A method for on-line measurement of mechanical characteristic time parameters in a closing process of a circuit breaker

Abstract

The application discloses an online measurement method of mechanical characteristic time parameters in a circuit breaker closing process, comprising the following steps: S1. signal acquisition: synchronously collecting vibration signals at at least two key mechanical components in the circuit breaker closing process; S2. signal processing: performing adaptive noise reduction purification on the collected vibration signals to enhance the signal-to-noise ratio in the signals; S3. feature extraction: calculating time-domain statistical characteristic functions of the vibration signals after noise reduction purification to highlight the impact event characteristics in the vibration signals; and S4. parameter measurement: based on the time-domain statistical characteristic functions, using an endpoint detection algorithm to locate the start and end points of the vibration events, and measuring the mechanical characteristic time parameters in the closing process according to the same. The application can non-invasively and highly accurately detect multiple key mechanical action time points in the circuit breaker closing process, and provides reliable data support for intelligent operation and maintenance of the circuit breaker.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, specifically to an online measurement method for the time parameters of mechanical characteristics during the circuit breaker closing process. Background Technology

[0002] High-voltage circuit breakers are critical control and protection devices in power systems, and their operational reliability directly affects the safe and stable operation of the power grid. The mechanical characteristic time parameters of circuit breakers (such as the timing of closing latch release, contact instantaneous closing, and overtravel) are core indicators for evaluating their mechanical condition and operational reliability.

[0003] Currently, obtaining these timing parameters typically relies on coil current signals, moving contact travel curves, and contact displacement signals. However, implementing these methods requires installing current sensors, displacement sensors, etc., which usually necessitates intrusion into the equipment, resulting in complex wiring, high costs, and difficulty in achieving long-term online real-time monitoring.

[0004] Vibration signal analysis provides a possibility for non-intrusive condition detection of circuit breakers. During the operation of a circuit breaker, the collision and friction of its internal mechanical components will excite vibration signals rich in state information. However, the vibration signal of a circuit breaker has strong non-stationary and nonlinear characteristics, and is a superposition of vibration events of multiple components, resulting in low signal-to-noise ratio and difficulty in feature extraction. Existing vibration signal-based methods mostly focus on qualitative discrimination of the state, making it difficult to accurately quantify and extract millisecond-level mechanical characteristic time parameters. Moreover, existing methods usually have the following problems: (1) insufficient signal preprocessing and serious noise interference; (2) weak ability of characteristic functions to represent vibration impact and easy to be interfered with by background noise; (3) poor robustness of event detection algorithms, resulting in low accuracy of time parameter extraction and high rates of misjudgment and missed judgment.

[0005] Therefore, there is an urgent need for a method that can measure the time parameters of the mechanical characteristics of circuit breakers with high precision and robustness from vibration signals. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an online measurement method for the mechanical characteristic time parameters of the circuit breaker closing process, which can non-invasively and with high precision detect multiple key mechanical action time points in the circuit breaker closing process, and provide reliable data support for the intelligent operation and maintenance of circuit breakers.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows.

[0008] An online measurement method for the time parameter of mechanical characteristics during the closing process of a circuit breaker includes the following steps: S1. Signal Acquisition: During the circuit breaker closing process, vibration signals at at least two key mechanical components are acquired simultaneously; S2. Signal Processing: Adaptive noise reduction and purification are performed on the acquired vibration signals to enhance the signal-to-noise ratio. S3. Feature Extraction: Calculate the time-domain statistical feature function of the vibration signal after noise reduction and purification to highlight the impact event characteristics in the vibration signal; S4. Parameter Measurement: Based on the time-domain statistical characteristic function, the endpoint detection algorithm is used to locate the start and end points of vibration events, and the mechanical characteristic time parameters of the closing process are measured accordingly.

