Arc prediction method of low-voltage circuit and related device

Abstract

The invention provides an arc prediction method of a low-voltage circuit and a related device. The method comprises the steps that a processor receives circuit data in a target time period from a current acquisition device; determining an effective current component and a noise current component based on the current data, and determining a reconstructed signal based on the effective current component, the noise current component, and a wavelet threshold denoising operation; determining a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal; determining a feature matrix based on the plurality of time domain features and the plurality of frequency domain features, and solving based on the feature matrix to obtain a feature contribution rate; and determining an arc prediction result based on the circuit data, the characteristic contribution rate and a pre-constructed arc prediction model. Therefore, the arc prediction model has strong anti-interference capability, can obtain a good effect in a severe environment, and can excavate the characteristics of the arc fault, thereby obtaining higher accuracy, reducing algorithm input layers due to characteristic optimization, and improving the detection speed and efficiency.

Description

Technical Field

[0001] The present invention relates to the field of circuit technology, and in particular to an arc prediction method and related device for a low-voltage circuit. Background Art

[0002] With the improvement of science and technology, the types of household appliances have increased, and the demand for electricity among ordinary household users has continued to rise. Most household products need to be driven by electricity. While electricity brings many conveniences, it also brings many safety hazards to people.

[0003] Some of the current related technologies are applied to fixed-point detection scenarios such as distribution cabinets, substations and switch cabinets of medium and low voltage distribution systems, and some establish time-varying relationship equations between two or more of current, power, voltage and energy. However, there are problems such as difficulty in being widely used in various complex low-voltage power consumption scenarios, there is a gap between the data generated by the model and the actual situation, and the response is not fast enough. Moreover, they all study whether a fault arc occurs at the current moment. However, once a fault arc occurs, it may cause an electrical fire. Summary of the invention

[0004] The embodiments of the present application provide an arc prediction method and related devices for a low-voltage circuit, which can be widely used in various complex low-voltage power usage scenarios, achieve early prediction of fault arcs and rapid response, and improve the safety factor of power usage.

[0005] In a first aspect, an embodiment of the present application provides an arc prediction method for a low-voltage circuit, which is applied to a processor of a low-voltage circuit system, comprising: Receiving circuit data within a target time period from the current acquisition device, the circuit data including current data, the current data including current data of the first circuit module and / or the second circuit module, the target time period including a current time period; Determine an effective current component and a noise current component based on the current data, and determine a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after denoising; Determine a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal; Determine a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtain a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model; An arc prediction result is determined based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, and the arc prediction result includes a first prediction result. If the arc prediction result is the first prediction result, a warning message is issued; the arc prediction result is used to characterize whether an arc may be generated in the low-voltage circuit during a sequence period.

[0006] In a possible embodiment, determining the effective current component and the noise current component based on the current data includes: determining normalized current data based on the current data; Performing modal decomposition based on the normalized current data to determine a j-order intrinsic modal component; Determine a plurality of signal parameters based on the j-order intrinsic mode component and / or the normalized current data, the signal parameters including a j-order variance of the j-order intrinsic mode component and / or a variance of the current data; A variance contribution rate is determined based on the signal parameter, and the effective current component and the noise current component are determined based on the variance contribution rate.

[0007] In a possible embodiment, determining the reconstructed signal based on the effective current component, the noise current component and the wavelet threshold denoising operation includes: determining a first type of frequency component based on the noise current component, wherein the first type of frequency component includes a high frequency component; Determine a j-order modal component threshold based on the j-order eigenmodal component; and determine a j-order denoised modal component based on the first type of frequency component and the j-order modal component threshold; The reconstructed signal is determined based on the j-order denoised modal component and the effective current component.

[0008] In a possible embodiment, determining the j-order modal component threshold based on the j-order eigenmode component includes: Determining a first-order eigenmode component based on the j-order eigenmode component, wherein the first-order eigenmode component is an eigenmode component when j=1; The first-order energy signal is determined based on the first-order eigenmode component and the following formula: ; in, represents the median value, represents the absolute value, is the first-order eigenmode component, and 0.6745 is the 75% quantile of the standard normal distribution; Determining a j-order energy signal based on the 1-order energy signal and a preset first empirical parameter; The j-order modal component threshold is determined based on the j-order energy signal and a plurality of preset second empirical parameters.

[0009] In a possible embodiment, the multiple time domain features include at least one of a root mean square parameter, a variance parameter, a kurtosis factor parameter, a pulse factor parameter, and a current change rate parameter, and the multiple frequency domain features include at least one of a harmonic current content rate and a total harmonic distortion rate; the determining of a feature matrix based on the multiple time domain features and the multiple frequency domain features, and solving the feature contribution rate based on the feature matrix includes: Determine a feature matrix with n rows and m columns based on the multiple time domain features and the multiple frequency domain features, where m corresponds to the number of the time domain features and / or the frequency domain features, and n corresponds to the number of the current data; Determine a correlation coefficient matrix based on the characteristic matrix with n rows and m columns; A plurality of eigenvalues ​​are determined based on the correlation coefficient matrix, and the feature contribution rate corresponding to each of the time domain features and / or the frequency domain features is determined based on the plurality of eigenvalues.

[0010] In a possible embodiment, determining the arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model includes: Determine the principal component feature and the number of principal component features based on the feature contribution rate, wherein the number of principal component features is the number of the principal component features; Based on the number of principal component features, the pre-built arc prediction model is updated so that the pre-built arc prediction model meets the current prediction condition; An arc prediction result is determined based on the circuit data and the pre-built arc prediction model.

[0011] In a possible embodiment, the circuit data further includes circuit environment data, and the circuit environment data includes at least one of circuit temperature data, circuit pressure data, and circuit illumination data within a target period; and determining the arc prediction result based on the circuit data and the pre-built arc prediction model includes: Perform data standardization processing based on the circuit temperature data and / or the circuit pressure data and / or the circuit illumination data to determine at least one standard environment data; Determine the environment data based on the standard environment data and perform linear fitting to determine a curve and a plurality of slope parameters corresponding to at least one of the circuit environment data; determining a deviation parameter based on the plurality of slope parameters and a preset slope parameter; Determining an environmental impact factor based on the deviation parameter; An arc prediction result is determined based on the environmental influencing factors, the current data and the pre-built arc prediction model.

[0012] In a second aspect, an embodiment of the present application provides an arc prediction device for a low-voltage circuit, which is applied to a processor of a low-voltage circuit system, wherein the low-voltage circuit system includes a first circuit module and a second circuit module, wherein the first circuit module includes an actual scenario circuit, and the second circuit module includes a simulated fault circuit. The low-voltage circuit system also includes a current acquisition device, wherein the device includes: A data acquisition module, configured to receive circuit data within a target period from the current acquisition device, wherein the circuit data includes current data, and the current data includes current data of the first circuit module and / or the second circuit module. The target period includes a current period; A first determination module, configured to determine an effective current component and a noise current component based on the current data, and to determine a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after denoising; A second determination module is used to determine a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal; A third determination module is used to determine a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtain a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model; An arc prediction module is used to determine an arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, the arc prediction result including a first prediction result. If the arc prediction result is the first prediction result, a warning message is issued; the arc prediction result is used to characterize whether an arc is likely to be generated in the low-voltage circuit during a sequence period.

