A security operation and maintenance system for intelligent chassis
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TIANJIN XINZHAN EXPRESSWAY CO LTD
- Filing Date
- 2026-02-05
- Publication Date
- 2026-06-02
Smart Images
Figure CN122137094A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power supply system technology, specifically to a smart chassis security operation and maintenance system. Background Technology
[0002] In the field of power supply system technology, intelligent distribution boxes are key equipment for power distribution and control. Real-time monitoring and safe operation and maintenance of their operating status are crucial for ensuring the stable operation of the power supply system. Real-time and accurate collection of operating data such as voltage, current, temperature, and humidity within the distribution box is fundamental for assessing equipment status, providing early warnings of anomalies, and handling faults.
[0003] In existing smart enclosure security operation and maintenance systems, data acquisition modules often directly acquire the raw operating data of the monitoring unit. Although some systems perform preliminary data processing, they often use simple static threshold filtering or mean filtering methods. These methods can only filter some low-frequency stable noise, and it is difficult to effectively suppress high-frequency noise or sudden interference signals generated in the distribution box due to electromagnetic interference, line contact fluctuations, sensor instantaneous errors, etc.
[0004] However, in the process of implementing the technical solution of this application, the inventors of this application discovered that the above-mentioned technology has at least the following technical problems: the noise contained in the original operating data will cause the measured values of key parameters such as voltage and current to deviate from the actual values, which will cause errors in the subsequent analysis process such as abnormal quantity extraction and behavior pattern recognition based on these data, and may easily lead to misjudgment or omission. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides a smart chassis security operation and maintenance system. This system solves the problem that noise in the raw operating data in existing technologies can cause deviations between the measured and actual values of key parameters such as voltage and current, leading to errors in subsequent analysis processes such as anomaly extraction and behavior pattern recognition based on this data, which can easily result in misjudgments or missed judgments.
[0006] To achieve the above objectives, the present invention provides the following technical solution: a security operation and maintenance system for an intelligent chassis, comprising:
[0007] The data acquisition module is used to collect real-time operating data from the monitoring unit inside the distribution box;
[0008] The data storage and investigation module is used to store the operational data and extract abnormal quantities;
[0009] The adaptive behavior pattern recognition module is used to identify the legitimacy or illegality of abnormal quantities through dynamic behavior profiling and weighted time correlation analysis.
[0010] An anomaly handling module is used to generate a temporary processing window based on the anomaly and compare it with historical data to generate analysis results.
[0011] The decentralized command propagation network module is used to propagate executed commands to neighboring nodes or the nearest site via a peer-to-peer communication network.
[0012] The data processing module is used to generate execution commands based on the analysis results;
[0013] The data feedback module is used to locate abnormal distribution boxes and send the execution command through the decentralized command propagation network module.
[0014] By adopting the above technical solution, the data acquisition module uses a moving average filtering algorithm to denoise the real-time acquired operating data such as voltage, current, temperature, and humidity. This effectively filters out noise caused by electromagnetic interference, line fluctuations, and instantaneous sensor errors, making the processed operating data closer to the actual values. This provides a reliable data foundation for subsequent analysis processes, such as accurate extraction of anomalies by the data storage and investigation module and precise identification of the legitimacy of anomalies by the adaptive behavior pattern recognition module. It reduces the interference of noise on anomaly detection, thereby improving the accuracy of anomaly detection, avoiding misjudgments or missed judgments caused by data noise, and ensuring accurate data support for the safe operation and maintenance of the intelligent chassis.
[0015] Preferably, the operational data collected by the data acquisition module includes voltage, current, temperature, and humidity. The operational data is processed by a moving average filtering algorithm to remove noise before being transmitted to the data storage and troubleshooting module.
[0016] Preferably, the data storage and investigation module stores 30 days of historical data and extracts anomalies by calculating the deviation between the real-time running data and the average of the historical data.