[0009] Preferably, the signal processing in step S2 specifically includes: S21. The improved Osprey optimization algorithm is used to adaptively optimize the parameter combination of variational mode decomposition to obtain the decomposition parameters that best match the current vibration signal. S22. Based on the optimal parameter combination obtained through optimization, variational mode decomposition is performed on the vibration signal to obtain several intrinsic mode function components; S23. Select the effective components containing the main vibration information from the intrinsic mode function components; S24. Perform stationary wavelet threshold denoising on the selected effective components and reconstruct the purified vibration signal.

[0010] Preferably, the improved Osprey optimization algorithm in step S21 specifically includes: The Latin hypercube sampling method was used to initialize the population in order to enhance population diversity; During the predation phase of the algorithm, the individuals with the best fitness values ​​are retained as elite individuals, while the remaining non-elite individuals are subjected to adaptive dynamic perturbation and random perturbation around the current global optimal solution in different proportions. When a population is detected to be trapped in a local optimum, a stagnation perturbation strategy is enforced to escape the local optimum.

[0011] Preferably, the specific process of initializing the population using the Latin hypercube sampling method is as follows: The initialization formula for the i-th individual parameter is:

[0012] in, K i and α i These represent the decomposition level represented by the i-th individual in the initial population. K and penalty factor α Candidate values; K min 、 、α min and Representing the number of decomposition levels K and penalty factor α The preset upper and lower limits; lhs (i,j) Represents the Latin hypercube sampled value, and lhs (i,1) Number of decomposition layers K The sampled value generated at the i-th individual, lhs (i,2) As a penalty factor α The sampled value generated at the i-th individual, lhs (i,j) The calculation formula is:

[0013] in, N Indicates population size; π j Indicates the first j Random permutation on the dimension; π j ( i ) indicates the first i The individual in the first j The interval number of the dimension; u (i,j) It represents a uniform random number on the interval (0,1).

[0014] Preferably, the perturbation factor of the adaptive dynamic perturbation dynamically shrinks with each iteration, and the number of decomposition layers is adjusted using the following formula. K and penalty factor α Perform perturbation and update:

[0015] in, perturb_K Indicates the number of decomposition layers K The adaptive perturbation factor; perturb_α Indicates the penalty factor α The adaptive perturbation factor; iter Indicates the current iteration number; Indicates the maximum number of iterations; Represents the maximum value function; round This represents the rounding function; The formula for updating individual location is:

[0016] in, K new and α new Indicates the current parameter; r Represents a random factor in the range [0, 0.7]. Ki , α i Indicates the current individual parameters; K best , α best This represents the current globally optimal parameter; This indicates that a random integer is taken from the integer interval [a, b].

[0017] Preferably, the effective component selection in step S23 is based on the product of the maximum cross-correlation between each component and the original signal and its spectral kurtosis integral.

[0018] Preferably, the time-domain statistical feature function in step S3 is the short-time robust difference kurtosis feature function, and its calculation formula is as follows:

[0019] in, F i These are the output feature values; x i It is the input signal frame; x i ( t ) is the first frame of the signal after being framed by the Hamming window. i Frame signal; T Frame length; For the first i The mean of the frame signal; This is a median function.

[0020] Preferably, the endpoint detection algorithm used in step S4 is an improved dual-threshold endpoint detection method, specifically: Set two detection thresholds, high and low, and introduce an event state machine to manage the detection logic; When the characteristic function value exceeds a high threshold during a period not in the event of a vibration event, the point where the nearest trough or the differential rate of change before the intersection point is greater than a preset threshold is determined as the starting point of the vibration event. During a vibration event, when the characteristic function value falls back to a low threshold, the nearest trough point after that intersection point is determined as the end point of the vibration event.

[0021] Preferably, in step S1, at least two key mechanical components include the crossbeam of the circuit breaker and the operating mechanism body; the moment when the contacts just close is extracted from the vibration signal at the crossbeam, and the moment when the closing latch unlocks, the moment when the cam is driven, the moment when the overtravel is reached, and the moment when the rebound is obtained are extracted from the vibration signal at the operating mechanism body.

