A building power system fault diagnosis method using digital twin technology

Abstract

The present invention discloses a method for diagnosing faults in a building power system using digital twin technology, including obtaining information of a target building, constructing a digital twin model, collecting sample data of power parameters of the target building, constructing a prediction model, collecting fault information of the building, constructing a fault classification model, collecting power parameters of the target building in real time, and inputting them into the prediction model to obtain predicted power parameters, inputting the predicted power parameters into the fault classification model to obtain fault prediction results, if the fault prediction result is that a fault has occurred, inputting the predicted power parameters and the fault prediction results into the digital twin model and updating the digital twin model, if the fault prediction result is that no fault has occurred, inputting the predicted power parameters and the fault prediction results into the digital twin model according to a preset update rule and updating the digital twin model, the present invention can effectively solve the hysteresis of data transmission and digital twin model update delay in the power system.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a building power system fault diagnosis method using digital twin technology. Background Art

[0002] As buildings expand in size and electricity demand increases, building power systems become more and more complex. The complex system structure increases the probability of failures and also increases the difficulty of troubleshooting and repairing.

[0003] A Chinese patent application with publication number CN116885858B discloses a distribution network fault handling method and system based on digital twin technology, which is used to solve the problem that the existing distribution network fault detection method cannot monitor all power equipment in the distribution network in real time through the digital twin model, obtain its abnormal conditions, and cannot simulate and repair the faults that occur, resulting in the inability to detect and handle the abnormalities of the distribution network in time, and it is difficult to ensure the stable power supply of the distribution network; the distribution network fault handling method can update the digital twin model of the distribution network in real time according to various operating data of the distribution network to reflect the real-time operating status of the actual distribution network. When a fault in the distribution network is detected, the corresponding fault handling method is simulated in the digital twin model according to the fault type to achieve simulation and preprocessing of the distribution network fault.

[0004] The solution of the above patent has the following defects in practical application:

[0005] First, the patent realizes fault monitoring and simulation of the distribution network by constructing a digital twin model of the distribution network. When a fault occurs in the distribution network, the digital twin model is used for simulation to obtain a fault handling method. In this method, the monitoring data of the distribution network needs to ensure that it can be transmitted to the digital twin model in real time, and the digital twin model needs to update the data and complete the simulation in real time. However, based on current communication and computing technologies, in the actual application of digital twin models, due to the lag in data transmission and the huge amount of simulation calculations, digital twin models generally have a certain lag, that is, the results of the simulation are relatively lagging compared to the actual physical scene. This defect is particularly fatal in the power scene. The fault in the power scene occurs immediately and is easy to induce other line faults, ultimately leading to serious consequences such as power outages and fires. Therefore, the method can be realized in a relatively ideal environment in which the digital twin model can be used to instantly complete the simulation and generate a fault handling method when a fault occurs, but the lag problem in the actual power monitoring process still needs to be solved;

[0006] Secondly, the patent obtains various operating data of the distribution network in real time and updates the digital twin model of the distribution network in real time. This method has high computing power requirements and computing power occupancy of the monitoring system, which is not conducive to actual deployment and application. Summary of the invention

[0007] In order to solve the above-mentioned technical problems, the present invention provides a building power system fault diagnosis method using digital twin technology.

[0008] The technical solution of the present invention is implemented as follows: a building power system fault diagnosis method using digital twin technology, comprising:

[0009] S1. Acquire building information of a target building, wherein the building information includes basic building data and power system data of the target building;

[0010] S2. constructing a digital twin model of the target building based on the building information;

[0011] S3. Collecting sample data of the power parameters of the target building and constructing a prediction model for the power parameters of the target building;

[0012] S4, collect building fault information and build a fault classification model for the power system;

[0013] S5. Collect the power parameters of the target building in real time, record them as real-time power parameters, input the real-time power parameters into the prediction model, and obtain predicted power parameters;

[0014] S6. Input the predicted power parameters into the fault classification model to obtain a fault prediction result of the power system. If the fault prediction result is that a fault has occurred, input the predicted power parameters and the fault prediction result into the digital twin model and update the digital twin model. If the fault prediction result is that no fault has occurred, input the predicted power parameters and the fault prediction result into the digital twin model and update the digital twin model according to a preset update rule.

[0015] Furthermore, the building basic data in step S1 includes the building type and building construction parameters, the power system data includes the building's power equipment parameters and distribution line parameters, the power equipment parameters include the type, quantity and rated power of the power equipment, and the distribution line parameters include the distribution location, maximum load voltage, maximum load current and maximum load power of the distribution line;

[0016] In the steps S3 and S5, the power parameters include the total operating power of the power equipment of the target building, and the voltage, current and power of the distribution line.

[0017] Furthermore, the building types in step S2 include residential buildings, commercial buildings, industrial buildings and office buildings.