[0013] In a third aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes some or all of the steps described in the first aspect.

[0014] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the program includes instructions for executing some or all of the steps described in the first aspect of the embodiment of the present application.

[0015] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application. The computer program product may be a software installation package.

[0016] By implementing the embodiment of the present application, the processor receives circuit data within a target time period from the current acquisition device, the circuit data includes current data, the current data includes current data of the first circuit module and / or the second circuit module, and the target time period includes the current time period; determines an effective current component and a noise current component based on the current data, and determines a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after noise reduction; determines multiple time domain features and multiple frequency domain features based on the reconstructed signal, wherein the frequency domain features include a frequency domain for characterizing the reconstructed signal waveform distortion characteristics; determining a feature matrix based on the multiple time domain features and the multiple frequency domain features, and solving the feature matrix to obtain a feature contribution rate, the feature contribution rate characterizing the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-built arc prediction model; determining an arc prediction result based on the circuit data, the feature contribution rate and the pre-built arc prediction model, the arc prediction result including a first prediction result, if the arc prediction result is the first prediction result, an early warning message is issued; the arc prediction result is used to characterize whether the low-voltage circuit may generate an arc during the sequence period. In this way, by performing empirical mode decomposition and wavelet threshold denoising on the current data of low-voltage power consumption monitoring, the arc prediction model has a strong anti-interference ability and can achieve good results in harsh environments. The time domain and frequency domain features of the denoised current data can be used to mine the features of the arc fault, thereby obtaining a higher accuracy rate. Feature optimization reduces the input layer of the algorithm, improves the detection speed and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0018] Figure 1a is a system architecture diagram of a low-voltage circuit system provided in an embodiment of the present application; Figure 1b is a system architecture diagram of another low-voltage circuit system provided in an embodiment of the present application; Figure 1c is a scenario schematic diagram of a low-voltage circuit system provided in an embodiment of the present application; Figure 1d is a structural schematic diagram of an electronic device provided in an embodiment of the present application; Figure 2 It is a flow chart of an arc prediction method for a low-voltage circuit provided in an embodiment of the present application; Figure 3 It is a coordinate diagram of a characteristic contribution rate provided in an embodiment of the present application; Figure 4 It is a schematic diagram of a model architecture of an arc prediction model provided in an embodiment of the present application; Figure 5 It is a schematic diagram of the hit rate of an arc prediction model proposed in an embodiment of the present application; Figure 6 It is a structural schematic diagram of an arc prediction device for a low-voltage circuit provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0020] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or electronic device that includes a series of steps or units is not limited to the listed steps or units, but in an optional example also includes steps or units that are not listed, or in an optional example also includes other steps or units inherent to these processes, methods, products, or electronic devices.

[0021] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0022] With the improvement of science and technology, the types of household appliances have increased, and the demand for electricity by ordinary household users has continued to rise. Most household products need to be driven by electricity. While electricity brings many conveniences, it also brings many safety hazards to people. There are more than 100,000 electrical fires every year, and 70% of the serious and major fires are electrical fires. Among them, more than 80% of electrical fires are caused by fault arcs, which pose a huge threat to people's economic property and personal safety. Some of the current related technologies are applied to fixed-point detection scenarios such as distribution cabinets, substations and switch cabinets of medium and low voltage distribution systems, and some establish time-varying relationship equations between two or more of current, power, voltage and energy, but there are problems such as difficulty in being widely used in various complex low-voltage power consumption scenarios, there is a gap between the data generated by the model and the actual situation, and the response is not fast enough. Moreover, they all study whether a fault arc occurs at the current moment. However, once a fault arc occurs, it may cause an electrical fire.

[0023] In response to the above problems, the embodiments of the present application provide an arc prediction method and related devices for a low-voltage circuit. By performing empirical mode decomposition and wavelet threshold denoising on the current data of low-voltage power monitoring, the arc prediction model has a strong anti-interference ability and can achieve good results in harsh environments. The time domain and frequency domain features of the denoised current data can be analyzed to mine the characteristics of arc faults, thereby obtaining a higher accuracy rate. Feature optimization reduces the input layer of the algorithm, thereby improving the detection speed and efficiency.

[0024] The arc prediction method and related device of the low voltage circuit provided in the embodiment of the present application can be applied to Figure 1a or Figure 1b In the low voltage circuit system shown.

[0025] See also Figure 1a , Figure 1a is a system architecture diagram of a low voltage circuit system provided in an embodiment of the present application, such as Figure 1a As shown, the low-voltage circuit system 100 includes: a circuit device 110 and a processor 120 , wherein the circuit device 110 further includes a circuit module 111 and a current acquisition device 112 .

[0026] Among them, the circuit device 110 is a low-voltage power usage scenario, for example, a circuit in a certain power usage area, the circuit module 111 included in the circuit device 110 is the actual circuit part, and the current acquisition device 112 includes a device for current acquisition, which can be set in a fixed position in the circuit or the circuit lead. The processor 120 is deployed with an arc prediction model for analyzing the arc generation situation in the circuit module 111. The processor 120 can also be used to collect data during the use of the model to facilitate the subsequent optimization of the arc prediction model.

[0027] The current acquisition device 112 may be a current sensor device, or a device with a current acquisition function, which is not limited here.

[0028] See also Figure 1b , Figure 1b is a system architecture diagram of another low-voltage circuit system provided in an embodiment of the present application, such as Figure 1b As shown, the low voltage circuit system 100 includes: a circuit device 110 and a processor 120 , wherein the circuit device 110 further includes a circuit module 111 and a current acquisition device 112 , wherein the circuit module 111 includes a first circuit module 1111 and a second circuit module 1112 .

[0029] Among them, the first circuit module is an actual scene circuit, and the second circuit module is a simulated fault circuit. Various low-voltage loads are provided in the first circuit module for power supply; the second circuit module is provided with a fault arc simulation device for generating fault data corresponding to the above low-voltage loads, which can simulate the fault arc situation. The high-frequency current signals generated in both cases are collected by the above current acquisition device 112 and transmitted to the above processor 120.

[0030] For examples, see Figure 1c , Figure 1c is a schematic diagram of a low-voltage circuit system provided in an embodiment of the present application, such as Figure 1c As shown, in the low-voltage circuit system, the circuit module may include various low-voltage loads such as fans, hair dryers, computers, etc., and the left side is a low-voltage power supply system. In one part, that is, in the first circuit module, the 220V low-voltage power supply system supplies power to various low-voltage loads such as fans, hair dryers, computers, etc. through switch A1; in the other part, that is, in the second circuit module, the 220V low-voltage power supply system is connected to the fault arc simulation device through switch A2, which can simulate the fault arc situation. The high-frequency current signals generated in both cases are collected by a high-frequency current sensor, that is, a current acquisition device. Then the current acquisition device can send the acquired circuit data to the processor, and the processor analyzes the current data to obtain the arc prediction result.

