A power distribution switchgear online monitoring system and method

Abstract

The application relates to the field of state monitoring, and discloses a power distribution switch device online monitoring system and method. The system comprises a state vector generation module, a feature decomposition module, a feature optimization module, a joint feature generation module and a fault alarm module. According to pre-acquired power distribution switch device operation data, an operation state estimation value of the power distribution switch device is generated, vector coding is carried out, an operation state vector is obtained, vector splicing, denoising and feature extraction are carried out on the operation state vector based on an EMD-WPD serial feature extraction algorithm, a joint feature matrix of the operation state is obtained, finally, the joint feature matrix of the power distribution switch device operation state is compared with a preset reference feature matrix, whether the power distribution switch device operation state is abnormal is judged based on the comparison result, and if the power distribution switch device operation state is abnormal, an alarm is given to the maintenance personnel. The EMD-WPD serial feature extraction algorithm improves the accuracy of the power distribution switch device online monitoring.

Description

Technical Field

[0001] This invention relates to the field of sensor data acquisition technology, and in particular to an online monitoring system and method for power distribution switchgear. Background Technology

[0002] With the continuous improvement of the construction of intelligent power distribution switchgear and the increasing requirements for power supply reliability, the real-time perception and fault early warning capabilities of power distribution switchgear, as a key node in the power system, directly affect the power supply quality and the safety of power distribution switchgear. The traditional operation and maintenance mode that relies on regular maintenance and manual inspection has problems such as strong lag and low response efficiency, making it difficult to detect potential defects such as insulation degradation and mechanical characteristic deviation in a timely manner, which may lead to unplanned power outages or even equipment damage accidents.

[0003] The condition of power distribution equipment is influenced by a combination of dynamic factors across multiple dimensions, including cyclic electrothermal stress caused by load fluctuations, material performance degradation due to changes in ambient temperature and humidity, and mechanical wear caused by the cumulative effects of mechanical operations. These factors exhibit nonlinear correlations, and their weights shift with the equipment's operating age and evolving conditions. Existing offline monitoring methods cannot capture the continuous trajectory of equipment condition changes, while single-parameter threshold alarm mechanisms are susceptible to interference signals, leading to biased condition assessments. Therefore, online monitoring systems for power distribution switchgear must possess the capability to fuse and analyze multi-source information to monitor equipment condition. Summary of the Invention

[0004] This invention provides an online monitoring system and method for power distribution switchgear, the main purpose of which is to solve the problem of low accuracy in online monitoring of power distribution switchgear.

[0005] To achieve the above objectives, the present invention provides an online monitoring system for power distribution switchgear, characterized in that the system includes a state vector generation module, a feature decomposition module, a feature optimization module, a joint feature generation module, and a fault alarm module, wherein: The state vector generation module is used to generate a state estimate of the power distribution switchgear in operation based on the pre-collected operating state data, and to perform vector encoding on the state estimate to obtain the state vector of the state estimate. The feature decomposition module is used to concatenate the state vector to obtain the state matrix of the state estimate, perform feature decomposition on the state matrix to obtain the intrinsic mode components of the state estimate, and filter the intrinsic mode components to obtain the key intrinsic mode components of the state estimate. The feature optimization module is used to perform wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate. The joint feature generation module is used to calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and then perform vector concatenation to obtain the joint feature matrix of the state estimate. The fault alarm module is used to compare the joint feature matrix with the preset state reference feature matrix, and determine whether there is an abnormality in the operating state based on the comparison result. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify the maintenance personnel.

[0006] Optionally, when the state vector generation module generates a state estimate of the power distribution switchgear in operation based on pre-collected operating state data, it includes: The operating status data is collected by data acquisition equipment pre-installed at the nodes of the power distribution switchgear. The operating status data includes voltage, current, power, power frequency, vibration amplitude, and partial discharge characteristics. A state estimate of the power distribution switchgear in its operating state is generated based on a preset state estimation algorithm and the operating state data, wherein the preset state estimation algorithm is:

[0007] in, It is the current moment. State estimate of the power distribution switchgear in operation. It was the previous moment State estimate of the power distribution switchgear in operation. It is the gain matrix. It is the current moment. The measured value, It is the observation matrix.