[0018] Furthermore, the construction of the prediction model in the S3 step is: collecting sample data of the power system operating parameters of the target building, cleaning the sample data, completing the processing of missing values ​​and outliers, extracting relevant features of the power system fault from the sample data, constructing a linear regression prediction model of the power parameters of the target building, and predicting the future change trend of the power parameters.

[0019] Furthermore, the process of constructing the fault classification model in step S4 includes: collecting basic building information and power system fault data of the target building, determining the building type and building construction parameters, as well as the fault parameters of the power equipment and the fault parameters of the distribution line, establishing a target building fault classification model through decision tree algorithms including ID3, C5, and CART, using information gain, information gain rate, and Gini index indicators for feature selection, generating a decision tree from top to bottom in a recursive manner, and optimizing it by pre-pruning or post-pruning to improve the generalization ability of the model.

[0020] Furthermore, as described in step S5: real-time power parameters are input into the prediction model to obtain predicted power parameters by arranging real-time sensors in the target building, collecting power system operating parameter data of the target building in real time, and transmitting the data to the prediction model in real time. The prediction model uses current data to predict the operating status of the power system of the target building in the future to obtain predicted power parameters.

[0021] Further, in step S6, according to the type of the target building, the frequency adjustment condition of the target building is preset, and the update rule is specifically: judging whether the target building meets the frequency adjustment condition, if so, adjusting the update frequency, and updating the digital twin model according to the adjusted update frequency, if not, maintaining the original update frequency to update the digital twin model;

[0022] According to the preset update rule, the predicted power parameter and the fault prediction result are input into the digital twin model and the digital twin model is updated specifically as follows:

[0023] When the target building is a residential building, a first update frequency is preset to determine whether the current time is in working hours or non-working hours. If the current time is in working hours, the digital twin model is updated according to the first update frequency. If the current time is in non-working hours, the first update frequency is increased, and the digital twin model is updated according to the increased first update frequency.

[0024] When the target building is a commercial building, a second update frequency is preset to determine whether the current time is in a holiday business period. If not, the digital twin model is updated at the second update frequency. If so, the second update frequency is increased, and the digital twin model is updated at the increased second update frequency.

[0025] When the target building is an industrial building, a third update frequency is preset, the production schedule of the production equipment in the target building is obtained, and then the total production schedule power of the production equipment is obtained, and it is determined whether the total production schedule power at the current moment is less than the preset production schedule power threshold. If so, the digital twin model is updated according to the third update frequency; if not, the third update frequency is increased, and the digital twin model is updated according to the increased third update frequency;

[0026] When the target building is an office building, the fourth update frequency is preset to determine whether the current time is in working hours or non-working hours. If the current time is in non-working hours, the digital twin model is updated according to the fourth update frequency. If the current time is in working hours, the fourth update frequency is increased and the digital twin model is updated according to the increased fourth update frequency.

[0027] Furthermore, in the step S6, when the target building is a residential building, the first update frequency is expressed as times / h, the first update frequency after the increase is expressed as Times / h:

[0028] ,

[0029] ;

[0030] in, and are all coefficients greater than 0. is a constant, represents the load fluctuation index of the target building, represents the average load of the residential building in the past certain period of time, represents the maximum load of the residential building;

[0031] When the target building is a commercial building, the second update frequency is expressed as times / h, the increased second update frequency is expressed as Times / h:

[0032] ,

[0033] ;

[0034] in, and are all coefficients greater than 0. is a constant, represents the commercial activity coefficient of the commercial building. If the commercial building is currently undertaking commercial activities, then , Indicates the number of commercial activities currently undertaken by the commercial building, otherwise , represents the average load of the commercial building in the past certain period of time, represents the maximum load of said commercial building;

[0035] When the target building is an industrial building, the third update frequency is expressed as times / h,

[0036] The increased third update frequency is expressed as Times / h:

[0037] ,

[0038] ;

[0039] in, and are all coefficients greater than 0. is a constant, Indicates the difference between the total production scheduling power at the current moment and the production scheduling power threshold. represents the average load of the industrial building in the past certain period of time, represents the maximum load of the industrial building in question;

[0040] When the target building is an office building, the fourth update frequency is expressed as times / h,

[0041] The increased fourth update frequency is expressed as Times / h:

[0042] ,

[0043] ;

[0044] in, and is a coefficient greater than 0, is a constant, represents the average working power of the office equipment in the office building within a certain period of time, represents the average working power of the fresh air system of the office building in a certain period of time, represents the average load of the office building in the past certain period of time, represents the maximum load of the office building. Beneficial Effects

[0045] The present invention predicts the power parameters of the target building by setting a prediction model to obtain predicted power parameters, predicts the faults of the target building based on the predicted power parameters, and updates the digital twin model in combination with the predicted power parameters and the fault prediction results. This allows the digital twin model to have a certain degree of advance compared to the actual physical scene, and can effectively make up for the lag caused by data transmission and digital twin model update delays, so that the digital twin model can reflect the power system status of the target building in real time.