[0017] Preferably, the adaptive behavior pattern recognition module identifies the legitimacy or illegality of abnormal quantities through the following steps:
[0018] Based on 30 days of historical data, a dynamic behavior profile is generated using the K-means clustering algorithm. The dynamic behavior profile contains multiple clusters of behavior patterns.
[0019] Calculate the distance between real-time running data and the center of the behavior pattern cluster, and generate dynamic thresholds based on weighted time correlation analysis;
[0020] If the distance is greater than the dynamic threshold, it is marked as an illegal anomaly.
[0021] Preferably, the dynamic threshold is calculated using the following formula:
[0022] ;
[0023] in, For dynamic thresholds, Let be the standard deviation of the behavioral pattern cluster. For time-related weights based on time and environmental factors, To adjust the parameters.
[0024] Preferably, the outlier processing module generates a temporary processing window containing historical data for 10 minutes before and after the outlier occurrence, and generates analysis results by calculating the similarity between the outlier and the average of the historical data.
[0025] Preferably, the similarity is calculated using the following formula:
[0026] ;
[0027] in, For similarity, This is an abnormal quantity. To temporarily process the historical data average within the window, It is the Euclidean norm.
[0028] Preferably, the decentralized command propagation network module implements command propagation through the following steps:
[0029] Each distribution box acts as a node, maintaining a local routing table and updating the distance to neighboring nodes and network status through periodic heartbeat signals;
[0030] A dynamic routing algorithm is used to select the optimal path, which takes into account both node distance and network latency.
[0031] The abnormal distribution box generates an encrypted command and broadcasts it to neighboring nodes, which then forward it to the nearest site.
[0032] Preferably, the dynamic routing algorithm selects the optimal path by minimizing the following objective function:
[0033] ;
[0034] in, The optimal path, The path distance. For path delay, These are the weighting coefficients.
[0035] Preferably, the data feedback module obtains the coordinates of the distribution box through the built-in GPS module, calculates the Euclidean distance to each station, selects the station with the smallest distance, and sends the execution command through the decentralized command propagation network module.
[0036] This invention provides a secure operation and maintenance system for an intelligent chassis. It has the following beneficial effects:
[0037] 1. The data acquisition module of this invention collects operating data such as voltage, current, temperature and humidity, and transmits them after noise reduction by a moving average filtering algorithm, which effectively reduces noise interference, ensures data accuracy, provides high-quality basic data for subsequent storage, investigation and anomaly identification, and improves the reliability of system analysis.
[0038] 2. The data storage and investigation module of this invention stores historical data and extracts anomalies by calculating the deviation between real-time data and historical averages. The adaptive behavior pattern recognition module uses dynamic thresholds to accurately determine the legality of anomalies, reducing the false judgment rate and significantly improving the accuracy of anomaly recognition.
[0039] 3. The anomaly processing module of this invention generates a temporary window containing data from 10 minutes before and after the anomaly. Through in-depth analysis using similarity calculation, it provides a basis for processing decisions and enhances the ability to conduct targeted analysis of anomalies. Attached Figure Description
[0040] Figure 1 This is a security operation and maintenance system architecture diagram of an intelligent chassis according to the present invention;
[0041] Figure 2 This is a schematic diagram of the data acquisition and filtering process of a security operation and maintenance system for an intelligent chassis according to the present invention;
[0042] Figure 3 This is a schematic diagram of the abnormal quantity extraction logic of a security operation and maintenance system for an intelligent chassis according to the present invention;
[0043] Figure 4 This is a schematic diagram of the adaptive behavior pattern recognition process of a security operation and maintenance system for an intelligent chassis according to the present invention;
[0044] Figure 5 This is a schematic diagram illustrating the anomaly handling and similarity calculation of a security operation and maintenance system for an intelligent chassis according to the present invention.
