device. By connecting the Bayesian network subunits through a topological structure diagram, collaborative analysis and prediction of energy consumption data between devices can be achieved, thereby improving the accuracy of the overall prediction. In addition, through the uncertainty model and the GRU model, the data quality and the efficiency of energy consumption prediction can be improved. Inputting the predicted energy consumption information into the Bayesian network unit for anomaly detection can improve the efficiency of discovering energy consumption anomalies and provide a basis for energy-saving optimization and troubleshooting. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments described in the present application. For those skilled in the art, other drawings can be obtained based on these drawings without inventive work. In the drawings:

[0019] Figure 1 A flow chart of a collaborative diagnosis method for abnormal building energy consumption provided in an embodiment of the present application;

[0020] Figure 2 A schematic structural diagram of a collaborative diagnostic device for abnormal building energy consumption provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The embodiments of the present application provide a method, device, and medium for collaborative diagnosis of abnormal building energy consumption.

[0022] In order to enable those skilled in the art to better understand the technical solutions in this application, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0023] The technical solutions proposed in the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0024] Figure 1 This is a flow chart of a collaborative diagnosis method for abnormal building energy consumption provided by the embodiment of the present application. Figure 1 As shown in FIG, the method for predicting abnormal building energy consumption includes the following steps:

[0025] S101. Obtain historical uncertainties corresponding to different historical energy consumption data in the building to be tested, and establish an uncertainty model based on the historical uncertainties.

[0026] In one embodiment of the present application, the data sources corresponding to different historical energy consumption data in the building to be measured are determined, and a first uncertainty is determined based on the different data sources. The type of the energy consumption measuring device is determined, and a first reference degree is determined based on the type. Furthermore, the service life and maintenance records corresponding to the energy consumption measuring device are determined, and a second reference degree is determined based on the service life and maintenance records. Furthermore, the placement environment information corresponding to the energy consumption measuring device is determined, and a third reference degree is determined based on the placement environment information. Based on the interaction between the first reference degree, the second reference degree, and the third reference degree, the influence degree is superimposed to determine the second uncertainty corresponding to the energy consumption measuring device. An uncertainty model is established based on the first uncertainty and the second uncertainty.

[0027] Specifically, the energy consumption data in the embodiments of this application may come from a variety of sources, such as building management systems, smart meters, and manual records. Different data sources may have varying degrees of accuracy and reliability. The primary uncertainty of the data is assessed based on factors such as the reliability and precision of the data source, and the presence of human error. For example, data from smart meters is generally more accurate than manual records, and therefore has a lower uncertainty.

[0028] Furthermore, different energy consumption measurement devices, such as high-precision electricity meters, standard electricity meters, and thermal energy meters, have different measurement accuracy and characteristics. Based on the device's technical specifications and industry standards, the contribution of each device to the accuracy of energy consumption data is determined (the first reference level). Secondly, based on the device's age and maintenance records, its potential impact on data accuracy is assessed (the second reference level). Furthermore, the device's operating environment, such as temperature, humidity, and electromagnetic interference, may also affect its measurement accuracy. Based on the device's actual operating environment, its potential impact on data accuracy is assessed (the third reference level).

[0029] Furthermore, the first, second, and third reference degrees are combined and their influences are superimposed using a weighted average method to calculate the second uncertainty of the energy consumption measurement device. Combining the first and second uncertainties, an uncertainty model is established to assess the overall uncertainty of historical energy consumption data. For example, the uncertainty corresponding to the uncertainty factors of the data source and the uncertainty factors of the equipment factors can be added together to obtain the uncertainty output by the uncertainty model.

[0030] S102: Acquire each energy-consuming device in the building to be tested, and generate a Bayesian network sub-unit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library.

[0031] In one embodiment of the present application, the corresponding energy-consuming devices in the building to be tested are obtained, and each energy-consuming device is compared with a preset building energy consumption anomaly detection library, and the energy-consuming devices that do not exist in the preset building energy consumption anomaly detection library are deleted. The node data corresponding to each energy-consuming device is obtained, and the node data is compared with the preset building energy consumption anomaly detection library, and the energy-consuming devices corresponding to the node data that do not exist in the preset building energy consumption anomaly detection library are deleted. The edge data corresponding to each energy-consuming device is obtained, and the edge data is compared with the preset building energy consumption anomaly detection library, and the energy-consuming devices corresponding to the edge data that do not exist in the preset building energy consumption anomaly detection library are deleted. Based on the remaining energy-consuming devices, a Bayesian network subunit corresponding to the energy-consuming device is generated.