[0031] See also Figure 1d , Figure 1dIt is a structural diagram of an electronic device provided in an embodiment of the present application, and the electronic device 200 includes a processor 120, a memory 130, a communication interface 140, and one or more programs 131, wherein the one or more programs 131 are stored in the above-mentioned memory 130 and are configured to be executed by the above-mentioned processor 120, and the one or more programs 131 include instructions for executing any step in the following method embodiment. The processor 120, the memory 130, and the communication interface 140 are interconnected and complete the communication work between each other; wherein the memory 130 can be a volatile memory such as a dynamic random access memory DRAM, or a non-volatile memory such as a mechanical hard disk. In a specific implementation, the processor 120 is used to execute any step performed by the infusion device in the following method embodiment, and when performing data transmission such as sending, the communication interface 140 can be selectively called to complete the corresponding operation.

[0032] See also Figure 2 , Figure 2 It is a flow chart of an arc prediction method for a low-voltage circuit provided by an embodiment of the present application. The method is applied to a processor of a low-voltage circuit system. The low-voltage circuit system includes a first circuit module and a second circuit module. The first circuit module includes an actual scenario circuit, and the second circuit module includes a simulated fault circuit. The low-voltage circuit system also includes a current acquisition device. The method includes: S210, receiving circuit data within a target time period from the current acquisition device, the circuit data including current data, the current data including current data of the first circuit module and / or the second circuit module, and the target time period including a current time period.

[0033] Among them, the current data includes the current data of the first circuit module and / or the second circuit module. The current data generated by the first circuit module is the actual current data generated by the actual application scenario, and the current data generated by the second circuit module is the simulated current data generated by the fault simulation device. In actual applications, it is possible to obtain the current data of the first circuit module, or to obtain the current data of the first circuit module and the second circuit module. The target time period can be the current time period, the time period length can be a time period length, and the specific period length can be preset, such as 1 second, 5 seconds, 1 millisecond, etc., which is not limited here.

[0034] S220, determining an effective current component and a noise current component based on the current data, and determining a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after noise reduction.

[0035] Among them, the effective current component and the noise current component are distinguished based on the current data. Specifically, data feature analysis can be performed first through one or more of time domain analysis, frequency domain analysis, and statistical characteristic analysis. Among them, time domain analysis can be to analyze the amplitude, pulse width, periodicity, etc. of the current waveform, and preliminarily distinguish steady-state signals, such as normal load current, and transient noise, such as spikes and glitches. Frequency domain analysis can be to convert the signal to the frequency domain through fast Fourier transform (FFT), draw a spectrum diagram, and identify the frequency band where the noise is mainly distributed. Statistical characteristics can be to calculate statistics such as mean, variance, kurtosis, etc., to determine whether the noise is Gaussian distribution or pulse noise. Then filtering is performed to distinguish the current data, mainly filtering the noise components to obtain the effective current component with less noise; specifically, wavelet filtering can be used for filtering, or other methods can be used, which are not limited here.

[0036] Then, the reconstructed signal is obtained by integrating the current component after filtering and removing part of the noise.

[0037] In a possible embodiment, determining the effective current component and the noise current component based on the current data includes: determining normalized current data based on the current data; performing modal decomposition based on the normalized current data to determine the j-order intrinsic modal component; determining multiple signal parameters based on the j-order intrinsic modal component and / or the normalized current data, the signal parameters including the j-order variance of the j-order intrinsic modal component and / or the variance of the current data; determining the variance contribution rate based on the signal parameters, and determining the effective current component and the noise current component based on the variance contribution rate.

[0038] Among them, since each low-voltage power current data collected (t=1, 2, ..., T), the size varies greatly under different loads. In order to eliminate the influence brought by it, the data is normalized first. Specifically, it can be normalized according to the following first formula: ; Among them, the above-mentioned That is, the current data Normalized current data obtained after normalization; for The minimum value, for The maximum value.

[0039] Among them, since the collected low-voltage power current data usually includes various noises, the existence of noise is not conducive to the accurate detection of fault arcs, so the current signal is first pre-processed, specifically, the current signal can be subjected to noise reduction processing. The specific process of noise reduction processing may include: first, the current data Perform empirical mode decomposition and transform the current data Decompose into h j-order eigenmode components arranged from high frequency to low frequency and a trend component Specifically, the empirical mode decomposition can be performed according to the following second formula: .

[0040] Among them, multiple signal parameters include j-order eigenmode component At least one of the j-order variance of the current data x(t) and the variance of the current data x(t), in some possible implementations, the j-order intrinsic mode component is determined based on the following third formula The j-order variance of the current data is determined by the fourth formula as follows The variance of The third formula: ; The fourth formula: ; Wherein, T is the length of the collected signal, that is, the target period.

[0041] Among them, the present invention uses the variance contribution rate to describe the contribution of each modal component to the fluctuation of the original current data, so as to characterize the influence of different periodic components on the original data. Therefore, the present invention uses the variance contribution rate as a screening condition. The variance contribution rate (VCR) of a component refers to the proportion of the variance that can be explained by a component to the total variance. The larger the value, the stronger the ability of the component to integrate the original current signal information. Specifically, the variance contribution rate is determined according to the following fifth formula: ; in, Represents the variance contribution rate of the j-th order modal component, j=1, 2,…, h.

[0042] Among them, when the variance contribution rate is not greater than the preset contribution rate threshold, the modal component can be considered to be the noise-dominated component. The preset contribution rate threshold can be 0.02. The preset contribution rate threshold can also be other values, which are not limited here. Then the qth-order modal component is calculated to be the boundary between the noise component and the effective component, that is, of If the qth order , at the q-1th order , then the qth to hth orders are effective current components, and the first q-1 orders are noise current components.

[0043] It can be seen that in this embodiment, through normalization, modal decomposition and signal parameter calculation, key features are gradually extracted from the original current data, avoiding the limitations of traditional methods for non-stationary signals. Quantitative analysis based on variance contribution rate can objectively distinguish effective current components and noise components, reducing reliance on human experience or subjective judgment. This method has strong robustness to nonlinear and non-stationary current signals and is suitable for a variety of low-voltage power usage scenarios.

[0044] In a possible embodiment, the reconstructed signal is determined based on the effective current component, the noise current component and the wavelet threshold denoising operation, including: determining a first type of frequency component based on the noise current component, the first type of frequency component including a high-frequency component; determining a j-order modal component threshold based on the j-order intrinsic modal component; and determining a j-order denoised modal component based on the first type of frequency component and the j-order modal component threshold; and determining the reconstructed signal based on the j-order denoised modal component and the effective current component.

[0045] Wherein, based on the preset base wavelet and the number of decomposition layers, the noise current component is decomposed to obtain high-frequency coefficients and low-frequency coefficients.

[0046] Specifically, the base wavelet is the core function of the wavelet transform, such as Daubechies wavelet, Symlets wavelet, etc., which are not limited here. Its characteristics, such as time-frequency localization ability, orthogonality, etc., will affect the effect of signal decomposition. The appropriate wavelet can be selected according to the signal characteristics (such as nonlinearity, non-stationarity) and analysis objectives (such as denoising, feature extraction). The number of decomposition layers determines the scale range of the signal to be decomposed. Each layer of decomposition will divide the signal into a low-frequency approximate component and a high-frequency detail component. Specifically, the number of decomposition layers is selected based on the main frequency range of the signal to be covered. Then, based on the number of decomposition layers, the noise current component is decomposed to obtain the first type of frequency component, that is, the high-frequency component, and the second type of frequency component, that is, the low-frequency component, can also be obtained.

