Method for predicting the life of a circuit breaker opening accessory based on the time of the movement and the energy of the drive

Abstract

The application is a circuit breaker opening accessory life prediction method based on action time and driving energy. Firstly, the original vibration is pretreated to obtain the denoised vibration signal. Then, the denoised vibration signal is framed, the short-time energy entropy ratio of each frame signal is calculated, each action stage in the vibration process is calibrated based on the short-time energy entropy ratio, and the action time and driving energy of each action stage are calculated. The action time and driving energy are used as performance degradation indicators. The performance degradation indicators with high correlation with the degradation ability are screened through the Spearman rank correlation coefficient method to obtain the key performance degradation indicators. Finally, according to the key performance degradation indicators, a circuit breaker opening accessory residual life prediction model is established based on the binary Wiener process. The application uses the action time and driving energy as the performance degradation indicators. The prediction model reflects the influence of multiple key performance degradation indicators on the overall degradation trend, and avoids the inaccurate prediction results caused by using the action time as a single indicator.

Description

Technical Field

[0001] The present invention belongs to the technical field of remaining life prediction of circuit breaker operating accessories, and in particular provides a method for predicting the life of circuit breaker opening accessories based on action time and driving energy. Background Art

[0002] Universal circuit breakers, as important electrical switching equipment, perform switching, control, and protection functions in power distribution systems. Their reliability impacts the safe and stable operation of the system. The contact system, the circuit breaker's actuator, is required to quickly disconnect the faulty circuit when a fault occurs in the distribution system, protecting the system. Considering the operating principle of universal circuit breakers, the proper movement of the energy storage accessory mechanism and the opening and closing accessories is a prerequisite for ensuring the normal operation of the circuit breaker contacts. Typically, as the number of circuit breaker operations increases, the opening and closing accessories are subject to mechanical wear and external environmental influences, causing their performance to degrade. When this degradation reaches a certain level, it can lead to sluggish opening and closing operations, or even cause circuit breaker failure. Therefore, considering the important role of circuit breaker opening and closing accessories, analyzing their degradation patterns through monitored data and timely predicting their remaining useful life (RUL) are of great significance for improving circuit breaker operational reliability.

[0003] In order to enable the circuit breaker to understand its operating status in time under normal working conditions and reduce the economic losses caused by power outages, it is necessary to perform real-time signal detection on the circuit breaker operating status. Signal measurement is the core of circuit breaker life prediction, which contains a large amount of internal status information of the circuit breaker. For multi-source signal monitoring, the following methods are currently mainly included: opening and closing coil current signal detection, vibration signal detection. Since the opening and closing coil current waveform can reflect the movement stroke of the moving iron core and the working conditions of the controlled latch and interlocking mechanism during operation, the change characteristics of the opening and closing coil current waveform can be extracted and analyzed to evaluate the operating status of the circuit breaker operating accessories. For example, Zhang Yongkui et al. (Zhang Yongkui, Zhao Zhizhong, Feng Xu, Guo Xue. Mechanical fault diagnosis of high-voltage circuit breakers based on opening and closing coil current signals [J]. High Voltage Electrical Appliances, 2013, 49(02): 37-42) collected the opening and closing coil current and used a combination of wavelet analysis and dynamic time warping to conduct a specific analysis of the signal, thereby accurately obtaining the circuit breaker status information. At the same time, the vibration signal is generated by a series of mechanism impacts during the operation of the circuit breaker, which contains rich and important mechanical state information and plays an important role in the status assessment of the equipment. For example, Zhao Shutao et al. (Zhao Shutao, Xu Wenjie, Li Yunpeng, Xia Xiaofei. Energy storage state identification method of vacuum circuit breaker spring mechanism based on optimal universal features [J]. High Voltage Technology, 2019, 55(11): 3777-3784) extracted the time-frequency spectrum characteristics of the vibration signal during the energy storage process of the circuit breaker and combined it with the K nearest neighbor algorithm to identify the state of the energy storage operating mechanism.

[0004] Existing literature on the research of circuit breaker performance degradation is still insufficient. For example, Sun Shuguang et al. (Sun Shuguang, Zhang Wei, Wang Jingqin, Du Taihang, Gao Hui. Prediction of mechanical life of low-voltage circuit breakers based on vibration detection during operation [J]. Chinese Journal of Scientific Instrument, 2020, 41(12): 146-157) analyzed the vibration signals generated by the circuit breaker during operation, extracted the deep features and parameterized features of the key vibration interval as the input of the life assessment model, and realized the prediction of the remaining mechanical life of the contact system. However, due to the complex internal structure of the circuit breaker and the changeable actual working conditions, its degradation process is complex and uncertain. It is even more necessary to explore multi-source degradation characteristics to fully reflect the mechanical degradation state and construct a life prediction model that conforms to its degradation characteristics to achieve accurate prediction of the remaining life of the circuit breaker operating accessories. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a method for predicting the life of circuit breaker trip accessories based on action time and driving energy.

