Method and system for circuit breaker anomaly detection, health index classification, and remaining useful life prediction

Abstract

A method and system for abnormality detection, health index classification, and remaining useful life prediction of circuit breakers includes the steps of capturing and collecting real-time trip coil waveform signals from each circuit breaker; transmitting the collected trip coil waveform signals to a circuit breaker health monitoring system; processing the collected signal data; analyzing the processed data using artificial intelligence and signal processing algorithms for abnormality detection, health index classification, and remaining useful life (RUL) prediction; and storing the data; performing trend analysis on the data; and predicting the remaining useful life (RUL).

Description

Technical Field

[0001] This invention relates to the field of circuit breaker anomaly detection, and more specifically, to methods and systems for circuit breaker anomaly detection, health index classification, and remaining useful life (RUL) prediction. Background Technology

[0002] The power system is vital to modern society because it provides electricity to homes, businesses, hospitals, factories, and other critical infrastructure. It plays a crucial role in supporting economic growth, public health, and quality of life. Therefore, a reliable and resilient power system is essential for the functioning of modern society.

[0003] Power outages can cause significant economic losses to power companies and their customers, and disrupt the basic functioning of society. Most power outages are caused by component failures in the distribution system; therefore, distribution systems are equipped with protection systems to detect faults caused by component failures and isolate the faulty section from the rest of the network, thereby minimizing the consequences of the fault. Among the many types of protection systems, circuit breakers are a widely used device to protect systems from system disturbances caused by component failures and overcurrent, overload, or short circuits.

[0004] Preventive maintenance and inspection of circuit breakers have always been an effective way to monitor their health and help ensure their reliable operation. Most preventive maintenance and inspection activities are performed regularly through scheduled maintenance, in accordance with the manufacturer's instructions and industry standards. Lack of maintenance can lead to unexpected delays in fault conditions, resulting in prolonged zeroing time, serious damage to electrical equipment, and greater risk of arc flashover.

[0005] However, such maintenance can be extremely expensive if more frequent and comprehensive testing and inspections are required. This not only burdens operating and maintenance budgets but also consumes a significant amount of time for engineers and is difficult to manage successfully. Because maintenance intervals are determined regardless of the condition of the circuit breaker, a component may develop serious problems before the next maintenance. Therefore, this can result in costly maintenance work that, in some cases, fails to improve reliability.

[0006] Some examples of this are discussed in the prior art below. Korean patent publication number 20220011052A discloses a circuit breaker fault detection method and an apparatus for performing the method. The circuit breaker fault diagnosis method includes the following steps: a circuit breaker management server receives circuit breaker data; the circuit breaker management server determines a circuit breaker data group; the circuit breaker management server considers the circuit breaker data group to determine a health coefficient of the circuit breaker; and the circuit breaker management server determines the state of the circuit breaker based on the health coefficient. This prior art demonstrates unsupervised learning based on anomaly detection. While unsupervised learning based on anomaly detection can identify anomalies throughout the waveform, it only supports binary classification, which is insufficient for multi-level health index classification.

[0007] Chinese patent publication number 107219457B discloses a method for fault diagnosis and severity assessment of frame-type circuit breakers based on operating accessory current. This method first determines the operating stage of the circuit breaker to be diagnosed and assessed, including an energy storage stage, a closing stage, and a opening stage. Then, it uses the energy storage motor current to diagnose faults in the energy storage stage, the closing coil current to diagnose faults and assess the severity of faults in the closing stage, and the opening coil current to diagnose faults and assess the severity of faults in the opening stage. It detects the energy storage motor current signal in the energy storage stage, the closing coil current signal in the closing stage, and the opening coil current signal in the opening stage, respectively. Simultaneously, it combines multi-core support vector machines for fault diagnosis. When a fault is diagnosed and a fault severity assessment is required, the fault severity characteristic curve can accurately determine the fault severity. However, this prior art uses a multi-core support vector machine, i.e., a traditional support vector machine (SVM), which may have limited performance when processing highly imbalanced data. The bias towards the majority class in the imbalanced data of this prior art may lead to poor processing of the minority class.