[0032] Specifically, first, comprehensive information on all energy-consuming devices within the building under test is collected. This includes, but is not limited to, electricity-consuming devices such as lighting systems, air conditioning systems, and elevators; water-consuming devices such as water supply and drainage systems; and gas-consuming devices such as gas boilers.

[0033] Furthermore, the information of each energy-consuming device obtained is compared with the preset building energy consumption anomaly detection library. The preset library in the embodiment of the present application contains various types of known equipment that may cause energy consumption anomalies and their characteristics. Secondly, the node data and edge data are compared. The "node data" and "edge data" in the embodiment of the present application refer to the position and mutual relationship of the energy-consuming equipment in the Bayesian network structure. The comparison process will be similar to the device comparison, that is, check whether these data match the information in the preset library. If they match, they will be retained. If they do not match, they will be deleted.

[0034] Furthermore, based on the remaining compared energy-consuming devices, a Bayesian network subunit can be constructed.

[0035] It's important to note that a Bayesian network is a graphical model used to represent probabilistic relationships between variables, making it suitable for handling uncertainty and correlation. In building energy anomaly detection, Bayesian networks can be used to represent the energy consumption relationships between various energy-consuming devices and their contribution to overall energy anomalies.

[0036] S103. Based on the distribution and influence of each energy-consuming device in the building to be tested, a topological structure diagram of the energy-consuming devices corresponding to the building to be tested is constructed, and based on the topological structure diagram of the energy-consuming devices, the Bayesian network subunits are connected to generate a Bayesian network unit of the building to be tested.

[0037] In one embodiment of the present application, based on the type of each energy-consuming device in the building to be tested, the energy-consuming devices are divided, and the distribution information corresponding to the divided energy-consuming devices is determined. Based on the connection relationship between each energy-consuming device, the energy-consuming devices are initially connected; wherein the connection relationship includes an Internet of Things connection and a logical association. Based on a preset energy consumption association table, the energy consumption impact relationship between different energy-consuming devices is determined, and based on the energy consumption impact relationship, the energy-consuming devices are secondarily connected; wherein, the preset energy consumption association table includes energy consumption impact relationships between multiple different energy-consuming devices, and also includes energy consumption impact values ​​corresponding to each energy consumption impact relationship. Based on the initial connection and the secondary connection, the corresponding connection relationship of each energy-consuming device is determined to construct a topological structure diagram of the energy-consuming device.

[0038] Specifically, energy-consuming equipment is divided into categories such as lighting systems, air conditioning systems, elevator systems, and heating systems, and the specific distribution locations of each type of equipment within the building, such as floors, rooms, or areas, are determined. The connection relationships between different energy-consuming equipment are determined, and based on these connection relationships, multiple energy-consuming equipment are initially connected, that is, the connection relationships of the topological result graph of the energy-consuming equipment are determined. This connection relationship includes the association or logical relationship between different equipment. For example, the activation of certain equipment may depend on the status of other equipment. When insufficient light and human activity are detected in the room, the lighting system is automatically turned on. At the same time, if a window is opened, the temperature setting of the air conditioning system may be automatically adjusted to reduce energy consumption.

[0039] Furthermore, the embodiment of the present application is provided with a preset energy consumption association table, which is a table containing the energy consumption impact relationship between different energy-consuming devices, including the type of impact relationship (positive, negative) and the specific impact value (such as the percentage of energy consumption increase or decrease). Based on this table, the complex energy consumption interactions between devices are further determined, so that these relationships are reflected in the topology diagram, that is, secondary connections are made according to the table. Combining the results of the initial connection and the secondary connection, a topological structure diagram containing all energy-consuming devices and their connection relationships is drawn.