[0022] Due to the adoption of the above technical solutions, the technical progress achieved by this invention is as follows.

[0023] This invention enables non-invasive measurement: detection can be completed simply by installing a universal vibration sensor on the outside of the circuit breaker, without intruding into the internal wiring of the equipment, which greatly reduces implementation costs, operational difficulty and safety risks.

[0024] This invention achieves adaptive signal processing: it adopts an improved Osprey optimization algorithm to adaptively optimize variational mode decomposition parameters, overcoming the shortcomings of traditional methods that rely on experience for parameters, and improving the adaptability of the noise reduction and purification process to different working conditions and signal characteristics.

[0025] This invention achieves robust feature extraction: the short-time robust difference kurtosis feature function combines the robustness of the median absolute deviation and the sensitivity of the higher-order moment to the impact, which can effectively enhance the vibration and impact characteristics and suppress noise interference.

[0026] This invention achieves high-precision event localization: the improved dual-threshold endpoint detection method significantly improves the localization accuracy of the start and end points of vibration events and reduces false positives and false negatives by introducing an event state machine and a trough / differential rate of change backtracking strategy.

[0027] This invention achieves comprehensive parameter extraction: it can completely extract multiple key mechanical characteristic time parameters such as the closing latch unlocking time, cam drive time, contact just-close time, overtravel time, and rebound time from the vibration signals of the crossbeam and the operating mechanism body, providing more comprehensive data support for circuit breaker condition assessment.

[0028] In summary, this invention enables non-invasive, high-precision, and robust online measurement of the mechanical characteristics and time parameters during the circuit breaker closing process, significantly improving the accuracy of circuit breaker condition monitoring and maintenance efficiency. Attached Figure Description

[0029] Figure 1 This is a flowchart of the present invention; Figure 2 This is a flowchart of step S2 of the present invention; Figure 3 This is a general flowchart of Embodiment 1 of the present invention; Figure 4 This is a flowchart of signal noise reduction and purification according to Embodiment 1 of the present invention; Figure 5 This is a comparison diagram of the time-domain waveforms of the vibration signal at the operating mechanism body before and after noise reduction in Embodiment 1 of the present invention; Figure 6 This is a waveform comparison diagram of the vibration signal at the operating mechanism body in Embodiment 1 of the present invention before and after short-time robust differential kurtosis processing; Figure 7 This is a comparison chart of the time parameter detection results of the beam vibration signal in Embodiment 1 of the present invention (traditional dual-threshold endpoint detection method and the improved dual-threshold endpoint detection method of the present invention). Figure 8 This is a comparison chart of the time parameter detection results of the vibration signal of the operating mechanism body in Embodiment 1 of the present invention (traditional dual-threshold endpoint detection method and the improved dual-threshold endpoint detection method of the present invention). Detailed Implementation

[0030] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0031] An online measurement method for the time parameters of mechanical characteristics during the closing process of a circuit breaker, combined with Figure 1 As shown, it includes the following steps: S1. Signal Acquisition: During the circuit breaker closing process, vibration signals at at least two key mechanical components are acquired simultaneously.

[0032] During the circuit breaker closing process, the operating mechanism drives the contact system to complete the closing action. During this process, the impact, friction, and movement of a series of mechanical components (such as latches, cams, connecting rods, and contacts) generate specific vibration signals. These vibration signals propagate through the circuit breaker structure, exhibiting different time-frequency characteristics at different locations. Simultaneous acquisition of vibration signals from multiple key mechanical components allows for the construction of a complete vibration spectrum of the closing process, thereby obtaining raw vibration data reflecting the mechanical actions at each stage of the circuit breaker closing process, providing an information source for subsequent processing. Furthermore, simultaneous acquisition from multiple locations ensures the ability to distinguish vibration events generated by different components, avoiding signal confusion.