[0047] Among them, soft threshold denoising is performed based on the following sixth formula, the noise component is partially filtered to obtain the j-order denoised modal component. Sixth formula: ; in, is the j-th order modal component after denoising, is the symbolic function, is the jth order first frequency component after wavelet decomposition, is the threshold corresponding to the j-th order modal component, i.e., the j-th order modal component threshold. For coefficients with amplitudes greater than the j-th order modal component threshold, the threshold is subtracted (the useful signal is retained but its amplitude is reduced). For coefficients with amplitudes less than the j-th order modal component threshold, they are directly set to zero (noise is removed).

[0048] The j-order modal component threshold may be determined by parameters such as signal length and noise standard deviation, or by parameters such as signal length and energy corresponding to the modal component.

[0049] Among them, in order to obtain the j-order denoised modal component After that, the j-order denoised modal component and the effective current component are integrated to obtain a reconstructed signal containing an effective signal. Specifically, the signal is reconstructed by the following seventh formula: ; in, is the current signal after noise reduction, is the j-th order modal component after wavelet denoising, is the effective amount to be retained, is the residual component.

[0050] It can be seen that in this embodiment, by accurately identifying the noise frequency range through wavelet decomposition, it can be determined that the noise is mainly distributed in the high-frequency area, so that the high-frequency component can be classified as a "first-order frequency component", and then the j-order modal component threshold is dynamically calculated according to the parameters related to the j-th order intrinsic modal component, so that the threshold is matched with the signal characteristics, so that the noise decomposition through the threshold is more accurate and meets the usage scenario, and the noise is processed separately at different scales (frequencies), thereby improving the denoising accuracy and avoiding the loss of high-frequency features caused by traditional low-pass filtering.

[0051] In a possible embodiment, the reconstructed signal is determined based on the effective current component, the noise current component and the wavelet threshold denoising operation, including: determining the j-order modal component threshold based on the j-order intrinsic modal component; and determining the j-order denoised modal component based on the j-order intrinsic modal component and the j-order modal component threshold; and determining the reconstructed signal based on the j-order denoised modal component and the effective current component.

[0052] Among them, based on the j-order intrinsic mode component, a soft threshold is used for denoising, and the soft threshold function is the following eighth formula: ; in, is the j-th order modal component after denoising, is the symbolic function, is the j-th order modal noise component before denoising. is the threshold corresponding to the j-th order modal component. The j-th order modal component threshold can be determined by parameters such as signal length and noise standard deviation, or by parameters such as signal length and energy corresponding to the modal component.

[0053] Among them, in order to obtain the j-order denoised modal component After that, the j-order denoised modal component and the effective current component are integrated to obtain a reconstructed signal containing an effective signal. Specifically, the signal is reconstructed by the following seventh formula: ; in, is the current signal after noise reduction, is the j-th order modal component after wavelet denoising, is the effective amount to be retained, is the residual component.

[0054] It can be seen that in this embodiment, the j-order modal component threshold is dynamically calculated through parameters related to the j-order intrinsic modal component, so that the threshold is matched with the signal characteristics, so that the noise decomposition through the threshold is more accurate and in line with the usage scenario, and the noise is processed separately at different scales (frequencies), thereby improving the denoising accuracy and avoiding the loss of high-frequency features caused by traditional low-pass filtering.

[0055] In a possible embodiment, the j-order modal component threshold is determined by the following steps: determining the first-order eigenmodal component based on the j-order eigenmodal component, the first-order eigenmodal component being the eigenmodal component when j=1; determining the first-order energy signal based on the first-order eigenmodal component and the following ninth formula: ; in, represents the median value, represents the absolute value, is the first-order eigenmodal component, and 0.6745 is the 75% quantile of the standard normal distribution; the j-order energy signal is determined based on the first-order energy signal and a preset first empirical parameter; the j-order modal component threshold is determined based on the j-order energy signal and multiple preset second empirical parameters.

[0056] Wherein, determining the j-order modal component threshold based on the j-order energy signal and a plurality of preset second empirical parameters specifically includes the following steps: determining the j-order energy signal based on the 1st-order energy signal and the preset first empirical parameter, specifically according to the following tenth formula: ; in, and is the first empirical parameter, is a constant, and is preset as 0.719, 2.01.

[0057] Wherein, determining the j-order modal component threshold based on the j-order energy signal and a plurality of preset second empirical parameters specifically comprises the following steps: Based on the j-th order energy signal and signal length and the preset second empirical parameter, it is determined according to the following eleventh formula: ; Wherein, D is the second empirical parameter, which is a constant and is generally taken as 0.3, and T is the signal length.

[0058] It should be noted that the above-mentioned first empirical parameter and second empirical parameter are variable and can be set based on actual application scenarios, and are not limited here.

[0059] It can be seen that in this embodiment, the signal statistical characteristics and engineering experience are combined to achieve the dual goals of efficient denoising and feature retention. The first empirical parameter and the second empirical parameter based on experience are introduced when calculating the threshold, so that the threshold is more in line with the actual scenario.

[0060] S230, determining a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal.

[0061] Among them, the current waveform changes greatly during normal operation of low-voltage electricity and when a fault arc occurs, and the time domain and frequency domain characteristics of the current change significantly before and after the arc fault occurs. Therefore, the present invention extracts the time domain and frequency domain characteristics of the current signal respectively.

[0062] In a possible embodiment, the multiple time domain features include at least one of a root mean square parameter, a variance parameter, a kurtosis factor parameter, a pulse factor parameter, and a current change rate parameter, and the multiple frequency domain features include at least one of a harmonic current content rate and a total harmonic distortion rate.

[0063] Among them, the time domain features can be reconstructed according to the seventh formula to obtain the signal Specifically, the root mean square parameter can be calculated according to the following twelfth formula: ; The variance parameter can be calculated according to the following thirteenth formula: ; The kurtosis factor parameter can be calculated according to the following fourteenth formula: ; The pulse factor parameter can be calculated according to the following fifteenth formula: The current change rate can be calculated according to the following sixteenth formula: . Where n represents the nth sample.

[0064] The frequency domain characteristics include at least one of a harmonic current content rate and a current total harmonic distortion rate.

[0065] Specifically, since the amplitude of the odd harmonics of the current increases significantly when a low-voltage arc fault occurs, and the increase in the amplitude of the odd harmonics is not affected by the load type, some harmonic current content can be selected as one of the frequency domain characteristic indicators for detecting arc faults. Specifically, the first harmonic current content rate and the third harmonic current content rate can be selected: Based on the signal obtained after reconstruction The first harmonic current content rate is calculated using the following seventeenth formula: ; Based on the reconstructed signal The third harmonic current content rate is calculated using the following eighteenth formula: ; in, and are the effective values ​​of the 1st and 3rd harmonic currents, is the effective value of the fundamental current.