[0040] In one embodiment of the present application, based on the distribution of energy-consuming equipment in the building to be tested, the Bayesian network subunits corresponding to the energy-consuming equipment are matched with the device nodes in the energy-consuming equipment topology diagram. Based on the connection relationship corresponding to the energy-consuming equipment topology diagram, the Bayesian network subunits corresponding to the energy-consuming equipment are connected to generate a reference Bayesian network unit of the building to be tested. Based on the energy consumption impact value and historical data in the connection relationship, the first probability value corresponding to each node is determined. And, based on the historical data and the connection relationship, the error data corresponding to the Bayesian network unit of the reference building to be tested is determined, and a second probability value is generated based on the error data. Based on the first probability value and the second probability value, the Bayesian network unit of the reference building to be tested is tested, and if the test passes, the Bayesian network unit of the building to be tested is generated.

[0041] Specifically, first, it is necessary to understand the distribution of all energy-consuming devices in the building under test, such as air conditioners, lighting, and elevators. Then, a corresponding Bayesian network subunit is created for each energy-consuming device. These subunits represent the device's independent role and status in the energy consumption system, such as on / off and power level. Based on the actual connections and dependencies between energy-consuming devices in the building, namely the topological structure diagram, the corresponding subunits are connected in the Bayesian network. This connection reflects the energy consumption impact between devices. For example, the use of air conditioning may increase the energy consumption of elevators because the elevator room needs to maintain a certain temperature.

[0042] Furthermore, using the energy consumption impact values ​​in the connection relationships, such as the energy consumption coefficient, and historical data, such as energy consumption records over a period of time, a first probability value is calculated for each node, that is, each energy-consuming device. These probability values ​​reflect the predicted probability of energy consumption of the device under a specific state. For example, based on historical data, when the outdoor temperature is above 30 degrees Celsius, the average probability of the air conditioner operating is 80%. This 80% is the first probability value for the air conditioner subunit to "turn on" in the "high temperature" state.

[0043] Furthermore, by comparing the predictions of the reference building's Bayesian network unit with historical data, error data for the reference building's Bayesian network unit is calculated. Based on this error data, a second probability value is generated to assess the uncertainty of the reference building's Bayesian network unit. For example, if the model predicts total energy consumption of 1000 kWh for a certain period, but the actual energy consumption is 1050 kWh, this 50 kWh difference can be used to calculate the error rate of the reference building's Bayesian network unit and further generate a second probability value reflecting the uncertainty of the prediction.

[0044] Furthermore, the Bayesian network unit of the reference building to be tested is tested based on the first probability value and the second probability value. If the Bayesian network unit of the reference building to be tested passes the test, that is, the prediction error is within an acceptable range, a final Bayesian network unit of the building to be tested is generated and used for future energy consumption prediction and optimization of the building.

[0045] S104. Based on the uncertainty model, the acquired energy consumption data corresponding to the building to be tested is adjusted, and the adjusted data is input into a preset GRU model to output the predicted energy consumption information corresponding to the building to be tested through the GRU model.

[0046] In one embodiment of the present application, the acquired energy consumption data corresponding to the building to be tested is input into an uncertainty model, and adjusted data corresponding to the energy consumption data of the building to be tested is output based on the uncertainty model. The adjusted data is input into a preset GRU model, which then outputs energy consumption information corresponding to different time scales. Weights are assigned to the energy consumption information corresponding to the different time scales, and a weighted calculation is performed on the multi-scale energy consumption information based on the assigned weights to obtain a weighted sum. Based on the weighted sum, predicted energy consumption information corresponding to the building to be tested within a preset future time period is determined.

[0047] Specifically, the actual energy consumption data of the building to be tested over a period of time is input into the uncertainty model. Based on the output of the uncertainty model, the original energy consumption data is adjusted or corrected to reduce the impact of errors and uncertainties. The adjusted data is closer to the actual situation, which helps to improve the accuracy of subsequent predictions. Secondly, the adjusted energy consumption data is input into the preset GRU model. It should be noted that GRU is a variant of a recurrent neural network that is suitable for processing time series data. It can capture the time dependency and long-term trends in the data, thereby making accurate predictions. The GRU model can output energy consumption forecast information at different time scales (such as day, week, and month).