[0008] Optionally, the gain matrix The calculation formula is:

[0009] in, It was the previous moment The state error covariance matrix, It is the transpose of the observation matrix. It is the observation matrix. It is the covariance matrix of the measured noise.

[0010] Optionally, when the state vector generation module performs vector encoding on the state estimate to obtain the state vector of the state estimate, it includes: The state estimates include the current state estimates, voltage state estimates, power state estimates, power frequency state estimates, vibration amplitude state estimates, and partial discharge characteristic estimates of the power distribution switchgear in operation. Each of the estimated current state, voltage state, power state, power frequency, vibration amplitude, and partial discharge characteristic state is vector-encoded to obtain a state vector of the estimated state value. The state vector... for:

[0011] in, It is a state vector. It is a voltage state estimate. It is a current state estimate. It is the power state estimate. It is an estimate of the power frequency state. It is an estimated value of the vibration amplitude state. It is an estimate of the characteristic state of partial discharge. It is a time marker.

[0012] Optionally, the feature decomposition module performs feature decomposition on the state vector to obtain the key intrinsic mode components of the state estimate, including: The state vector is eigenvalued by performing feature decomposition according to a preset empirical mode algorithm to generate the intrinsic mode components of the state estimate, wherein the preset empirical mode algorithm is:

[0013] in, It is a state vector. It is the initial signal for the k-th decomposition. It is the residual after the (k-1)th decomposition. It is the i-th intrinsic mode component after the eigenvalue decomposition of the state vector. The mean function after the i-th sieve iteration in the k-th decomposition. It is obtained by cubic spline interpolation. The mean of the upper and lower envelopes, It is the standard deviation, which measures the stability of the signal between two iterations.

[0014] Optionally, when the eigenvalue decomposition module performs filtering on the intrinsic mode components to obtain the key intrinsic mode components of the state estimate, it includes: High-frequency noise is removed from the intrinsic mode components based on preset kurtosis and energy percentage to obtain the key intrinsic mode components of the state estimate. The preset kurtosis and energy percentage are calculated using the following formulas:

[0015] in, It is the kth The steepness, It is the key intrinsic mode component after the kth screening. It is the expected value. represent The mean, It is the standard deviation, used to eliminate the influence of dimensions, making kurtosis a dimensionless quantity. It is the kth The proportion of energy, It is the state vector.

[0016] Optionally, when the feature optimization module performs wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate, it includes: The optimal basis of the wavelet packet decomposition is obtained by minimizing the cost function and the key intrinsic mode components according to the preset entropy criterion, wherein the entropy criterion minimization cost function is:

[0017] in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the first Layer Individual belt position wavelet coefficients at It is coefficient energy. Total energy of the sub-band; Based on comparing the entropy values ​​of all wavelet bases generated by wavelet packet decomposition, the base that minimizes the preset entropy criterion cost function is selected.

[0018] The maximum number of wavelet packet decomposition layers is obtained based on a preset layering algorithm and the key intrinsic mode components, wherein the preset layering algorithm is:

[0019] in, It's the sampling rate. Decomposition layer number, It is the highest effective frequency. It is the optimal number of decomposition levels selected dynamically; The high-frequency fault sub-band of the key intrinsic mode components is generated according to the preset wavelet packet recursive decomposition formula and the key intrinsic mode components, wherein the preset wavelet packet recursive decomposition formula is:

[0020] in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the nth wavelet basis function, representing the wavelet position. It is a normalization factor that ensures consistent wavelet energy across different scales. Control the scaling and translation operations of wavelet compression, stretching, and position shift. These are the coefficients of a low-pass filter, used to extract the low-frequency components of a signal. These are high-pass filter coefficients, used to extract the high-frequency components of a signal. It is the first Layer The wavelet coefficients at position m in the sub-band It is the position index of the current layer.