[0066] Specifically, since the current waveform is distorted when an arc fault occurs, the total harmonic distortion rate of the current can be selected as one of the frequency domain characteristic indicators of the arc current. Specifically, based on the signal obtained after reconstruction The total harmonic distortion rate of current is calculated using the following formula: ; in, is the effective value of the pth harmonic current, is the effective value of the fundamental current, L is the specified maximum harmonic order. In actual use, L can be 30, and in some cases it can also be other values, which are not limited here.

[0067] It can be seen that in this embodiment, the time domain and frequency domain features are combined to cover the multi-dimensional information of the signal. The time domain features are sensitive to faults, and the frequency domain features are suitable for harmonic analysis. The overall method has high practicality and reliability and is suitable for the diversified needs of power system monitoring.

[0068] S240, determining a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtaining a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model.

[0069] The extracted multiple time domain features and multiple frequency domain features are combined into a matrix, which is the feature matrix. Assuming that a time domain features and b frequency domain features are extracted, the size of the feature matrix is ​​(a+b)×n, where n is the number of samples, that is, the number of current data acquired by the current acquisition device and sent to the processor for analysis.

[0070] Among them, the feature contribution rate based on the feature matrix can be determined by principal component analysis. Through linear transformation, the original feature space is mapped to a new coordinate system, so that the coordinate axes (i.e., principal components) in the new coordinate system are arranged in sequence according to the magnitude of the data change. In the new coordinate system, the first principal component can explain the largest variance in the data, the second principal component explains the largest part of the remaining variance, and so on. In some cases, the feature contribution rate can also be determined by various methods such as mutual information, feature importance analysis, regularization analysis, variance analysis, neural network analysis, etc., which are not limited here.

[0071] In a possible embodiment, the multiple time domain features include at least one of a root mean square parameter, a variance parameter, a kurtosis factor parameter, a pulse factor parameter, and a current change rate parameter, and the multiple frequency domain features include at least one of a harmonic current content rate and a total harmonic distortion rate of the flow; determining a feature matrix based on the multiple time domain features and the multiple frequency domain features, and solving to obtain a feature contribution rate based on the feature matrix includes: determining a feature matrix of n rows and m columns based on the multiple time domain features and the multiple frequency domain features, wherein m corresponds to the number of the time domain features and / or the frequency domain features, and n corresponds to the number of the current data; determining a correlation coefficient matrix based on the feature matrix of n rows and m columns; determining multiple eigenvalues ​​based on the correlation coefficient matrix, and determining the feature contribution rate corresponding to each of the time domain features and / or the frequency domain features based on the multiple eigenvalues.

[0072] Among them, the multiple time domain features can be a, the multiple frequency domain features can be b, the above m=a+b, based on the a time domain features and b frequency domain features corresponding to each sample to form an n×m feature matrix, each row represents the m features corresponding to a sample, specifically including the values ​​of a time domain features and b frequency domain features, there are a total of n rows, in this embodiment m=a+b=8. Specifically, the n×m feature matrix can be determined according to the twentieth formula: .

[0073] Among them, the 20th formula calculates the correlation coefficient corresponding to the feature matrix. Specifically, the correlation coefficient can be calculated according to the following 21st formula: : ; Then the correlation coefficient calculated based on the above formula 21 is Establish the correlation coefficient matrix R, see the following formula 22: ; Among them, the above , Represents time domain features or frequency domain features , The mean of .

[0074] Among them, based on the above correlation coefficient matrix R through Solve the characteristic equation, where I is the unit matrix of the same dimension as the correlation coefficient matrix R, which is 8×8 here. is the eigenvalue, and finally at least one Value, that is , here can be , , , , , , , ; Then sort the eigenvalues ​​so that ; Based on the solution To calculate the characteristic contribution rate, it can be calculated by the following formula 23: ; Among them, we get To correspond to , corresponding to The characteristic contribution rate.

[0075] It can be seen that in this embodiment, through eigenvalue decomposition and contribution rate calculation, the importance of each feature can be quantified, which is suitable for multi-feature analysis and feature screening scenarios, thereby obtaining a higher accuracy rate. Feature optimization reduces the input layer of the algorithm and improves the detection speed and efficiency.

[0076] S250, determining an arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, the arc prediction result including a first prediction result, and issuing a warning message if the arc prediction result is the first prediction result; the arc prediction result is used to characterize whether an arc may be generated in the low-voltage circuit during a sequence period.

[0077] Among them, based on the feature contribution rate, at least one feature with a larger contribution rate can be determined, and then the arc prediction model is optimized and constructed based on the number of features with a larger contribution rate, so that the dimension of the arc prediction model is more streamlined under the current usage. After optimizing the pre-built arc prediction model, the arc generation situation is analyzed based on the current data of the first circuit module to obtain a first prediction result, which characterizes that the current circuit may generate an arc. In some cases, a second prediction result can also be obtained, and the second prediction result characterizes that the current circuit is unlikely to generate an arc. When the first prediction result is obtained, a warning message is issued; if the second prediction result is obtained, no information may be sent or information indicating that no arc is generated may be sent.

[0078] In a possible embodiment, determining the arc prediction result based on the circuit data, the feature contribution rate and the pre-built arc prediction model includes: determining the principal component features and the number of principal component features based on the feature contribution rate, wherein the number of principal component features is the number of principal component features; updating the pre-built arc prediction model based on the number of principal component features so that the pre-built arc prediction model meets the current prediction conditions; and determining the arc prediction result based on the circuit data and the pre-built arc prediction model.

[0079] Among them, a cumulative contribution rate threshold (such as 90% or 95%) is set, and the first k features whose cumulative contribution rate reaches the threshold are selected as principal component features. If the cumulative contribution rate of the first 3 features is 0.3+0.25+0.2=0.75 (75%), which does not reach 90%, continue to add features; if the cumulative contribution rate of the first 4 features is 0.75+0.15=0.9 (90%), then the first 4 features are selected as principal component features. The number of principal component features (k) is the number of selected features. The principal component corresponding to the contribution rate exceeding the cumulative contribution rate threshold can effectively cover the main information of a time domain features and b frequency domain features. In this embodiment, the cumulative contribution rate threshold can be set to 85%.

[0080] For examples, see Figure 3 , Figure 3 is a coordinate diagram of a characteristic contribution rate provided in an embodiment of the present application. In this embodiment, if , , , , , , , Since the preset cumulative contribution rate threshold is 85%, the cumulative contribution rate of the first three principal components is more than 90%, so that the use of the first three principal components can cover the main information of the selected 8 time domain and frequency domain characteristic parameters.

