Method and system for detecting state of low-voltage circuit breaker

Abstract

The invention relates to a low-voltage circuit breaker state detection method, which comprises the following steps of: synchronously acquiring multi-source state data of a low-voltage circuit breaker in an operation process through a plurality of sensors, the multi-source state data at least comprising a vibration signal, an opening and closing coil current signal, a moving contact displacement signal, a temperature signal and a partial discharge signal; the collected signals are preprocessed, and feature vectors related to the health state of the circuit breaker are extracted; inputting the feature vector into an improved weighted D-S evidence theory fusion diagnosis model; the improved weighted D-S evidence theory fusion diagnosis model is used for fusing evidences from a plurality of sensors in a mode of distributing weights for evidences of different sensors and calculating weighted average evidences; and determining the comprehensive health state grade of the low-voltage circuit breaker according to an output result of the fusion diagnosis model. According to the invention, comprehensive evaluation of the multi-dimensional working condition of the circuit breaker is realized, and the method has the advantages of comprehensively reflecting the multi-dimensional working condition of the circuit breaker, reducing the risk of misjudgment and missed judgment, and improving the performance degradation pre-judgment capability of equipment.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, and in particular to a method and system for detecting the condition of low-voltage circuit breakers. Background Technology

[0002] As a key protection and control component in power distribution systems, the operational reliability of low-voltage circuit breakers directly affects the safety and stability of the entire power supply system. Currently, the industry generally adopts periodic inspections and reactive maintenance as the main maintenance methods. This passive maintenance model has obvious drawbacks: on the one hand, the setting of periodic inspection cycles lacks scientific basis, which can easily lead to over-maintenance or under-maintenance; on the other hand, maintenance after a fault has occurred often results in serious consequences such as power outages.

[0003] While existing online monitoring technologies have achieved condition monitoring to some extent, most are limited to the detection and analysis of single parameters, such as monitoring only the current of the opening and closing coils or mechanical characteristic parameters. These methods suffer from three significant technical bottlenecks: First, monitoring a single parameter cannot comprehensively reflect the multi-dimensional integrated operating conditions of the circuit breaker, including its mechanical structure, electrical performance, and insulation status, leading to blind spots in condition assessment. Second, single parameters are easily affected by environmental factors such as electromagnetic interference and temperature changes, and early fault characteristic signals are weak, easily causing misjudgments or missed diagnoses. Most importantly, existing methods lack the ability to predict equipment performance degradation trends, making it difficult to provide effective support for preventative maintenance.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] The purpose of this application is to provide a low-voltage circuit breaker condition detection method and system, which has the advantages of comprehensively reflecting the multi-dimensional operating conditions of the circuit breaker, reducing the risk of misjudgment and missed judgment, and improving the ability to predict equipment performance degradation.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for condition detection of a low-voltage circuit breaker includes the following steps: Step S100: Simultaneously collect multi-source status data of the low-voltage circuit breaker during operation using multiple sensors. The multi-source status data includes at least vibration signals, opening and closing coil current signals, moving contact displacement signals, temperature signals, and partial discharge signals. Step S200: Preprocess the various signals collected and extract the feature vectors related to the health status of the circuit breaker. Step S300: Input the extracted feature vectors into the improved weighted DS evidence theory fusion diagnostic model; the improved weighted DS evidence theory fusion diagnostic model fuses evidence from multiple sensors by assigning weights to evidence from different sensors and calculating a weighted average evidence. Step S400: Determine the overall health status level of the low-voltage circuit breaker based on the output of the fusion diagnostic model.

[0007] In a further embodiment of the present invention, in step S100, the vibration signal is acquired by a vibration sensor installed on the housing of the operating mechanism; the opening and closing coil current signal is acquired by a current sensor connected to the coil circuit; the moving contact displacement signal is acquired by a linear displacement sensor connected to the moving contact; the temperature signal is acquired by a temperature sensor installed on the conductive circuit of the circuit breaker; and the partial discharge signal is acquired by an ultra-high frequency sensor installed near the arc-extinguishing chamber.