[0006] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0007] The present invention provides a method for predicting the life of a circuit breaker opening accessory based on operation time and driving energy, comprising the following steps:

[0008] The first step is to collect the current, voltage and original vibration signals during the circuit breaker opening process;

[0009] The second step is to reduce the noise of the collected original vibration signal to obtain a noise-reduced vibration signal;

[0010] The third step is to divide the vibration signal after noise reduction into frames and calculate the short-time energy entropy ratio of the vibration signal of each frame; based on the short-time energy entropy ratio, calibrate the start and end times of each vibration sub-event, and divide the vibration stages that occur during the circuit breaker opening process according to the start and end times of the vibration sub-event; then, calculate the action time and driving energy of each vibration stage, where the action time refers to the time difference between the occurrence and end times of the vibration sub-event, and the driving energy refers to the energy consumed by driving the opening accessories during the vibration stage; use each action time and driving energy as performance degradation indicators, and use the Spearman rank correlation coefficient method to screen out performance degradation indicators with high correlation with degradation capacity, thereby obtaining key performance degradation indicators;

[0011] Step 4: Based on the binary Wiener process, a remaining life prediction model for circuit breaker opening accessories is established to predict the remaining life of circuit breaker opening accessories;

[0012] 4-1. Verify whether the relative increments of each key performance degradation indicator conform to the normal distribution. If they conform to the normal distribution, it indicates that the key performance degradation indicator obeys the Wiener process. If the change of the key performance degradation indicator X(t) conforms to the Wiener process, the performance degradation amount at time t is expressed as:

[0013] X(t)=x0+ηt+σ B B(t) (18)

[0014] Where x0 is the initial value of the degradation; η is the drift coefficient, which represents the speed of performance degradation; σ B is the diffusion coefficient, and σ B >0; B(t) is the standard Brownian motion;

[0015] For any time t j , j=1,2,···,J, performance degradation increment ΔX(t j )=X(t j )-X(t j-1 ), according to the Wiener property:

[0016]

[0017] Where Δt j =tj -t j-1 is the interval between adjacent moments, and the relative increments in the degradation process are ΔX(t1), ΔX(t2), ···, ΔX(t J ) is regarded as a sample taken from a normal population, then the probability density function of the sample is:

[0018]

[0019] Through formula (20), the likelihood function of the key performance degradation index X(t) is obtained as:

[0020]

[0021] Taking the logarithm of the likelihood function of equation (21) and taking the partial derivative, we can obtain the drift coefficient η and the diffusion coefficient σ respectively. B The maximum likelihood estimate of is:

[0022]

[0023]

[0024] When the performance degradation {X(t), t≥0} reaches or exceeds the failure threshold h, it is usually considered to be a performance failure. Therefore, the remaining life of the circuit breaker trip accessory at time t is T RUL for:

[0025] T RUL ={inf:X(t)>h} (24)

[0026] In the formula, inf represents the infimum of the function;

[0027] The probability density function expression of the remaining life of the circuit breaker opening accessories corresponding to the key performance degradation index X(t) is:

[0028]

[0029] Where f(s) represents the probability of reaching the failure threshold h after time s, and the time corresponding to the maximum value of formula (25) is taken as the remaining life of the circuit breaker trip accessory;

[0030] Repeat step 4-1 for each key performance degradation indicator to obtain the probability density function of the remaining life of the circuit breaker opening accessories based on each key performance degradation indicator;

[0031] 4-2. Based on the marginal distribution functions of each key performance degradation indicator, a joint distribution function is obtained;

[0032] F(X1(t),X2(t),…,X r (t),…X R(t))=C(g1(t),g2(t),…,g r (t),…,g R (t))(26)

[0033] Where, F(X1(t),X2(t),…,X r (t),…X R (t)) is the joint distribution function; g1(t),g2(t),…,g r (t),…,g R (t) is the marginal distribution function of each key performance degradation indicator; C(·) is the Copula function; R is the number of key performance degradation indicators;

[0034] The Copula function is used to analyze the correlation of each key performance degradation indicator, and then the derivative of formula (26) is taken to obtain the joint probability density function of all key performance degradation indicators;

[0035]

[0036] Where: c(·) is the probability density function of the Copula function, represents the probability density function of the remaining life of the circuit breaker trip accessories corresponding to the rth key performance degradation indicator;

[0037] 4-3. Calculate the remaining life of the circuit breaker opening accessories according to formula (27). The number of operations corresponding to the maximum value of the function is the remaining life of the circuit breaker opening accessories.