[0033] Specifically, during the circuit breaker closing process, vibration sensors are used to simultaneously collect vibration signals from at least two key mechanical components, preferably the circuit breaker's crossbeam and the operating mechanism body. The sampling frequency should be high enough to capture millisecond-level transient impacts, and the sampling duration should cover the entire closing process. The moment the contacts just close is extracted from the vibration signal at the crossbeam, and the moments of closing latch unlocking, cam drive, overtravel, and rebound are extracted from the vibration signal at the operating mechanism body. This step is fundamental to achieving non-invasive online measurement. Synchronous acquisition ensures that signals from different measuring points are strictly aligned in time, providing raw data for subsequent accurate correlation and extraction of various mechanical characteristic time parameters.

[0034] S2. Signal Processing: Adaptive noise reduction and purification are performed on the acquired vibration signals to enhance the signal-to-noise ratio.

[0035] This step aims to adaptively extract the effective vibration components generated by mechanical motion from strong background noise, such as... Figure 2 As shown, the specific steps include: S21. An improved Osprey optimization algorithm is used to adaptively optimize the parameter combination of variational mode decomposition to obtain the decomposition parameters that best match the current vibration signal.

[0036] The performance of variational mode decomposition (VMD) is highly dependent on the number of decomposition layers. K and penalty factor α The selection of parameters is crucial. Traditional methods rely on fixed parameters or empirical selection, which struggles to adapt to the signal characteristics of different circuit breaker types and operating conditions. The improved Osprey optimization algorithm simulates the predation behavior of an osprey, using a population evolution mechanism to find the optimal solution in the parameter space. This adaptively determines the optimal VMD decomposition parameters based on the characteristics of the currently acquired vibration signal, ensuring the best signal decomposition effect and laying the foundation for subsequent component selection and noise reduction.

[0037] The improved Osprey Optimization Algorithm (IOOA) specifically includes: The Latin hypercube sampling method was used to initialize the population in order to enhance population diversity.

[0038] During the predation phase of the algorithm, the individuals with the best fitness values ​​are retained as elite individuals, while the remaining non-elite individuals are subjected to adaptive dynamic perturbations and random perturbations around the current global optimal solution in different proportions.

[0039] When a population is detected to be trapped in a local optimum, a stagnation perturbation strategy is enforced to escape the local optimum.

[0040] Specifically, the process of initializing the population using the Latin hypercube sampling method is as follows: The initialization formula for the i-th individual parameter is:

[0041] in, K i and α i These represent the decomposition level represented by the i-th individual in the initial population. K and penalty factor α Candidate values; K min 、 、α min and Representing the number of decomposition levels K and penalty factor α The preset upper and lower limits; lhs (i,j) Represents the Latin hypercube sampled value, and lhs (i,1) Number of decomposition layers K The sampled value generated at the i-th individual, lhs (i,2) As a penalty factor α The sampled value generated at the i-th individual, lhs (i,j) The calculation formula is:

[0042] in, N Indicates population size; π j Indicates the first j Random permutation on the dimension; π j ( i ) indicates the first i The individual in the first j The interval number of the dimension; u (i,j) It represents a uniform random number on the interval (0,1).

[0043] A hybrid perturbation strategy is adopted during the predation phase: elite individuals are retained, adaptive dynamic perturbation is applied to some non-elite individuals, and random perturbation is applied to another group of non-elite individuals, with the perturbation factor dynamically shrinking with iteration.

[0044] Specifically, the perturbation factor of the adaptive dynamic perturbation dynamically shrinks with each iteration, and the number of decomposition layers is calculated using the following formula. K and penalty factor α Perform perturbation and update:

[0045] in, perturb_K Indicates the number of decomposition layers K The adaptive perturbation factor; perturb_α Indicates the penalty factor α The adaptive perturbation factor; iter Indicates the current iteration number; Indicates the maximum number of iterations; Represents the maximum value function; round This represents the rounding function; The formula for updating individual location is:

[0046] in, K new and α new Indicates the current parameter; r Represents a random factor in the range [0, 0.7]. K i , α i Indicates the current individual parameters; K best , α best This represents the current globally optimal parameter; This indicates that a random integer is taken from the integer interval [a, b].