[0046] Furthermore, the present invention can realize the prediction of power system faults through prediction models and fault classification models. Compared with the prior art solution of performing simulation after the fault occurs, the present invention can predict the faults of the power system in advance and intuitively display them through a digital twin model, which helps maintenance personnel to intervene in advance and avoid faults.

[0047] In addition, the present invention also sets the update rules of the digital twin model. When a fault is predicted, it indicates that the target building is in a high-risk state where a fault may occur. The digital twin model is updated in real time to display the predicted fault, prompting maintenance personnel to investigate and intervene in time. When no fault is predicted, it indicates that the target building is in a relatively safe low-risk state and normal monitoring can be maintained. In combination with the characteristics of each type of building, a reasonable update frequency is set, so that the digital twin model does not need to be updated in real time, thereby reducing the computing power requirements and computing power occupancy of the system, which is more conducive to actual deployment and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a structural block diagram of a building power system fault diagnosis method using digital twin technology in an embodiment of the present invention;

[0049] Figure 2 The present invention is a flowchart of the steps of a building power system fault diagnosis method using digital twin technology in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] The preferred implementation methods of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these implementation methods are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0052] It can be understood that the terms "first", "second", etc. used in the present application can be used in this article to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from another element.

[0053] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0054] See also Figure 1-Figure 2 As shown, a building power system fault diagnosis method using digital twin technology includes:

[0055] S1. Acquire building information of a target building, wherein the building information includes basic building data and power system data of the target building;

[0056] S2. constructing a digital twin model of the target building based on the building information;

[0057] S3. Collecting sample data of the power parameters of the target building and constructing a prediction model for the power parameters of the target building;

[0058] S4, collect building fault information and build a fault classification model for the power system;

[0059] S5. Collect the power parameters of the target building in real time, record them as real-time power parameters, input the real-time power parameters into the prediction model, and obtain predicted power parameters;

[0060] S6. Input the predicted power parameters into the fault classification model to obtain a fault prediction result of the power system. If the fault prediction result is that a fault has occurred, input the predicted power parameters and the fault prediction result into the digital twin model and update the digital twin model. If the fault prediction result is that no fault has occurred, input the predicted power parameters and the fault prediction result into the digital twin model and update the digital twin model according to a preset update rule.

[0061] like Figure 1-Figure 2 As shown, in the step S1, the building information of the target building is obtained, and the building information includes basic building data and power system data of the target building;

[0062] Specifically, in step S1, the basic building data includes the building type and building construction parameters, the power system data includes the building's power equipment parameters and distribution line parameters, the power equipment parameters include the type, quantity and rated power of the power equipment, and the distribution line parameters include the distribution location, maximum load voltage, maximum load current and maximum load power of the distribution line;

[0063] In the steps S3 and S5, the power parameters include the total operating power of the power equipment of the target building, and the voltage, current and power of the distribution line.

[0064] like Figure 1-Figure 2 As shown, in the step S2, a digital twin model of the target building is constructed based on the building information;

[0065] Specifically, the building types described in step S2 include residential buildings, commercial buildings, industrial buildings and office buildings;

[0066] The process of constructing the digital twin model of the target building includes: using modeling techniques and tools to obtain the physical model of the building power system based on the physical principles of the acquired building basic data and power system data, describing the parameters of the power system power equipment and distribution line parameters, and then translating the physical model into a mathematical model to simulate the dynamic operation behavior of the power system, and based on the data related to the power system and combined with the mathematical model, obtaining the digital twin model of the target building.

[0067] like Figure 1-Figure 2 As shown, in the step S3, sample data of the power parameters of the target building are collected to construct a prediction model for the power parameters of the target building;

[0068] Specifically, the step of building a prediction model in step S3 is: collecting sample data of the power system operating parameters of the target building, cleaning the sample data, completing the processing of missing values ​​and abnormal values, extracting relevant features of power system failures from the sample data, building a prediction model of the power parameters of the target building, and predicting the future change trend of the power parameters; wherein, the processing of missing values ​​of the data is to calculate the average value of the sample data of the power system operating parameters of the target building, and fill the missing values ​​with the average value, and the calculation formula is:

[0069]

[0070] in, The raw values ​​of the sample data representing the power system operating parameters of the target building, represents the average value of the sample data of the power system operating parameters of the target building, The number of sample data representing the power system operating parameters of the target building;

[0071] Data outliers are processed by calculating the Z-Score for the sample data of the power system operating parameters of the target building. The Z-Score is the difference between the data and the mean value divided by the standard deviation. The standardized Z-Score calculation formula is:

[0072]

[0073] in, is the original value of the sample data of the power system operating parameters of the target building, is the mean value of the sample data of the power system operating parameters of the target building, σ is the standard deviation of the sample data of the power system operating parameters of the target building;

[0074] Define outliers as values ​​with a Z-Score greater than 3 or less than -3, and then delete or replace the outliers;

[0075] For outliers, the absolute value of the Z-Score is used to determine the range of the outlier:

[0076]

[0077] Then, delete or replace the values ​​where abs_Z>3.