[0008] Chinese patent publication number 109270442B discloses a high-voltage circuit breaker fault detection method based on a DBN-GA neural network. The method follows this process: current data monitored by an online monitoring system is used as input variables; then, a fault type prediction model is constructed using a deep learning algorithm based on a deep belief neural network, defining a restricted Boltzmann machine (RBM) model. A portion of the current data samples are extracted and used to build and train this model; after training the RBM, the entire deep belief neural network model is trained; finally, all data is input into the trained fault type prediction model, which processes the input opening and closing coil current data to complete the high-voltage circuit breaker fault detection. However, this method only predicts and detects the fault type of the circuit breaker, without estimating the time of fault occurrence.

[0009] Therefore, a solution is still needed in this field to address the problems described in this article. Summary of the Invention

[0010] The main objective of this invention is to provide a method and system that analyzes the trip coil current waveform of a circuit breaker by applying signal processing technology combined with unsupervised machine learning methods, achieving better performance when processing unbalanced datasets, thereby enabling anomaly detection and health index classification of circuit breakers in power distribution systems.

[0011] Another object of the present invention is to provide a method and system that can perform multi-level health index classification with high accuracy.

[0012] Another objective of this invention is to provide a simple, accurate, and efficient calculation method and system for predicting the remaining useful life (RUL) of circuit breakers, which helps to optimize maintenance plans and reduce operating costs.

[0013] Another objective of this invention is to provide a cloud-based online health monitoring system. This system can centralize data, enabling users to monitor the current and past performance of all circuit breakers in the network, and automatically issue early warnings when anomalies or poor health conditions are detected, while maintaining a reliable and secure power supply network. To at least achieve the above-mentioned main objectives, the technical solution adopted by this invention is: a method for circuit breaker anomaly detection, health index classification, and remaining useful life (RUL) prediction, comprising the following steps: capturing and collecting real-time trip coil waveform signals from each circuit breaker; transmitting the collected trip coil waveform signals to the circuit breaker health monitoring system; Process the collected signal data; analyze the processed data using artificial intelligence and signal processing algorithms for anomaly detection, health index classification, and circuit breaker remaining service life prediction; store the data; perform trend analysis on the data; and predict remaining service life. Attached Figure Description

[0014] The features of the invention will be more readily understood and recognized when reading the following detailed description in conjunction with the accompanying drawings of preferred embodiments of the invention, wherein: Figure 1 The operating environment of the system of the present invention is shown; Figure 2 a~f in the figure respectively show the situation when the circuit breaker with spring operating mechanism is performing a disconnecting switch operation; Figure 3 An example of a typical capture trip coil current profile is shown; Figure 4 A health standard chart predicting remaining lifespan is shown; and; Figure 5 An example flowchart of the method for detecting anomalies in the circuit breaker system of the present invention is shown.

[0015] Marker explanation: 102: Power distribution system; 104: Circuit breaker; 106: Waveform data acquisition system; 108: Circuit breaker health monitoring system; 110: Artificial intelligence; 112: Signal processing algorithm; 114: Storage module; 116: User interface; 120: Log server; 202: Spring; 204: Main contact; 206: Trip coil; 208: Latch; 210: Plunger; 502: Circuit breaker; 504: Current curve data; 506: Curve data matching; 508: Signal filtering; 510: User; 512: Trend analysis; Detailed Implementation

[0016] Specific embodiments of the present invention are disclosed herein as requested. However, it should be understood that the disclosed embodiments are merely examples of the invention, which may be implemented in many different forms. Therefore, the specific structural and functional details disclosed herein should not be construed as limiting, but rather serve as the basis for the claims. It should be understood that the accompanying drawings and their detailed description are not intended to limit the invention to the specific forms disclosed herein; rather, the invention covers all modifications, equivalents, and alternatives falling within the scope defined by the claims. As used throughout this application, the word “may” indicates optional (i.e., possible) rather than mandatory (i.e., required). Similarly, the words “comprising” and “including” mean including but not limited to. Furthermore, unless otherwise stated, the word “a” means “at least one” and the word “a plurality” means one or more. When using abbreviations or technical terms, these refer to their generally accepted meanings known in the art.

[0017] In one embodiment of the present invention, a method for anomaly detection, health index classification, and remaining useful life (RUL) prediction of circuit breaker 104 is provided, comprising the following steps: capturing and collecting real-time trip coil 206 waveform signals from each circuit breaker 104; transmitting the collected trip coil 206 waveform signals to a circuit breaker health monitoring system 108; processing the collected data; analyzing the processed data using artificial intelligence 110 and signal processing algorithm 112 to perform anomaly detection, health index classification, and remaining useful life (RUL) prediction for the circuit breaker; storing the data; performing trend analysis on the data; and predicting the remaining useful life (RUL).

