BMS battery management method based on BIM model
Through the BMS battery management method based on the BIM model, combined with sensors and Internet of Things technology, and using machine learning algorithms to monitor and optimize battery status, the shortcomings of traditional BMS in battery management are solved, and the fine monitoring of battery status and life extension are achieved. Improves system efficiency and reliability.
Patent Information
- Application Number
- CN202410918809.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-05
AI Technical Summary
Traditional BMSs find it difficult to fully consider the mutual influence between building power systems and batteries, and lack comprehensive health assessment and dynamic optimization capabilities, resulting in insufficient battery management, making it difficult to extend battery life and improve system efficiency.
Using BMS battery management method based on BIM model, through data integration, real-time monitoring, timing analysis, abnormal evaluation and pattern recognition, combined with sensor and Internet of Things technology, machine learning algorithms are used to establish a time series model and support vector machine model to realize battery state real-time monitoring and optimization strategies.
It realizes comprehensive and accurate monitoring and management of battery status, predicts battery performance and health status in advance, recognizes abnormal situations, optimizes charging and discharging strategies, extends battery life, and improves system efficiency and reliability.
Smart Images

Figure CN118899558A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of battery management, and in particular to a BMS battery management method based on a BIM model. Background Art
[0002] With the widespread use of batteries in power systems, the importance of battery management systems has become increasingly prominent. Traditional BMS mainly relies on a single data source for battery status monitoring, which makes it difficult to fully consider the mutual influence between the building power system and the battery. Therefore, a battery management system that integrates building information models and BMS came into being. By combining building power system information with battery data, more comprehensive battery status monitoring and management can be achieved.
[0003] At this stage, traditional battery management methods mainly rely on single data monitoring and simple abnormal alarm mechanisms. They lack comprehensive health assessment and dynamic optimization capabilities, and are not easy to conduct real-time analysis of abnormal conditions of the battery in the future. After detecting abnormal conditions, it is not easy to distinguish the specific causes of the battery failure during operation. At the same time, due to the lack of in-depth analysis of battery operation modes, traditional BMS is not easy to provide effective charging and discharging optimization strategies, making it difficult to fully extend battery life and improve the overall efficiency of the system. Summary of the invention
[0004] In view of the deficiencies in the prior art, the present invention provides a BMS battery management method based on a BIM model, which solves the problems mentioned in the background technology.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A BMS battery management method based on a BIM model, comprising the following steps:
[0006] S1. Data integration: Integrate the building power system information in the BIM model with the battery data in the BMS system and establish a joint database;
[0007] S2. Real-time monitoring: Use sensors and IoT technology to monitor battery status data in real time, and then use the visualization function of the BIM model to display and collect battery status and the operation of the building's power system in real time, and clean and pre-process the collected data;
[0008] S3, time series analysis: establish a time series model and use machine learning algorithms to analyze historical data to predict future battery performance and health status, and then input the real-time collected battery status data into the time series model for calculation to obtain the battery health status Jk;
[0009] S4, abnormality assessment: preset the first battery abnormality threshold value Y1 and the second battery abnormality threshold value Y2 and compare and evaluate the battery health status Jk obtained based on the prediction result to analyze the abnormal condition of the battery at the future moment;
[0010] S5. Pattern recognition: When the battery health is in an abnormal state, the support vector machine (SVM) model is used to identify different battery operating modes, analyze the causes of the battery health abnormality, and optimize the charging and discharging strategy based on dynamic programming.
[0011] Preferably, according to step S1, data integration specifically includes the following steps:
[0012] S11. Collect the building power system data and battery management system data in real time through sensors, smart meters and BMS system equipment, and use the MQTT transmission protocol of the Internet of Things technology to transmit the collected power system data and battery management system data to the central database, and then convert and standardize the power system data and battery management system data from the BIM model and BMS system. The standardization includes data exchange in JSON and XML data formats, and using a relational database to store the integrated data;
[0013] S12. Clean the collected data, remove outliers and noise, and fill in missing values. Then map and match the building power system information in the BIM model with the battery data in the BMS system. Use BIM software and visualization tools to visualize the integrated power system data and battery management system data.