[0048] Furthermore, because energy consumption information at different timescales may contribute differently to the prediction results, it is necessary to assign weights to each of them. Then, a weighted sum is calculated based on these assigned weights, which serves as the basis for the final prediction results. Based on this weighted sum, the predicted energy consumption information for the building under test can be determined for a preset time period in the future (e.g., the next week, month, etc.).

[0049] S105: Input the predicted energy consumption information into the Bayesian network unit of the building to be tested, so as to output energy consumption abnormality prediction information corresponding to the building to be tested.

[0050] In one embodiment of the present application, predicted energy consumption information is mapped to corresponding nodes in a Bayesian network for the building under test. Based on the input predicted energy consumption information, the probability distribution of the corresponding nodes is updated to obtain the posterior probability corresponding to each node. Based on the posterior probability distribution corresponding to the Bayesian network for the building under test, energy consumption anomaly information for the building under test is determined. Based on the predicted time period corresponding to the energy consumption anomaly information and the predicted energy consumption data, predicted energy consumption anomaly information corresponding to the building under test is determined.

[0051] Specifically, the predicted energy consumption information, such as the predicted energy consumption value for a certain time period in the future, is mapped to the corresponding nodes in the Bayesian network. Based on the input predicted energy consumption information, the probability distribution of the corresponding nodes is updated using the Bayesian theorem, that is, the posterior probability distribution of these nodes under the given predicted energy consumption information is calculated. Based on the updated posterior probability distribution, by setting a threshold or comparing with other statistical indicators, such as the historical average and standard deviation, it is determined whether there is an energy consumption anomaly. If the posterior probability distribution of a node shows that its value exceeds the normal range, such as the energy consumption is too high or too low, it is considered that the energy consumption corresponding to the node is abnormal. Based on the predicted time period corresponding to the energy consumption anomaly information and the predicted energy consumption data, it is determined in which time periods the building under test may have energy consumption anomalies, thereby helping building managers take measures in advance, such as adjusting the operating status of equipment, optimizing energy use plans, etc., to reduce unnecessary energy waste.

[0052] In one embodiment of the present application, the number and type of abnormal energy-consuming devices are determined based on the abnormal node distribution information in the energy consumption anomaly information. The number and type are input into a preset anomaly cause detection model, which then outputs corresponding anomaly information. The anomaly information is sorted based on the predicted time period. Based on the sorted anomaly information, anomaly handling is performed sequentially before an anomaly occurs.

[0053] Specifically, the number and type of abnormal energy-consuming devices are input into a preset abnormality cause detection model, which is used to diagnose the specific causes of abnormal energy consumption. The model conducts a comprehensive analysis based on the input information, combined with historical data, equipment performance parameters, environmental factors, and other information, and outputs corresponding abnormality information, including possible causes of the abnormality and the degree of impact. The abnormality information is sorted according to the predicted time period. The sorting can be based on various factors such as the likelihood of the abnormality occurring, the degree of impact, and the urgency. The purpose of sorting is to prioritize those abnormalities that are about to occur and have a greater impact. Based on the sorted abnormality information, abnormality handling is carried out in sequence before the abnormality occurs. Among them, the handling measures may include adjusting the operating status of the equipment, optimizing the energy usage plan, notifying relevant personnel for inspection or maintenance, etc.

[0054] Figure 2 This is a schematic diagram of the structure of a building energy consumption abnormality collaborative diagnosis device provided in an embodiment of the present application. Figure 2 As shown, a building energy consumption anomaly prediction device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed 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: obtain historical uncertainties corresponding to different historical energy consumption data in the building to be tested, and establish an uncertainty model based on the historical uncertainties; obtain each energy-consuming device in the building to be tested, and generate a Bayesian network sub-unit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library; construct a topological structure diagram of the energy-consuming device corresponding to the building to be tested based on the distribution and influence of each energy-consuming device in the building to be tested, and connect the Bayesian network sub-units based on the energy consumption device topological structure diagram to generate a Bayesian network unit of the building to be tested; adjust the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model, and input the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; input the predicted energy consumption information into the Bayesian network unit of the building to be tested to output energy consumption anomaly prediction information corresponding to the building to be tested.