[0018] In a preferred embodiment of the present invention, the analysis of the processed data using artificial intelligence 110 includes the following steps: training and developing an unsupervised machine learning model using the waveform of the trip coil 206 of the circuit breaker 104; and identifying anomalies in the waveform.

[0019] In a preferred embodiment of the present invention, the analysis of the processed data using signal processing algorithm 112 includes the following steps: filtering and smoothing the signal; extracting features for calculating health standards.

[0020] In a preferred embodiment of the present invention, the method further includes the following steps: simultaneously providing multi-level health index classification; and automatically sending alarm information to the user when a circuit breaker malfunction is detected, so as to take timely follow-up actions.

[0021] In a preferred embodiment of the present invention, identifying waveform anomalies includes the following steps: recording the results as historical data for trend analysis 512 of circuit breaker 104; setting an end-of-life (EOL) threshold; and predicting the remaining useful life (RUL) of circuit breaker 104.

[0022] In a preferred embodiment of the present invention, the method further includes the following steps: online monitoring, and centralizing the collected signal data, data analysis, and data storage in the circuit breaker health monitoring system 108. The method further includes the step of storing the data in a server, including but not limited to a local server or a cloud server.

[0023] In a preferred embodiment of the invention, the method further includes the step of communicating with other data sources, including but not limited to an event log server 120, to determine the identity (ID) of the circuit breaker 104 associated with the diagnostic signal.

[0024] The present invention also provides a circuit breaker diagnostic system, the system comprising: a power distribution system 102 with circuit breakers 104, and equipped with: a data acquisition module configured to acquire the waveform signal of the real-time trip coil 206 of each circuit breaker 104; a processing module; a monitoring module equipped with artificial intelligence 110 and signal processing algorithm 112; and a storage module 114.

[0025] In a preferred embodiment of the present invention, the data acquisition module includes a waveform data acquisition system 106 for the trip coil 206. The waveform data acquisition system 106 for the trip coil 206 includes a current probe, an oscilloscope, and a microcontroller.

[0026] In a preferred embodiment of the invention, the monitoring module includes a circuit breaker health monitoring system 108, configured to perform data analysis and data storage. The circuit breaker health monitoring system 108 is a cloud-based, online, centralized system configured to monitor the health status and performance of circuit breakers 104 located in different locations and can transmit data to the cloud in real time. This allows users to manage tens of thousands of circuit breakers 104 in multiple locations using a single website.

[0027] In a preferred embodiment of the invention, artificial intelligence 110 includes an unsupervised machine learning model. The unsupervised machine learning model is developed and trained using the waveform of the entire trip coil 206 of the circuit breaker 104 to identify anomalies in the waveform.

[0028] In a preferred embodiment of the present invention, the signal processing algorithm 112 is configured to analyze the characteristics of the current waveform of the trip coil 206, diagnose the health status of the circuit breaker 104, and determine the health standard of the circuit breaker 104.

[0029] In a preferred embodiment of the present invention, the storage module 114 is configured to record the results as historical data for trend analysis 512.

[0030] In a preferred embodiment of the invention, the system is configured to predict the remaining useful life (RUL) of the circuit breaker 104.

[0031] As an embodiment of the present invention, Figure 1 illustrates the operating environment of the system of the present invention. The system includes a power supply circuit or power distribution system 102, in which a circuit breaker 104 is installed as a protection system. When an overcurrent, overload, or short circuit occurs, the circuit breaker 104 will trip, isolating the faulty part from the rest of the network. The system also includes a trip coil waveform data acquisition system 106, which serves as a data acquisition module and includes, but is not limited to, a current probe, an oscilloscope, and a microcontroller. These components are configured to collect real-time trip coil 206 waveform signals from all circuit breakers 104.