[0014] Preferably, the real-time monitoring according to step S2 specifically includes the following steps:
[0015] S21. Install current sensors, voltage sensors, temperature sensors and Hall effect sensors around the power system and batteries inside the building, collect battery status data collected by the sensors through the Internet of Things platform, and transmit the battery status data to the database, wherein the battery status data includes voltage V, temperature W, charging current Ic and discharging current If, and import the power system model of the building into Autodesk Revit, and implement real-time update and display of sensor data through Revit API;
[0016] S22. After the battery status data is stored in the database, data preprocessing is performed using a Python script. The preprocessing method includes removing outliers, filling missing values, data denoising, and normalization.
[0017] Preferably, the timing analysis according to step S3 specifically includes the following steps:
[0018] S31. Select a linear regression algorithm to build a time series model, and then divide the historical battery status data into a training set, a test set, and a validation set. Use the training set to train the time series model of the model and determine the regression coefficient a. Use the test set to test the prediction performance of the time series model through mean square error, root mean square error, and mean absolute error. Based on the test results obtained from the test set, verify the time series model through the validation set to evaluate the accuracy of the time series model.
[0019] Preferably, S32, inputting the real-time collected battery status data into a verified time series model, predicting the future battery health status Jk, and analyzing the battery health status Jk(t+n) at the future moment;
[0020] The battery health status Jk(t+n) is obtained by the following algorithm formula;
[0021] Jk(t+n)=a1+a2×t+a3×V(t)+a4×W(t)+a5×(Ic+If)(t);
[0022] Where V(t) represents the voltage at time t, W(t) represents the temperature at time t, (Ic+If)(t) represents the charging current and discharging current at time t, t represents time, and n represents the prediction time step;
[0023] a1 represents the intercept term, which indicates the value of the dependent variable when all independent variables are zero;
[0024] a2 represents the regression coefficient related to time t, indicating the influence of time on the battery health status;
[0025] a3 represents the regression coefficient related to the voltage V(t), which indicates the influence of the voltage on the battery health status;
[0026] a4 represents the regression coefficient related to temperature W(t), which indicates the influence of temperature on the battery health status;
[0027] a5 represents the regression coefficient related to the charging current and the discharging current (Ic+If)(t), which indicates the influence of the current on the battery health status.
[0028] Preferably, the abnormality assessment according to step S4 specifically includes the following steps:
[0029] S41, based on the analysis of the historical normal battery status data, the maximum safety value and the minimum safety value of the normal battery, the first battery abnormal threshold Y1 and the second battery abnormal threshold Y2 are preset, and then the abnormality assessment is performed according to the battery health status Jk(t+n) at the future moment output by the time series model. The specific assessment scheme is as follows;
[0030] When the battery health status Jk(t+n)> the first battery abnormality threshold Y1, it means that the battery health status is abnormal at the future time t+n. At this time, the pattern recognition is started to further analyze the cause of the battery health abnormality;
[0031] When the second battery abnormality threshold Y2≤battery health status Jk(t+n)≤first battery abnormality threshold Y1, it indicates that the battery health status at the future moment is in a normal state, and no further operation is required at this time;
[0032] When the battery health state Jk(t+n) is less than the second battery abnormality threshold Y2, it means that at the future time t+n, the battery health state is abnormal. At this time, pattern recognition is started to further analyze the cause of the battery health abnormality.
[0033] Preferably, the pattern recognition according to step S5 specifically includes the following steps:
[0034] S51. When the battery health status Jk(t+n) is predicted to be in an abnormal state at a future moment through the time series model, the pattern recognition function is automatically started to analyze the cause of the abnormal battery health by converting the data in the battery status data into a feature vector xi, wherein the feature vector xi includes {voltage vector x1, temperature vector x2, charging current vector x3 and discharging current vector x4}.
[0035] Preferably, S52, using a support vector machine (SVM) algorithm in a machine learning algorithm to classify the extracted feature vectors, identify the battery in different operating modes, and then identify the failure mode of the battery according to the classification result F(xi), and analyze the cause of the failure;
[0036] The classification result F(xi) is obtained by the following algorithm formula;
[0037] F(x)=sign(Q×xi×b);
[0038] In the formula, xi represents the currently monitored feature vector, Q represents the weight vector, and b represents the bias.
[0039] Preferably, S53, analyzing the cause of the current battery abnormality according to the result output by the classification result F(xi), the specific analysis scheme is as follows;
[0040] When the classification result F(xi)>1, it means that the current feature vector xi causes battery operation failure, and the optimized charging and discharging strategy is automatically started at this time;
[0041] When the classification result F(xi)≤1, it means that the current feature vector xi does not cause battery operation failure, and the next feature vector xi is further analyzed.