[0078] Specifically, the power parameter of the target building can be power. The power time series data of the same type of building in the past period of time, such as a week, is obtained as sample data. For the same type of building, the change in power has a certain regularity, which makes it possible to predict the power of the building power system. Specifically, the power at future times can be predicted based on the power value and change trend in a certain period of time.

[0079] In a feasible embodiment, a linear regression model is used as a prediction model for the power parameters of the target building, and the specific steps are as follows:

[0080] Obtain sample data: Deploy power sensors in actual buildings to obtain power data from the building power system over a continuous period of time. Alternatively, obtain sample data through simulation and perform expansion processing to ensure that the sample data volume is sufficient.

[0081] Data preprocessing: Clean, denoise and normalize the collected data to ensure data quality. At the same time, the data needs to be labeled in time series to facilitate subsequent model training.

[0082] Model training: A linear regression model is trained using the preprocessed power data.

[0083] Model validation and optimization: Evaluate the prediction performance of the model through methods such as cross-validation, and adjust the model parameters according to actual needs to improve the prediction accuracy.

[0084] Model deployment: Deploy the trained model to the building power system. This can be achieved in the following ways:

[0085] A server is set up in the building to run the prediction model, receive power sensor data in real time and make predictions.

[0086] Alternatively, the model can be integrated into the building automation system (BAS) for real-time monitoring and prediction.

[0087] Real-time prediction and monitoring: After the model is deployed, the building’s power data can be collected in real time, and the power of the building’s power system can be predicted through the model.

[0088] In other optional embodiments, the prediction model may also adopt time series analysis (such as ARIMA model), machine learning methods (such as random forest, support vector machine) and deep learning methods (such as recurrent neural network RNN, long short-term memory network LSTM, etc.) other models, which will not be repeated here.

[0089] like Figure 1-Figure 2 As shown, in the step S4, the fault information of the building is collected to construct a fault classification model for the power system;

[0090] Specifically, the process of constructing a fault classification model described in step S4 includes: collecting basic building information and power system fault data of the target building, determining the building type and building construction parameters, as well as power equipment fault parameters and distribution line fault parameters, establishing a target building fault classification model through decision tree algorithms including ID3, C5, and CART, using information gain, information gain rate, and Gini index indicators for feature selection, generating a decision tree from top to bottom through a recursive method, and optimizing it by pre-pruning or post-pruning to improve the generalization ability of the model, generating a classification model for power system fault classification, and the classification result is a fault or no fault.

[0091] like Figure 1-Figure 2 As shown, in the step S5, the power parameters of the target building are collected in real time, recorded as real-time power parameters, and the real-time power parameters are input into the prediction model to obtain predicted power parameters;

[0092] Specifically, in step S5, the real-time power parameters are input into the prediction model to obtain the predicted power parameters: real-time sensors are arranged in the target building to collect power parameter data of the power system operation of the target building in real time, and the power parameter data is transmitted to the prediction model in real time. The prediction model uses the current power data to predict the operating status of the power system of the target building in the future to obtain the predicted power parameters.

[0093] like Figure 1-Figure 2 As shown, in the S6 step, the predicted power parameters are input into the fault classification model to obtain a fault prediction result of the power system. If the fault prediction result is that a fault has occurred, the predicted power parameters and the fault prediction result are input into the digital twin model and the digital twin model is updated. If the fault prediction result is that no fault has occurred, the predicted power parameters and the fault prediction result are input into the digital twin model and the digital twin model is updated according to a preset update rule.

[0094] Specifically, in step S6, according to the type of the target building, the frequency adjustment condition of the target building is preset, and the update rule is specifically: determine whether the target building meets the frequency adjustment condition, if so, adjust the update frequency, and update the digital twin model according to the adjusted update frequency, if not, keep the original update frequency to update the digital twin model;

[0095] In step S6, according to the preset update rule, the predicted power parameters and the fault prediction results are input into the digital twin model and the digital twin model is updated as follows:

[0096] When the target building is a residential building, a first update frequency is preset to determine whether the current time is working time or non-working time. If the current time is working time, the number of residents at home is small, the power consumption is at a low peak, and the possibility of failure of the residential building power system is low. The digital twin model is updated at a lower first update frequency. If the current time is non-working time, the number of residents at home increases, resulting in an increase in residents' electricity demand, power consumption is at a peak period, and the possibility of failure of the residential power system increases, the lower first update frequency is increased, and the digital twin model is updated at the increased first update frequency;