[0047] In this step, Latin hypercube sampling ensures that the initial population is uniformly distributed in the parameter space, avoiding blind spots in the search; a hybrid perturbation strategy balances global exploration and local exploitation capabilities; and a dynamically shrinking perturbation factor improves convergence accuracy.

[0048] S22. Based on the optimal parameter combination obtained through optimization, variational mode decomposition is performed on the vibration signal to obtain several intrinsic mode function components.

[0049] Variational Mode Decomposition (VMD) adaptively decomposes an input signal into a specified number of intrinsic mode functions (IMFs) by constructing and solving a variational problem. Each IMF is a band-limited amplitude-frequency modulated signal with a specific center frequency. This step can decompose complex multi-component vibration signals into a series of relatively simple single-component signals, achieving effective separation of different frequency components and facilitating the identification and extraction of vibration components related to mechanical motion.

[0050] S23. Select the effective components containing the main vibration information from the intrinsic mode function components.

[0051] The IMF components obtained from the decomposition contain some of the main mechanical vibration information, while others are mainly noise. The maximum cross-correlation reflects the degree of correlation between the component and the original signal, and the spectral kurtosis characterizes the impact characteristics of the signal. The product of the two comprehensively evaluates the information content and feature quality of the component.

[0052] In this step, the effective components are selected based on the product of the maximum cross-correlation between each component and the original signal and its spectral kurtosis integral. This allows for the identification and retention of effective components containing information about major vibration events from numerous IMF components, while removing noise-dominant and irrelevant components, thereby improving the efficiency and accuracy of subsequent processing.

[0053] S24. Perform stationary wavelet threshold denoising on the selected effective components and reconstruct the purified vibration signal.

[0054] Wavelet transform possesses multi-resolution analysis capabilities, enabling the separation of signal and noise at different scales. Stationary wavelet transform maintains translation invariance by eliminating downsampling operations, while BayesShrink (a Bayesian-based method for automatically calculating the optimal threshold for wavelet thresholding) adaptively determines the threshold size based on subband statistical characteristics. This step allows for fine-tuning of the selected effective components, removing residual noise interference while preserving useful impulse characteristics, further improving signal quality.

[0055] S3. Feature Extraction: Calculate the time-domain statistical feature function of the vibration signal after noise reduction and purification to highlight the impact event characteristics in the vibration signal.

[0056] Mechanical impact events during the circuit breaker closing process manifest as brief pulse signals in the time domain, characterized by high peak value and short duration.

[0057] Specifically, the time-domain statistical feature function is the short-time robust difference kurtosis feature function, and its calculation formula is as follows:

[0058] in, F i These are the output feature values; x i It is the input signal frame; x i ( t ) is the first frame of the signal after being framed by the Hamming window. i Frame signal; T Frame length; For the first i The mean of the frame signal; This is a median function.

[0059] The short-time robust differential kurtosis feature function combines the sensitivity of higher-order statistics to shocks with the resistance of robust statistics to outliers. It can enhance the shock characteristics in vibration signals, suppress background noise and slowly changing interference, and generate feature waveforms suitable for event detection.

[0060] S4. Parameter Measurement: Based on the time-domain statistical characteristic function, the endpoint detection algorithm is used to locate the start and end points of vibration events, and the mechanical characteristic time parameters of the closing process are measured accordingly.

[0061] Each mechanical action during the circuit breaker closing process is represented by a characteristic peak on the characteristic function waveform. By accurately detecting the start and end points of these characteristic peaks, the occurrence time of the corresponding mechanical action can be obtained.