[0042] Preferably, S54, when it is monitored that the current characteristic vector xi causes a battery operation failure, then the charging and discharging strategy is further optimized based on dynamic programming through a dynamic programming algorithm, so that the battery can extend its life while ensuring the power supply demand. The specific planning content is as follows;
[0043]
[0044] Where U represents the utility function, C(t) represents the charging strategy at time t, D(t) represents the discharging strategy at time t, and T represents the total length of the time range.
[0045] The present invention provides a BMS battery management method based on the BIM model, which has the following beneficial effects:
[0046] (1) This method integrates the building power system information in the BIM model with the battery data in the BMS system to establish a joint database to achieve the docking and unified management of the power system and battery data. The building power system data and battery management system data are collected in real time, and the collected data is transmitted to the database using the Internet of Things technology. The collected data is converted and standardized in the database. Combined with the real-time monitoring of the Internet of Things technology and sensors, the battery status and the operation of the power system can be fully and accurately collected and displayed. Sensors are installed around the power system and batteries inside the building, and the battery status data collected by the sensors is collected through the Internet of Things platform and transmitted to the database. At the same time, the real-time update and display of sensor data is realized through the Revit API. By cleaning and preprocessing the collected data, these steps improve the accuracy and comprehensiveness of battery management, so that the power system and battery data can be more effectively monitored and managed, providing a solid data foundation for subsequent analysis and optimization.
[0047] (2) This method uses machine learning algorithms to build a time series model, analyze and predict historical data, and predict the performance and health status of the battery in advance. By real-time monitoring and analysis of battery status data, battery health anomalies can be discovered in a timely manner and appropriate measures can be taken to deal with them. The abnormal evaluation step can effectively identify abnormal battery conditions in the future by presetting battery abnormality thresholds, and activate pattern recognition functions for in-depth analysis.
[0048] (3) This method can accurately identify different operating modes and fault causes of batteries through the pattern recognition function based on the support vector machine (SVM) model, and provide detailed analysis and solutions for battery abnormalities. By optimizing the charging and discharging strategy through dynamic programming, the battery can maintain the best operating state while meeting the power supply demand. This method not only improves the operating efficiency of the battery management system, but also greatly improves the reliability of the system, reduces the probability of battery failure, and ensures the stable and safe operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 A schematic diagram of the steps of a BMS battery management method based on a BIM model of the present invention;
[0050] Figure 2 The present invention is a flow chart of a BMS battery management method based on a BIM model. DETAILED DESCRIPTION
[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0052] Example 1
[0053] The present invention provides a BMS battery management method based on the BIM model, please refer to Figure 1 and Figure 2 , including the following steps:
[0054] S1. Data integration: Integrate the building power system information in the BIM model with the battery data in the BMS system and establish a joint database;
[0055] S2. Real-time monitoring: Use sensors and IoT technology to monitor battery status data in real time, and then use the visualization function of the BIM model to display and collect battery status and the operation of the building's power system in real time, and clean and pre-process the collected data;
[0056] S3, time series analysis: establish a time series model and use machine learning algorithms to analyze historical data to predict future battery performance and health status, and then input the real-time collected battery status data into the time series model for calculation to obtain the battery health status Jk;
[0057] S4, abnormality assessment: preset the first battery abnormality threshold value Y1 and the second battery abnormality threshold value Y2 and compare and evaluate the battery health status Jk obtained based on the prediction result to analyze the abnormal condition of the battery at the future moment;
[0058] S5. Pattern recognition: When the battery health is in an abnormal state, the support vector machine (SVM) model is used to identify different battery operating modes, analyze the causes of the battery health abnormality, and optimize the charging and discharging strategy based on dynamic programming.
[0059] In this embodiment, the method integrates the building power system information with the battery data in the data integration stage to establish a joint database, thereby achieving unified data management and improving data accessibility and accuracy. Secondly, in the real-time monitoring link, the battery status data is monitored in real time through sensors and Internet of Things technology, and the battery status and the operation of the building power system are displayed through the BIM model, thereby achieving comprehensive monitoring and management of the power system. Then, in the time series analysis stage, a machine learning algorithm is used to establish a time series model to predict the performance and health status of the battery in the future, making battery management more intelligent and accurate. In the abnormal evaluation stage, by comparing and evaluating the preset battery abnormality threshold with the prediction results, the abnormal battery condition at the future moment can be discovered and analyzed in time, thereby improving the efficiency of fault detection and processing. Finally, in the pattern recognition stage, the support vector machine SVM model is used to identify different operating modes of the battery, and the charging and discharging strategy is optimized through dynamic programming, effectively extending the battery life.
