5G Leakage Current Fault Monitoring Method Based on Improved VMD and LMS

Abstract

The present invention discloses a 5G leakage current fault monitoring method based on improved VMD and LMS, including: acquiring load leakage current information and temperature information; using the firefly algorithm to optimize through the leakage current signal to obtain the optimal decomposition number M and penalty factor β of the signal VMD; using the values of M and β to perform VMD decomposition on the original leakage current signal, and dividing the modes in the time domain into effective IMF, mixed IMF, and noise IMF according to the ratio E of the cross-correlation coefficient R and the time-domain energy entropy; using the LMS algorithm to denoise the mixed IMF, and superimposing the denoised mode with the effective IMF to obtain a reconstructed signal; monitoring the reconstructed signal and the temperature signal, and performing fault diagnosis. When a fault occurs, the protector respectively executes an alarm and an action, and in real time transmits the information to the master station through the 5G communication module for analysis and processing. The present invention can improve the denoising effect of the leakage current signal and has a fast optimization speed.

Description

Technical Field

[0001] The present invention relates to the technical field of leakage current fault monitoring. More specifically, the present invention relates to a 5G leakage current fault monitoring method based on improved VMD and LMS. Background Art

[0002] According to the national fire situation of residential places in the past 10 years released by the Fire and Rescue Bureau of the Ministry of Emergency Management. According to statistics, from 2012 to 2021, a total of 1.324 million residential place fires occurred across the country, causing 11,634 deaths, 6,738 injuries, and direct property losses of 7.77 billion yuan. Among them, electrical fires accounted for 42.7% (data source: Fire and Rescue Bureau of the Ministry of Emergency Management, February 18, 2022, https: / / www.119.gov.cn / article / 46rcva01Vzg). Among the reasons for electrical fires, wire short circuits or grounding arcing short circuits account for the largest proportion, which have the remarkable characteristics of suddenness and concealment. And the increase in leakage current during a short circuit is a typical fault criterion. When an electrical device fails or the circuit ages, the leakage current increases slightly. At this time, the leakage current changes little compared with the normal situation and is not easily detected. As the leakage current increases, the accumulation of its energy will accelerate the aging of the circuit, which will cause the insulation layer of the cable to break down (electrical breakdown) at a certain moment and cause a grounding arcing short circuit at the breakdown point. Therefore, it is precisely the high concealment of the leakage current that leads to the frequent occurrence of electrical fire accidents. For electrical fires caused by circuit short circuits, installing an overcurrent protection device in the circuit can actively cut off the circuit when the circuit current is too large, although it can effectively avoid the occurrence of such fires, but its probability of false alarms and missed alarms is also relatively high, which has an adverse impact on the smart grid and economic development.

[0003] The leakage current signal is a weak non-linear signal with serious noise interference. Therefore, while ensuring the effectiveness of the detection signal, it is necessary to achieve fast and accurate leakage current detection. Accurately extracting the effective components from the leakage current signal is the basis for ensuring the rapid and correct operation of the leakage current protector. At present, a variety of methods have been proposed to improve the accuracy and speed of leakage current detection. Compared with traditional leakage current protectors that only act according to the peak current, it is easy to cause false actions and rejection actions. Modern signal processing methods such as wavelet analysis and neural networks overcome the shortcomings of traditional leakage current protectors and can further improve the reliability of leakage current protectors. However, restricted by environmental factors and real-time requirements, their denoising effect and recognition accuracy cannot be guaranteed. Variational mode decomposition (VMD) is an adaptive signal processing method proposed in recent years and is widely used in signal denoising and feature extraction. And how to apply variational mode decomposition to the leakage current protector to improve the detection ability of the leakage current protector has become an urgent problem to be solved. Summary of the Invention

[0004] One object of the present invention is to solve at least the above-mentioned deficiencies and provide at least the advantages described hereinafter.

[0005] Another object of the present invention is to provide a 5G leakage current fault monitoring method based on improved VMD and LMS, to solve the problems of poor current leakage denoising effect and low recognition accuracy in high-temperature environments in summer. By improving VMD and LMS to perform denoising processing on the leakage current signal, and at the same time, according to the comprehensive trend recognition method and the 5G communication module, together perform leakage current fault diagnosis and real-time communication, so as to efficiently and accurately identify the leakage current signal, providing guarantee for its fault diagnosis and reducing the occurrence of electrical fires.