[0062] In this step, the endpoint detection algorithm used is an improved dual-threshold endpoint detection method, specifically: Set two detection thresholds, high and low, and introduce an event state machine to manage the detection logic; When the characteristic function value exceeds a high threshold during a period not in the event of a vibration event, the point where the nearest trough or the differential rate of change before the intersection point is greater than a preset threshold is determined as the starting point of the vibration event. During a vibration event, when the characteristic function value falls back to a low threshold, the nearest trough point after that intersection point is determined as the end point of the vibration event.

[0063] The improved dual-threshold endpoint detection method overcomes the positioning bias of traditional methods by introducing an event state machine and a trough backtracking strategy. It can accurately identify the timing information of each vibration event from the characteristic function waveform and extract the mechanical characteristic time parameters of the circuit breaker closing process. Example 1

[0064] The method of this invention was implemented using a 35kV SF6 high-voltage circuit breaker equipped with a CT14 type spring operating mechanism as the test object. Figure 3 As shown, the specific implementation steps are as follows: S1. Signal Acquisition: Acquire vibration signals from the crossbeam and the operating mechanism body.

[0065] Two DH131E piezoelectric accelerometers were installed at the circuit breaker crossbeam (measuring point 1) and the operating mechanism body (measuring point 2), respectively. The sampling frequency was set to 10kHz and the sampling duration to 300ms to ensure complete capture of vibration events during the closing process. Simultaneously, a high-voltage switch characteristic tester was used to synchronously record the moving contact travel signal as a reference for the true value of the time parameter.

[0066] S2. Signal processing.

[0067] like Figure 4 The signal processing flow shown performs noise reduction processing on the vibration signals collected from the measuring points at the crossbeam and the operating mechanism body, respectively. Specifically: S21. Parameter adaptive optimization.

[0068] We use an improved Osprey optimization algorithm (IOOA) to optimize the key parameters of variational mode decomposition.

[0069] During the population initialization phase, Latin hypercube sampling is used to initialize the population using the following formula:

[0070] in, K i and α i These represent the decomposition level represented by the i-th individual in the initial population. K and penalty factor α Candidate values; K min 、 、α min and Representing the number of decomposition levels K and penalty factor α The preset upper and lower limits; lhs (i,j) Represents the Latin hypercube sampled value, and lhs (i,1) Number of decomposition layers K The sampled value generated at the i-th individual, lhs (i,2) As a penalty factor α The sampled value generated at the i-th individual, lhs (i,j) The calculation formula is:

[0071] in, N Indicates population size; π j Indicates the first j Random permutation on the dimension; π j ( i ) indicates the first i The individual in the first j The interval number of the dimension; u (i,j) It represents a uniform random number on the interval (0,1).

[0072] During the predation phase, the number of decomposition layers is determined using the following formula. K and penalty factor α Perform adaptive dynamic perturbation and update: Disturbance factor calculation:

[0073] in, perturb_K Indicates the number of decomposition layers K The adaptive perturbation factor; perturb_α Indicates the penalty factor α The adaptive perturbation factor; iter Indicates the current iteration number; Indicates the maximum number of iterations; Represents the maximum value function; round This represents the rounding function; The formula for updating individual location is:

[0074] in, K new and α new Indicates the current parameter; r Represents a random factor in the range [0, 0.7]. K i , α i Indicates the current individual parameters; K best ,α best This represents the current globally optimal parameter; This indicates that a random integer is taken from the integer interval [a, b].

[0075] The parameters are set as follows: fixed number of decomposition layers. K The range is [3, 8], and the central penalty factor is... α The range is [400, 2500], the population size is 20, the maximum number of iterations is 10, and the frame length is 3ms. The fitness function is the product of the maximum cross-correlation and the spectral kurtosis integral. The optimal parameter combination of the decomposition layer number and the center penalty factor is (7, 1440.7).