[0045] Figure 6 This is a schematic diagram of a decentralized command propagation network for a security operation and maintenance system of an intelligent chassis according to the present invention. Detailed Implementation
[0046] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0047] Please see the appendix Figure 1 - Appendix Figure 6 This invention provides a security operation and maintenance system for an intelligent chassis, comprising:
[0048] The data acquisition module is used to collect real-time operating data from the monitoring unit inside the distribution box;
[0049] The data storage and troubleshooting module is used to store operational data and extract abnormal data.
[0050] The adaptive behavior pattern recognition module is used to identify the legitimacy or illegality of abnormal quantities through dynamic behavior profiling and weighted time correlation analysis.
[0051] The anomaly handling module is used to generate temporary processing windows based on anomalies and compare them with historical data to generate analysis results.
[0052] The decentralized command propagation network module is used to propagate executed commands to neighboring nodes or the nearest site via a peer-to-peer communication network.
[0053] The data processing module is used to generate execution commands based on the analysis results;
[0054] The data feedback module is used to locate abnormal distribution boxes and send execution commands through the decentralized command propagation network module.
[0055] Specifically, the system uses an embedded switch deployed in the distribution box. The switch connects to voltage transformers, current transformers, temperature and humidity sensors, and a GPS module via an I2C bus to collect operational data and perform calculation tasks.
[0056] The data acquisition module obtains voltage, current, temperature and humidity data in the distribution box through sensors, generates time series data vectors, and transmits them to the data storage and investigation module after moving average filtering. The data storage and investigation module stores historical data, calculates the deviation between real-time data and historical averages, and extracts outliers.
[0057] The adaptive behavior pattern recognition module generates dynamic behavior profiles based on historical data using the K-means clustering algorithm, which includes 5 behavior pattern clusters. It calculates dynamic thresholds by combining time and environmental factors through weighted time correlation analysis to determine whether anomalies are illegal behaviors. The anomaly processing module generates a temporary processing window containing data from 10 minutes before and after the anomaly, calculates the similarity between the anomaly and the historical data mean, and generates analysis results.
[0058] The data processing module generates execution commands based on the analysis results. Command types include locking the distribution box, issuing an alarm, or continuing monitoring. The decentralized command propagation network module uses a point-to-point communication network and a dynamic routing algorithm to select the optimal path and propagate the command to neighboring nodes or the nearest station. The data feedback module obtains the coordinates of the distribution box through the GPS module, calculates the Euclidean distance to the station, and selects the nearest station to send the command.
[0059] The data acquisition module generates real-time data, the data storage and investigation module extracts anomalies, the adaptive behavior pattern recognition module determines the legality of anomalies, the anomaly processing module generates analysis results, the data processing module generates execution commands, the decentralized command propagation network module propagates commands, and the data feedback module locates and sends commands to the site.
[0060] The system integrates data acquisition, storage, analysis, and command propagation functions through an embedded switch to achieve precise monitoring and efficient operation and maintenance of distribution boxes. The data acquisition module acquires real-time operating data such as voltage, current, temperature, and humidity. After noise reduction via moving average filtering, a time-series data vector is generated and transmitted to the data storage and investigation module to store 30 days of historical data and extract anomalies. The adaptive behavior pattern recognition module uses a K-means clustering algorithm to generate a dynamic behavior profile containing five behavior pattern clusters. Combined with weighted time correlation analysis, a dynamic threshold is calculated to accurately determine the legality or illegality of anomalies. The anomaly handling module generates a temporary processing window containing data from 10 minutes before and after the anomaly. Analysis results are generated through similarity calculation, and the data processing module uses this to generate execution commands such as locking the distribution box, issuing an alarm, or continuing monitoring. The decentralized command propagation network module uses a point-to-point communication network and a dynamic routing algorithm to select the optimal path to propagate encrypted commands to neighboring nodes or the nearest station. The data feedback module uses a GPS module to locate the abnormal distribution box, calculates the Euclidean distance to the station, and selects the nearest station to send the command. By working together with noise reduction, anomaly detection, behavior analysis, and decentralized command propagation, the accuracy of anomaly identification is significantly improved, the false positive rate is reduced, and commands are delivered quickly and reliably, enabling efficient and secure operation and maintenance of the intelligent chassis.