[0021] Optionally, when the joint feature generation module calculates the energy proportion of the key intrinsic mode components and the energy entropy of the high-frequency fault sub-band and performs vector concatenation to obtain the joint feature matrix of the state estimate, it includes: The energy proportion of the key intrinsic mode components of the state estimate is calculated based on a preset energy proportion formula and the key intrinsic mode components, wherein the preset energy proportion formula is:

[0022] in, It is the key intrinsic mode component of the i-th state estimate. It is the total energy; The energy entropy of the high-frequency fault sub-band of the state estimate is calculated based on the preset energy entropy formula and the high-frequency fault sub-band, wherein the preset energy entropy formula is:

[0023] in, It is the sum of coefficient energy. It is the first Layer Individual belt position wavelet coefficients at the location, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter; The joint feature matrix of the state estimate is obtained by vector concatenation of the energy proportion of the key intrinsic mode components and the energy entropy of the high-frequency fault sub-band.

[0024] Optionally, after performing the comparison of the joint feature matrix of the state estimate and the preset benchmark feature matrix, the fault alarm module further includes: Based on the joint feature matrix of the state estimates, an operating state prediction model for the power distribution switchgear is established to predict the future operating state and probability of failure of the equipment.

[0025] To address the above problems, the present invention also provides an online monitoring method for power distribution switchgear, the method comprising: A state estimate of the state estimate is generated based on the pre-collected data of the state estimate, and the state estimate is vector-encoded to obtain the state vector of the state estimate; The state vector is concatenated to obtain the state matrix of the state estimate. The state matrix is ​​then subjected to eigenvalue decomposition to obtain the intrinsic mode components of the state estimate. The intrinsic mode components are then filtered to obtain the key intrinsic mode components of the state estimate. Wavelet packet decomposition is performed on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate; Calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and concatenate the vectors to obtain the joint feature matrix of the state estimate; The joint feature matrix is ​​compared with the baseline feature matrix of the preset state. Based on the comparison result, it is determined whether there is an abnormality in the operating state. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify the maintenance personnel.

[0026] Beneficial effects This invention generates a state estimate of the operating status of power distribution switchgear by real-time data acquisition and vectorizes it, providing an accurate representation of the operating status of the switchgear. In power distribution switchgear monitoring, real-time data acquisition is susceptible to interference from noise and other factors. Traditional power distribution switchgear data acquisition often relies on hardware filtering and frequency domain analysis, requiring preset filter parameters or assumptions of signal stationarity, which can easily lead to distortion of the effective signal or residual noise. The preset empirical mode algorithm achieves high-precision separation of noise and effective components in complex signals through adaptive multi-scale decomposition of the data, and then filters to obtain the key intrinsic mode components of the state estimate. Wavelet packet decomposition is then performed on the key intrinsic mode components. Breaking through the limitations of traditional wavelet transform in high-frequency signal decomposition, this invention refines high-frequency sub-bands layer by layer, accurately separating noise and weak fault features to obtain high-frequency fault sub-bands for state estimation, preserving key details of transient faults, and achieving a balance between noise reduction and signal integrity. The energy entropy of the high-frequency fault sub-bands and the energy proportion of key intrinsic mode components are calculated and vector-concatenated to obtain a joint feature matrix of the state estimation. The joint feature matrix is ​​compared with a preset benchmark feature matrix, and the comparison results determine whether there is an abnormality in the operating state. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify maintenance personnel. Therefore, the online monitoring system and method for power distribution switchgear proposed in this invention can improve the accuracy of monitoring the operating state of power distribution switchgear. Attached Figure Description

[0027] Figure 1 This is a system architecture diagram of an online monitoring system for power distribution switchgear provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating an online monitoring method for power distribution switchgear according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0030] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0031] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0032] In practice, the server-side equipment deployed in the online monitoring system for power distribution switchgear may consist of one or more devices. This online monitoring system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node to provide online monitoring services to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage each user terminal. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide online monitoring services to various user terminals.