[0032] The trip coil waveform data acquisition system 106 sends the acquired signals to the monitoring module, also referred to as the circuit breaker health monitoring system 108 in this invention. This module can also act as a data storage server, and the data is analyzed by the installed artificial intelligence 110 and signal processing algorithm 112. The current waveform of the trip coil 206 of the circuit breaker 104 is used as a diagnostic signal in this invention for real-time monitoring of the health status of the circuit breaker 104. The behavior of the trip coil 206 current is directly affected by the coil actuator system of the circuit breaker 104. By capturing and analyzing the current curve of the trip coil 206 of the circuit breaker 104 under switching operation, a predictive maintenance alarm for the circuit breaker 104 can be activated. The specific mechanisms of the circuit breaker 104 produced by different manufacturers vary. Figure 2af shows a general diagram of the circuit breaker 104 with a spring-operated mechanism when performing a disconnecting switch operation, while Figure 3 shows a typical captured trip coil 206 current curve.

[0033] When the trip coil 206 is energized, the increased current causes a magnetic field to act on the iron plunger 210. Figure 2 As shown in Figure 3, when the force on plunger 210 exceeds the hysteresis, plunger 210 begins to move at point "A". The movement of plunger 210 generates an electromagnetic field in the coil by reducing the current flowing through it. The current reaches its first peak value I at point "B". pk The current then decreases until the plunger 210 strikes the latch 208 mechanism at point "C," causing a sudden drop in speed that creates a "corner" in the current curve. The combined mass of the plunger 210 and latch 208 reduces the momentum of the plunger 210, causing the coil current to decrease further from point "D" to point "E" until it strikes the buffer, bringing it to a stop. The latch 208 then unlocks the spring 202 operating mechanism, opening the main contact 204 at point "G." Simultaneously, the current increases to its maximum value I at point "F." maxThe current continues until the main contact 204 opens, while the plunger 210 remains stationary. When the main contact 204 opens, the current will drop significantly at point "H".

[0034] All these feature points are treated as events. Each event has an expected value range and its relationship to other events. Outliers can be used to identify potential problems with the mechanism. For example, if the times of "A" and "B" are as expected, but the times of "C" and all subsequent events are delayed, it can be inferred that the problem lies with latch 208. Domain knowledge helps to identify possible scenarios, such as poor lubrication causing latch 208 to become too stiff.

[0035] like Figure 1 As shown, signal processing algorithm 112 is used to analyze the characteristics of the current waveform of the trip coil 206, diagnose the health status of the circuit breaker 104, and determine the health standard of the circuit breaker 104. The current signal of the trip coil 206 of each circuit breaker 104 is acquired by the trip coil waveform data acquisition system 106 and then sent to the circuit breaker health monitoring system 108 for analysis. The circuit breaker health monitoring system 108 first processes the input signal using signal processing techniques, including signal filtering 508, signal smoothing, and feature extraction. Then, it performs peak-valley detection-based feature extraction on the processed signal to obtain features such as... Figure 3 The T shown B T E T F T H I B and I F Features such as these are used for health index classification.

[0036] These signals are also analyzed using machine learning algorithms. For anomaly detection, the dataset is highly imbalanced because anomalies are rare and represent a small proportion of the data. Supervised machine learning models often have limited performance on highly imbalanced datasets. In contrast, unsupervised machine learning models for anomaly detection do not rely on labeled data to discover patterns or clusters in the data. These models are trained on datasets containing mostly normal samples, thus outputting smaller errors when predicting normal data and larger errors when predicting anomalous data. When the dataset is highly imbalanced, unsupervised machine learning models for anomaly detection consistently achieve significantly higher performance than supervised models and support binary classification. Therefore, the machine learning algorithm in this invention is based on unsupervised machine learning methods. The entire trip coil waveform of circuit breaker 104 is used to train and develop a machine learning model capable of identifying any anomalies in the waveform.

[0037] Furthermore, compared to signal processing methods that only detect anomalies at certain feature points, the unsupervised machine learning in this invention can identify anomalies in the entire waveform, thus achieving higher accuracy. However, it only supports binary classification and is insufficient for multi-level health index classification. Therefore, as a preferred embodiment of this invention, it combines signal processing-based health standards with unsupervised machine learning, which can not only diagnose the health status of circuit breaker 104 and determine health index classification standards, but also simultaneously achieve multi-level classification with high accuracy.

[0038] In addition to the feature points detected by machine learning output and signal processing, the health standard also includes other domain data such as the aging and maintenance time of the circuit breaker 104. It includes, but is not limited to, [other data]. T B , T E , T F , T H , I B , I F , age , n 维护 and R 损失 Some or all of the parameters in the file.