[0076] S22. Based on the optimal parameters, variational mode decomposition is performed on the vibration signal to obtain 7 IMF components.

[0077] S23. By calculating the product of the cross-correlation between each IMF component and the original signal and the spectral kurtosis integral, IMF4, IMF6, and IMF7 are selected as effective components.

[0078] S24. Perform stationary wavelet thresholding denoising on these three components (using the sym5 wavelet basis, 5-level decomposition, Firm shrinkage function, and BayesShrink thresholding rule), and finally reconstruct the purified vibration signal. For example... Figure 5 As shown, the time-domain waveforms of the vibration signal at the operating mechanism body before and after noise reduction are compared. After noise reduction, the abnormal impact interference in the background section is significantly suppressed, and the weak impact is highlighted.

[0079] S3. Feature extraction.

[0080] Calculate the short-time robust difference kurtosis characteristic function of the purified signal:

[0081] in, F i These are the output feature values; x i It is the input signal frame; x i ( t ) is the first frame of the signal after being framed by the Hamming window. i Frame signal; T Frame length; For the first i The mean of the frame signal; This is a median function.

[0082] like Figure 6The image shows a waveform comparison of the vibration signal at the operating mechanism body before and after short-time robust differential kurtosis processing. The original signal is the waveform before short-time robust differential kurtosis processing, which shows that the amplitude of the weak event impact is similar to the amplitude of the interference signal, making it easy to misjudge. The noise-reduced signal is the waveform after short-time robust differential kurtosis processing, which shows that the vibration event signal can be highlighted and the interference signal can be suppressed.

[0083] S4. Parameter measurement.

[0084] An improved dual-threshold endpoint detection method is used to locate vibration events. Vibration signals from the crossbeam are measured at the moment the contacts just close (T3), and vibration signals from the operating mechanism body are measured at the moment the closing latch unlocks (T1), the cam drive moment (T2), the overtravel moment (T4), and the rebound moment (T5). Figure 7 and Figure 8 As shown, compared with the traditional dual-threshold endpoint detection method, the improved dual-threshold endpoint detection method uses the nearest trough before the intersection of the feature function waveform and the high threshold as the starting point of the impact event, thus achieving more accurate extraction of time parameters.

[0085] The final extracted closing process time parameters are shown in Table 1. The maximum relative error is only 2.63%, which verifies the effectiveness of this method.

[0086] Table 1. Results of time parameter extraction T1 38.0 38.5 1.32% T2 49.2 49.5 0.61% T3 88.3 87.5 0.91% T4 99.3 100.5 1.21% T5 110.4 107.5 2.63%

Claims

1. An online measurement method for the time parameter of mechanical characteristics during the closing process of a circuit breaker, characterized in that: Includes the following steps: S1. Signal Acquisition: During the circuit breaker closing process, vibration signals at at least two key mechanical components are acquired simultaneously; S2. Signal Processing: Adaptive noise reduction and purification are performed on the acquired vibration signals to enhance the signal-to-noise ratio; specifically including: S21. An improved Osprey optimization algorithm is used to adaptively optimize the parameter combination of variational mode decomposition to obtain the decomposition parameters that best match the current vibration signal; the improved Osprey optimization algorithm specifically includes: The Latin hypercube sampling method was used to initialize the population in order to enhance population diversity; During the predation phase of the algorithm, the individuals with the best fitness values ​​are retained as elite individuals, while the remaining non-elite individuals are subjected to adaptive dynamic perturbation and random perturbation around the current global optimal solution in different proportions. When the population is detected to be trapped in a local optimum, a stagnation perturbation strategy is forcibly implemented to escape the local optimum; S22. Based on the optimal parameter combination obtained through optimization, variational mode decomposition is performed on the vibration signal to obtain several intrinsic mode function components; S23. Select the effective components containing the main vibration information from the intrinsic mode function components; In step S23, the effective components are selected based on the product of the maximum cross-correlation between each component and the original signal and its spectral kurtosis integral. S24. Perform stationary wavelet threshold denoising on the selected effective components and reconstruct the purified vibration signal. S3. Feature Extraction: Calculate the time-domain statistical feature function of the vibration signal after noise reduction and purification to highlight the impact event characteristics in the vibration signal; S4. Parameter Measurement: Based on the time-domain statistical characteristic function, the endpoint detection algorithm is used to locate the start and end points of vibration events, and the mechanical characteristic time parameters of the closing process are measured accordingly.