[0033] In terms of implementation, the online monitoring system for power distribution switchgear and the user terminal are mutually compatible. That is, if the online monitoring system for power distribution switchgear is an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the online monitoring system for power distribution switchgear is implemented as a website, then the user terminal is implemented as a webpage; or if the online monitoring system for power distribution switchgear is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0034] like Figure 1 The figure shown is a system architecture diagram of an online monitoring system for power distribution switchgear provided in an embodiment of the present invention.

[0035] The online monitoring system 100 for power distribution switchgear described in this invention can be located on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed on the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed as a website. Depending on the functions implemented, the online monitoring system 100 for power distribution switchgear may include a state vector generation module 101, a feature decomposition module 102, a feature optimization module 103, a joint feature generation module 104, and a fault alarm module 105. The module described in this invention can also be called a unit, referring to a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0036] In this embodiment of the invention, each of the above-mentioned modules in the online monitoring system for power distribution switchgear can be implemented independently and can be called upon with other modules. This "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the online monitoring system for power distribution switchgear provided by this embodiment of the invention, the applicable scope of the system architecture can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the online monitoring system for power distribution switchgear. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0037] The following describes the components and specific workflow of the online monitoring system for power distribution switchgear, using specific embodiments as examples: The state vector generation module 101 is used to generate a state estimate of the power distribution switchgear in operation based on the pre-collected operating state data, and to perform vector encoding on the state estimate to obtain the state vector of the state estimate. In this embodiment of the invention, when the state vector generation module generates a power distribution switchgear state estimate based on pre-collected power distribution switchgear data, it includes: The operating status data of the power distribution switchgear is collected by the data acquisition equipment pre-installed at the nodes of the power distribution switchgear. The operating status data of the power distribution switchgear includes voltage, current, power, power frequency, vibration amplitude and partial discharge characteristics. A state estimate of the power distribution switchgear in its operating state is generated based on a preset state estimation algorithm and the operating state data, wherein the preset state estimation algorithm is:

[0038] in, It is the current moment. State estimate of the power distribution switchgear in operation. It was the previous moment State estimate of the power distribution switchgear in operation. It is the gain matrix. It is the current moment. The measured value, It is the observation matrix.

[0039] In detail, assume that the following data has been collected from the power distribution switchgear (assuming it is the measurement data at a certain time t): voltage (V) is 380V, current (I) is 630A, power (P) is 600kW, power frequency (f) is 50Hz, vibration amplitude (vb) is 100μm, and partial discharge characteristic (PD) is discharge amplitude: 50pC.

[0040] Furthermore, based on the given state estimation algorithm, it is assumed that the previous time step is known. The estimated value and measured values Through the gain matrix Estimate the current state and finally calculate the current time. Estimated value of the operating status of the power distribution switchgear.

[0041] In detail, the gain matrix for:

[0042] in, It was the previous moment The state error covariance matrix, It is the transpose of the observation matrix. It is the observation matrix. It is the covariance matrix of the measured noise.

[0043] In detail, assume the state error covariance matrix of the previous time step... for:

[0044] Observation matrix for:

[0045] Measurement noise covariance matrix for:

[0046] Assuming the voltage and current at the previous moment were 380V and 630A respectively, then the state estimate at the previous moment would be... for:

[0047] Assuming the current voltage and current are 385V and 632A respectively, then the measured values ​​at the current moment are... for:

[0048] Further, the gain matrix is ​​calculated. Substitute these data into the formula and first calculate... for:

[0049] because It is an identity matrix, so ,so:

[0050] Next, calculate :

[0051] Then calculate the gain matrix. :

[0052] In detail, the gain matrix is ​​a diagonal matrix, where each diagonal element represents the gain of a state variable. In this example, it corresponds to the gain of the two state variables of the power distribution switchgear (let's say voltage and current).

[0053] The gain value of 0.7143 indicates the degree to which the system responds to the observed data. The closer it is to 1, the more the system trusts the current observed data, and vice versa. In this example, the gains for voltage and current are the same, indicating that the system gives equal weight to the observed data for voltage and current.