[0006] To achieve these objects and other advantages of the present invention, the present invention provides a 5G leakage current fault monitoring method based on improved VMD and LMS, including:

[0007] S1: Obtain the load leakage current information and temperature information, and use them as the original signals;

[0008] S2: Use the firefly algorithm to optimize the original leakage current signal to obtain the optimal decomposition number M and penalty factor β of the signal VMD;

[0009] S3: Use the obtained optimal M and β values to perform VMD decomposition on the original leakage current signal, and use the ratio E of the cross-correlation coefficient R between the reconstructed signal and the original signal and the time-domain energy entropy obtained by the decomposition as the classification index, divide the modes in the time domain into three categories: effective IMF, mixed IMF, and noise IMF, and retain the effective IMF and remove the noise IMF;

[0010] S4: Use the LMS algorithm to denoise the mixed IMF, and superimpose the denoised mode and the effective IMF to obtain the reconstructed leakage current signal;

[0011] S5: Monitor the reconstructed leakage current signal and the temperature signal, and perform fault diagnosis. When a fault occurs, the protector executes alarm and action respectively, and during this period, the information is transmitted to the master station through the 5G communication module in real time for analysis and processing to avoid false alarms.

[0012] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, the leakage current information is the vector sum of the ground leakage of each phase wire.

[0013] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, specifically, the firefly algorithm uses the envelope entropy value S g = f(M0, β0) as the fitness function. After the leakage current signal undergoes VMD decomposition and the firefly algorithm optimization process, S gThe minimum value is also obtained for the optimal parameters M and β of VMD.

[0014] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, the process of signal VMD decomposition is as follows:

[0015] 1) Initialize {u 1 m}, {ω 1 m}, τ 1 m and n = 0;

[0016] 2) Enter the loop;

[0017] 3) Introduce the quadratic penalty factor β and the Lagrange multiplier τ, and transform it into an unconstrained variational problem;

[0018]

[0019] where u m is the M single-component amplitude-frequency modulated signals after decomposition; ω m is the center frequency of each single-component amplitude-frequency modulated signal; n is the number of iterations; τ is the Lagrange multiplier; χ is the noise tolerance parameter; t is time; j is the imaginary unit; δ(t) is the impulse function; is the partial derivative of the function with respect to t;

[0020] 4) Update {u 1 m}, {ω 1 m}, τ 1 m according to the update formula, and the update formula is as follows:

[0021]

[0022] 5) Given the accuracy σ, if the stop condition is satisfied, output the M modal components;

[0023]

[0024] 6) Stop the loop, otherwise go back to step 2).

[0025] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, the process of the firefly algorithm is as follows:

[0026] S21: Initialize the parameters of the firefly algorithm, including the total number of firefly individuals n, the maximum number of iterations t max of the population, the step size factor α of the perturbation at t = 1, the attraction β, and the light absorption coefficient θ;

[0027] S22: Calculate the relative brightness I of the fireflies, and let the fireflies with lower brightness move towards those with higher brightness;

[0028] S23: Calculate the attraction β and update the brightness of the fireflies;

[0029] S24: Record the current solution and compare it with the optimal solution;

[0030] S25: Check whether the stopping condition is met. If it is met, exit and output the result; otherwise, repeat S22 - S24.

[0031] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, in step S3, the calculation formula for the cross - correlation relationship R between the reconstructed signal x′(t) obtained by VMD decomposition and the original leakage current signal x(t) is as follows:

[0032]

[0033] The ratio E of the time - domain energy entropy k is calculated as:

[0034]

[0035] where, for effective IMFs, E k ≥ 8% and R ≥ 0.9; for mixed IMFs, E k ≥ 8% and R < 0.9; for noise IMFs, E k < 8%.

[0036] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, the LMS algorithm has a simple structure and strong robustness. In step S4, the execution process of using the LMS algorithm to denoise the mixed IMF is as follows:

[0037] S41: Filter initialization: Determine the initial weight vector w(0), the initial input signal p(0), η as the step - size factor, and the filter order n

[0038] S42: For each new input sample p(n), calculate the output signal q(n);

[0039] S43: Use the desired output d(n) to calculate the error signal e(n) and obtain the gradient

[0040] S44: Update the weight vector using the LMS update formula:

[0041] S45: Return to S42 until the end, and the output sequence and error sequence can be obtained.