2. The online measurement method for the mechanical characteristic time parameter of a circuit breaker during the closing process according to claim 1, characterized in that: The specific process of initializing the population using the Latin hypercubic sampling method is as follows: The initialization formula for the i-th individual parameter is: in, K i and α i These represent the decomposition level represented by the i-th individual in the initial population. K and penalty factor α Candidate values; K min 、 、α min and Representing the number of decomposition levels K and penalty factor α The preset upper and lower limits; lhs (i,j) Represents the Latin hypercube sampled value, and lhs (i,1) Number of decomposition layers K The sampled value generated at the i-th individual, lhs (i,2) As a penalty factor α The sampled value generated at the i-th individual, lhs (i,j) The calculation formula is: in, N Indicates population size; π j Indicates the first j Random permutation on the dimension; π j ( i ) indicates the first i The individual in the first j The interval number of the dimension; u (i,j) It represents a uniform random number on the interval (0,1).

3. The online measurement method for the mechanical characteristic time parameter of a circuit breaker during the closing process according to claim 1, characterized in that: The perturbation factor of the adaptive dynamic perturbation dynamically shrinks with each iteration, and the number of decomposition layers is calculated using the following formula. K and penalty factor α Perform perturbation and update: in, perturb_K Indicates the number of decomposition layers K The adaptive perturbation factor; perturb_α Indicates the penalty factor α The adaptive perturbation factor; iter Indicates the current iteration number; Indicates the maximum number of iterations; Represents the maximum value function; round This represents the rounding function; The formula for updating individual location is: in, K new and α new Indicates the current parameter; r Represents a random factor in the range [0, 0.7]. K i , α i Indicates the current individual parameters; K best , α best This represents the current globally optimal parameter; This indicates that a random integer is taken from the integer interval [a, b].

4. The online measurement method for the mechanical characteristic time parameter of a circuit breaker during the closing process according to claim 1, characterized in that: In step S3, the time-domain statistical feature function is the short-time robust difference kurtosis feature function, and its calculation formula is as follows: in, F i These are the output feature values; x i It is the input signal frame; x i ( t ) is the first frame of the signal after being framed by the Hamming window. i Frame signal; T Frame length; For the first i The mean of the frame signal; This is a median function.

5. The online measurement method for the time parameter of mechanical characteristics during the closing process of a circuit breaker according to claim 1, characterized in that: The endpoint detection algorithm used in step S4 is an improved dual-threshold endpoint detection method, specifically: Set two detection thresholds, high and low, and introduce an event state machine to manage the detection logic; When the characteristic function value exceeds a high threshold during a period not in the event of a vibration event, the point where the nearest trough or the differential rate of change before the intersection point is greater than a preset threshold is determined as the starting point of the vibration event. During a vibration event, when the characteristic function value falls back to a low threshold, the nearest trough point after that intersection point is determined as the end point of the vibration event.

6. A method for online measurement of the time parameter of mechanical characteristics during the closing process of a circuit breaker according to any one of claims 1 to 5, characterized in that: In step S1, at least two key mechanical components include the circuit breaker's crossbeam and the operating mechanism body; the moment when the contacts just close is extracted from the vibration signal at the crossbeam, and the moment when the closing latch unlocks, the moment when the cam is driven, the moment when the overtravel is reached, and the moment when the rebound is obtained from the vibration signal at the operating mechanism body.

Images