[0054] Furthermore, based on the formula, the final calculation is... The matrix is ​​substituted into a preset state estimation algorithm to obtain the state estimate of the power distribution switchgear.

[0055] First, calculate :

[0056] Next, the state estimate update is calculated:

[0057] Therefore, the state estimate of the power distribution switchgear is:

[0058] Here, 383.57 is the estimated voltage value in volts (V), and 631.43 represents the estimated current value in amperes (A). These values ​​are the optimal estimates of the state of the power distribution switchgear after combining system predictions (the state at the previous moment) and current measurements (such as the outputs of voltage and current sensors).

[0059] In detail, the gain matrix describes the weighting distribution when combining observations and predictions. The larger the value, the more the system trusts the observation data; the smaller the value, the more the system trusts the prediction data. The state estimate is the system's best estimate of the state (such as voltage and current) of the power distribution switchgear at the current moment, combining the system's predictions and current measurements.

[0060] Similarly, when the observed values ​​include electric power, electric frequency, vibration amplitude, and partial discharge characteristics, the same calculation method is used.

[0061] In this embodiment of the invention, when the state vector generation module performs vector encoding on the state estimate to obtain the state vector of the state estimate, it includes: Each of the estimated current state, voltage state, power state, power frequency, vibration amplitude, and partial discharge characteristic state values ​​is vector-encoded to obtain a state vector for the estimated state values. The state vector... for:

[0062] in, It is a state vector. It is a voltage state estimate. It is a current state estimate. It is the power state estimate. It is an estimate of the power frequency state. It is an estimated value of the vibration amplitude state. It is an estimate of the characteristic state of partial discharge. It is a time marker.

[0063] In this embodiment of the invention, based on the estimated values ​​of various states of the power distribution switchgear (current, voltage, power, power frequency, vibration amplitude, and partial discharge characteristics), they can be integrated into a single state vector.

[0064] For example, assuming the following states of a power distribution switchgear have been estimated using a state estimation algorithm: voltage 380V; current 630A; power 600kW; power frequency 50Hz; vibration amplitude 100μm; partial discharge characteristic 50pC, then the state vector of the power distribution switchgear is: .

[0065] The feature decomposition module 102 is used to perform feature decomposition on the state vector to obtain the intrinsic mode components of the state estimate, and to filter the intrinsic mode components to obtain the key intrinsic mode components of the state estimate. In this embodiment of the invention, the feature decomposition module performs feature decomposition on the state vector to obtain the intrinsic mode components of the state estimate, including: The state vector is eigenvalued by performing feature decomposition according to a preset empirical mode algorithm to generate the intrinsic mode components of the state estimate, wherein the preset empirical mode algorithm is:

[0066] in, It is a state vector. It is the initial signal of the k-th decomposition. It is the residual after the (k-1)th decomposition. It is the i-th intrinsic mode component after the eigenvalue decomposition of the state vector. The mean function after the i-th sieve iteration in the k-th decomposition. It is obtained by cubic spline interpolation. The mean of the upper and lower envelopes, It is the standard deviation, which measures the stability of the signal between two iterations.

[0067] Specifically, the timestamps of the state vectors are determined, and the state vectors are concatenated based on the timestamps to obtain the state matrix. The state matrix is ​​as follows:

[0068] in, Indexes representing different points in time. It is a vector matrix formed by concatenating the state vectors. It represents the total number of points in time.

[0069] In this embodiment of the invention, it is assumed that at multiple time points ( , ,. . . , At three time points ( ), the different state vectors of the state estimate, where , , , The three states correspond to three state vectors respectively. These state vectors are concatenated to obtain the state matrix of the estimated state value. The state matrix X formed by concatenating these state vectors is:

[0070] Among them, at three time points ( , , The values ​​corresponding to the three voltage state estimates are 380, 386, and 378, the three current state estimates are 630, 632, and 631, the three electric power state estimates are 600, 620, and 612, the three electric frequency state estimates are 50, 49.9, and 50.1, the vibration amplitude estimates are 100, 98, and 101, and the partial discharge characteristics are 50, 51, and 53.