[0061] The data acquisition module collects operational data including voltage, current, temperature, and humidity. After noise reduction processing using a moving average filtering algorithm, the operational data is transmitted to the data storage and investigation module. The data storage and investigation module stores 30 days of historical data and extracts anomalies by calculating the deviation between the real-time operational data and the average of the historical data.
[0062] Specifically, the data acquisition module is deployed inside the distribution box, collecting operational data through voltage transformers, current transformers, and temperature and humidity sensors, and generating time-series data vectors.
[0063] ;
[0064] in, Let be the voltage at time t. For current, For temperature, To determine humidity, the collected data is then processed using a moving average filtering algorithm to reduce noise interference. The calculation formula for the filtering algorithm is as follows:
[0065] ;
[0066] in, This is the filtered data vector. This is the size of the sliding window.
[0067] The filtering process is executed in the embedded program of the switch. The filtered data is transmitted to the data storage and investigation module via the bus. The data storage and investigation module stores 30 days of historical data in time-series format, with each record containing a timestamp and a data vector. The historical dataset is defined as follows:
[0068] ;
[0069] in, This is the starting point of the current time.
[0070] Then, outliers are extracted. Outlier extraction is achieved by calculating the deviation between the real-time data and the historical data mean. The formula for calculating the historical mean vector is:
[0071] ;
[0072] in, Let represent the mean values of voltage, current, temperature, and humidity, respectively. The formula for calculating the deviation vector is: When the Euclidean norm of the deviation satisfies: If it is an outlier, then it is marked as an outlier. The preset threshold is adjusted according to the operating environment of the distribution box. The abnormal quantity set is defined as follows: The anomaly extraction algorithm runs on the switch processor, and the anomalies are transmitted to the adaptive behavior pattern recognition module via the internal bus.
[0073] The real-time data acquisition and moving average filtering of the data acquisition module ensure the accuracy of the operating data and reduce noise interference. The data storage and investigation module stores 30 days of historical data and extracts anomalies through deviation calculation, providing a reliable basis for anomaly detection. The anomalies are efficiently transmitted to subsequent modules through the internal bus, supporting adaptive behavior pattern recognition, which improves the overall system's detection accuracy and response efficiency for distribution box anomalies.
[0074] The adaptive behavior pattern recognition module identifies the legitimacy or illegality of abnormal quantities through the following steps:
[0075] Based on 30 days of historical data, a dynamic behavior profile is generated using the K-means clustering algorithm. The dynamic behavior profile contains multiple clusters of behavior patterns.
[0076] Calculate the distance between real-time running data and the center of the behavior pattern cluster, and generate dynamic thresholds based on weighted time correlation analysis;
[0077] If the distance is greater than the dynamic threshold, it is marked as an illegal anomaly.
[0078] The dynamic threshold is calculated using the following formula:
[0079] ;
[0080] in, For dynamic thresholds, Let be the standard deviation of the behavioral pattern cluster. For time-related weights based on time and environmental factors, To adjust the parameters.
[0081] Specifically, the adaptive behavior pattern recognition module is deployed inside the power distribution box. The adaptive behavior module receives the abnormal quantity set from the data storage and investigation module, and generates a dynamic behavior profile based on the historical dataset to determine the legality or illegality of the abnormal quantity. The historical dataset contains 30 days of filtered operating data.
[0082] The dynamic behavior profile is generated using the K-means clustering algorithm, with K=5 clusters corresponding to five typical electricity consumption patterns, categorized as daytime peak, nighttime off-peak, and holidays. Each data record includes voltage, current, temperature, humidity, hour, day of the week, and environmental factors, forming a feature vector.