[0008] The present invention further includes the following steps: In step S200, the energy values ​​of each frequency band after wavelet packet decomposition are extracted from the vibration signal as feature vectors; the key time points and holding current values ​​in the current waveform are extracted from the opening and closing coil current signal as feature vectors; and one or more of the average speed, maximum speed, and bounce time of the moving contact are calculated from the moving contact displacement signal as feature vectors.

[0009] The present invention further specifies that, in step S300, the construction of the improved weighted DS evidence theory fusion diagnostic model includes the following sub-steps: S31: Establish an identification framework that includes all possible health status propositions of the circuit breaker; S32: Based on the Mahalanobis distance between each feature vector and the standard state sample library, generate a basic probability assignment function for each type of sensor evidence; S33: Assign different weighting coefficients to each sensor based on its historical reliability or importance; S34: Using the aforementioned weighting coefficients, calculate the weighted average evidence of all sensor evidence; S35: Perform multiple self-fusions on the weighted average evidence to obtain the final comprehensive basic probability allocation result.

[0010] In a further embodiment of the present invention, in step S32, the calculated Mahalanobis distance is mapped to the value of a basic probability allocation function using an exponential function. This function value represents the degree of support and uncertainty of the corresponding sensor evidence for each health state proposition.

[0011] In a further embodiment of the present invention, in step S34, the weighted average evidence is obtained by weighting and summing the basic probability allocation functions of each sensor evidence according to their weight coefficients.

[0012] The present invention further specifies that, in step S400, the comprehensive health status level includes health, early warning, and fault; during decision-making, propositions whose basic probability allocation value in the final fusion result exceeds a preset confidence threshold are determined as the current status of the circuit breaker.

[0013] The present invention also provides a low-voltage circuit breaker condition detection system for implementing the method, comprising: The data acquisition module includes the aforementioned multiple sensors; The data processing module is used to perform signal preprocessing and feature extraction; The data fusion and diagnosis module, which incorporates the improved weighted DS evidence theory fusion and diagnosis model, is used to perform evidence fusion and state diagnosis. The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0015] The beneficial effects of the present invention are as follows: The low-voltage circuit breaker condition detection method and system provided in this application realizes a comprehensive evaluation of the multi-dimensional operating conditions of the circuit breaker, including mechanical structure, electrical performance, and insulation status, through synchronous acquisition of multi-source condition data, feature vector extraction and improved weighted DS evidence theory fusion diagnosis. It has the advantages of comprehensively reflecting the multi-dimensional operating conditions of the circuit breaker, reducing the risk of misjudgment and omission, and improving the ability to predict equipment performance degradation. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the process of an embodiment of the present invention. Detailed Implementation

[0017] The following will describe in detail the implementation of this application with reference to the accompanying drawings and embodiments, so that the implementation process of how this application uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0018] In existing technologies, low-voltage circuit breakers, as core protection devices in power distribution systems, directly impact power supply safety due to their operational reliability. Traditional maintenance methods rely on periodic inspections and reactive repairs, which suffer from issues such as unreasonable maintenance cycles and delayed fault response. Existing online monitoring technologies often employ single-parameter analysis methods, such as monitoring only the current of the opening and closing coils or mechanical vibration signals, making it difficult to comprehensively reflect multi-dimensional information such as the circuit breaker's mechanical structure, electrical performance, and insulation status. Single parameters are susceptible to environmental interference, and early fault characteristics, when subtle, can easily lead to misjudgments. Furthermore, they cannot effectively warn of performance degradation trends, failing to meet the smart grid's requirement for comprehensive equipment status perception.