[0097] When the target building is a commercial building, the second update frequency is preset to determine whether the current time is in the holiday business period. If it is not in the holiday business period, the commercial building is in the normal period, the power consumption is in the low peak period, and the possibility of failure of the commercial building power system is low. The digital twin model is updated at a lower second update frequency. If the commercial building is in the holiday business period, the commercial building has undertaken large-scale commercial activities at this time, the electricity demand increases, the power consumption is in the peak period, and the possibility of failure of the commercial building power system increases, then the lower second update frequency is increased, and the digital twin model is updated at the increased second update frequency;

[0098] When the target building is an industrial building, a third update frequency is preset, the production schedule of the production equipment in the target building is obtained, and then the total production power of the production equipment is obtained, and it is determined whether the total production power at the current moment is less than the preset production power threshold. If the total production power at the previous moment is less than the preset production power threshold, the power consumption is in a low peak period, and the possibility of failure of the power system of the industrial building is low, then the digital twin model is updated according to the lower third update frequency. If the total production power at the previous moment is not less than the preset production power threshold, it means that the production activities of the industrial building are concentrated, the electricity demand is increased, the power consumption is in a peak period, and the possibility of failure of the power system of the industrial building is increased, then the lower third update frequency is increased, and the digital twin model is updated according to the increased third update frequency;

[0099] When the target building is an office building, the fourth update frequency is preset to determine whether the current time is in working hours or non-working hours. If the current time is in non-working hours, the office equipment in the office building is less used, and it is in a low-peak electricity consumption period. The possibility of failure of the office building power system is low. The digital twin model is updated according to the lower fourth update frequency. If the current time is in working hours, the office equipment in the office building is more used, the electricity demand increases, the electricity consumption is in a peak period, and the possibility of failure of the office building power system increases. The lower fourth update frequency is increased, and the digital twin model is updated according to the increased fourth update frequency.

[0100] In step S6, when the target building is a residential building, in the case of a power system failure, the average load reflects the overall load level of the power system of the target building over a period of time, and is an important reference for fault diagnosis. The maximum load reflects the load bearing capacity of the power system of the target building under extreme conditions. If the maximum load occurs frequently or exceeds the design range, the risk of failure increases. For residential buildings, since residents' electricity consumption is highly random, the safety impact of load fluctuation on the power system is usually more prominent. The load fluctuation index reflects the fluctuation of the load of the building power system. The greater the fluctuation, the worse the stability of the power system and the greater the risk of failure. The first update frequency formula and the calculation formula of the first update frequency after increase are as follows. The first update frequency is expressed as times / h, the first update frequency after the increase is expressed as Times / h:

[0101] ,

[0102] ;

[0103] in, and are all coefficients greater than 0. is a constant, represents the load fluctuation index of the target building, represents the average load of the residential building in the past certain period of time, represents the maximum load of the residential building;

[0104] Load fluctuation index of target building By calculating the standard deviation, the formula is:

[0105]

[0106] Among them, r i is the load value of the i-th period, α is the average load value, and N is the number of data;

[0107] In the above formula, during the non-peak period of electricity consumption in residential buildings, the initial first update frequency is designed by considering the relative relationship between the average load and the maximum load of residential electricity consumption. When the residential buildings are in the peak period of electricity consumption, the load volatility index is considered to have the most prominent impact on the residential building power system. This index is used as a calculation factor for the increased first update frequency, so that the set first update frequency is reasonable, and under the premise of reducing the system computing power usage as much as possible, it is ensured that the residential building power system can play an effective role in fault monitoring and diagnosis. ‌

[0108] When the target building is a commercial building, in the case of a power system failure, the average load reflects the overall load level of the power system of the target building over a period of time, and is an important reference for fault diagnosis. The maximum load reflects the load bearing capacity of the power system of the target building under extreme conditions. If the maximum load occurs frequently or exceeds the design range, the risk of failure increases. The commercial activity coefficient is correlated with the power load of the commercial building. The more frequent and more numerous the commercial activities, the greater the load on the power system. Therefore, in the power system fault diagnosis, the commercial activity coefficient is used as an important reference factor. The second update frequency formula and the increased second update frequency calculation formula are as follows. The second update frequency is expressed as times / h, the increased second update frequency is expressed as Times / h:

[0109] ,

[0110] ;

[0111] in, and are all coefficients greater than 0. is a constant, represents the commercial activity coefficient of the commercial building. If the commercial building is currently undertaking commercial activities, then , Indicates the number of commercial activities currently undertaken by the commercial building, otherwise , represents the average load of the commercial building in the past certain period of time, represents the maximum load of said commercial building;