[0060] Example 2
[0061] This embodiment is explained in Example 1, please refer to Figure 1 and Figure 2 According to step S1, data integration specifically includes the following steps:
[0062] S11. Collect the building power system data and battery management system data in real time through sensors, smart meters and BMS system equipment, and use the MQTT transmission protocol of the Internet of Things technology to transmit the collected power system data and battery management system data to the central database, and then convert and standardize the power system data and battery management system data from the BIM model and BMS system. The standardization includes data exchange in JSON and XML data formats, and using a relational database to store the integrated data;
[0063] S12. Clean the collected data, remove outliers and noise, and fill in missing values. Then map and match the building power system information in the BIM model with the battery data in the BMS system. Use BIM software and visualization tools to visualize the integrated power system data and battery management system data.
[0064] The real-time monitoring according to step S2 specifically includes the following steps:
[0065] S21. Install current sensors, voltage sensors, temperature sensors and Hall effect sensors around the power system and batteries inside the building, collect battery status data collected by the sensors through the Internet of Things platform, and transmit the battery status data to the database. The battery status data includes voltage V, temperature W, charging current Ic and discharging current If. Import the power system model of the building into Autodesk Revit, and implement real-time update and display of sensor data through Revit API.
[0066] S22. After the battery status data is stored in the database, the data is preprocessed through a Python script. The preprocessing method includes removing outliers, filling missing values, data denoising and normalization.
[0067] In this embodiment, in terms of data integration, the method collects building power system data and battery management system data in real time by using sensors, smart meters and BMS system equipment, and transmits them to the database, thereby realizing efficient data integration and unified management. At the same time, the collected data is cleaned and formatted, outliers and noise are removed, the accuracy and completeness of the data are ensured, and a reliable data basis is provided for subsequent analysis. Secondly, in terms of real-time monitoring, by installing various sensors to monitor battery status data in real time, and storing the battery status data in the database, comprehensive and accurate monitoring of the battery status is realized. Through data preprocessing, the data quality is further improved, providing reliable data support for subsequent analysis and decision-making.
[0068] Example 3
[0069] This embodiment is explained in Example 2. Please refer to Figure 1 and Figure 2 According to step S3, the timing analysis specifically includes the following steps:
[0070] S31. Select a linear regression algorithm to build a time series model, and then divide the historical battery status data into a training set, a test set, and a validation set. Use the training set to train the time series model of the model and determine the regression coefficient a. Use the test set to test the prediction performance of the time series model through mean square error, root mean square error, and mean absolute error. Based on the test results obtained from the test set, verify the time series model through the validation set to evaluate the accuracy of the time series model.
[0071] S32, inputting the real-time collected battery status data into the verified time series model, predicting the future battery health status Jk, and analyzing the battery health status Jk(t+n) at the future moment;
[0072] The battery health status Jk(t+n) is obtained by the following algorithm formula;
[0073] Jk(t+n)=a1+a2×t+a3×V(t)+a4×W(t)+a5×(Ic+If)(t);
[0074] Where V(t) represents the voltage at time t, W(t) represents the temperature at time t, (Ic+If)(t) represents the charging current and discharging current at time t, t represents time, and n represents the prediction time step;
[0075] a1 represents the intercept term, which indicates the value of the dependent variable when all independent variables are zero;
[0076] a2 represents the regression coefficient related to time t, indicating the influence of time on the battery health status;
[0077] a3 represents the regression coefficient related to the voltage V(t), which indicates the influence of the voltage on the battery health status;
[0078] a4 represents the regression coefficient related to temperature W(t), which indicates the influence of temperature on the battery health status;
[0079] a5 represents the regression coefficient related to the charging current and the discharging current (Ic+If)(t), which indicates the influence of the current on the battery health status.