[0019] To address these issues, the R&D team conducted an in-depth analysis of circuit breaker fault mechanisms, discovering that typical defects such as mechanical jamming, contact wear, and insulation degradation can trigger anomalies in multiple physical quantities, but a single sensor cannot capture the complete fault characteristics. Through research on multi-source information fusion theory, they found that traditional DS evidence theory carries the risk of decision failure when handling highly conflicting evidence. To solve this problem, a dynamic weight allocation mechanism was proposed to adjust the credibility of evidence based on the historical reliability of the sensors. Simultaneously, for the feature extraction stage, differentiated processing methods were designed based on the physical characteristics of different signals. For example, wavelet packet decomposition was used to extract frequency domain energy features from vibration signals, and timing key points were analyzed for current signals. By constructing a multi-level feature system and improving the evidence fusion algorithm, a diagnostic model capable of comprehensively assessing the health status of circuit breakers was formed.

[0020] like Figure 1 As shown, the present invention provides a low-voltage circuit breaker condition detection method, comprising the following steps: Step S100: Simultaneously collect multi-source status data of the low-voltage circuit breaker during operation using multiple sensors. The multi-source status data includes at least vibration signals, opening and closing coil current signals, moving contact displacement signals, temperature signals, and partial discharge signals. The multi-source status data is collected synchronously by deploying multiple sensors on the circuit breaker to collect signals during operation. The sensors include: a vibration sensor: installed on the housing of the operating mechanism, used to collect vibration acceleration signals during opening and closing.

[0021] Current sensor: Connected to the opening and closing coil circuits to collect current waveform signals during operation.

[0022] Displacement sensor: A linear potentiometer or laser rangefinder is used in conjunction with the moving contact to collect the travel-time curve of the contact.

[0023] Temperature sensor: A wireless or passive sensor is used, installed on the stationary contact or terminal, to monitor the operating temperature.

[0024] Ultra-high frequency (UHF) sensor: Installed near the arc-extinguishing chamber to collect internal partial discharge signals.

[0025] During a single closing operation, all data acquisition devices are synchronously triggered, with sampling rates set as follows: vibration signal 25.6 kHz, current signal 10 kHz, displacement signal 10 kHz, temperature signal 1 Hz, and UHF signal 1 GHz. Specifically, all sensor channels are controlled by a multi-channel synchronous data acquisition card (DAQ) (such as the NI USB-6366). A rising edge TTL pulse is emitted from the auxiliary contact (S1) of the closing / opening command as a synchronous trigger signal to initiate synchronous acquisition of all channels, ensuring strict alignment of data on the time axis.

[0026] Step S200: Preprocess the various signals collected and extract the feature vectors related to the health status of the circuit breaker. For the vibration signal, wavelet packet decomposition was performed, and vibration data for 100ms before and after the closing operation were extracted. A 4-level wavelet packet decomposition was performed using the db5 wavelet basis, extracting the frequency band energy of 16 nodes from low to high frequencies (0-16kHz) in the 4th level. After normalization, a 16-dimensional feature vector was constructed. .

[0027] Coil current analysis: Analyze the closing coil current waveform to extract the key time point: the current rise time, i.e., the start-up time. The time when the iron core begins to move Main contact time Assisted touch switching and current stability point eigenvectors .

[0028] For displacement signals: The average velocity is calculated by performing differentiation and other processing on the travel-time curve. Maximum speed Features such as overtravel time constitute the feature vector. .

[0029] For the temperature signal, calculate the three-phase temperature difference ΔT and the highest temperature. , constitute the feature vector .

[0030] For partial discharge signals: Extract statistical features such as amplitude, frequency, and phase distribution of the discharge signal to construct a feature vector. .

[0031] Step S300: Input the extracted feature vectors into the improved weighted DS evidence theory fusion diagnostic model; the improved weighted DS evidence theory fusion diagnostic model fuses evidence from multiple sensors by assigning weights to evidence from different sensors and calculating a weighted average evidence. Step S400: Determine the overall health status level of the low-voltage circuit breaker based on the output of the fusion diagnostic model.

[0032] In a further embodiment of the present invention, in step S100, the vibration signal is acquired by a vibration sensor installed on the housing of the operating mechanism; the opening and closing coil current signal is acquired by a current sensor connected to the coil circuit; the moving contact displacement signal is acquired by a linear displacement sensor connected to the moving contact; the temperature signal is acquired by a temperature sensor installed on the conductive circuit of the circuit breaker; and the partial discharge signal is acquired by an ultra-high frequency sensor installed near the arc-extinguishing chamber.