[0039] The health standard H (based on signal processing) is shown in the following formula (1): H = ）] ..................(1) in: T E :arrive Figure 5 Time at point E (weight: ω 1 Time constant: p 1 ) T H Time to reach point H in Figure 5 (weight: ω 2 Time constant: p 2 ) I B The amplitude of point B (weight: ω 3 ) age : Age of the circuit breaker (weight: ω 4 Age constant: p 4 ) n 维护 Total number of maintenance tasks (weight: ω 5 ,constant: p 5 ) R 损失 Machine learning output, reconstruction loss (weighted): ω 6 ,constant: p 6 Some factors in the formula are not essential and can be added or removed according to needs during implementation. Health standards can be divided into n levels, where n can be any number. Here, we take n=3 as an example. For n=3, the health standards are divided into 3 ranges: Range 1: H <= c_1 Range 2: c_1 <H<=c_2 Range 3: H > c_2 constant c 1 and c 2 It was determined through statistical analysis based on the actual trip coil 206 current waveform dataset provided by the local power supply company and the health standard H calculated based on the actual health status of circuit breaker 104. In one case, c 1 and c 2 The values ​​are 1 and 1.6, respectively. The Health Index (HI) is determined by machine learning results and the range of health standards in Table 1.

[0040] Table 1: Range of Health Standards

[0041] If the machine learning algorithm diagnoses circuit breaker 104 as "normal" and the health criterion H is less than c1 (when n=3), then circuit breaker 104 will be classified as Class 1. If the machine learning algorithm diagnoses the circuit breaker as "abnormal", it will be classified as Class 2 or Class 3 according to the health criterion H.

[0042] Most samples belong to Level 1 and are not particularly noticeable in the maintenance plan. To ensure that anomalous samples are not incorrectly classified into this category, high accuracy is required. Therefore, to improve accuracy, both machine learning models and signal processing techniques with health criteria are used to classify samples into Level 1. Samples classified as anomalous by the machine learning model are further classified into Level 2 or Level 3 according to the health criterion H shown in Formula (1). The failure probability can also be determined accordingly.

[0043] When HI is level 1, circuit breaker 104 is in normal condition. When HI is level 2, the performance of circuit breaker 104 deteriorates and requires special attention. When HI is level 3, the circuit breaker is in poor health and requires immediate temporary repair or maintenance to prevent further degradation or failure of circuit breaker 104.

[0044] As shown in Figure 1, the circuit breaker health monitoring system 108 can further communicate with other data sources, including but not limited to the event log server 120, to identify the circuit breaker 104 that issued the diagnostic signal. The trip coil waveform data capture system 106 and the circuit breaker health monitoring system 108 are connected to other data sources via network devices (such as an Ethernet switch 118). If any anomaly is detected, the circuit breaker 104 anomaly diagnosis system of this invention will issue an alarm message. The user interface 116 allows the user to browse the data in the system.

[0045] As shown in Figure 4, the predicted value of RUL can be obtained by tracking the upward trend of health standards and setting thresholds. When a poor health index is detected, an alarm message, such as a text message or email, will be automatically sent to the engineer to provide early warning before circuit breaker 104 fails.

[0046] Figure 5 illustrates a flowchart of a method for detecting the health status of circuit breaker 104 in this invention. Circuit breaker 502 trips, triggering a trip coil waveform data acquisition system 106 to collect the current curve of trip coil 206. The method for detecting health status includes creating trip coil 206 current curve data 504, communicating with an event log server 120, and matching curve data with the corresponding circuit breaker 104 identity when there is no circuit breaker 104 identity from the trip coil waveform data acquisition system 106. The method further includes sending a signal to a circuit breaker health monitoring system 108 for analysis using a signal processing algorithm 112, including signal filtering 508, smoothing, and feature extraction. Furthermore, the data is analyzed using a machine learning algorithm for health diagnosis. If the algorithm diagnosis result is abnormal, the system automatically sends an alarm message to users such as engineers 510; if the algorithm diagnosis result is normal, the system records the result as historical data for trend analysis 512.

[0047] The above explanation of the present invention is not limited to the foregoing embodiments and drawings, and it will be apparent to those skilled in the art that various substitutions, modifications and alterations can be made without departing from the scope of the present invention.