[0083] ;
[0084] in, It refers to the number of hours in a day. What day of the week is it? The environmental factor and seasonal index are used, with values of 0-1 in spring, 1 in summer, and 0.5 in winter. The clustering algorithm determines the cluster centers by minimizing the intra-cluster variance. For real-time running data... The module calculates the Euclidean distance between itself and the center of each behavioral pattern cluster, and selects the cluster corresponding to the smallest distance. The dynamic threshold is calculated based on the cluster standard deviation and time-related weights, using the following formula: ,in, ,in, and Weighting coefficients The time period factor is used to simulate daily electricity consumption fluctuations. As an environmental factor, when the cluster distance of real-time data is greater than... If it is, it will be marked as an illegal abnormal quantity and transmitted to the abnormal quantity handling module.
[0085] This module, deployed within the distribution box, receives a set of anomalies from the data storage and investigation module. This set includes filtered operational data such as voltage, current, temperature, and humidity. These anomalies are then combined with hourly data, day of the week, and environmental factors to form a feature vector. A clustering algorithm determines the cluster centers by minimizing the intra-cluster variance. The Euclidean distance between the real-time operational data and each cluster center is calculated, and the cluster corresponding to the minimum distance is selected. When the distance between the real-time data and the cluster center exceeds this threshold, it is marked as an illegal anomaly and transmitted to the anomaly processing module. Through dynamic behavior profiling and weighted time correlation analysis, comprehensively considering time and environmental factors, the accuracy of anomaly validity judgment is significantly improved, and the false positive rate is reduced.
[0086] The outlier handling module generates a temporary processing window, which contains historical data for 10 minutes before and after the outlier occurrence. The module also generates analysis results by calculating the similarity between the outlier and the mean of the historical data.
[0087] Similarity is calculated using the following formula:
[0088] ;
[0089] in, For similarity, This is an abnormal quantity. To temporarily process the historical data average within the window, It is the Euclidean norm.
[0090] Specifically, the anomaly handling module receives a set of anomalies transmitted by the adaptive behavior pattern recognition module, and each anomaly... This indicates the deviation of real-time data from historical averages, including voltage, current, temperature, and humidity components. ,in This is the historical mean. For each outlier... The module generates a temporary processing window containing filtered historical data from 10 minutes before and after the anomaly occurred. The window data is extracted from the database, and the dataset is defined as follows: Then, the similarity formula is used to calculate and generate analysis results.
[0091] The anomaly handling module generates a temporary processing window containing data from 10 minutes before and after the anomaly occurred. Combined with Euclidean norm similarity calculation, it quantifies the difference between the anomaly and the historical data mean, generates reliable analysis results, reduces the false positive rate, and improves the targeting of anomaly handling and the system response efficiency.
[0092] The decentralized command propagation network module achieves command propagation through the following steps:
[0093] Each distribution box acts as a node, maintaining a local routing table and updating the distance to neighboring nodes and network status through periodic heartbeat signals;
[0094] The optimal path is selected using a dynamic routing algorithm, which takes into account both node distance and network latency.
[0095] The abnormal distribution box generates an encrypted command and broadcasts it to nearby nodes, which then forward it to the nearest site.
[0096] Dynamic routing algorithms select the optimal path by minimizing the following objective function:
[0097] ;
[0098] in, The optimal path, The path distance. For path delay, These are the weighting coefficients.
[0099] Specifically, the decentralized command propagation network module is deployed within distribution boxes. Each distribution box acts as a network node, equipped with a unique identifier. It communicates with neighboring nodes via LoRa modules, forming a point-to-point communication network. Each distribution box maintains a local routing table, storing neighboring node IDs, physical distances, and network latency. The routing table is updated periodically via heartbeat signals. Receiving nodes calculate the latency based on signal propagation time. A dynamic routing algorithm selects the optimal path based on node distance and network latency. The objective function is: The path selection algorithm runs on the switch processor. After receiving the execution command from the data processing module, the faulty distribution box generates an encrypted command containing the command type, timestamp, and distribution box ID. The command is broadcast to neighboring nodes via the LoRa module. Upon receiving the command, the neighboring nodes verify the encrypted signature and select the optimal path to forward it to the next node until the nearest station is reached. If the path is interrupted, the node recalculates the optimal path and switches to a backup node to ensure command delivery.