[0038] Furthermore, the specific process of the second step is: using the CEEMDAN algorithm to decompose the original vibration signal to obtain the residual components and the eigenmode components arranged in descending frequency; calculating the correlation coefficient between the original vibration signal and each eigenmode component, performing wavelet denoising on the eigenmode components with a correlation coefficient greater than 0.5, and reconstructing the residual components and the denoised eigenmode components to obtain the denoised vibration signal.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. As the number of trip accessory actuations increases, transmission mechanism jamming, trip device failure, and loose mechanical components may occur, leading to increased friction during the actuation process and changes in load force. Consequently, the actuation time and drive energy increase significantly. Therefore, both actuation time and drive energy can reflect the degradation performance of the trip accessory. However, in the early stages of a trip action, the actuation time changes relatively steadily. As the number of actuations increases, the local volatility of the actuation time increases. Existing technologies mostly use actuation time as a single performance degradation indicator, which can lead to inaccurate predictions and small predicted values. The local volatility of drive energy, on the other hand, is relatively gentle, which can improve the accuracy of the prediction results. Therefore, the present invention uses actuation time and drive energy as performance degradation indicators.

[0041] 2. Since the mechanical structures of the circuit breaker collide with each other to generate complex vibration signals, and considering that the vibration signals have non-stationary and time-varying characteristics, the present invention analyzes the vibration signals and current signals collected during the opening process. Since the operation of the opening accessory is time-sequential, the various vibration sub-events in the vibration process are calibrated based on the short-time energy entropy ratio and the double threshold method. The short-time energy entropy ratio has a strong noise suppression ability, which can take into account the tiny vibration signals generated by the opening accessory during the operation process, and then highlight the waveform characteristics of the vibration event. By extracting the driving energy and time parameters in each stage as the performance degradation index of the opening accessory, the degree of performance degradation of the opening accessory can be reflected. Since the action time and driving energy of each action stage have different effects on the remaining life of the circuit breaker opening accessory, the Spearman rank correlation coefficient method is used to screen the indicators that characterize the strong degradation ability of the opening accessory, which plays a vital role in improving the accuracy of the remaining life prediction.

[0042] 3. This invention constructs a binary Wiener model, using selected key performance degradation indicators as inputs. The prediction model reflects the impact of multiple key performance degradation indicators on the overall degradation trend, effectively improving the accuracy of predicting the remaining life of circuit breaker trip accessories. Life prediction tests were conducted on circuit breaker samples from three different manufacturers. The results show that the proposed model improves prediction accuracy compared to a univariate Wiener process and exhibits better prediction transfer performance.

[0043] 4. The CEEMDAN algorithm of the present invention effectively solves the problems of modal aliasing and endpoint effects generated by EMD and EEMD during the vibration signal decomposition process, reduces the residual noise in the intrinsic modal components, effectively reduces the reconstruction error, and has a global stopping standard at each stage of decomposition, which makes the decomposition efficiency higher and more suitable for modal analysis of vibration signals. By calculating the cross-correlation coefficient between the intrinsic modal components and the original signal, the high-frequency noisy components are screened out, and then processed and reconstructed using wavelet denoising. The reconstructed signal is the noise-reduced signal. Using CEEMDAN and wavelet denoising to combine noise reduction reduces the interference of noise and makes the impact characteristics of the vibration signal more obvious, which is conducive to the calibration of subsequent action events. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the overall flow chart of the present invention;

[0045] Figure 2 This is a schematic structural diagram of a circuit breaker life test system according to the present invention;

[0046] Figure 3 A physical diagram of the circuit breaker tripping accessories and a schematic diagram of the sensor installation position of the present invention;

[0047] Figure 4 It is the waveform diagram of each component after the original vibration signal is decomposed;

[0048] Figure 5 This is a comparison chart of the original vibration signal and the vibration signal after noise reduction;

[0049] Figure 6 Schematic diagram for oscillator event calibration;

[0050] Figure 7 Schematic diagram of vibration stage division;

[0051] Figure 8 The Spearman rank correlation heat map of performance degradation indicators and remaining life;

[0052] Figure 9 This is a degradation trend diagram of circuit breaker trip accessories based on operation time;

[0053] Figure 10 This is a degradation trend diagram of circuit breaker trip accessories based on driving energy;

[0054] Figure 11 It is the relative incremental cumulative probability curve of the key performance degradation indicator W3;

[0055] Figure 12 The relative incremental cumulative probability curve of the key performance degradation indicator T2

[0056] Figure 13 The probability density and remaining life prediction results of sample 1 in each test cycle are shown in the figure;

[0057] Figure 14 The probability density and remaining life prediction results of sample 2 in each test cycle are shown in the figure;

[0058] Figure 15 The probability density and remaining life prediction results of sample 3 in each test cycle are shown in the figure. DETAILED DESCRIPTION

[0059] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of protection of the present application is not limited thereto.