[0055] A non-volatile computer storage medium provided by an embodiment of the present application stores computer-executable instructions, wherein the computer-executable instructions are configured to: obtain historical uncertainties corresponding to different historical energy consumption data in a building to be tested, and establish an uncertainty model based on the historical uncertainties; obtain each energy-consuming device in the building to be tested, and generate a Bayesian network sub-unit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library; construct a topological structure diagram of the energy-consuming device corresponding to the building to be tested based on the distribution and influence of each energy-consuming device in the building to be tested, and connect the Bayesian network sub-units based on the energy consumption device topological structure diagram to generate a Bayesian network unit of the building to be tested; adjust the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model, and input the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; and input the predicted energy consumption information into the Bayesian network unit of the building to be tested to output energy consumption anomaly prediction information corresponding to the building to be tested.

[0056] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the device, apparatus, and non-volatile computer storage medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0057] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. It will be apparent to those skilled in the art that various modifications and variations may be made to the embodiments of the present application. However, such modifications or substitutions do not deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A collaborative diagnosis method for abnormal building energy consumption, characterized in that: The method comprises: Obtaining historical uncertainties corresponding to different historical energy consumption data in the building to be tested, and establishing an uncertainty model based on the historical uncertainties; Obtain each energy-consuming device in the building to be tested, and generate a Bayesian network subunit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library; Based on the distribution and influence of each energy-consuming device in the building to be tested, a topological structure diagram of the energy-consuming devices corresponding to the building to be tested is constructed, and based on the topological structure diagram of the energy-consuming devices, the Bayesian network subunits are connected to generate a Bayesian network unit of the building to be tested; Based on the uncertainty model, the acquired energy consumption data corresponding to the building to be tested is adjusted, and the adjusted data is input into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model; Inputting the predicted energy consumption information into the Bayesian network unit of the building to be tested to output abnormal energy consumption prediction information corresponding to the building to be tested; The obtaining of historical uncertainties corresponding to different historical energy consumption data of the building to be tested and establishing an uncertainty model based on the historical uncertainties specifically includes: Determining data sources corresponding to different historical energy consumption data in the building to be tested, and determining a first uncertainty based on the different data sources; Determining a type of the energy consumption measurement device, and determining a first reference degree based on the type; and determining the service life and maintenance record corresponding to the energy consumption measuring device, and determining a second reference degree based on the service life and the maintenance record; and determining placement environment information corresponding to the energy consumption measuring device, and determining a third reference degree based on the placement environment information; Based on the interaction between the first reference degree, the second reference degree and the third reference degree, influence superposition is performed to determine a second uncertainty corresponding to the energy consumption measuring device; The uncertainty model is established based on the first uncertainty and the second uncertainty.

2. A collaborative diagnosis method for abnormal building energy consumption according to claim 1, characterized in that: The process of obtaining each energy-consuming device in the building to be tested and generating a Bayesian network subunit corresponding to the energy-consuming device based on a preset building energy consumption anomaly detection library specifically includes: Obtaining each energy-consuming device corresponding to the building to be tested, and comparing each energy-consuming device with the preset building energy consumption anomaly detection library, and deleting the energy-consuming devices that do not exist in the preset building energy consumption anomaly detection library; and obtaining node data corresponding to each of the energy-consuming devices, and comparing the node data with the preset building energy consumption anomaly detection library, and deleting energy-consuming devices corresponding to the node data that does not exist in the preset building energy consumption anomaly detection library; and obtaining edge data corresponding to each of the energy-consuming devices, and comparing the edge data with the preset building energy consumption anomaly detection library, and deleting energy-consuming devices corresponding to edge data that do not exist in the preset building energy consumption anomaly detection library; Based on the remaining energy-consuming devices, a Bayesian network subunit corresponding to the energy-consuming devices is generated.