[0100] The decentralized command propagation network module achieves efficient and reliable command propagation through LoRa's point-to-point communication and dynamic routing algorithms. Its optimal path selection based on node distance and network latency, combined with periodic heartbeat signals to update the routing table, ensures real-time adaptation to network changes. Encrypted command broadcasting and forwarding mechanisms, along with backup node switching in case of path interruption, guarantee high reliability and security of command transmission, significantly improving the real-time response and operational efficiency of the intelligent chassis in case of anomalies.
[0101] The data feedback module obtains the coordinates of the distribution box through the built-in GPS module, calculates the Euclidean distance to each station, selects the station with the smallest distance, and sends the execution command through the decentralized command propagation network module.
[0102] Specifically, the data feedback module is integrated into the distribution box. It reads the latitude and longitude coordinates of the distribution box in real time through the built-in GPS module. Then, it calculates the Euclidean distance between the coordinates and the pre-stored coordinates of each maintenance site, and automatically selects the site with the smallest distance as the target. Then, it encapsulates the execution command generated by the data processing module into a package with a timestamp, distribution box ID and instruction content and sends it directly to the decentralized command propagation network module. The command is then forwarded hop by hop through the point-to-point LoRa network according to the optimal path of the dynamic routing algorithm.
[0103] The data feedback module, through its built-in GPS module and Euclidean distance calculation, enables rapid location of abnormal distribution boxes and accurate delivery of commands. Utilizing LoRa network and dynamic routing algorithms, it ensures that executed commands are reliably transmitted to the nearest site via the optimal path, significantly improving the system's response speed to anomalies and operational efficiency. This guarantees real-time monitoring and rapid handling of the distribution box's operating status, and enhances the overall reliability of the intelligent enclosure's safe operation and maintenance.
[0104] Example 2
[0105] The system is deployed in a city's power distribution network. Ten distribution boxes are selected as test nodes. Each distribution box contains an embedded switch that connects to voltage transformers, current transformers, temperature and humidity sensors, and a GPS module via an I2C bus, forming a data acquisition module. The switch is responsible for data acquisition, filtering, storage, analysis, and command propagation. The distribution boxes are interconnected via LoRa modules to form a decentralized point-to-point communication network, connecting to three maintenance sites. The site coordinates are pre-stored in the system database.
[0106] Data acquisition and denoising: The data acquisition module collects voltage, current, temperature, and humidity data in real time, generating time-series data vectors. For example, the data collected by a distribution box at time t is: voltage 220V, current 10A, temperature 25°C, humidity 60%, forming a vector [220, 10, 25, 60]. The data is processed by a moving average filtering algorithm, and the filtered data vector is [219.8, 10.1, 24.9, 59.8], effectively reducing instantaneous fluctuations caused by electromagnetic interference. The filtered data is transmitted to the data storage and troubleshooting module via the I2C bus.
[0107] The outlier extraction, data storage, and investigation module stores 30 days of historical data. The historical mean vector is calculated as [220.5, 10.2, 24.5, 60.2], and the deviation of real-time data from the mean is calculated as [0.7, 0.1, 0.4, 0.4]. The Euclidean norm is 0.92, which is less than the preset threshold of 1.5, so it is not marked as an outlier. When the voltage suddenly changes to 230V at a certain moment, the deviation vector is [9.5, 0.1, 0.4, 0.4], and the Euclidean norm is 9.51, which is greater than the threshold, so it is marked as an outlier and transmitted to the adaptive behavior pattern recognition module.