[0033] The vibration sensor installed on the housing of the operating mechanism means that the mechanical vibration waveform during the operation of the circuit breaker is directly captured by fixing it near the mechanical vibration source. Specifically, a piezoelectric accelerometer can be used. This position can effectively avoid external vibration interference and accurately reflect the state changes of the moving parts inside the mechanism.

[0034] The acquisition of the opening and closing coil current signal by a current sensor connected in the coil circuit refers to the real-time monitoring of dynamic current changes during the coil energization process using a non-intrusive current transformer. Specifically, a Hall effect sensor can be used to achieve this. This method can accurately acquire the current waveform characteristics without affecting the integrity of the circuit.

[0035] The acquisition of the moving contact displacement signal by a linear displacement sensor connected to the moving contact refers to the direct measurement of the linear motion trajectory of the moving contact through mechanical or optical means. Specifically, a linear variable differential transformer can be used to achieve this. This connection method can eliminate transmission backlash error and improve displacement measurement accuracy.

[0036] Temperature signal acquisition by temperature sensor installed on the conductive circuit of circuit breaker refers to monitoring the temperature distribution of current-carrying conductor through contact temperature sensing element. Specifically, a platinum resistance temperature sensor can be used. This installation location can directly reflect the abnormal temperature rise of conductive parts caused by increased contact resistance or overload.

[0037] The acquisition of partial discharge signals by a UHF sensor installed near the arc-extinguishing chamber refers to receiving the electromagnetic radiation signals generated by partial discharge through a wideband antenna. Specifically, a UHF sensor with a frequency range of 300MHz to 3GHz can be used. This location can effectively capture the discharge pulse signals caused by insulation degradation inside the arc-extinguishing chamber.

[0038] Specifically, the operating mechanism housing, as the load-bearing structure of the circuit breaker's mechanical transmission components, directly reflects the motion state of internal components such as linkages and springs through its vibration characteristics. Installing a vibration sensor at this location avoids signal attenuation caused by excessively long mechanical transmission paths. Current changes in the opening and closing coil circuits are directly related to the synchronization of core movement and contact action. A socket-type current sensor can completely record the dynamic characteristics of the current rise, hold, and fall phases. Real-time measurement of the moving contact displacement requires eliminating hysteresis errors caused by transmission mechanism gaps. Direct connection of a linear displacement sensor to the moving contact ensures a strict correspondence between the displacement signal and the actual contact position. Temperature monitoring points on the conductive circuit are selected at the contact points of the moving and stationary contacts and at busbar connections to accurately capture localized overheating caused by poor contact. As the main area where partial discharge occurs, the arc-extinguishing chamber is equipped with UHF sensors to maximize the reception of discharge signals and suppress external electromagnetic interference.

[0039] Compared to existing technologies, traditional methods often involve vibration sensors installed on the circuit breaker base or far from the operating mechanism, resulting in vibration signals containing significant environmental noise and failing to accurately reflect the internal state of the mechanism. Current sensors are typically connected in series in the control circuit rather than in the main circuit of the opening and closing coils, making it difficult to capture the true waveform of the drive current. Displacement measurement relies on rotary encoders to indirectly calculate contact travel, making it susceptible to mechanical transmission errors. Temperature monitoring points are scattered on the outer casing surface or far from conductive circuits, failing to effectively identify abnormal temperature rises in current-carrying components. Partial discharge detection uses low-frequency current transformers or ultrasonic sensors, which are susceptible to electromagnetic interference and lack sufficient sensitivity. This solution significantly improves the accuracy and correlation of multi-source state data by optimizing the spatial arrangement of each sensor and the signal acquisition method.