[0042] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, by performing shape extraction and recognition analysis on the leakage current parameter information and environmental temperature information, the leakage current can be monitored. In step S5, by combining the recombined leakage current signal information and temperature information and using the comprehensive trend recognition method to respectively determine the adaptive slopes of the leakage current and temperature curves, when the slope exceeds a predetermined value, it is determined as a sudden increase trend, and alarms and actions are respectively executed, thereby finally realizing the fault diagnosis and monitoring of the leakage current.

[0043] Preferably, in the 5G leakage current fault monitoring method based on improved VMD and LMS, the specific calculation process for the fault diagnosis and monitoring of the leakage current is as follows:

[0044] S51: Calculate the average current fc and the average temperature ft based on the first s data uploaded in S45;

[0045] S52: Determine the initial current setting value oc and the initial temperature setting value ot, then add the next two sampling values to the sequence respectively, and delete the first two sampling values of the sequence;

[0046] S53: Update oc and ot again;

[0047] S54: Fit the leakage current and temperature curves and calculate the adaptive slopes Vc and Vt; when the absolute value of Vc exceeds the predetermined value, it is transmitted to the master station through 5G communication for alarm and real-time analysis and diagnosis of faults. When the absolute value of Vt exceeds the predetermined value, the protection device's actuator operates to disconnect the circuit to achieve leakage current protection.

[0048] A 5G leakage current protector that executes the 5G leakage current fault monitoring method based on improved VMD and LMS, which includes:

[0049] A detection element for obtaining the load leakage current information and temperature information;

[0050] A denoising element for the optimization and denoising step of the 5G leakage current fault monitoring method based on improved VMD and LMS;

[0051] An amplification element for amplifying the recombined leakage current signal and temperature signal;

[0052] A threshold comparison element for monitoring and fault diagnosis of the amplified recombined leakage current signal and temperature signal;

[0053] An actuator for executing alarms and actions according to the fault diagnosis results;

[0054] A 5G communication element for communication between the leakage current protector and the master station.

[0055] The present invention has at least the following beneficial effects:

[0056] 1. The firefly algorithm is used to optimize the optimal decomposition number M and the quadratic penalty factor β of VMD, and the modes in the VMD time domain are classified according to the ratio E of the cross-correlation coefficient R and the time-domain energy entropy. Finally, the LMS algorithm is used to perform secondary denoising on the mixed modes, improving the denoising effect of the leakage current signal, with high optimization accuracy, fast optimization speed, and excellent performance in terms of accuracy and speed.

[0057] 2. The leakage current fault is identified according to the comprehensive trend recognition method, and the alarm and action of the leakage current protector are realized according to the adaptive slopes Vc and Vt of the current and temperature respectively, reducing the probability of false alarms and missed alarms of the leakage current protector in high-temperature environments.

[0058] 3. Aiming at the problems of insufficient detection ability of the original leakage current protector and the inability of the communication module to transmit information in real time, the present invention uses a 5G communication module to transmit the leakage current signal in real time, which not only reduces the delay but also facilitates the master station to perform predictive analysis and real-time processing on the signal.

[0059] Other advantages, objectives, and features of the present invention will be partially reflected by the following description and partially understood by those skilled in the art through the research and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flowchart of the 5G leakage current fault monitoring method based on improved VMD and LMS of the present invention;

[0061] Figure 2 is a structural diagram of the 5G leakage current protector of the present invention;

[0062] Figure 3 is a flowchart of the firefly algorithm of the present invention;

[0063] Figure 4 is a flowchart of the LMS algorithm of the present invention;

[0064] Figure 5 is a flowchart of the comprehensive trend recognition method of the present invention

[0065] Figure 6 is a convergence comparison graph of the firefly algorithm, particle swarm optimization algorithm, and grey wolf search algorithm of the present invention;

[0066] Figure 7 is an optimization curve graph of the penalty factor based on FA-VMD of the present invention;

[0067] Figure 8 is an optimization curve graph of the decomposition mode number based on FA-VMD of the present invention;

[0068] Figure 9This is a comparison chart of each modal component of FA-VMD of the present invention and the initial leakage current signal. Detailed implementation manners

[0069] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, so that those skilled in the art can implement it according to the description in the specification.