[0081] Among them, please refer to Figure 4 , Figure 4 is a schematic diagram of a model architecture of an arc prediction model provided in an embodiment of the present application, such as Figure 4 As shown, the above-mentioned pre-built arc prediction model can be a multi-layer neural network model, specifically, it can be a three-layer BP neural network model, which can include an input layer, a hidden layer and an output layer, wherein the number of neurons in the input layer is multiple. In this embodiment, the number of neurons in the input layer can be determined by the number of principal component features determined above. Specifically, the number of neurons in the input layer can be equal to the number of principal component features; the number of neurons in the output layer can also be multiple, specifically, it is related to the situation of the output result. For example, the output arc prediction result includes two situations, the first output result and the second output result, then the number of neurons in the output layer is 2. If the output arc prediction result has multiple situations, the number of neurons in the output layer can also be updated to the number of types of the above arc prediction results. The randomly initialized weights from the input layer to the hidden layer of the arc prediction model can be And the weight from hidden layer to output layer can be . Initialize bias and . Then from the input layer to the hidden layer: , where f is the ReLU activation function. Hidden layer to output layer: .

[0082] The number of neurons in the hidden layer is also related to the number of principal component features. Specifically, it can be determined by the twenty-fourth formula: ; in, is the number of neurons in the hidden layer, p is, is a number between 1 and 10; It is an empirical formula used to estimate the benchmark value of the number of hidden layer neurons. Used for fine-tuning to avoid insufficient model complexity or overfitting, and to select the best For example, grid search is performed in the range of 1 to 10. If the model is underfitting, increase ; If overfitting, you can reduce .

[0083] The output result of the model, i.e., the arc prediction result, includes a first prediction result and a second prediction result, wherein the first prediction result indicates that an arc may be generated, and the second prediction result indicates that an arc may not be generated. Specifically, the first prediction result can be represented by 1, and the second prediction result can be represented by 0.

[0084] It should be noted that this model will calculate the hit rate every time it runs for a period of time. The hit rate is calculated according to the following formula: ; in is the number of low-voltage arcs that are accurately identified, and B is the total number of system model identifications. Figure 5 , Figure 5 is a schematic diagram of the hit rate of an arc prediction model provided in an embodiment of the present application, such as Figure 5 As shown in the figure, in the results output by the model, "0" indicates fault, i.e. the first prediction result, "1" indicates normal, i.e. the second prediction result, the value of the upper left cell is 103, which indicates the number of samples that are actually normal and correctly predicted to be normal, the value of the upper right cell is 2, which indicates the number of samples that are actually normal but incorrectly predicted to be faulty, the value of the lower left cell is 0, i.e. the number of samples that are actually faulty but incorrectly predicted to be normal, and the value of the lower right cell is 95, which indicates the number of samples that are actually faulty and correctly predicted to be faulty. The color depth represents the value size, and the color scale on the right indicates that the darker the color, the larger the value, which can intuitively reflect the value of each cell. It can be seen from the matrix that the overall prediction effect of the arc prediction model is good, and there are fewer misjudgments.

[0085] It can be seen that in this embodiment, through reasonable structural design, feature optimization, and dynamic adjustment of the number of hidden layer neurons, the number of layers of the model is streamlined and redundant parts are eliminated, the computational complexity is low, the amount of calculation is reduced, and it is suitable for real-time monitoring tasks, making the arc more efficient, accurate, and flexible in model analysis, thereby improving analysis efficiency and robustness.

[0086] In a possible embodiment, the circuit data also includes circuit environment data, and the circuit environment data includes at least one of circuit temperature data, circuit pressure data, and circuit illumination data within a target time period; the method of determining the arc prediction result based on the circuit data and the pre-constructed arc prediction model includes: performing data standardization processing based on the circuit temperature data and / or circuit pressure data and / or circuit illumination data to determine at least one standard environment data; performing linear fitting on the environment data determined based on the standard environment data to determine a curve and multiple slope parameters corresponding to at least one of the circuit environment data; determining a deviation parameter based on the multiple slope parameters and a preset slope parameter; determining an environmental impact factor based on the deviation parameter; and determining the arc prediction result based on the environmental impact factor, the current data, and the pre-constructed arc prediction model.

[0087] The circuit environment data includes the circuit temperature data, circuit pressure data, and circuit light data within the target period. Assume that the environment data is T (temperature), P (pressure), and L (light). Each type of environment data is normalized, and the formula is: ; Where X is the original environment data, and are the minimum and maximum values ​​of the data respectively. The standardized data is recorded as , , The standardized environmental data is used as a new input feature and together with the current data, it forms an input vector: , where x is the current data.

[0088] Among them, a linear fit is performed for each environmental data (temperature, pressure, light), and the fitting formula can be: y=kt+b, or other curve formulas. Among them, y is environmental data (such as temperature T); t is time or sampling point; k is the slope parameter, which indicates the trend of environmental data; b is the intercept. The slope parameter k of each environmental data is calculated by the least squares method or other fitting methods. T , k P , k L Then, the deviation parameter is determined based on a preset normal slope parameter, wherein the preset normal slope parameter may be a parameter range, for example, a normal temperature change slope , Normal pressure change slope , Normal illumination change slope Then calculate the deviation δ between the actual slope parameter and the preset slope parameter T ,δ P ,δ L , which can be determined based on the following formula: , , ; Then, the total deviation parameter is determined based on the above deviation parameter. Specifically, the environmental impact factor can be determined based on the deviation size of each deviation to characterize the degree of environmental impact. Specifically, each deviation can be assigned a different weight coefficient to obtain the final environmental impact factor G(t). The environmental impact factor G(t) is used as a new input feature and together with the current data, it constitutes an extended input vector: The number of neurons in the original input layer is the number of principal component features, and the number of neurons in the expanded input layer is the number of principal component features + 1 (adding environmental influence factors).

[0089] The arc prediction result includes a first output result and a second output result. In addition to outputting 1, the first output result also outputs the current environmental impact factor, which is used to characterize the possibility or severity of the arc.

[0090] It can be seen that in this embodiment, by integrating circuit environment data (temperature, pressure, light) into the neural network model, and combining linear fitting, deviation calculation, environmental influencing factors and other steps, an arc prediction model with multi-dimensional feature fusion is realized. This method not only improves the accuracy and robustness of the model, but also enhances the model's adaptability to complex environments, and has high engineering practical value.

[0091] It can be seen that through the embodiment of the present application, the processor receives circuit data within a target time period from the current acquisition device, the circuit data includes current data, the current data includes current data of the first circuit module and / or the second circuit module, and the target time period includes the current time period; determines an effective current component and a noise current component based on the current data, and determines a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after noise reduction; determines multiple time domain features and multiple frequency domain features based on the reconstructed signal, and the frequency domain features include a frequency domain for characterizing the reconstructed signal waveform distortion characteristics; determining a feature matrix based on the multiple time domain features and the multiple frequency domain features, and solving the feature matrix to obtain a feature contribution rate, the feature contribution rate characterizing the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-built arc prediction model; determining an arc prediction result based on the circuit data, the feature contribution rate and the pre-built arc prediction model, the arc prediction result including a first prediction result, if the arc prediction result is the first prediction result, an early warning message is issued; the arc prediction result is used to characterize whether the low-voltage circuit may generate an arc during the sequence period. In this way, by performing empirical mode decomposition and wavelet threshold denoising on the current data of low-voltage power consumption monitoring, the arc prediction model has a strong anti-interference ability and can achieve good results in harsh environments. The time domain and frequency domain features of the denoised current data can be used to mine the features of the arc fault, thereby obtaining a higher accuracy rate. Feature optimization reduces the input layer of the algorithm, improves the detection speed and efficiency.