[0071] Specifically, suppose the estimated amplitude of the vibration at a certain moment is [value missing].

[0072] Specifically, suppose the estimated amplitude of the vibration at a certain moment is:

[0073] First, determine the upper and lower envelopes, and then perform cubic spline interpolation on the maxima and minima to obtain the upper envelope U(t) and the lower envelope (t): The envelope of the maximum point is:

[0074] The envelope of the minimum point is:

[0075] Calculate the envelope mean:

[0076] In detail, in the vibration signal of power distribution equipment, the mean envelope can reflect the low-frequency mechanical vibration trend of the equipment operation (such as slow vibration changes caused by bearing wear), and after removing it, the high-frequency fault characteristics (such as instantaneous impact vibration) can be highlighted.

[0077] Next, initialization, during the first iteration. :

[0078] in, It is a state vector. It is the initial signal for the k-th decomposition. It is the residual after the (k-1)th decomposition. It is the i-th intrinsic mode component after the eigenvalue decomposition of the state vector.

[0079] Furthermore, the signal is updated through multiple iterations:

[0080] in, It is obtained by cubic spline interpolation. The mean of the upper and lower envelopes, The mean function after the i-th sieve iteration in the k-th decomposition.

[0081] Finally, the iteration stops when the threshold is reached:

[0082] in, It is the standard deviation, which measures the stability of the signal between two iterations. The value is usually 0.2-0.3.

[0083] In detail, if Greater than This indicates that the signal still contains significant trend components (such as noise interference); if it satisfies Less than If the current component meets the IMF's quasi-periodicity and symmetry requirements, then screening can be stopped. = Otherwise, repeat the above process.

[0084] Assume the decomposition result is

[0085] In this embodiment of the invention, when the eigenvalue decomposition module performs screening of the intrinsic mode components to obtain the key intrinsic mode components of the state estimate, it includes: High-frequency noise is removed from the intrinsic mode components based on preset kurtosis and energy percentage to obtain the key intrinsic mode components of the state estimate. The preset kurtosis and energy percentage are calculated using the following formulas:

[0086] in, It is the kth The steepness, It is the key intrinsic mode component after the kth screening. It is the expected value. represent The mean, It is the standard deviation, used to eliminate the influence of dimensions, making kurtosis a dimensionless quantity. It is the kth The proportion of energy, It is the state vector.

[0087] Furthermore, high-frequency noise intrinsic mode components are removed, while key intrinsic mode components are retained. For example:

[0088] First, calculate the mean:

[0089] Next, calculate the standard deviation and kurtosis:

[0090]

[0091] Finally, calculate Ability percentage:

[0092] Total energy of the state matrix:

[0093] so, Energy percentage:

[0094] In detail, the intrinsic mode components (IMFs) are screened based on kurtosis greater than 3 and the percentage of retained energy greater than 5% to obtain the key intrinsic mode components (S_IMFs) that are non-Gaussian and non-noise-dominated.

[0095] Assuming the screening results are:

[0096] The feature optimization module 103 is used to perform wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate. In this embodiment of the invention, when the feature optimization module performs wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate, it includes: The optimal basis of the wavelet packet decomposition is obtained by minimizing the cost function and the key intrinsic mode components according to the preset entropy criterion, wherein the entropy criterion minimization cost function is:

[0097] in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the first Layer Individual belt position wavelet coefficients at It is coefficient energy. Total energy of the sub-band; In this embodiment of the invention, it is assumed that the key intrinsic mode components Center frequency 800Hz, energy ratio 60%, kurtosis 4.2.

[0098] Candidate wavelet bases: db4, sym8, coif3. The entropy criterion minimizes the cost function to calculate the entropy value after decomposition for each wavelet base.

[0099]

[0100]

[0101] Therefore, based on comparing the entropy values ​​of all wavelet bases generated by wavelet packet decomposition, the base that minimizes the preset entropy criterion cost function is selected. As the optimal wavelet basis.