[0040] Through the above technical solution, this application solves the problem of data partiality caused by unreasonable sensor placement in traditional methods, and realizes comprehensive monitoring of the circuit breaker's mechanical movement, electrical characteristics, temperature distribution, and insulation status. The precise installation of each sensor at key locations ensures the physical representativeness and anti-interference capability of the signal acquisition, providing highly reliable basic data for subsequent multi-source data fusion, thereby effectively reducing the risk of diagnostic errors caused by misjudgment of a single parameter.

[0041] Multi-source state data refers to a set of physical quantities reflecting different dimensions of the circuit breaker's operating state acquired through heterogeneous sensors. Specifically, this can be achieved using a combination of vibration sensors, current sensors, displacement sensors, temperature sensors, and UHF sensors, covering key indicators such as mechanical motion characteristics, electrical parameter changes, and insulation status monitoring. Preprocessing involves operations such as noise reduction, normalization, and time-frequency transformation on the original signal. This can be achieved using methods such as wavelet threshold denoising and moving average filtering to eliminate the impact of environmental interference on data quality. Feature vectors are a set of quantified indicators extracted from the preprocessed signal. This can be achieved by calculating the frequency band energy ratio through wavelet packet decomposition, extracting current rise time and holding current values ​​through waveform analysis, and calculating motion velocity parameters through displacement curves, forming a feature space characterizing the equipment state. The improved weighted DS evidence theory fusion diagnostic model refers to an evidence fusion algorithm that introduces a dynamic weight allocation mechanism. Specifically, it can assign weight coefficients to each sensor's evidence based on historical diagnostic accuracy, and improve decision reliability through weighted averaging and multiple self-fusion processing of conflicting evidence.

[0042] Specifically, during the data acquisition phase, five types of sensors simultaneously record the changes in multidimensional physical quantities during circuit breaker operation. For example, vibration sensors capture the impact waveforms of the mechanism's movements, current sensors monitor the coil's energizing characteristics, and displacement sensors track the contact's movement trajectory. The preprocessing stage optimizes for different signal characteristics; for instance, wavelet packet decomposition is performed on non-stationary vibration signals, and baseline correction is applied to current signals. The feature extraction process filters out fault-sensitive parameters from each signal; for example, the energy distribution of the vibration frequency band reflects the wear degree of mechanical components, and the displacement velocity curve characterizes the lubrication status of the operating mechanism. In the fusion diagnosis phase, initial evidence is generated by calculating the Mahalanobis distance between each feature vector and the standard state library. Dynamic weights are assigned based on the historical reliability of the sensors, and evidence conflicts are eliminated through weighted averaging and iterative self-fusion. Finally, a comprehensive health status level is output. These stages form a closed-loop link of data acquisition, feature extraction, and fusion decision-making, achieving complementary verification of multi-source information.

[0043] Compared to existing technologies, traditional methods relying on single-parameter analysis, such as judging the mechanism's state solely based on the current of the opening and closing coils, are prone to misjudgments due to changes in contact resistance. This solution, through multi-dimensional data fusion, such as simultaneously analyzing the correlation characteristics of vibration signals and displacement curves, can effectively distinguish between mechanical jamming and electrical control faults. Existing DS evidence theory, when directly fusing multi-sensor evidence, can cause fusion conflicts when a sensor drifts. This solution reduces the influence of anomalous evidence through dynamic weight allocation; for example, setting a higher weight coefficient for displacement sensors ensures the dominant decision-making role of high-precision sensors. Traditional threshold judgment methods can only identify explicit faults; this solution, through a three-level state classification and confidence assessment, can identify early performance degradation trends based on the probability distribution of the fusion results.

[0044] Through the above technical solutions, this application achieves coordinated monitoring of the mechanical, electrical, and insulation conditions of low-voltage circuit breakers, solving the problem of information loss caused by monitoring a single parameter. Multi-source data complementarity verification reduces the risk of misjudgment caused by environmental interference, an improved evidence fusion algorithm enhances conflict evidence processing capabilities, and a hierarchical strategy based on confidence thresholds enables progressive condition assessment from healthy to faulty, providing a reliable decision-making basis for preventative maintenance.