[0108] The outlier validity assessment module uses an adaptive behavior pattern recognition module based on 30 days of historical data. It generates five behavior pattern clusters (K=5) using a K-means clustering algorithm, corresponding to peak, off-peak, and holiday electricity consumption patterns. The feature vectors include voltage, current, temperature, humidity, hours, day of the week, and seasonal index. For example, if an outlier's feature vector is [230, 10.1, 24.9, 59.8, 14, 3, 1], the Euclidean distance to each cluster center is calculated, and the cluster with the smallest distance is selected. The cluster standard deviation σ=2.5, the weighted time correlation weight ω=1.2, the adjustment parameter α=1.5, and a dynamic threshold is calculated. When the distance is 5.2 (greater than 4.5), it is marked as an illegal outlier and transmitted to the outlier processing module.
[0109] Anomaly handling: The anomaly handling module generates a temporary processing window containing 1200 historical data points from 10 minutes before and after the anomaly occurred, calculating the historical mean vector [220.4, 10.3, 24.7, 60.1]. The anomaly vector is [230, 10.1, 24.9, 59.8], and the results are analyzed using similarity calculations and then transmitted to the data processing module.
[0110] Command generation and propagation: The data processing module generates and executes commands based on the analysis results to continue monitoring. The decentralized command propagation network module broadcasts encrypted commands via the LoRa module, including timestamps, distribution box IDs, and command types. It selects the optimal path and broadcasts the commands to neighboring nodes, which then forward them to the nearest station.
[0111] The command feedback and data feedback module obtains the coordinates of the distribution box through the GPS module, calculates the Euclidean distance to each station, and sends the command via the LoRa network along the optimal path. After receiving the command, station A records and continuously monitors the distribution box.
[0112] Example 3
[0113] The system is deployed in another city's power distribution network, selecting 15 distribution boxes as test nodes. Each distribution box is equipped with an embedded switch, which connects to voltage transformers, current transformers, temperature and humidity sensors, and a GPS module via an I2C bus, forming a data acquisition module. The switch is used to handle data acquisition, filtering, storage, analysis, and command propagation tasks. The distribution boxes form a decentralized point-to-point communication network via LoRa modules, connecting to four maintenance sites. The site coordinates are pre-stored in the system database.
[0114] Data acquisition and denoising: The data acquisition module collects voltage, current, temperature, and humidity data once per second, generating a time-series data vector. For example, the data collected by a distribution box at a certain moment is: voltage 230V, current 12A, temperature 28°C, humidity 55%, forming the vector [230, 12, 28, 55]. The data is processed using a moving average filtering algorithm with a moving window size of 7. After filtering, the data vector is [229.7, 11.9, 27.8, 54.9]. The filtered data is transmitted to the data storage and troubleshooting module via the I2C bus.
[0115] The outlier extraction, data storage, and investigation module saves 30 days of historical data. The historical data mean vector is calculated as [230.2, 12.1, 27.5, 55.3], and the deviation of real-time data from the mean is calculated as [0.5, 0.2, 0.3, 0.4]. The deviation is small and is not marked as an outlier. If the current suddenly changes to 18A at a certain moment, the deviation vector is [0.5, 5.9, 0.3, 0.4]. The deviation is large and is marked as an outlier, then transmitted to the adaptive behavior pattern recognition module.
[0116] The outlier validity assessment module uses 30 days of historical data and a K-means clustering algorithm to generate six behavioral pattern clusters, corresponding to different time periods and electricity consumption scenarios, such as peak hours on weekdays and off-peak hours on non-working days. Feature vectors include voltage, current, temperature, humidity, hour, day of the week, and seasonal index. For example, a certain outlier's feature vector might be [230, 18, 27.8, 54.9, 16, 5, 0.8]. The module calculates the distance between real-time data and the center of each cluster, selecting the cluster with the smallest distance. If the distance exceeds a dynamic threshold, it is marked as an illegal outlier and transmitted to the outlier processing module.
[0117] Outlier handling: The outlier handling module generates a temporary processing window and calculates the historical mean vector [230.1, 12.2, 27.6, 55.2]. The outlier vector is [230, 18, 27.8, 54.9]. The difference between the outliers and the historical mean is analyzed through similarity calculation, and the results are transmitted to the data processing module.