[0060] The present invention provides a method for predicting the life of circuit breaker opening accessories based on action time and driving energy (hereinafter referred to as the method, see Figures 1 to 15 ), including the following steps:

[0061] The first step is to collect the current, voltage and original vibration signals during the circuit breaker opening process;

[0062] The second step is to use the CEEMDAN algorithm to decompose the original vibration signal to obtain the residual components and the eigenmode components arranged in descending frequency. The correlation coefficient between the original vibration signal and each eigenmode component is calculated. The eigenmode components with a correlation coefficient greater than 0.5 are subjected to wavelet denoising. The residual components and the denoised eigenmode components are then reconstructed to obtain the denoised vibration signal.

[0063] 2-1. Define operator E k (·) is the k-th order IMF component obtained by EMD decomposition of the signal; first, M white noise components w are added to the original vibration signal x(n) with a data length of N. m (n), and get the noisy signal x m (n), its expression is:

[0064] x m (n) = x(n) + β0w m (n) (1)

[0065] Where β0 is the noise coefficient, n=1,2,…,N, m=1,2,···,M;

[0066] Each noisy signal is decomposed by the EMD algorithm, and the first-order eigenmode component of each decomposition is selected Calculate the average value of all first-order eigenmode components to obtain the mean value of the first-order eigenmode components

[0067]

[0068] Then the first residual component r1(n) is:

[0069]

[0070] Add a white noise sequence to the first residual component r1(n) Then decomposed by EMD, the mean of the second-order eigenmode component is obtained

[0071]

[0072] Where β1 is the noise coefficient, E1(w m (n)) is the white noise component w m (n) The first-order IMF component obtained by EMD decomposition;

[0073] When k = 2, 3, ···, K, the kth residual component r is calculated by formula (5) respectively k (n);

[0074]

[0075] Where r k-1 (n) is the k-1th residual component, is the mean value of the k-th order eigenmode component;

[0076] In the kth residual component r k Add white noise sequence β to (n) k E k (w m (n)), and then decomposed by EMD to obtain the mean of the k+1th order eigenmode component

[0077]

[0078] Repeat the above operation until the residual signal cannot be decomposed, that is, the residual component has only one extreme value, and the final residual component R(n) is obtained;

[0079]

[0080] In summary, the original vibration signal x(n) can be expressed as follows after decomposition by CEEMDAN:

[0081]

[0082] 2-2. The correlation coefficient is a method that describes the closeness of the relationship between objective things and judges the similarity and correlation between things through certain processing. The correlation coefficient between the original vibration signal and each eigenmode component is calculated, and the eigenmode components with a correlation coefficient greater than 0.5 are selected for wavelet denoising. The remaining components are reconstructed with the denoised eigenmode components to obtain the noise-reduced vibration signal.

[0083] The correlation coefficient between the original vibration signal x(n) and the signal y(n) is defined as shown in formula (9), where the signal y(n) should be each eigenmode component;

[0084]

[0085] The third step is to frame the denoised vibration signal and calculate the short-time energy entropy ratio of each frame. Based on the short-time energy entropy ratio, the double threshold method is used to calibrate the start and end times of each vibration sub-event. The vibration stages that occur during the circuit breaker opening process are divided according to the start and end times of the vibration sub-events. Then, the action time and drive energy of each vibration stage are calculated. Each action time and drive energy is used as a performance degradation indicator. The performance degradation indicators with high correlation with degradation capability are screened using the Spearman rank correlation coefficient method.

[0086] 3-1. The noise-reduced vibration signal is framed and the short-time energy entropy ratio of each frame is calculated. The short-time energy entropy ratio has strong noise suppression capabilities and can take into account the tiny vibration signals generated by the tripping accessories during operation, thereby highlighting the waveform characteristics of the vibration event.