[0045] In this embodiment, the energy values ​​of each frequency band after wavelet packet decomposition are extracted from the vibration signal as feature vectors; the key time points and holding current values ​​in the current waveform are extracted from the opening and closing coil current signal as feature vectors; and one or more of the average speed, maximum speed, and bounce time of the moving contact are calculated from the moving contact displacement signal as feature vectors.

[0046] The energy values ​​of each frequency band after wavelet packet decomposition refer to the energy proportion of each frequency band after the vibration signal is decomposed into different frequency bands through wavelet packet transform. Specifically, a 5-level decomposition can be performed using the db4 wavelet basis function to divide the vibration signal into 32 frequency bands and calculate the proportion of energy in each frequency band to the total energy. This feature can reflect the energy distribution of mechanical vibration in different frequency bands and is used to detect abnormal vibration modes caused by mechanism jamming or loose parts.

[0047] The key time points in the current waveform refer to the characteristic moments on the current curve of the opening and closing coil that reflect the movement stage of the iron core. Specifically, the starting moment and the iron core's engagement point can be detected by the extreme points of the first derivative of the current curve. The holding current value refers to the current amplitude after the coil is energized and stabilized. Specifically, it can be obtained by using a sliding window mean filter and taking the average value of the steady-state interval. This feature can quantify the electromagnet's operating timing characteristics and coil resistance changes, and is used to assess coil aging or power supply abnormalities.

[0048] The average velocity of the moving contact refers to the rate of displacement change of the moving contact from its initial position to complete closure or discontinuity, which can be calculated differentially from the displacement signal. The maximum velocity refers to the peak velocity during this process. The bounce time refers to the duration of vibration caused by mechanical collision after the contact is made, which can be measured by the duration of high-frequency oscillation of the displacement signal after the contact is closed. This feature can characterize the transmission efficiency of the operating mechanism and the degree of contact wear, and is used to identify mechanical jamming or contact ablation.

[0049] Specifically, after wavelet packet decomposition, the energy distribution of vibration signals in each frequency band can reflect the vibration characteristics of different components of the circuit breaker operating mechanism. For example, abnormal energy in the high-frequency band may indicate loose parts, while changes in energy in the low-frequency band may indicate jamming of the mechanism. The critical time point delay of the opening and closing coil current can characterize the lag in electromagnet action, and a decrease in the holding current value may indicate an inter-turn short circuit in the coil. A decrease in the moving contact speed directly reflects poor lubrication of the operating mechanism or spring fatigue, while a prolonged bounce time indicates accelerated contact wear. By simultaneously extracting three types of features—mechanical vibration, electromagnetic characteristics, and motion dynamics—it is possible to cover the state information of the three key links of the circuit breaker: mechanical transmission, electromagnetic drive, and contact, forming multi-dimensional monitoring indicators.

[0050] Compared to existing technologies, traditional methods typically extract features from only a single signal source, such as analyzing only the time-domain statistics of vibration signals or the amplitude parameters of opening and closing currents. This makes it impossible to distinguish the coupled effects of mechanical and electrical faults. In contrast, this solution designs feature extraction methods for different physical processes. For example, it uses wavelet packet decomposition to reveal the frequency domain characteristics of vibration, detects the timing of core movements based on current derivatives, and calculates dynamic motion parameters using displacement differentials. This allows for the independent quantification and mutual verification of three types of state information: mechanical impact, electromagnetic response, and motion performance. This solves the problem of incomplete characterization of complex faults by single-signal analysis.

[0051] Through the above technical solution, this application can extract complementary features with clear physical meaning from multi-source heterogeneous signals. For example, it can detect mechanical anomalies through vibration frequency energy, evaluate the state of electromagnetic systems through current time-series parameters, and quantify mechanism performance through motion velocity. This overcomes the problem that single-parameter analysis cannot fully characterize mechanical, electrical, and motion characteristics. The three types of feature vectors correspond to state information in different physical dimensions, providing a quantitative basis for subsequent fusion diagnosis that can distinguish fault modes, effectively reducing the risk of misjudgment due to incomplete information.