[0118] Command generation and propagation: The data processing module generates alarm commands based on the analysis results. The decentralized command propagation network module broadcasts encrypted commands via the LoRa module, which include timestamps, distribution box IDs, and command types. The dynamic routing algorithm considers node distance and network latency to select the optimal path. The command is broadcast to neighboring nodes, and the nodes forward it to the nearest station.
[0119] The command feedback and data feedback module obtains the coordinates of the distribution box via the GPS module, calculates the distance to each site, selects site B, and sends the command via the LoRa network along the optimal path. Upon receiving the command, site B records it, triggers an alarm, and notifies maintenance personnel to check.
[0120] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these 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 security operation and maintenance system for an intelligent chassis, characterized in that, include: The data acquisition module is used to collect real-time operating data from the monitoring unit inside the distribution box; The data storage and investigation module is used to store the operational data and extract abnormal quantities; The adaptive behavior pattern recognition module is used to identify the legitimacy or illegality of abnormal quantities through dynamic behavior profiling and weighted time correlation analysis. An anomaly handling module is used to generate a temporary processing window based on the anomaly and compare it with historical data to generate analysis results. The decentralized command propagation network module is used to propagate executed commands to neighboring nodes or the nearest site via a peer-to-peer communication network. The data processing module is used to generate execution commands based on the analysis results; The data feedback module is used to locate abnormal distribution boxes and send the execution command through the decentralized command propagation network module.
2. The security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The data acquisition module collects operational data including voltage, current, temperature, and humidity. The operational data is then processed by a moving average filtering algorithm to remove noise before being transmitted to the data storage and troubleshooting module.
3. The security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The data storage and investigation module stores 30 days of historical data and extracts anomalies by calculating the deviation between real-time operating data and the average of historical data.
4. The security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The adaptive behavior pattern recognition module identifies the legitimacy or illegality of abnormal quantities through the following steps: Based on 30 days of historical data, a dynamic behavior profile is generated using the K-means clustering algorithm. The dynamic behavior profile contains multiple clusters of behavior patterns. Calculate the distance between real-time running data and the center of the behavior pattern cluster, and generate dynamic thresholds based on weighted time correlation analysis; If the distance is greater than the dynamic threshold, it is marked as an illegal anomaly.
5. The security operation and maintenance system for an intelligent chassis according to claim 4, characterized in that, The dynamic threshold is calculated using the following formula: ； in, For dynamic thresholds, Let be the standard deviation of the behavioral pattern cluster. For time-related weights based on time and environmental factors, To adjust the parameters.
6. The security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The outlier processing module generates a temporary processing window containing historical data for 10 minutes before and after the outlier occurrence, and generates analysis results by calculating the similarity between the outlier and the mean of the historical data.
7. The security operation and maintenance system for an intelligent chassis according to claim 6, characterized in that, The similarity is calculated using the following formula: ； in, For similarity, This is an abnormal quantity. To temporarily process the historical data average within the window, It is the Euclidean norm.
8. The security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The decentralized command propagation network module achieves command propagation through the following steps: Each distribution box acts as a node, maintaining a local routing table and updating the distance to neighboring nodes and network status through periodic heartbeat signals; A dynamic routing algorithm is used to select the optimal path, which takes into account both node distance and network latency. The abnormal distribution box generates an encrypted command and broadcasts it to neighboring nodes, which then forward it to the nearest site.
9. The security operation and maintenance system for an intelligent chassis according to claim 8, characterized in that, The dynamic routing algorithm selects the optimal path by minimizing the following objective function: ； in, The optimal path, The path distance. For path delay, These are the weighting coefficients.
10. A security operation and maintenance system for an intelligent chassis according to claim 1, characterized in that, The data feedback module obtains the coordinates of the distribution box through the built-in GPS module, calculates the Euclidean distance to each station, selects the station with the smallest distance, and sends the execution command through the decentralized command propagation network module.