[0080] The abnormality assessment according to step S4 specifically includes the following steps:
[0081] S41, based on the analysis of the historical normal battery status data, the maximum safety value and the minimum safety value of the normal battery, the first battery abnormal threshold Y1 and the second battery abnormal threshold Y2 are preset, and then the abnormality assessment is performed according to the battery health status Jk(t+n) at the future moment output by the time series model. The specific assessment scheme is as follows;
[0082] When the battery health status Jk(t+n)> the first battery abnormality threshold Y1, it means that the battery health status is abnormal at the future time t+n. At this time, the pattern recognition is started to further analyze the cause of the battery health abnormality;
[0083] When the second battery abnormality threshold Y2≤battery health status Jk(t+n)≤first battery abnormality threshold Y1, it indicates that the battery health status at the future moment is in a normal state, and no further operation is required at this time;
[0084] When the battery health state Jk(t+n) is less than the second battery abnormality threshold Y2, it means that at the future time t+n, the battery health state is abnormal. At this time, pattern recognition is started to further analyze the cause of the battery health abnormality.
[0085] In this embodiment, the method establishes a time series model by selecting a linear regression algorithm. The purpose of this model is to use the battery status data collected in real time to analyze trends and predict the future battery health status Jk(t+n). The historical data is divided into a training set, a test set and a validation set. The model is trained by the training set to determine the regression coefficient a. Then, the test set is used to evaluate the prediction performance of the model, and the accuracy of the model is evaluated by indicators such as mean square error, root mean square error and mean absolute error. Finally, the accuracy of the model is verified by the validation set to ensure that the model can effectively predict the battery health status. The analysis is based on the historical normal battery status data, and the first battery abnormal threshold Y1 and the second battery abnormal threshold Y2 are preset. Help to determine whether there is an abnormality in the battery health status. Specifically, when the battery health status at a future moment exceeds or is lower than the interval of the set first battery abnormal threshold Y1 and the second battery abnormal threshold Y2, the pattern recognition mechanism will be triggered. Different measures are taken according to different situations of the battery health status.
[0086] Example 4
[0087] This embodiment is explained in Example 3, please refer to Figure 1 and Figure 2 According to step S5, the pattern recognition specifically includes the following steps:
[0088] S51. When the battery health status Jk(t+n) is predicted to be in an abnormal state at a future moment through the time series model, the pattern recognition function is automatically started to analyze the cause of the abnormal battery health by converting the data in the battery status data into a feature vector xi, which includes {voltage vector x1, temperature vector x2, charging current vector x3 and discharging current vector x4}.
[0089] S52, using the support vector machine (SVM) algorithm in the machine learning algorithm to classify the extracted feature vectors, identify the battery in different operating modes, and then identify the failure mode of the battery according to the classification result F(xi), and analyze the cause of the failure;
[0090] The classification result F(xi) is obtained by the following algorithm formula;
[0091] F(x)=sign(Q×xi×b);
[0092] In the formula, xi represents the currently monitored feature vector, Q represents the weight vector, and b represents the bias.
[0093] S53, analyzing the cause of the current battery abnormality according to the result output by the classification result F(xi), the specific analysis scheme is as follows;
[0094] When the classification result F(xi)>1, it means that the current feature vector xi causes battery operation failure, and the optimized charging and discharging strategy is automatically started at this time;
[0095] When the classification result F(xi)≤1, it means that the current feature vector xi does not cause battery operation failure, and the next feature vector xi is further analyzed.
[0096] In this embodiment, the method predicts whether the battery health state is in an abnormal state at a future moment through a time series model. Once an abnormality is predicted, the pattern recognition function is immediately started to deeply analyze the cause of the battery health abnormality. This analysis process includes converting the battery status data into a feature vector xi, and then classifying the extracted feature vector xi using the support vector machine SVM machine learning algorithm to identify the state of the battery under different operating modes. According to the classification results, the failure mode of the battery can be further identified and the cause of the failure can be analyzed. In the specific analysis scheme, the classification result F is used as a judgment criterion. If the classification result is greater than 1, it indicates that the current feature vector xi causes a battery operation failure, and the optimized charging and discharging strategy will be immediately started to ensure the safety and stability of the battery operation. If the classification result is less than or equal to 1, it means that the current feature vector xi does not cause a battery operation failure, and the next feature vector xi will be further analyzed to fully understand the battery operation status. This pattern recognition method effectively improves the efficiency of identifying and processing battery health anomalies, and helps to ensure the reliability and continuous operation of the battery system.
[0097] Example 5
[0098] This embodiment is explained in Example 4. Please refer to Figure 1 and Figure 2 , S54, when it is monitored that the current characteristic vector xi causes a battery operation failure, then the charging and discharging strategy is further optimized based on dynamic programming through a dynamic programming algorithm, so that the battery can extend its life while ensuring the power supply demand. The specific planning content is as follows;
[0099]
[0100] Where U represents the utility function, C(t) represents the charging strategy at time t, D(t) represents the discharging strategy at time t, and T represents the total length of the time range.