[0087] A window function is used to divide the denoised vibration signal into a time series in frames, that is, the denoised vibration signal is framed. Since the Haining window can effectively reduce spectrum leakage and high-frequency interference during signal processing, this embodiment processes the denoised vibration signal using the Haining window. The definition of the Haining window is as follows:

[0088] ω(v)=0.5-(1-cos(2πv / (L-1)))0≤v≤L-1 (10)

[0089] Where L is the frame length;

[0090] After framing, the vibration signal z of frame i i (l) is expressed as:

[0091] z i (l)=ω(v)*x((i-1)*inc+l) (11)

[0092] Where inc is the frame shift length, l = 1, 2, ···, L, i = 1, 2, ···, I, and I is the total number of frames after the frame shift;

[0093] For the vibration signal z of frame i i (l) Perform fast Fourier transform and assume that the energy spectrum of the qth frequency component is S i (q), then the normalized spectral probability density function p of the qth frequency component is i (q) is defined as:

[0094]

[0095] The vibration signal z of the i-th frame i The short-time spectral entropy of (l) is defined as:

[0096]

[0097] The vibration signal z of the i-th frame i The short-time energy of (l) is:

[0098]

[0099] In the formula, a is a constant, and the value of a should make the effective features more obvious;

[0100] The vibration signal z of the i-th frame i The short-time energy entropy ratio of (l) is:

[0101]

[0102] Repeat the above operation to calculate the short-time energy entropy ratio of each frame of vibration signal; after the vibration signal is processed by the short-time energy entropy ratio, the impact change of the signal can be effectively enhanced;

[0103] 3-2. Based on the short-time energy entropy ratio, a double-threshold method is used for endpoint detection to calibrate the start and end times of each vibration sub-event. The vibration stages that occur during the circuit breaker opening process are divided according to the start and end times of the vibration sub-events. The start time of the overall vibration event during the circuit breaker opening process is determined by the current generation time, and the start time of each vibration stage is determined by the vibration signal. The action time and drive energy of each vibration stage are calculated and used as performance degradation indicators.

[0104] Action time T: the time t when each vibration sub-event occurs a and the end time t b The time difference is:

[0105] T=t b -t a (16)

[0106] Driving energy W: The energy consumed by driving the opening accessories during the vibration stage, that is:

[0107]

[0108] Where u(n) and i(n) are the voltage and current signals corresponding to each vibration stage, respectively, and D is the signal length;

[0109] 3-3. Calculate the correlation between each performance degradation indicator and the remaining life of the circuit breaker opening accessories using the Spearman rank correlation coefficient method. If there is a significant correlation or above, it indicates that the performance degradation indicator is highly correlated with the degradation capacity of the circuit breaker opening accessories. The performance degradation indicator with a high correlation with the degradation capacity of the circuit breaker opening accessories is used as the key performance degradation indicator for analysis in the fourth step;

[0110] Step 4: Based on the binary Wiener process, a remaining life prediction model for circuit breaker opening accessories is established to predict the remaining life of circuit breaker opening accessories;

[0111] 4-1. First, verify whether the relative increments of each key performance degradation indicator conform to the normal distribution. If they conform to the normal distribution, it indicates that the key performance degradation indicator obeys the one-dimensional Wiener process. If the change of the key performance degradation indicator X(t) conforms to the Wiener process, the performance degradation amount at time t can be expressed as:

[0112] X(t)=x0+ηt+σ B B(t) (18)

[0113] Where x0 is the initial value of the degradation; η is the drift coefficient, which represents the speed of performance degradation; σ B is the diffusion coefficient, and σ B >0; B(t) is the standard Brownian motion, which characterizes the time-varying uncertainty in the performance degradation process;

[0114] For any time t j , j=1,2,···,J, performance degradation increment ΔX(t j )=X(t j )-X(t j-1 ), according to the Wiener property:

[0115]

[0116] Where Δt j =t j -t j-1 is the interval between adjacent moments, and the relative increments in the degradation process are ΔX(t1), ΔX(t2), ···, ΔX(t J ) is regarded as a sample taken from a normal population, then the probability density function of the sample is:

[0117]

[0118] Through formula (20), the likelihood function of the key performance degradation index X(t) is obtained as:

[0119]

[0120] Taking the logarithm of the likelihood function of equation (21) and taking the partial derivative, we can obtain the drift coefficient η and the diffusion coefficient σ respectively. B The maximum likelihood estimate of is:

[0121]

[0122]

[0123] When the performance degradation {X(t), t≥0} reaches or exceeds the failure threshold h, it is usually considered to be a performance failure. Therefore, the remaining life of the circuit breaker trip accessory at time t is T RUL for:

[0124] T RUL ={inf:X(t)>h} (24)

[0125] In the formula, inf represents the infimum of the function;

[0126] When the parameters η, σ are determined B After determining the value of h, the probability density function expression of the remaining life of the circuit breaker opening accessories corresponding to the key performance degradation index X(t) is obtained as follows:

[0127]

[0128] Where f(s) represents the probability of reaching the failure threshold h after time s, and the time corresponding to the maximum value of formula (25) is taken as the remaining life of the circuit breaker trip accessory;

[0129] Repeat step 4-1 for each key performance degradation indicator to obtain the probability density function of the remaining life of the circuit breaker opening accessories based on each key performance degradation indicator;