[0052] In step S300, the construction of the improved weighted DS evidence theory fusion diagnostic model includes the following sub-steps: S31: Establish an identification framework that includes all possible health status propositions of the circuit breaker; The identification framework is defined as θ = {θ1, θ2, θ3}, where θ1 represents the "healthy" state, θ2 represents the "warning" state, and θ3 represents the "fault" state. This framework contains all the propositions that need to be identified.

[0053] S32: Based on the Mahalanobis distance between each feature vector and the standard state sample library, generate a basic probability assignment function for each type of sensor evidence; Based on Mahalanobis distance, a BPA is generated for each sensor piece of evidence. The Mahalanobis distance calculation formula is: ;in This is the feature vector of a certain sensor of the circuit breaker under test; In a pre-established standard state sample library, all states belonging to state θ i The mean vector of the sample feature vectors. Let be the covariance matrix of the above sample features.

[0054] The distance is then mapped to BPA using an exponential function: in, This evidence supports the conclusion that the circuit breaker is in good condition. The basic probability; The uncertainty probability of this evidence indicates that it cannot be determined. The adjustment coefficient, used to control the sensitivity of the influence of distance on probability, is usually determined through experimental optimization. A preset distance threshold is used to control the range of uncertainty. This process is repeated for each sensor piece of evidence to obtain its respective BPA function: , , , as well as .

[0055] S33: Assign different weighting coefficients to each sensor based on its historical reliability or importance; That is, assign a weighting coefficient to each sensor piece of evidence. ,satisfy = 1. The weight can be determined based on the diagnostic accuracy of historical data for this type of sensor or its correlation with core conditions (such as mechanical conditions).

[0056] S34: Using the aforementioned weighting coefficients, calculate the weighted average evidence of all sensor evidence. : ; A: Any subset of the identification framework θ; N is the total number of sensor evidences; The BPA value of the j-th piece of evidence for proposition A.

[0057] S35: Perform multiple self-fusions on the weighted average evidence to obtain the final comprehensive basic probability allocation result. Specifically: The weighted average evidence... Treating it as a unified body of evidence, Dempster's combination rule is used to fuse it (N-1) times to obtain the final comprehensive basic probability allocation result. And according to the final Make a decision. Set a confidence threshold λ (e.g., λ = 0.7).

[0058] like ( If ) > λ, then the circuit breaker is determined to be in a state of... state.

[0059] If the confidence level of all propositions is lower than λ, output "Uncertain" and suggest manual intervention.

[0060] The system's final output includes the overall state level, confidence level, and the main contributing sources of sensor evidence.

[0061] Example 1

[0062] The present invention also provides a low-voltage circuit breaker condition detection system for implementing the method, comprising: The data acquisition module includes the aforementioned multiple sensors; The data processing module is used to perform signal preprocessing and feature extraction; The data fusion and diagnosis module, which incorporates the improved weighted DS evidence theory fusion and diagnosis model, is used to perform evidence fusion and state diagnosis.

[0063] Example 2

[0064] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method.

[0065] Example 3

[0066] The present invention also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0067] Through the above technical solutions, this application effectively reduces the probability of misjudgment caused by fluctuations in a single parameter, and achieves early identification of equipment performance degradation trends through the setting of early warning states. The application of dynamic confidence thresholds enables the diagnostic system to adapt to different environmental noise levels and sensor accuracy differences, improving the detection sensitivity of weak abnormal features while ensuring the reliability of fault alarms. The probability allocation comparison mechanism after multi-source information fusion solves the problem of difficulty in decision-making when evidence conflicts occur in traditional methods, providing a quantitative basis for state classification.

[0068] As used in the specification and claims, certain terms refer to specific components. Those skilled in the art will understand that hardware manufacturers may use different names to refer to the same component. This specification and claims do not distinguish components based on differences in name, but rather on differences in function. The term "comprising" throughout the specification and claims is an open-ended term and should be interpreted as "comprising but not limited to." "Approximately" means that within an acceptable margin of error, those skilled in the art can solve the technical problem and substantially achieve the technical effect within a certain margin of error.