[0112] In the above formula, during the non-peak period of commercial buildings, the initial second update frequency is designed by considering the relative relationship between the average load and the maximum load of the commercial buildings. When the commercial buildings are in the peak period of electricity consumption, the commercial activity coefficient is highly correlated with the power load of the commercial buildings. The more frequent and numerous the commercial activities are, the greater the load on the power system. The commercial activity coefficient is taken into consideration in the calculation of the increased second update frequency, so that the set second update frequency is reasonable, and under the premise of minimizing the system computing power usage, it is ensured that the commercial building power system can be effectively monitored and diagnosed for faults. ‌

[0113] When the target building is an industrial building, in the event of a power system failure, the average load reflects the overall load level of the power system of the target building over a period of time, and is an important reference for fault diagnosis. The maximum load reflects the load-bearing capacity of the power system of the target building under extreme conditions. If the maximum load occurs frequently or exceeds the design range, the risk of failure increases. In addition, for manufacturing enterprises inside industrial buildings, they usually have corresponding production plans, that is, for an industrial building, the load situation of the power system is highly correlated with the production plans of each enterprise. In this embodiment, the difference between the total production power of the industrial building and the production power threshold is calculated to reflect the power load situation of the industrial building. The production power threshold can be understood as the early warning threshold of the power system of the industrial building. When the difference is positive, it means that the power load exceeds the early warning threshold. The larger the value is, the more the total scheduled power exceeds the warning threshold, and the greater the possibility of system failure. Therefore, the third update frequency is set to a higher value to achieve effective fault monitoring. Specifically, the third update frequency formula and the calculation formula of the increased third update frequency are as follows: times / h, the increased third update frequency is expressed as Times / h:

[0114] ,

[0115] ;

[0116] in, and are all coefficients greater than 0. is a constant, Indicates the difference between the total production scheduling power at the current moment and the production scheduling power threshold. represents the average load of the industrial building in the past certain period of time, represents the maximum load of the industrial building in question;

[0117] In the above formula, during the off-peak period of industrial buildings, the initial third update frequency is designed by considering the relative relationship between the average load and the maximum load of industrial electricity. When residential buildings are in peak electricity consumption, the load situation of the power system is considered to be highly correlated with the production schedule. Therefore, in the process of setting the third update frequency, the difference between the total production power of industrial buildings and the production power threshold is introduced as a calculation factor for the third update frequency after the industrial building is enlarged. This makes the set third update frequency reasonable, and ensures effective fault monitoring and diagnosis of the industrial building power system while minimizing the system computing power usage.

[0118] When the target building is an office building, the high power usage of office equipment on weekdays leads to an increase in the load of the power system, thereby increasing the risk of power system failure. At the same time, an analysis of the power system of the office building shows that the factor that has the greatest impact on the power system in the office field is the fresh air system. Specifically, the fresh air system occupies the largest power of the power system. Therefore, the working power of the fresh air system is introduced as a consideration in the setting process of the fourth update frequency. It is easy to understand that the greater the working power of the fresh air system, the greater the impact and possible risks it will have on the building power system during its operation. The fourth update frequency is expressed as times / h, the increased fourth update frequency is expressed as Times / h:

[0119] ,

[0120] ;

[0121] in, and is a coefficient greater than 0, is a constant, represents the average working power of the office equipment in the office building within a certain period of time, represents the average working power of the fresh air system of the office building in a certain period of time, represents the average load of the office building in the past certain period of time, represents the maximum load of said office building;

[0122] In the above formula, during the non-peak period of electricity consumption in the office building, the initial first update frequency is designed by considering the relative relationship between the average load and the maximum load of the office equipment electricity consumption. When the office building is in the peak period of electricity consumption, considering that the fresh air system occupies the largest power of the power system, this indicator is used as a calculation factor for the increased fourth update frequency, so that the set fourth new frequency is reasonable. Under the premise of reducing the system computing power occupancy as much as possible, it is ensured that the office building power system can play an effective fault monitoring and diagnosis role.

[0123] like Figure 1-Figure 2 As shown, the present invention solves the shortcomings of data transmission lag and huge simulation calculation of digital twin models in the prior art by working as follows:

[0124] Through the analysis of four types of buildings, namely residential buildings, commercial buildings, industrial buildings and office buildings, during the non-peak period of electricity consumption, the relative relationship between the average load and the maximum load of electricity consumption is considered to design the initial lower first update frequency, the initial lower second update frequency, the initial lower third update frequency and the initial lower fourth update frequency. When the building is in the peak period of electricity consumption, for residential buildings, since residential electricity consumption is highly random, the safety impact of load fluctuation on the power system of residential buildings is usually more prominent, and this indicator is used as a calculation factor for the first update frequency after the increase of residential buildings. For commercial buildings, the frequency and quantity of commercial activities have a greater impact on the load of the power system. Therefore, in the power system fault diagnosis, the commercial activity coefficient is used as a calculation factor for the second update frequency after the increase of commercial buildings. For an industrial building in this embodiment, For industrial buildings, the load of the power system is highly correlated with the production schedule. The difference between the total production power of industrial buildings and the production power threshold is calculated to reflect the power load of industrial buildings. Therefore, the difference between the total production power of industrial buildings and the production power threshold is introduced in the setting process of the third update frequency. This indicator is used as a calculation factor for the third update frequency after the industrial building is increased. For office buildings, analysis shows that the factor that has the greatest impact on the power system in the office field is the fresh air system. The power occupancy of the fresh air system is the largest. Therefore, this indicator is used as a calculation factor for the fourth update frequency after the office building is increased, so that the corresponding settings of the increased first update frequency, second update frequency, third update frequency, and fourth update frequency are reasonable. Under the premise of reducing the system computing power occupancy as much as possible, it is ensured that the power systems of various types of buildings can be effectively monitored and diagnosed.

[0125] In one embodiment, the present invention works as follows:

[0126] Get the target building information:

[0127] Obtaining target building basic data and power system data. The building basic data includes building parameters of residential buildings, commercial buildings, industrial buildings, and office buildings. The power system data includes the type, quantity, and rated power of electrical equipment, as well as the distribution location, maximum load voltage, maximum load current, and maximum composite power of distribution lines.

[0128] Building a digital twin model:

[0129] Using modeling techniques and tools, based on the physical principles of the acquired building basic data and power system data, the physical model of the building power system is obtained to describe the parameters of the power system's electrical equipment and distribution lines. The physical model is then converted into a mathematical model to simulate the dynamic operation of the power system. Based on the power system-related data and combined with the mathematical model, a digital twin model of the target building is obtained.

[0130] Building a predictive model:

[0131] Collect sample data of power parameters of the target building, clean, denoise and normalize the power sample data, use the preprocessed power data to train a linear regression model, and evaluate the prediction performance of the model through cross-validation. After adjusting the model parameters according to actual needs, deploy the trained prediction model to the building power system through the server to predict whether the power system has a fault.

[0132] Build a fault classification model:

[0133] Collect building operation data to obtain predicted power parameter fault information, including basic building information and power system fault parameters of electrical equipment and distribution line faults, and build a fault classification model for the power system through a decision tree algorithm;

[0134] Get predicted power parameters:

[0135] Arrange real-time sensors in the target building to collect the power parameter data of the power system operation of the target building in real time, and transmit the power data to the prediction model in real time. The prediction model uses the current power data to predict the operation status of the power system of the target building in the future and obtain the predicted power parameters;

[0136] Get the fault prediction results:

[0137] Input the predicted power parameters into the fault classification model to obtain the fault prediction results of the power system.

[0138] The following fault conditions are preset: the target building where the fault occurs is an office building, this time period is working hours, the average working power of the office equipment in the office building in a week is 60W, the average working power of the fresh air system in the office building in a week is 200W, the average load in the past week before the fault occurs is 240kVA, and the maximum load of the office building is 1600kVA. and 1. and , is 2;

[0139] Calculate update frequency:

[0140] When the target building is an office building, the fourth update frequency is expressed as times / h, the increased fourth update frequency is expressed as times / h, the calculation process is as follows:

[0141] ,

[0142] ;

[0143] At this time, the update frequency of the digital twin model of the office building is 18.15 times / h. In the digital twin model of the office building, the digital twin model is updated according to the increased update frequency of 18.15 times / h.

[0144] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0145] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and rules of the present invention shall be included in the protection scope of the present invention.