[0130] 4-2. When the marginal distribution functions of each key performance degradation indicator are known, the joint distribution function is obtained according to Sklar's theorem;

[0131] F(X1(t),X2(t),…,X r (t),…X R (t))=C(g1(t),g2(t),…,g r (t),…,g R (t))(26)

[0132] Where, F(X1(t),X2(t),…,X r (t),…X R (t)) is the joint distribution function; g1(t),g2(t),…,g r (t),…,g R (t) is the marginal distribution function of each key performance degradation indicator; C(·) is the Copula function; R is the number of key performance degradation indicators;

[0133] The Copula function is used to analyze the correlation of each key performance degradation indicator, and then the derivative of formula (26) is taken to obtain the joint probability density function of all key performance degradation indicators;

[0134]

[0135] Where: c(·) is the probability density function of the Copula function, represents the probability density function of the remaining life of the circuit breaker trip accessories corresponding to the rth key performance degradation indicator;

[0136] The above analysis shows that the Copula function is crucial for constructing the joint probability density function. Commonly used Copula functions include Gaussian, Frank, Gumbel, and Clayton. Different Copula functions may produce different results. Therefore, the appropriate Copula function is selected based on the actual performance degradation of the circuit breaker trip accessories. The Akaike information criterion (AIC) is used to judge the quality of the Copula function fit. Its expression is as follows:

[0137] E AIC =-2ln(A)+2e (28)

[0138] Where ln(A) is the log-likelihood function, e is the number of unknown parameters in the log-likelihood function, and when E AIC The smaller the value of , the better the fitting effect;

[0139] 4-3. Formula (27) is the remaining life prediction model of the circuit breaker opening accessories. The remaining life of the circuit breaker opening accessories is calculated according to Formula (27). The number of operations corresponding to the maximum value of the function is the remaining life of the circuit breaker opening accessories, and the remaining life prediction is completed.

[0140] Example

[0141] The following specific embodiments are given to further illustrate the technical solution of the present invention, which include the following steps:

[0142] The first step is to use three DW15-1600 universal circuit breakers of the same specification but from different manufacturers as test pieces. In order to comprehensively analyze the operating status of the tripping accessory, i.e., the shunt release, a circuit breaker life test system is designed to collect signals during the mechanical life test of the circuit breaker. The principle of the test system is as follows: Figure 2 As shown in Figure 2, in this test system, the current, voltage and vibration signals generated during the circuit breaker opening process are collected respectively; considering that the opening accessory structure is small and easily disturbed by the vibration of other mechanisms, Figure 3 LC0159 acceleration sensors are installed at different positions shown to collect vibration signals of the tripping accessories.

[0143] The second step is to perform CEEMDAN decomposition on the original vibration signal, and decompose the original vibration signal into the intrinsic mode components IMF1~IMF12 arranged from high to low frequency and the residual component R1. Figure 4 The waveforms of each component are shown in Table 1. The correlation coefficients between the original vibration signal and each intrinsic mode component are calculated, and the results shown in Table 1 are obtained. Since high-frequency vibration signals are easily affected by noise, the intrinsic mode components IMF1 to IMF4 with correlation coefficients greater than 0.5 are subjected to wavelet denoising. The denoised intrinsic mode components and the remaining components are reconstructed to obtain the denoised vibration signal. Figure 5 The waveform comparison diagram of the original vibration signal and the vibration signal after noise reduction.

[0144] Table 1 Correlation coefficients between each eigenmode component and the original vibration signal

[0145]

[0146] The third step is to divide the vibration signal after noise reduction into frames and calculate the short-time energy entropy ratio of each frame vibration signal; based on the short-time energy entropy ratio, the double threshold method is used to calibrate the start and end time of each vibration sub-event, such as Figure 6 As shown in Figure 2, the vibration stages that occur during the circuit breaker opening process are divided according to the start and end times of the vibration sub-events, as shown in Figure 2. Figure 7As shown, the vibration that occurs during the circuit breaker opening process is divided into four stages. Stage t0-t1: After the opening command is issued, the coil begins to energize. Before time t1, the electromagnetic force generated by the current is less than the external resistance on the moving iron core, causing it to remain stationary. Stage t1-t2: As the current increases, the moving iron core transitions from a stationary state to a moving state. During this time, a vibration sensor attached to the opening attachment detects a weak vibration signal. This stage can indicate any jamming of the moving iron core during movement. Stage t2-t3: At time t2, the ejector rod strikes the baffle, generating slight vibration. At time t3, the transmission mechanism's latch opens, releasing the opening spring's constraint and generating noticeable vibration. The moving contact begins to move, and the contact change voltage signal jumps. This stage reflects mechanical degradation of the transmission mechanism and possible malfunctions in the tripping device. During the t3-t4 phase, the moving contact reaches its maximum opening distance at time t4 and strikes the stop pin, generating intense vibration. The auxiliary contact operates, and the current in the trip accessory coil drops to zero, completing the trip operation. The operating time and drive energy for each vibration phase were calculated from both the vibration and electrical aspects. Using these time parameters and drive energy as performance degradation indicators, the results are shown in Table 2.