Claims

1. A method for anomaly detection, health index classification, and remaining service life prediction of circuit breakers, characterized in that, Includes the following steps: Capture and collect real-time trip coil waveform signals from each circuit breaker; The collected trip coil waveform signal is transmitted to the circuit breaker health monitoring system; Process the collected signal data; Artificial intelligence and signal processing algorithms are used to analyze the processed data for anomaly detection, health index classification, and prediction of the remaining service life of circuit breakers. Store data; Perform trend analysis on the data; as well as Predict remaining useful life.

2. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 1, characterized in that, The data processed using artificial intelligence includes the following steps: Training and developing unsupervised machine learning models using circuit breaker trip coil waveforms; and Identify anomalies in waveforms.

3. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 1, characterized in that, The analysis and processing of the data using signal processing algorithms includes the following steps: Filtering and smoothing signals; and Extract features for health diagnosis and health standard calculation; The health standard is based on the following formula: H = ）]..........（1） in: T E Time to reach point E in Figure 5 (weight: ω 1 Time constant: p 1 ) T H Time to reach point H in Figure 5 (weight: ω 2 Time constant: p 2 ) I B The amplitude of point B (weight: ω 3 ) age : Age of the circuit breaker (weight: ω 4 Age constant: p 4 ) n 维护 Total number of maintenance jobs (weight: ω 5 ,constant: p 5 ) R 损失 Machine learning output, reconstruction loss (weighted): ω 6 ,constant: p 6 ).

4. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 2 or 3, characterized in that, Further steps include: It also provides multi-level health index classification; and When a circuit breaker malfunction is detected, an alarm message is automatically sent to the user so that subsequent actions can be taken in a timely manner.

5. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 2, characterized in that, The identification of anomalies in the waveform includes the following steps: The results are recorded as historical data for trend analysis of circuit breakers; Set a threshold for the end of lifespan; Predict the remaining service life of the circuit breaker.

6. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 1, characterized in that, Further steps include: Online monitoring is performed, and the collected signal data, data analysis, and data storage are centralized in the circuit breaker health monitoring system.

7. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 6, characterized in that, It further includes the step of storing data in a server.

8. The method for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 1, characterized in that, Further steps include: It communicates with other data sources to identify and diagnose the individual circuit breakers.

9. A system for circuit breaker anomaly detection, health index classification, and remaining service life prediction, characterized in that, include: A power distribution system with circuit breakers; The power distribution system equipped with circuit breakers is deployed as follows: The data acquisition module is configured to acquire the real-time trip coil waveform signal of each circuit breaker; Processing module; The monitoring module is equipped with artificial intelligence and signal processing algorithms; and Storage module.

10. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The data acquisition module includes a trip coil waveform data acquisition system.

11. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 10, characterized in that, The trip coil waveform data capture system includes a current probe, an oscilloscope, and a microcontroller.

12. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The monitoring module includes a circuit breaker health monitoring system, which is configured to perform data analysis and data storage.

13. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 12, characterized in that, The circuit breaker health monitoring system is a cloud-based online centralized system configured to monitor the health status and performance of circuit breakers located in different locations and to transmit data to the cloud in real time.

14. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The artificial intelligence mentioned includes unsupervised machine learning models.

15. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 14, characterized in that, The unsupervised machine learning model is developed and trained using the entire trip coil waveform of the circuit breaker to identify anomalies in the waveform.

16. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The signal processing algorithm is configured to analyze the characteristics of the trip coil current waveform, diagnose the health status of each circuit breaker based on health criteria, and determine the health index of the circuit breaker. The health standard is based on the following formula: H = ）].....................（1） in: T E Time to reach point E in Figure 5 (weight: ω 1 Time constant: p 1 ) T H Time to reach point H in Figure 5 (weight: ω 2 Time constant: p 2 ) I B The amplitude of point B (weight: ω 3 ) age : Age of the circuit breaker (weight: ω 4 Age constant: p 4 ) n 维护 Total number of maintenance jobs (weight: ω 5 ,constant: p 5 ) R 损失 Machine learning output, reconstruction loss (weighted): ω 6 ,constant: p 6 ).

17. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 16, characterized in that, The signal processing algorithm is further configured to simultaneously provide multi-level health index classification.

18. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The storage module is configured to record the results as historical data for trend analysis.

19. The system for circuit breaker anomaly detection, health index classification, and remaining service life prediction according to claim 9, characterized in that, The system is configured to predict the remaining service life of the circuit breaker.

Images