[0101] In this embodiment, when the method detects that the current characteristic vector xi causes a battery operation failure, it further adopts a method based on dynamic programming to optimize the charging and discharging strategy. The goal of this optimization strategy is to extend the service life of the battery as much as possible while ensuring that the power supply demand is met. Specifically, through the dynamic programming algorithm, a set of effective charging and discharging strategies are designed to maximize the utility function U of the battery. In this process, the charging strategy and the discharging strategy within the time range T are considered. By dynamically adjusting the charging strategy and the discharging strategy, the loss rate of the battery can be reduced and its service life can be extended while ensuring the power supply demand. This optimization strategy based on dynamic programming brings significant benefits to the battery management system, which not only improves the reliability and stability of the battery, but also reduces maintenance costs and energy consumption, thereby further improving the performance and efficiency of the system.
[0102] Specific example:
[0103] Real-time monitoring of battery voltage, temperature and current data, through time series analysis, found that the battery voltage will have a peak between 8 and 10 every morning, which may be due to the solar panels on the roof of the building charging efficiency is the highest during this period. Through pattern recognition, it can be found that when the battery temperature exceeds 50 degrees Celsius, the battery charging current will suddenly drop, which may be the result of the activation of the battery overheating protection mechanism. Through these analyses, the charging and discharging strategy can be optimized to prevent battery overheating and extend battery life.
[0104] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A BMS battery management method based on a BIM model, characterized in that: The following steps are involved: S1. Data integration: Integrate the building power system information in the BIM model with the battery data in the BMS system and establish a joint database; S2. Real-time monitoring: Use sensors and IoT technology to monitor battery status data in real time, and then use the visualization function of the BIM model to display and collect battery status and the operation of the building's power system in real time, and clean and pre-process the collected data; S3, time series analysis: establish a time series model and use machine learning algorithms to analyze historical data to predict future battery performance and health status, and then input the real-time collected battery status data into the time series model for calculation to obtain the battery health status Jk; S4, abnormality assessment: preset the first battery abnormality threshold value Y1 and the second battery abnormality threshold value Y2 and compare and evaluate the battery health status Jk obtained based on the prediction result to analyze the abnormal condition of the battery at the future moment; S5. Pattern recognition: When the battery health is in an abnormal state, the support vector machine (SVM) model is used to identify different battery operating modes, analyze the causes of the battery health abnormality, and optimize the charging and discharging strategy based on dynamic programming.
2. According to a BMS battery management method based on a BIM model according to claim 1, it is characterized in that: According to step S1, data integration specifically includes the following steps: S11. Collect the building power system data and battery management system data in real time through sensors, smart meters and BMS system equipment, and use the MQTT transmission protocol of the Internet of Things technology to transmit the collected power system data and battery management system data to the central database, and then convert and standardize the power system data and battery management system data from the BIM model and BMS system. The standardization includes data exchange in JSON and XML data formats, and using a relational database to store the integrated data; S12. Clean the collected data, remove outliers and noise, and fill in missing values. Then map and match the building power system information in the BIM model with the battery data in the BMS system. Use BIM software and visualization tools to visualize the integrated power system data and battery management system data.
3. The BMS battery management method based on the BIM model according to claim 1, characterized in that: The real-time monitoring according to step S2 specifically includes the following steps: S21. Install current sensors, voltage sensors, temperature sensors and Hall effect sensors around the power system and batteries inside the building, collect battery status data collected by the sensors through the Internet of Things platform, and transmit the battery status data to the database, wherein the battery status data includes voltage V, temperature W, charging current Ic and discharging current If, and import the power system model of the building into Autodesk Revit, and implement real-time update and display of sensor data through Revit API; S22. After the battery status data is stored in the database, data preprocessing is performed using a Python script. The preprocessing method includes removing outliers, filling missing values, data denoising, and normalization.
4. The BMS battery management method based on the BIM model according to claim 3 is characterized in that: The timing analysis according to step S3 specifically includes the following steps: S31. Select a linear regression algorithm to build a time series model, and then divide the historical battery status data into a training set, a test set, and a validation set. Use the training set to train the time series model of the model and determine the regression coefficient a. Use the test set to test the prediction performance of the time series model through mean square error, root mean square error, and mean absolute error. Based on the test results obtained from the test set, verify the time series model through the validation set to evaluate the accuracy of the time series model.