[0147] Table 2 Performance degradation indicators at each vibration stage

[0148]

[0149] The Spearman rank correlation coefficient method is used to calculate the correlation between each performance degradation index and the remaining life of the circuit breaker trip accessory under each number of operations. In this embodiment, the remaining life is measured in units of the number of operations, which is the full life of the trip accessory minus the number of operations in the current state. Based on the analysis of the three samples, the visualization results of the Spearman rank correlation coefficient are as follows: Figure 8 As shown in the figure, the three performance degradation indicators W3, T2, and T3 are highly correlated with the remaining life of the trip accessories. Due to the coupling relationship between the indicators, the correlation between the indicators needs to be calculated to remove redundant performance degradation indicators.

[0150] Through the analysis of the tripping accessories action process, we know that in the t2-t3 stage, the tripping lock is the most important mechanism in the tripping process. Its degradation degree and triggering timeliness will directly affect whether the tripping operation can be performed normally. Therefore, T2 is selected as the performance degradation index for analyzing the tripping accessories and as the basis for judging the mechanical life failure. Finally, W3 and T2 are selected as performance degradation indicators. The overall degradation trend is as follows: Figure 9 、 10As shown in the figure, analysis of indicator T2 reveals that the trip accessory is relatively stable in the initial stages of operation, but enters a degradation phase as the number of operations increases. Furthermore, T2 is highly sensitive to life degradation during this phase, with significant local fluctuations. Analyzing the remaining life prediction based solely on T2 would result in inaccurate predictions. However, the W3 indicator for individual test pieces is less sensitive to life degradation but exhibits relatively gentle local fluctuations. Therefore, T2 and W3 can be considered key performance degradation indicators to improve the accuracy of trip accessory life predictions.

[0151] The fourth step is to establish a remaining life prediction model for circuit breaker opening accessories based on the binary Wiener process; according to the relative increments of the two key performance degradation indicators T2 and W3, calculate the relative increment cumulative probability, such as Figure 11 、 12 As shown in Figure 2, since the relative increments are basically linear, it shows that the relative increments conform to the normal distribution, that is, both key performance degradation indicators obey the Wiener process. The drift coefficient and diffusion coefficient of the two key performance degradation indicators in each detection cycle are calculated using equations (22) and (23), respectively. The AIC Akaike information criterion method is used to select an appropriate copula function to describe the correlation between the two key performance degradation indicators. The results are shown in Table 3.

[0152] Table 3 AIC values ​​of four Copula function models

[0153]

[0154] The Gaussian Copula function is used to describe the correlation between the two key degradation indicators. The joint probability density function of the two key degradation indicators is obtained by combining formula (27). The number of operations corresponding to the maximum value of the function is the remaining life of the circuit breaker opening accessories corresponding to the detection cycle. The probability density and remaining life prediction results of the three test samples in each detection cycle are compared in

[15] . Figures 13-15 .

[0155] As the number of circuit breaker operations increases, the probability density function based on the time parameter and the drive energy performance degradation index fluctuates significantly within different detection cycles, resulting in inaccurate prediction results. This is due to the uncertainty in the degradation of the tripping accessories, which leads to the low accuracy of the univariate Wiener prediction. However, the peak value of the joint probability density function based on the binary Wiener process shows a steady growth trend, and the interval corresponding to this peak value gradually narrows, overcoming the influence of degradation uncertainty in the life prediction process.

[0156] The average relative errors of the life predictions for the three specimens were 4.3%, 9.77%, and 8.46%, respectively. For specimen 1, the average relative errors of the time- and energy-based univariate Wiener processes were 23.3% and 28.2%, respectively. Compared with the univariate Wiener method based on the time parameter, this method reduced the average relative error by 19%. Similarly, for the other two specimens, the error reductions were 5.83% and 8.94%, respectively. Therefore, the combined vibration-electric analysis and binary Wiener prediction significantly improved RUL prediction accuracy.

[0157] The data acquisition software of this embodiment is partially implemented based on LabVIEW software, and the data processing and life prediction are partially implemented based on Matlab. The LabVIEW and Matlab languages ​​used in this embodiment are well known to those skilled in the art. The percentages in this embodiment are all numerical percentages.