[0092] See also Figure 6 , Figure 61 is a schematic diagram of the structure of an arc prediction device for a low-voltage circuit proposed in an embodiment of the present application, and is applied to a processor of a low-voltage circuit system. The low-voltage circuit system includes a first circuit module and a second circuit module. The first circuit module includes an actual scenario circuit, and the second circuit module includes a simulated fault circuit. The low-voltage circuit system also includes a current acquisition device. The arc prediction device 600 for a low-voltage circuit includes: A data acquisition module 610 is used to receive circuit data within a target period from the current acquisition device, wherein the circuit data includes current data, and the current data includes current data of the first circuit module and / or the second circuit module. The target period includes a current period; A first determination module 620, configured to determine an effective current component and a noise current component based on the current data, and determine a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after denoising; A second determination module 630, configured to determine a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal; A third determination module 640 is used to determine a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtain a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model; The arc prediction module 650 is used to determine the arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, and the arc prediction result includes a first prediction result. If the arc prediction result is the first prediction result, a warning message is issued; the arc prediction result is used to characterize whether the low-voltage circuit is likely to generate an arc during the sequence period.

[0093] In a possible embodiment, the first determination module 620, in determining the effective current component and the noise current component based on the current data, is specifically configured to: determining normalized current data based on the current data; Performing modal decomposition based on the normalized current data to determine a j-order intrinsic modal component; Determine a plurality of signal parameters based on the j-order intrinsic mode component and / or the normalized current data, the signal parameters including a j-order variance of the j-order intrinsic mode component and / or a variance of the current data; A variance contribution rate is determined based on the signal parameter, and the effective current component and the noise current component are determined based on the variance contribution rate.

[0094] In a possible embodiment, the second determination module 630, in determining the reconstructed signal based on the effective current component, the noise current component and the wavelet threshold denoising operation, is specifically configured to: determining a first type of frequency component based on the noise current component, wherein the first type of frequency component includes a high frequency component; Determine a j-order modal component threshold based on the j-order eigenmodal component; and determine a j-order denoised modal component based on the first type of frequency component and the j-order modal component threshold; The reconstructed signal is determined based on the j-order denoised modal component and the effective current component.

[0095] In a possible embodiment, the second determination module 630, in determining the j-order modal component threshold based on the j-order eigenmode component, is specifically configured to: Determining a first-order eigenmode component based on the j-order eigenmode component, wherein the first-order eigenmode component is an eigenmode component when j=1; The first-order energy signal is determined based on the first-order eigenmode component and the following formula: ; in, represents the median value, represents the absolute value, is the first-order eigenmode component, and 0.6745 is the 75% quantile of the standard normal distribution; Determining a j-order energy signal based on the 1-order energy signal and a preset first empirical parameter; The j-order modal component threshold is determined based on the j-order energy signal and a plurality of preset second empirical parameters.

[0096] In a possible embodiment, the multiple time domain features include at least one of a root mean square parameter, a variance parameter, a kurtosis factor parameter, a pulse factor parameter, and a current change rate parameter, and the multiple frequency domain features include at least one of a harmonic current content rate and a total harmonic distortion rate; the third determination module 640 determines a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtains a feature contribution rate based on the feature matrix, specifically for: Determine a feature matrix with n rows and m columns based on the multiple time domain features and the multiple frequency domain features, where m corresponds to the number of the time domain features and / or the frequency domain features, and n corresponds to the number of the current data; Determine a correlation coefficient matrix based on the characteristic matrix with n rows and m columns; A plurality of eigenvalues ​​are determined based on the correlation coefficient matrix, and the feature contribution rate corresponding to each of the time domain features and / or the frequency domain features is determined based on the plurality of eigenvalues.

[0097] In a possible embodiment, the arc prediction module 650 is specifically used to determine the arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model: Determine the principal component feature and the number of principal component features based on the feature contribution rate, wherein the number of principal component features is the number of the principal component features; Based on the number of principal component features, the pre-built arc prediction model is updated so that the pre-built arc prediction model meets the current prediction condition; An arc prediction result is determined based on the circuit data and the pre-built arc prediction model.

[0098] In a possible embodiment, the circuit data further includes circuit environment data, and the circuit environment data includes at least one of circuit temperature data, circuit pressure data, and circuit illumination data within a target period; the arc prediction module 650 is specifically used to determine the arc prediction result based on the circuit data, the feature contribution rate, and the pre-built arc prediction model: Perform data standardization processing based on the circuit temperature data and / or the circuit pressure data and / or the circuit illumination data to determine at least one standard environment data; Determine the environment data based on the standard environment data and perform linear fitting to determine a curve and a plurality of slope parameters corresponding to at least one of the circuit environment data; determining a deviation parameter based on the plurality of slope parameters and a preset slope parameter; Determining an environmental impact factor based on the deviation parameter; An arc prediction result is determined based on the environmental influencing factors, the current data and the pre-built arc prediction model.

[0099] It is worth noting that the specific functional implementation of the arc prediction device 600 for the low voltage circuit can be found in the above Figure 2 The description of the arc prediction method for the low-voltage circuit shown in the figure, for example, the data acquisition module 610 is used to implement the relevant content of executing S210. The various units or modules in the arc prediction device 600 for the low-voltage circuit can be respectively or completely combined into one or several other units or modules to constitute, or one (some) of the units or modules can also be divided into multiple functionally smaller units or modules to constitute, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present invention. The above-mentioned units or modules are divided based on logical functions. In actual applications, the functions of one unit (or module) are implemented by multiple units (or modules), or the functions of multiple units (or modules) are implemented by one unit (or module).

[0100] It can be seen that the arc prediction device for a low-voltage circuit described in the embodiment of the present application receives circuit data within a target time period from the current acquisition device, the circuit data includes current data, and the current data includes current data of the first circuit module and / or the second circuit module, and the target time period includes the current time period; determines an effective current component and a noise current component based on the current data, and determines a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after noise reduction; determines multiple time domain features and multiple frequency domain features based on the reconstructed signal, and the frequency domain features include features for characterizing the Reconstruct the waveform distortion characteristics of the frequency domain of the signal; determine the characteristic matrix based on the multiple time domain characteristics and the multiple frequency domain characteristics, and solve the characteristic contribution rate based on the characteristic matrix, the characteristic contribution rate characterizes the importance of each time domain feature or the frequency domain feature, so as to constrain the characteristic dimension of the pre-constructed arc prediction model; determine the arc prediction result based on the circuit data, the characteristic contribution rate and the pre-constructed arc prediction model, the arc prediction result includes the first prediction result, if the arc prediction result is the first prediction result, then issue a warning message; the arc prediction result is used to characterize whether the low-voltage circuit may generate an arc during the sequence period. In this way, by performing empirical mode decomposition and wavelet threshold denoising on the current data of low-voltage power consumption monitoring, the arc prediction model has a strong anti-interference ability and can achieve good results in harsh environments. The time domain and frequency domain characteristics of the denoised current data can be used to mine the characteristics of the arc fault, thereby obtaining a higher accuracy rate. The feature optimization reduces the input layer of the algorithm, improves the detection speed and efficiency.