3. The method for collaborative diagnosis of abnormal building energy consumption according to claim 1, characterized in that: The step of constructing a topological structure diagram of energy-consuming devices corresponding to the building to be tested based on the distribution and influence of each energy-consuming device in the building to be tested specifically includes: Based on the types of the energy-consuming devices in the building to be tested, the energy-consuming devices are divided, and distribution information corresponding to the divided energy-consuming devices is determined; Based on the connection relationship between the energy-consuming devices, the energy-consuming devices are initially connected; wherein the connection relationship includes an Internet of Things connection and a logical association; Based on a preset energy consumption association table, determining the energy consumption impact relationship between different energy consuming devices, and performing secondary connection on the energy consuming devices based on the energy consumption impact relationship; wherein the preset energy consumption association table includes energy consumption impact relationships between multiple different energy consuming devices, and also includes energy consumption impact values ​​corresponding to each energy consumption impact relationship; Based on the primary connection and the secondary connection, the connection relationship corresponding to each of the energy-consuming devices is determined to construct a topological structure diagram of the energy-consuming devices.

4. A collaborative diagnosis method for abnormal building energy consumption according to claim 3, characterized in that: The connecting of the Bayesian network subunits based on the energy consumption equipment topology diagram to generate a Bayesian network unit for the building to be tested specifically includes: Based on the distribution of energy-consuming devices in the building to be tested, matching the Bayesian network subunits corresponding to the energy-consuming devices with the device nodes in the energy-consuming device topology diagram; Based on the connection relationship corresponding to the energy-consuming equipment topology diagram, connecting the Bayesian network subunits corresponding to the energy-consuming equipment to generate a reference Bayesian network unit of the building to be tested; Determining a first probability value corresponding to each node based on the energy consumption impact value and historical data in the connection relationship; and, based on the historical data and the connection relationship, determining error data corresponding to the Bayesian network unit of the reference building to be tested, and generating a second probability value based on the error data; The reference building Bayesian network unit to be tested is tested based on the first probability value and the second probability value, and the building Bayesian network unit to be tested is generated if the test passes.

5. The method for collaborative diagnosis of abnormal building energy consumption according to claim 1, characterized in that: The step of adjusting the acquired energy consumption data corresponding to the building to be tested based on the uncertainty model and inputting the adjusted data into a preset GRU model to output predicted energy consumption information corresponding to the building to be tested through the GRU model specifically includes: Inputting the acquired energy consumption data corresponding to the building to be measured into the uncertainty model, and outputting adjusted data corresponding to the energy consumption data of the building to be measured based on the uncertainty model; Inputting the adjusted data into the preset GRU model, so as to output energy consumption information corresponding to different time scales through the preset GRU model; Assign weights to the energy consumption information corresponding to different time scales, and perform weighted calculation on the multi-scale energy consumption information based on the assigned weights to obtain a weighted sum; Based on the weighted sum, predicted energy consumption information corresponding to the building to be tested within a future preset time period is determined.

6. A collaborative diagnosis method for abnormal building energy consumption according to claim 1, characterized in that: Inputting the predicted energy consumption information into the Bayesian network unit of the building to be tested to output abnormal energy consumption prediction information corresponding to the building to be tested specifically includes: Mapping the predicted energy consumption information to corresponding nodes in the Bayesian network of the building to be tested; According to the input predicted energy consumption information, the probability distribution of the corresponding node is updated to obtain the posterior probability corresponding to each node; Determining abnormal energy consumption information of the building to be tested based on the posterior probability distribution corresponding to the Bayesian network of the building to be tested; Based on the energy consumption anomaly information and the predicted time period corresponding to the predicted energy consumption data, energy consumption anomaly prediction information corresponding to the building to be tested is determined.

7. A collaborative diagnosis method for abnormal building energy consumption according to claim 6, characterized in that: After determining the energy consumption anomaly prediction information corresponding to the building to be tested based on the energy consumption anomaly information and the predicted time period corresponding to the predicted energy consumption data, the method further includes: Determining the number and type of abnormal energy-consuming devices based on abnormal node distribution information in the abnormal energy consumption information; Inputting the quantity and the type into a preset abnormality cause detection model, so as to output corresponding abnormality information through the preset abnormality cause detection model; sorting the abnormal information based on the predicted time period; Based on the sorted exception information, exception handling is performed in sequence before the exception occurs.

8. A collaborative diagnostic device for abnormal building energy consumption, characterized in that: The device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the method according to any one of claims 1 to 7.

9. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 7.

Images