[0102] The maximum number of wavelet packet decomposition layers is obtained based on a preset layering algorithm and the key intrinsic mode components, wherein the preset layering algorithm is:

[0103] in, It is the sampling rate. Decomposition layer number, It is the highest effective frequency. It is the optimal number of decomposition layers selected dynamically.

[0104] Assuming sampling rate =1MHz, highest effective frequency =100kHz, the maximum number of decomposition layers is determined based on a hierarchical algorithm.

[0105] Calculate the maximum number of decomposition levels:

[0106] The high-frequency fault sub-bands of the key intrinsic mode components are generated according to the preset wavelet packet recursive decomposition formula and the key intrinsic mode components, wherein the preset wavelet packet recursive decomposition formula is:

[0107] in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the nth wavelet basis function, representing the wavelet position. It is a normalization factor that ensures consistent wavelet energy across different scales. Control the scaling and translation operations of wavelet compression, stretching, and position shift. These are the coefficients of a low-pass filter, used to extract the low-frequency components of a signal. These are high-pass filter coefficients, used to extract the high-frequency components of a signal. It is the first Layer The wavelet coefficients at position m in the sub-band It is the position index of the current layer.

[0108] The detailed decomposition process: Level 1 decomposition:

[0109]

[0110] Level 2 decomposition (for) ):

[0111]

[0112] Level 3 decomposition (for) )

[0113]

[0114] Finally, based on the partial discharge characteristic frequency containing 800Hz, which has the highest energy proportion (40%), the sub-band was selected. As a high-frequency fault sub-band.

[0115] In detail, the feature decomposition module removes noise and retains fault-sensitive components through empirical mode decomposition and kurtosis screening, while the feature optimization module accurately extracts high-frequency fault sub-bands through wavelet packet decomposition and entropy criterion screening, providing key inputs for subsequent joint feature generation; together, they improve the system's ability to identify early faults in power distribution switchgear.

[0116] The joint feature generation module 104 is used to calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and then perform vector concatenation to obtain the joint feature matrix of the state estimate.

[0117] In detail, energy entropy describes the disorder of energy distribution within a sub-band. The lower the entropy value, the more concentrated the energy is at the fault frequency, and the more obvious the fault characteristics are.

[0118] In detail, the proportion of energy of a single key component to the total energy reflects the significance of that component in the overall signal. The higher the energy proportion, the stronger the correlation between that component and the fault.

[0119] Assume that wavelet packet decomposition yields 4 sub-bands:

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133] Joint eigenvectors:

[0134] The fault alarm module 105, after comparing the joint feature matrix of the state estimate with a preset benchmark feature matrix, further includes: In this embodiment of the invention, the fault alarm module executes a predictive model of the operating status of the power distribution switchgear based on the joint feature matrix of the state estimates, and predicts the future operating status and probability of fault occurrence of the equipment.

[0135] In detail, the similarity calculation is performed between the currently monitored joint feature matrix and the preset benchmark feature matrix to determine whether the equipment's operating status deviates from the normal mode.

[0136] The joint feature matrix, output by the joint feature generation module, includes the energy proportion of key intrinsic mode components and the energy entropy of high-frequency fault sub-bands. Physical meaning: It reflects the multidimensional characteristics of the current equipment operating state (such as vibration energy distribution and partial discharge band energy entropy).

[0137] Baseline Feature Matrix: Constructed based on historical data during normal equipment operation and stored in the system database. Physical Meaning: Represents a "standard template" of features when the equipment is fault-free.

[0138] Assuming a reference matrix for:

[0139] Current joint feature matrix for:

[0140] Furthermore, based on the similarity calculation results, it is determined whether the device's operating status is abnormal. If the similarity is lower than a preset threshold, an anomaly detection is triggered. Physical meaning: The threshold reflects the system's tolerance for a "normal state"; the higher the threshold, the more stringent the anomaly detection.

[0141] The threshold is determined through verification using historical data. For example, 100 sets of normal data are collected from the same model of equipment, and the similarity distribution between them and the benchmark matrix is ​​calculated. The lower bound of the distribution (such as the 95th percentile) is taken as the threshold.