[0101] An embodiment of the present application also provides a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute part or all of the steps of any method described in the above method embodiments, and the above computer includes an electronic device.

[0102] The embodiment of the present application also provides a computer program product, the computer program product includes a non-transitory computer-readable storage medium storing a computer program, the computer program is operable to cause a computer to execute some or all of the steps of any method described in the method embodiment. The computer program product may be a software installation package, and the computer includes an electronic device.

[0103] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0104] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0105] In the several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above-mentioned units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0106] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0107] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0108] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including a number of instructions to enable a computer electronic device (which can be a personal computer, electronic device or network electronic device, etc.) to perform all or part of the steps of the above-mentioned methods in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0109] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable memory, which may include a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0110] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for predicting arc in a low voltage circuit, characterized in that: A processor applied to a low-voltage circuit system, the low-voltage circuit system comprising a first circuit module and a second circuit module, the first circuit module comprising an actual scenario circuit, the second circuit module comprising a simulated fault circuit, the low-voltage circuit system further comprising a current acquisition device, the method comprising: receiving circuit data within a target period from the current acquisition device, the circuit data including current data, the current data including current data of the first circuit module and / or the second circuit module, and the target period including a current period; Determine an effective current component and a noise current component based on the current data, and determine a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after denoising; Determine a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal; Determine a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtain a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model; An arc prediction result is determined based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, and the arc prediction result includes a first prediction result. If the arc prediction result is the first prediction result, a warning message is issued; the arc prediction result is used to characterize whether an arc may be generated in the low-voltage circuit during a sequence period.

2. The method according to claim 1, characterized in that: The determining of the effective current component and the noise current component based on the current data comprises: determining normalized current data based on the current data; Performing modal decomposition based on the normalized current data to determine a j-order intrinsic modal component; Determine a plurality of signal parameters based on the j-order intrinsic mode component and / or the normalized current data, the signal parameters including a j-order variance of the j-order intrinsic mode component and / or a variance of the current data; A variance contribution rate is determined based on the signal parameter, and the effective current component and the noise current component are determined based on the variance contribution rate.

3. The method according to claim 2, characterized in that The determining of the reconstructed signal based on the effective current component, the noise current component and the wavelet threshold denoising operation comprises: determining a first type of frequency component based on the noise current component, wherein the first type of frequency component includes a high frequency component; Determine a j-order modal component threshold based on the j-order eigenmodal component; and determine a j-order denoised modal component based on the first type of frequency component and the j-order modal component threshold; The reconstructed signal is determined based on the j-order denoised modal component and the effective current component.

4. The method according to claim 3, characterized in that The determining of the j-order modal component threshold based on the j-order eigenmodal component comprises: Determining a first-order eigenmode component based on the j-order eigenmode component, wherein the first-order eigenmode component is an eigenmode component when j=1; The first-order energy signal is determined based on the first-order eigenmode component and the following formula: ； in, represents the median value, represents the absolute value, is the first-order eigenmode component, and 0.6745 is the 75% quantile of the standard normal distribution; Determining a j-order energy signal based on the 1-order energy signal and a preset first empirical parameter; The j-order modal component threshold is determined based on the j-order energy signal and a plurality of preset second empirical parameters.

5. The method according to claim 1 or 4, characterized in that: The multiple time domain features include at least one of a root mean square parameter, a variance parameter, a kurtosis factor parameter, a pulse factor parameter, and a current change rate parameter; the multiple frequency domain features include at least one of a harmonic current content rate and a total harmonic distortion rate; The determining of a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtaining a feature contribution rate based on the feature matrix comprises: Determine a feature matrix with n rows and m columns based on the multiple time domain features and the multiple frequency domain features, where m corresponds to the number of the time domain features and / or the frequency domain features, and n corresponds to the number of the current data; Determine a correlation coefficient matrix based on the characteristic matrix with n rows and m columns; A plurality of eigenvalues ​​are determined based on the correlation coefficient matrix, and the feature contribution rate corresponding to each of the time domain features and / or the frequency domain features is determined based on the plurality of eigenvalues.

6. The method according to claim 5, characterized in that Determining the arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model includes: Determine the principal component feature and the number of principal component features based on the feature contribution rate, wherein the number of principal component features is the number of the principal component features; Based on the number of principal component features, the pre-built arc prediction model is updated so that the pre-built arc prediction model meets the current prediction condition; An arc prediction result is determined based on the circuit data and the pre-built arc prediction model.

7. The method according to claim 6, characterized in that The circuit data further includes circuit environment data, and the circuit environment data includes at least one of circuit temperature data, circuit pressure data, and circuit illumination data within a target period; Determining the arc prediction result based on the circuit data and the pre-built arc prediction model includes: Perform data standardization processing based on the circuit temperature data and / or the circuit pressure data and / or the circuit illumination data to determine at least one standard environment data; Determine the environment data based on the standard environment data and perform linear fitting to determine a curve and a plurality of slope parameters corresponding to at least one of the circuit environment data; determining a deviation parameter based on the plurality of slope parameters and a preset slope parameter; Determining an environmental impact factor based on the deviation parameter; An arc prediction result is determined based on the environmental influencing factors, the current data and the pre-built arc prediction model.

8. An arc prediction device for a low voltage circuit, characterized in that: A processor applied to a low-voltage circuit system, the low-voltage circuit system comprising a first circuit module and a second circuit module, the first circuit module comprising an actual scenario circuit, the second circuit module comprising a simulated fault circuit, the low-voltage circuit system further comprising a current acquisition device, the device comprising: A data acquisition module, configured to receive circuit data within a target period from the current acquisition device, wherein the circuit data includes current data, and the current data includes current data of the first circuit module and / or the second circuit module. The target period includes a current period; A first determination module, configured to determine an effective current component and a noise current component based on the current data, and to determine a reconstructed signal based on the effective current component, the noise current component and a wavelet threshold denoising operation, wherein the reconstructed signal is a current signal after denoising; A second determination module is used to determine a plurality of time domain features and a plurality of frequency domain features based on the reconstructed signal, wherein the frequency domain features include waveform distortion features in the frequency domain for characterizing the reconstructed signal; A third determination module is used to determine a feature matrix based on the multiple time domain features and the multiple frequency domain features, and obtain a feature contribution rate based on the feature matrix, wherein the feature contribution rate represents the importance of each of the time domain features or the frequency domain features, so as to constrain the feature dimension of the pre-constructed arc prediction model; An arc prediction module is used to determine an arc prediction result based on the circuit data, the characteristic contribution rate and the pre-built arc prediction model, the arc prediction result including a first prediction result. If the arc prediction result is the first prediction result, a warning message is issued; the arc prediction result is used to characterize whether an arc is likely to be generated in the low-voltage circuit during a sequence period.

9. A computer-readable storage medium, characterized in that: An arc prediction program for a low-voltage circuit is stored, including execution instructions. When a processor of an electronic device executes the execution instructions, the processor executes the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It comprises a processor and a memory storing execution instructions, wherein the memory stores one or more programs; when the processor executes the execution instructions stored in the memory, the processor executes the method according to any one of claims 1 to 7.

Images