[0070] It should be noted that, unless otherwise specified, the experimental methods described in the following implementation schemes are all conventional methods, and the reagents and materials, unless otherwise specified, can all be obtained from commercial channels.

[0071] At present, leakage current protectors are widely used at home and abroad. The use of leakage current protectors is effective in preventing electrical fires, damage to electrical equipment, and electric shock to the human body, and has become an important part of the safety protection of low-voltage power grids. Leakage current protectors are widely used at home and abroad, and most developed countries focus on the prevention of electric shock to the human body and fires. However, with the increasingly complex power system environment and the continuous improvement of the public's daily power safety requirements, existing leakage current protectors cannot meet their needs. Many electric shock accidents and electrical fires are caused by the insufficient detection ability of leakage current protectors.

[0072] The structure of the leakage current protector of the present invention mainly consists of four basic parts: a detection element, a denoising element, an amplification element, a threshold comparison element, and an execution element. As shown in the attached Figure 1 legend, specifically including: a leakage current and temperature detection element, an improved VMD and LMS denoising element, an amplification element, a leakage current and temperature threshold comparison element, and an execution element. Among them, the execution element includes an alarm device, a switch, etc. The leakage current detection element can be a leakage current sensor, and its current acquisition frequency is set according to needs, generally 10 - 100 times per second; the temperature detection element can be a temperature sensor, and the signal acquisition frequency is also set according to needs, also 10 - 100 times per second.

[0073] When the leakage current protector executes the 5G leakage current fault monitoring method based on improved VMD and LMS:

[0074] When the trends of the leakage current are all determined to be sudden increases, the alarm device alarms the master station through 5G communication and conducts real-time diagnostic analysis. When the trends of the temperature are all determined to be sudden increases, the execution element acts and the main switch disconnects the circuit. Compared with the existing 4G network, 5G can meet the requirements of ultra-high bandwidth, ultra-low latency, and ultra-large-scale connection, and is used to further realize the interconnection of all things, and then serve various vertical industries. Applying the 5G communication module to the leakage current protector can effectively solve the high latency between it and the master station, and realize the real-time monitoring and analysis and decision-making of the line leakage current.

[0075] VMD is an adaptive signal processing method proposed in recent years and is widely used in signal denoising and feature extraction.

[0076] The present invention improves VMD, mainly in the following two points:

[0077] 1. The firefly algorithm is used to optimize the optimal decomposition number M and the quadratic penalty factor β of VMD.

[0078] 2. According to the ratio E of the cross-correlation coefficient R and the time-domain energy entropy as the classification index, the modes in the time domain are divided into three categories: effective IMF, mixed IMF, and noise IMF. The effective IMF is retained and the noise IMF is removed. The LMS algorithm is used to denoise the mixed IMF, and the denoised modes are superimposed with the effective modes to obtain the reconstructed leakage current signal.

[0079] The firefly algorithm is a swarm optimization algorithm that mimics the information exchange and attraction aggregation behavior between fireflies. Its principle is simple, but it has not been proposed for a long time.

[0080] In the present invention, the algorithm rules are idealized into the following three points:

[0081] (1) Do not distinguish the genders of all fireflies. Each firefly can be attracted by any other firefly;

[0082] (2) The brightness of a firefly is only related to the objective function. To solve the brightness optimization problem, the brightness is proportional to the value of the objective function. For example, a method similar to the fitness function is used to establish an optional brightness form.

[0083] (3) The attraction of a firefly is only related to the brightness of the firefly. The darker firefly will move towards the brighter firefly. In addition, the relative brightness will decrease as the distance between fireflies increases. If there is no brighter firefly to be found, the firefly will move randomly within the search space.

[0084] (4) Both the brightness I and the attraction ψ change with the distance r.

[0085] Specifically, it can be given by formulas (1) and (2):

[0086]

[0087]

[0088] I0 and ψ0 are the initial brightness and the attraction when the distance is 0 respectively, θ is the light absorption coefficient, and r is the distance between fireflies. The distance r between two fireflies i and j is given by the equation expressed in (3)

[0089]

[0090] The position update formula of the firefly at each subsequent moment is given by the following formula (4):

[0091]

[0092] The first term in formula (4) represents the position of the firefly at the current moment t, the second term represents the distance between two fireflies due to their attraction, and the last term represents the random perturbation of the firefly, which is beneficial to expanding the search area and avoiding premature stagnation of the algorithm. Among them, α is the step size factor of the perturbation and is a constant between 0 and 1, and G i is a variable quantity obeying the Gaussian distribution. If the brightness of the fireflies is the same, the fireflies move randomly respectively. Through the continuous update of the positions of the fireflies, the group will eventually gather at the position of the firefly with the highest brightness to achieve target optimization. The basic process of the firefly algorithm is as shown in the appendix Figure 3 as follows.