5. The BMS battery management method based on the BIM model according to claim 4 is characterized in that: S32, inputting the real-time collected battery status data into the verified time series model, predicting the future battery health status Jk, and analyzing the battery health status Jk(t+n) at the future moment; The battery health status Jk(t+n) is obtained by the following algorithm formula; Jk(t+n)=a1+a2×t+a3×V(t)+a4×W(t)+a5×(Ic+If)(t); Where V(t) represents the voltage at time t, W(t) represents the temperature at time t, (Ic+If)(t) represents the charging current and discharging current at time t, t represents time, and n represents the prediction time step; a1 represents the intercept term, which indicates the value of the dependent variable when all independent variables are zero; a2 represents the regression coefficient related to time t, indicating the influence of time on the battery health status; a3 represents the regression coefficient related to the voltage V(t), which indicates the influence of the voltage on the battery health status; a4 represents the regression coefficient related to temperature W(t), which indicates the influence of temperature on the battery health status; a5 represents the regression coefficient related to the charging current and the discharging current (Ic+If)(t), which indicates the influence of the current on the battery health status.
6. The BMS battery management method based on the BIM model according to claim 4 is characterized in that: The abnormality assessment according to step S4 specifically includes the following steps: S41, based on the analysis of the historical normal battery status data, the maximum safety value and the minimum safety value of the normal battery, the first battery abnormal threshold Y1 and the second battery abnormal threshold Y2 are preset, and then the abnormality assessment is performed according to the battery health status Jk(t+n) at the future moment output by the time series model. The specific assessment scheme is as follows; When the battery health status Jk(t+n)> the first battery abnormality threshold Y1, it means that the battery health status is abnormal at the future time t+n. At this time, the pattern recognition is started to further analyze the cause of the battery health abnormality; When the second battery abnormality threshold Y2≤battery health status Jk(t+n)≤first battery abnormality threshold Y1, it indicates that the battery health status at the future moment is in a normal state, and no further operation is required at this time; When the battery health state Jk(t+n) is less than the second battery abnormality threshold Y2, it means that at the future time t+n, the battery health state is abnormal. At this time, pattern recognition is started to further analyze the cause of the battery health abnormality.
7. The BMS battery management method based on the BIM model according to claim 6 is characterized in that: According to step S5, pattern recognition specifically includes the following steps: S51. When the battery health status Jk(t+n) is predicted to be in an abnormal state at a future moment through the time series model, the pattern recognition function is automatically started to analyze the cause of the abnormal battery health by converting the data in the battery status data into a feature vector xi, wherein the feature vector xi includes {voltage vector x1, temperature vector x2, charging current vector x3 and discharging current vector x4}.
8. The BMS battery management method based on the BIM model according to claim 7 is characterized in that: S52, using the support vector machine (SVM) algorithm in the machine learning algorithm to classify the extracted feature vectors, identify the battery in different operating modes, and then identify the failure mode of the battery according to the classification result F(xi), and analyze the cause of the failure; The classification result F(xi) is obtained by the following algorithm formula; F(x)=sign(Q×xi×b); In the formula, xi represents the currently monitored feature vector, Q represents the weight vector, and b represents the bias.
9. The BMS battery management method based on the BIM model according to claim 8, characterized in that: S53, analyzing the cause of the current battery abnormality according to the result output by the classification result F(xi), the specific analysis scheme is as follows; When the classification result F(xi)>1, it means that the current feature vector xi causes battery operation failure, and the optimized charging and discharging strategy is automatically started at this time; When the classification result F(xi)≤1, it means that the current feature vector xi does not cause battery operation failure. At this time, the next feature vector xi is further analyzed until an abnormality is identified and execution is stopped.
10. A BMS battery management method based on a BIM model according to claim 9, characterized in that: S54, when it is monitored that the current characteristic vector xi causes a battery operation failure, then the charging and discharging strategy is further optimized based on dynamic programming through a dynamic programming algorithm, so that the battery can extend its life while ensuring the power supply demand. The specific planning content is as follows; Where U represents the utility function, C(t) represents the charging strategy at time t, D(t) represents the discharging strategy at time t, and T represents the total length of the time range.
Citation Information
Cited By
Electric vehicle battery state real-time monitoring system based on deep learning
CN120044415A