[0158] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A method for predicting the life of circuit breaker trip accessories based on action time and driving energy, characterized in that: The method comprises the following steps: The first step is to collect the current, voltage and original vibration signals during the circuit breaker opening process; The second step is to reduce the noise of the collected original vibration signal to obtain a noise-reduced vibration signal; The third step is to divide the vibration signal after noise reduction into frames and calculate the short-time energy entropy ratio of the vibration signal of each frame; based on the short-time energy entropy ratio, calibrate the start and end times of each vibration sub-event, and divide the vibration stages that occur during the circuit breaker opening process according to the start and end times of the vibration sub-event; then, calculate the action time and driving energy of each vibration stage, where the action time refers to the time difference between the occurrence and end times of the vibration sub-event, and the driving energy refers to the energy consumed by driving the opening accessories during the vibration stage; use each action time and driving energy as performance degradation indicators, and use the Spearman rank correlation coefficient method to screen out performance degradation indicators with high correlation with degradation capacity, thereby obtaining key performance degradation indicators; Step 4: Based on the binary Wiener process, a remaining life prediction model for circuit breaker opening accessories is established to predict the remaining life of circuit breaker opening accessories; 4-1. Verify whether the relative increments of each key performance degradation indicator conform to the normal distribution. If they conform to the normal distribution, it indicates that the key performance degradation indicator obeys the Wiener process. If the change of the key performance degradation indicator X(t) conforms to the Wiener process, the performance degradation amount at time t is expressed as: X(t)=x0+ηt+σ B B(t) (18) Where x0 is the initial value of the degradation; η is the drift coefficient, which represents the speed of performance degradation; σ B is the diffusion coefficient, and σ B >0; B(t) is the standard Brownian motion; For any time t j , j=1,2,···,J, performance degradation increment ΔX(t j )=X(t j )-X(t j-1 ), according to the Wiener property: Where Δt j =t j -t j-1 is the interval between adjacent moments, and the relative increments in the degradation process are ΔX(t1), ΔX(t2), ···, ΔX(t J ) is regarded as a sample taken from a normal population, then the probability density function of the sample is: Through formula (20), the likelihood function of the key performance degradation index X(t) is obtained as: Taking the logarithm of the likelihood function of equation (21) and taking the partial derivative, we can obtain the drift coefficient η and the diffusion coefficient σ respectively. B The maximum likelihood estimate of is: When the performance degradation {X(t), t≥0} reaches or exceeds the failure threshold h, it is usually considered to be a performance failure. Therefore, the remaining life of the circuit breaker trip accessory at time t is T RUL for: T RUL ={inf:X(t)>h} (24) In the formula, inf represents the infimum of the function; The probability density function expression of the remaining life of the circuit breaker opening accessories corresponding to the key performance degradation index X(t) is: Where f(s) represents the probability of reaching the failure threshold h after time s, and the time corresponding to the maximum value of formula (25) is taken as the remaining life of the circuit breaker trip accessory; Repeat step 4-1 for each key performance degradation indicator to obtain the probability density function of the remaining life of the circuit breaker opening accessories based on each key performance degradation indicator; 4-2. Based on the marginal distribution functions of each key performance degradation indicator, a joint distribution function is obtained; F(X1(t),X2(t),…,X r (t),…X R (t))=C(g1(t),g2(t),…,g r (t),…,g R (t)) (26) Where, F(X1(t),X2(t),…,X r (t),…X R (t)) is the joint distribution function; g1(t),g2(t),…,g r (t),…,g R (t) is the marginal distribution function of each key performance degradation indicator; C(·) is the Copula function; R is the number of key performance degradation indicators; The Copula function is used to analyze the correlation of each key performance degradation indicator, and then the derivative of formula (26) is taken to obtain the joint probability density function of all key performance degradation indicators; Where: c(·) is the probability density function of the Copula function, represents the probability density function of the remaining life of the circuit breaker trip accessories corresponding to the rth key performance degradation indicator; 4-3. Calculate the remaining life of the circuit breaker opening accessories according to formula (27). The number of operations corresponding to the maximum value of the function is the remaining life of the circuit breaker opening accessories.

2. The method for predicting the life of circuit breaker opening accessories based on action time and driving energy according to claim 1, characterized in that: The specific process of the second step is as follows: the original vibration signal is decomposed using the CEEMDAN algorithm to obtain the residual components and the eigenmode components arranged in descending frequency; the correlation coefficient between the original vibration signal and each eigenmode component is calculated, the eigenmode components with a correlation coefficient greater than 0.5 are subjected to wavelet denoising, and the residual components and the denoised eigenmode components are reconstructed to obtain the denoised vibration signal.

Images