Claims

1. A building power system fault diagnosis method using digital twin technology, characterized in that: The following steps are involved: S1. Acquire building information of a target building, wherein the building information includes basic building data and power system data of the target building; S2. constructing a digital twin model of the target building based on the building information; S3. Collecting sample data of the power parameters of the target building and constructing a prediction model for the power parameters of the target building; S4, collect building fault information and build a fault classification model for the power system; S5. Collect the power parameters of the target building in real time, record them as real-time power parameters, input the real-time power parameters into the prediction model, and obtain predicted power parameters; S6. Input the predicted power parameter into the fault classification model to obtain a fault prediction result of the power system. If the fault prediction result is that a fault occurs, input the predicted power parameter and the fault prediction result into the digital twin model and update the digital twin model. If the fault prediction result is that no fault occurs, input the predicted power parameter and the fault prediction result into the digital twin model according to a preset update rule and update the digital twin model. In step S6, according to the preset update rules, the predicted power parameters and the fault prediction results are input into the digital twin model and the digital twin model is updated as follows: When the target building is a residential building, a first update frequency is preset to determine whether the current time is in working hours or non-working hours. If the current time is in working hours, the digital twin model is updated according to the first update frequency. If the current time is in non-working hours, the first update frequency is increased, and the digital twin model is updated according to the increased first update frequency. When the target building is a commercial building, a second update frequency is preset to determine whether the current time is in a holiday business period. If not, the digital twin model is updated at the second update frequency. If so, the second update frequency is increased, and the digital twin model is updated at the increased second update frequency. When the target building is an industrial building, a third update frequency is preset, the production schedule of the production equipment in the target building is obtained, and then the total production schedule power of the production equipment is obtained, and it is determined whether the total production schedule power at the current moment is less than the preset production schedule power threshold. If so, the digital twin model is updated according to the third update frequency; if not, the third update frequency is increased, and the digital twin model is updated according to the increased third update frequency; When the target building is an office building, a fourth update frequency is preset to determine whether the current time is in working hours or non-working hours. If the current time is in non-working hours, the digital twin model is updated according to the fourth update frequency. If the current time is in working hours, the fourth update frequency is increased, and the digital twin model is updated according to the increased fourth update frequency. When the target building is a residential building, the first update frequency is expressed as times / h, the first update frequency after the increase is expressed as Times / h: ， ； in, and are all coefficients greater than 0. is a constant, represents the load volatility index of the target building, the load volatility index of the target building By calculating the standard deviation, the formula is: Among them, r i is the load value of the i-th period, α is the average load value, N is the number of data, represents the average load of the residential building in the past certain period of time, represents the maximum load of the residential building; When the target building is a commercial building, the second update frequency is expressed as times / h, the increased second update frequency is expressed as Times / h: ， ； in, and are all coefficients greater than 0. is a constant, represents the commercial activity coefficient of the commercial building. If the commercial building is currently undertaking commercial activities, then , Indicates the number of commercial activities currently undertaken by the commercial building, otherwise , represents the average load of the commercial building in the past certain period of time, represents the maximum load of said commercial building; When the target building is an industrial building, the third update frequency is expressed as times / h, The increased third update frequency is expressed as Times / h: ， ； in, and are all coefficients greater than 0. is a constant, Indicates the difference between the total production scheduling power at the current moment and the production scheduling power threshold. represents the average load of the industrial building in the past certain period of time, represents the maximum load of the industrial building in question; When the target building is an office building, the fourth update frequency is expressed as times / h, The increased fourth update frequency is expressed as Times / h: ， ； in, and is a coefficient greater than 0, is a constant, represents the average working power of the office equipment in the office building within a certain period of time, represents the average working power of the fresh air system of the office building in a certain period of time, represents the average load of the office building in the past certain period of time, represents the maximum load of the office building.

2. A building power system fault diagnosis method using digital twin technology according to claim 1, characterized in that: In step S1, the basic building data includes the building type and building construction parameters, the power system data includes the building's power equipment parameters and distribution line parameters, the power equipment parameters include the type, quantity and rated power of the power equipment, and the distribution line parameters include the distribution location, maximum load voltage, maximum load current and maximum load power of the distribution line; In steps S3 and S5, the power parameters include the total operating power of the power equipment of the target building, and the voltage, current and power of the distribution lines.

3. The method for diagnosing a fault in a building power system using digital twin technology according to claim 1, characterized in that: In step S2, the process of constructing the digital twin model of the target building includes: using modeling techniques and tools to obtain the physical model of the building power system based on the physical principles of the acquired building basic data and power system data, describing the parameters of the power system power equipment and distribution line parameters, and then translating the physical model into a mathematical model to simulate the dynamic operation behavior of the power system, and based on the data related to the power system and combined with the mathematical model, obtain the digital twin model of the target building.

4. The method for diagnosing a fault in a building power system using digital twin technology according to claim 1, characterized in that: In step S3, the steps for constructing a prediction model for the power parameters of the target building are: collecting sample data of the power system operating parameters of the target building, cleaning the sample data, completing the processing of missing values ​​and outliers, extracting relevant features of power system failures from the sample data, and constructing a linear regression prediction model for the power parameters of the target building.

5. The method for diagnosing faults in a building power system using digital twin technology according to claim 1, characterized in that: In step S4, basic building information and power system fault data of the target building are collected, the building type and building construction parameters, as well as power equipment fault parameters and distribution line fault parameters are determined, and a target building fault classification model is established through decision tree algorithms including ID3, C5, and CART. Information gain, information gain rate, and Gini index indicators are used for feature selection, and a decision tree is generated from top to bottom in a recursive manner. The tree is optimized by pre-pruning or post-pruning to generate a classification model for power system fault classification.

6. The method for diagnosing a fault in a building power system using digital twin technology according to claim 1, characterized in that: In step S5, the real-time power parameters are input into the prediction model, and the predicted power parameters are obtained by arranging real-time sensors in the target building, collecting the power system operation parameter data of the target building in real time, and transmitting the data to the prediction model in real time. The prediction model uses the current data to predict the operating status of the power system of the target building in the future and obtain the predicted power parameters.

Images