[0093] In the present invention, the firefly uses the envelope entropy value S g = f(M0, β0) as the fitness function. After the signal goes through the VMD and FA optimization processes, the minimum value of S g is obtained, and the optimal parameter decomposition number M and the quadratic penalty factor β of the VMD decomposition are also obtained.

[0094] In the present invention, the specific process of the VMD decomposition is as follows:

[0095] 1) Initialize {u 1 m}, {ω 1 m}, τ 1 m and n = 0;

[0096] 2) Enter the loop;

[0097] 3) Introduce the quadratic penalty factor β and the Lagrange multiplier τ, and convert it into an unconstrained variational problem;

[0098]

[0099] Among them, u m is the M single-component amplitude-frequency modulation signals after decomposition; ω m is the center frequency of each single-component amplitude-frequency modulation signal; n is the number of iterations; τ is the Lagrange multiplier; χ is the noise tolerance parameter; t is the time; j is the imaginary unit; δ(t) is the impulse function; is the partial derivative of the function with respect to t;

[0100] 4) For {u 1 m}, {ω 1 m}, τ 1 m Update according to the update formula, and the update formula is as follows:

[0101]

[0102] 5) Given the precision σ, if the stopping condition is met, output M modal components;

[0103]

[0104] 6) Stop the loop, otherwise go back to step 2).

[0105] Perform VMD decomposition on the original leakage current signal using the optimal M and β values obtained by optimization, and use the ratio E of the cross-correlation coefficient R and the time-domain energy entropy as the classification index to classify the modes in the time domain into three categories: effective IMF, mixed IMF, and noise IMF. The calculation formula of R between the reconstructed signal x′(t) and the original signal x(t) obtained by VMD decomposition is as follows:

[0106]

[0107] Furthermore, the ratio E of the time-domain energy entropy k The calculation formula is:

[0108]

[0109] where x k (t) are K IMF components. Based on the analysis of the leakage current signal, the modes in the time domain can be classified into effective IMF (Ek≥10% and R≥0.8), mixed IMF (Ek≥10% and R<0.8), and noise IMF (Ek<10%). After classification, retain the effective IMF, remove the noise IMF, and perform secondary denoising on the mixed IMF through the LMS algorithm.

[0110] The LMS algorithm is derived based on Wiener filtering and with the help of the steepest descent algorithm. The Wiener solution obtained by Wiener filtering must be determined under the condition of knowing the prior statistical information of the input signal and the desired signal, and then performing an inverse operation on the autocorrelation matrix of the input signal. Based on the criterion of minimum mean square error, the LMS algorithm minimizes the mean square error between the output signal of the filter and the desired output signal, and its vector signal flow diagram is as shown in the appendix Figure 4 as shown.

[0111] where p(n) and w(n) are the input leakage current signal vector and weight vector at time n respectively, and d(n) is the desired output value. The LMS algorithm has a simple structure and strong robustness, so it can be used for secondary denoising of mixed IMF. Its execution process is:

[0112] 1) Filter initialization: Determine the initial weight vector w(0), the initial input signal p(0), the step size factor η, and the filter order n

[0113] 2) For each new input sample p(n), calculate the output signal q(n)

[0114] q(n) = w T (n)p(n) (7)

[0115] 3) Using the desired output d(n), calculate the error signal e(n) to obtain the gradient

[0116]

[0117] 4) Update the weight vector using the LMS update formula:

[0118] w(n + 1) = w(n) + ηe(n)p(n) (9)

[0119] 5) Return to step 2) until the end to obtain the output sequence and the error sequence.