[0069] The foregoing description illustrates and describes several preferred embodiments of the present invention. However, as previously stated, it should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for detecting the condition of a low-voltage circuit breaker, characterized in that, Includes the following steps: Step S100: Simultaneously collect multi-source status data of the low-voltage circuit breaker during operation using multiple sensors. The multi-source status data includes at least vibration signals, opening and closing coil current signals, moving contact displacement signals, temperature signals, and partial discharge signals. Step S200: Preprocess the various signals collected and extract the feature vectors related to the health status of the circuit breaker. Step S300: Input the extracted feature vectors into the improved weighted DS evidence theory fusion diagnostic model; the improved weighted DS evidence theory fusion diagnostic model fuses evidence from multiple sensors by assigning weights to evidence from different sensors and calculating a weighted average evidence. Step S400: Determine the overall health status level of the low-voltage circuit breaker based on the output of the fusion diagnostic model.

2. The low-voltage circuit breaker condition detection method according to claim 1, characterized in that, In step S100, the vibration signal is acquired by a vibration sensor installed on the housing of the operating mechanism; the opening and closing coil current signal is acquired by a current sensor connected to the coil circuit; the moving contact displacement signal is acquired by a linear displacement sensor connected to the moving contact; the temperature signal is acquired by a temperature sensor installed on the conductive circuit of the circuit breaker; and the partial discharge signal is acquired by an ultra-high frequency sensor installed near the arc-extinguishing chamber.

3. The low-voltage circuit breaker condition detection method according to claim 1, characterized in that, In step S200, the energy values ​​of each frequency band after wavelet packet decomposition are extracted from the vibration signal as feature vectors; the key time points and holding current values ​​in the current waveform are extracted from the opening and closing coil current signal as feature vectors. One or more of the following parameters are calculated from the moving contact displacement signal: average velocity, maximum velocity, and bounce time of the moving contact, and used as the feature vector.

4. The low-voltage circuit breaker condition detection method according to claim 1, characterized in that, In step S300, the construction of the improved weighted DS evidence theory fusion diagnostic model includes the following sub-steps: S31: Establish an identification framework that includes all possible health status propositions of the circuit breaker; S32: Based on the Mahalanobis distance between each feature vector and the standard state sample library, generate a basic probability assignment function for each type of sensor evidence; S33: Assign different weighting coefficients to each sensor based on its historical reliability or importance; S34: Using the aforementioned weighting coefficients, calculate the weighted average evidence of all sensor evidence; S35: Perform multiple self-fusions on the weighted average evidence to obtain the final comprehensive basic probability allocation result.

5. The low-voltage circuit breaker condition detection method according to claim 4, characterized in that, In step S32, the calculated Mahalanobis distance is mapped to the value of the basic probability assignment function using an exponential function. This function value represents the degree of support and uncertainty of the corresponding sensor evidence for each health state proposition.

6. The low-voltage circuit breaker condition detection method according to claim 4, characterized in that, In step S34, the weighted average evidence is obtained by weighting and summing the basic probability allocation functions of each sensor evidence according to their weight coefficients.

7. The low-voltage circuit breaker condition detection method according to claim 1, characterized in that, In step S400, the comprehensive health status level includes healthy, early warning, and fault; during decision-making, propositions whose basic probability allocation value in the final fusion result exceeds the preset confidence threshold are judged as the current status of the circuit breaker.

8. A low-voltage circuit breaker condition monitoring system, used to implement the method described in any one of claims 1-7, characterized in that, include: The data acquisition module includes the aforementioned multiple sensors; The data processing module is used to perform signal preprocessing and feature extraction; The data fusion and diagnosis module, which incorporates the improved weighted DS evidence theory fusion and diagnosis model, is used to perform evidence fusion and state diagnosis.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-7.

Images