[0120] After obtaining the denoised current sequence, the fault monitoring of the leakage current is realized by combining the data of the temperature sensor. Under normal circumstances, the leakage current and the temperature trend remain stable. Only when a fault occurs, the leakage current will show irregular fluctuations up and down. Therefore, a comprehensive trend recognition algorithm is selected as shown in the appendix Figure 5 When the trend of the leakage current is determined to be a sudden increase, the alarm device alerts the master station through 5G communication and performs real-time diagnostic analysis. When the trend of the temperature is determined to be a sudden increase, the actuator operates and the main switch disconnects the circuit. The calculation process is as follows:

[0121] First, calculate the average value f c and f t ;

[0122] Determine the initial current and temperature setting values o c and o t

[0123] o c = f c + Δl (10)

[0124] o t = f t + Δl (11)

[0125] where Δl is the margin preset according to past fault data.

[0126] Then, the next two sampling values are respectively added to the sequence, and the first two sampling values of the sequence are deleted. Update o again. c and o t , fit the current and temperature curves and find their adaptive slopes V c and V t , when the absolute value of V c exceeds the predetermined value, it is transmitted to the master station for alarm through 5G communication and the fault is analyzed and diagnosed in real time. When the absolute value of V t exceeds the predetermined value, the protection device's actuator operates to disconnect the circuit to achieve leakage current protection.

[0127] Performance test and comparison

[0128] The FA firefly algorithm is compared with PSO (particle swarm optimization algorithm) and GWO (gray wolf search algorithm) through the fitness function S g = f(M0,β0), and the algorithm convergence comparison is as Figure 6 shown.

[0129] According to the analysis and comparison with the other two traditional swarm intelligence optimization algorithms, it can be concluded that the firefly algorithm has high performance in local search, with high optimization accuracy, fast optimization speed, and excellent performance in both accuracy and optimization speed.

[0130] The optimization curves of the penalty factor β and the number of modes M obtained through the FA-VMD process of the data collected by the leakage current protector are as Figure 7 and Figure 8 shown. It can be seen that the optimal parameter of M = 7 and the optimal parameter of β = 95.38 are finally determined after 10 and 2 iterations respectively.

[0131] Substitute the optimal parameters M and β into the VMD decomposition process, and the comparison diagram of each mode component of FA-VMD and the initial leakage current signal is as Figure 9 shown. In the figure, the sampling frequency is 10 times per second;

[0132] Among them, IMF1 is the leakage current signal after denoising by the FA-VMD of the present invention, which reflects the change trend of the leakage current signal in different details. It can be seen from the comparison with the initial leakage current signal that its denoising effect is obvious and it can more effectively reflect the leakage current information of the line. The remaining components IMF2-IMF7 reflect the randomness of the initial leakage current signal, and all components are stably distributed on both sides, more clearly showing the characteristics of the leakage current signal.

[0133] In summary, the 5G leakage current fault monitoring method based on improved VMD and LMS of the present invention can solve the disadvantages of poor current leakage denoising effect and low recognition accuracy at present. The leakage current signal is denoised by improving VMD and LMS. At the same time, combined with the 5G communication module and the comprehensive trend recognition method, leakage current fault warning, diagnosis and real-time communication are carried out. It can also efficiently and accurately identify the leakage current signal, providing a guarantee for reducing the occurrence of electrical fires.

[0134] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily implemented.

Claims

1. A 5G leakage current fault monitoring method based on improved VMD and LMS, characterized in that Including: S1: Obtain the load leakage current information and temperature information as the original signals; S2: Use the firefly algorithm to optimize the leakage current original signal, and obtain the optimal decomposition number M and penalty factor β of the signal VMD; S3: Use the obtained optimal M and β values to perform VMD decomposition on the leakage current original signal, and use the cross-correlation coefficient R between the reconstructed signal and the original signal obtained by the decomposition and the ratio E of the time-domain energy entropy as the classification index. Classify the modes in the time domain into three categories: effective IMF, mixed IMF, and noise IMF, and retain the effective IMF and remove the noise IMF; S4: Use the LMS algorithm to denoise the mixed IMF, and superimpose the denoised mode with the effective IMF to obtain the reconstructed leakage current signal; S5: Monitor the reconstructed leakage current signal and temperature signal, and perform fault diagnosis. When a fault occurs, the protector performs alarm and action respectively. During this period, the information is transmitted to the master station for analysis and processing through the 5G communication module in real time to avoid false alarms.

2. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 1, characterized in that, The leakage current information is the vector sum of the leakage of each phase line to the ground.

3. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 1, wherein Specifically, the firefly algorithm uses the envelope entropy value S g = f(M0, β0) as the fitness function. After the leakage current signal undergoes VMD decomposition and the firefly algorithm optimization process, the minimum value of S g is obtained, and the optimal parameters M and β of VMD are also obtained.

4. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 3, characterized in that, The process of VMD decomposition is as follows: 1) Initialize {u 1 m}, {ω 1 m}, τ 1 m and n = 0; 2) Enter the loop; 3) Introduce the quadratic penalty factor β and the Lagrange multiplier τ, and convert it into an unconstrained variational problem; Among them, u m is the M single-component amplitude-frequency modulation signals after decomposition; ω m is the center frequency of each single-component amplitude-frequency modulation signal; n is the number of iterations; τ is the Lagrange multiplier; χ is the noise tolerance parameter; t is time; j is the imaginary unit; δ(t) is the impulse function; is the partial derivative of the function with respect to t; 4) Update {u 1 m}, {ω 1 m}, and τ 1 m according to the update formula as follows: 5) Given the accuracy σ, if the stopping condition is satisfied, output M modal components; 6) Stop the loop, otherwise return to step 2).

5. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 4, characterized in that, The process of the firefly algorithm is as follows: S21: Initialize the parameters of the firefly algorithm, including the total number of firefly individuals n, the maximum number of iterations t of the population, max , the step size factor α of the perturbation when t = 1, the attractiveness ψ, and the light absorption coefficient θ; S22: Calculate the relative brightness I of the fireflies, and let the fireflies with lower brightness move towards the fireflies with higher brightness; S23: Calculate the attractiveness ψ and update the brightness of the fireflies; S24: Record the current solution and compare it with the optimal solution; S25: Whether the stopping condition is satisfied. If satisfied, exit and output the result. Otherwise, repeat S22 - S24.

6. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 1, wherein, In step S3, the calculation formula of the cross-correlation relationship R between the reconstructed signal x′(t) obtained by VMD decomposition and the leakage current original signal x(t) is as follows: The ratio E of the time-domain energy entropy k The calculation formula is as follows: Among them, E k ≥ 8% and R ≥ 0.9 are valid IMFs; E k ≥ 8% and R < 0.9 are mixed IMFs; E k < 8% are noise IMFs.

7. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 1, characterized in that, In step S4, the execution process of using the LMS algorithm to denoise the mixed IMF is as follows: S41: Filter initialization: Determine the initial weight vector w(0), initial input signal p(0), step size factor η, and filter order n; S42: For each new input sample p(n), calculate the output signal q(n); S43: Calculate the error signal e(n) using the expected output d(n) to obtain the gradient S44: Update the weight vector using the LMS update formula: S45: Return to S42 until the end, and the output sequence and error sequence can be obtained.

8. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 7, wherein, In step S5, combine the information of the reconstructed leakage current signal and the temperature information, and use the comprehensive trend recognition method to determine the adaptive slopes of the leakage current and temperature curves respectively. When the slope exceeds the predetermined value, it is determined as a sudden increase trend, and alarm and action are performed respectively, so as to finally realize the fault diagnosis and monitoring of the leakage current.

9. The 5G leakage current fault monitoring method based on improved VMD and LMS according to claim 8, wherein The specific calculation process of the fault diagnosis and monitoring of the leakage current is as follows: S51: Calculate the average current fc and average temperature ft based on the first s data uploaded in S45; S52: Determine the initial current setting value oc and temperature setting value ot, then add the next two sampling values to the sequence respectively, and delete the first two sampling values of the sequence; S53: Update oc and ot again; S54: Fit the leakage current and temperature curves and obtain the adaptive slopes Vc and Vt; when the absolute value of Vc exceeds the predetermined value, transmit a warning to the master station through 5G communication and analyze and diagnose the fault in real time. When the absolute value of Vt exceeds the predetermined value, the protection device's actuator operates to disconnect the circuit to achieve leakage current protection.

10. 5G leakage protector, characterized in that, Including the following required for implementing the 5G leakage current fault monitoring method based on improved VMD and LMS described in claim 1: A detection component for obtaining load leakage current information and temperature information; A denoising component for optimizing denoising; An amplification component for amplifying the signal; A threshold comparison component for monitoring and fault diagnosing the amplified signal; An actuator for performing warnings and actions according to the fault diagnosis results; A 5G communication component for the communication between the 5G leakage protector and the master station.

Images