[0142] Calculate similarity:

[0143] With a threshold set at 0.9, and a current similarity of 0.87 < 0.9, the operating status of the power distribution switchgear is determined to be abnormal. Alarm levels: Alarms can be classified according to the severity of the anomaly, corresponding to different fault probability ranges. Physical meaning: Classified alarms help maintenance personnel prioritize handling high-risk faults.

[0144] Failure probability distribution:

[0145]

[0146]

[0147] In summary, the fault alarm module achieves quantitative assessment and tiered early warning of faults in power distribution switchgear through similarity analysis and probabilistic models. This module not only promptly detects abnormal states but also predicts fault development trends, providing data-driven decision support for equipment maintenance. The complementary multi-dimensional features in the joint feature matrix (such as energy percentage and energy entropy) significantly improve the accuracy of early fault identification.

[0148] Reference Figure 2 The diagram shown is a flowchart illustrating an online monitoring method for power distribution switchgear according to an embodiment of the present invention. In this embodiment, the online monitoring method for power distribution switchgear includes: S1. Based on the pre-collected operating status data, generate an estimated state value of the power distribution switchgear in the operating state, and perform vector encoding on the estimated state value to obtain the state vector of the estimated state value; S2. Concatenate the state vectors to obtain the state matrix of the state estimate, perform eigenvalue decomposition on the state matrix to obtain the intrinsic mode components of the state estimate, and filter the intrinsic mode components to obtain the key intrinsic mode components of the state estimate. S3. Perform wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate; S4. Calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and concatenate the vectors to obtain the joint feature matrix of the state estimate; S5. Compare the joint feature matrix with the baseline feature matrix of the preset state, and determine whether there is an abnormality in the operating state based on the comparison result. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify the maintenance personnel.

[0149] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0150] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0151] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0152] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0153] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An online monitoring system for power distribution switchgear, characterized in that, The system includes a state vector generation module, a feature decomposition module, a feature optimization module, a joint feature generation module, and a fault alarm module, wherein: The state vector generation module is used to generate a state estimate of the power distribution switchgear in operation based on the pre-collected operating state data, and to perform vector encoding on the state estimate to obtain the state vector of the state estimate. The feature decomposition module is used to concatenate the state vector to obtain the state matrix of the state estimate, perform feature decomposition on the state matrix to obtain the intrinsic mode components of the state estimate, and filter the intrinsic mode components to obtain the key intrinsic mode components of the state estimate. The feature optimization module is used to perform wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate. The joint feature generation module is used to calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and then perform vector concatenation to obtain the joint feature matrix of the state estimate. The fault alarm module is used to compare the joint feature matrix with the preset state reference feature matrix, and determine whether there is an abnormality in the operating state based on the comparison result. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify the maintenance personnel.

2. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, When the state vector generation module generates an estimated state value of the power distribution switchgear in operation based on pre-collected operating state data, it includes: The operating status data is collected by data acquisition equipment pre-installed at the nodes of the power distribution switchgear. The operating status data includes voltage, current, power, power frequency, vibration amplitude, and partial discharge characteristics. A state estimate of the power distribution switchgear in its operating state is generated based on a preset state estimation algorithm and the operating state data, wherein the preset state estimation algorithm is: in, It is the current moment. State estimate of the power distribution switchgear in operation. It was the previous moment State estimate of the power distribution switchgear in operation. It is the gain matrix. It is the current moment. The measured value, It is the observation matrix.

3. The online monitoring system for power distribution switchgear as described in claim 2, characterized in that, The gain matrix The calculation formula is: in, It was the previous moment The state error covariance matrix, It is the transpose of the observation matrix. It is the observation matrix. It is the covariance matrix of the measured noise.

4. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, When the state vector generation module performs vector encoding on the state estimate to obtain the state vector of the state estimate, it includes: The state estimates include the current state estimates, voltage state estimates, power state estimates, power frequency state estimates, vibration amplitude state estimates, and partial discharge characteristic estimates of the power distribution switchgear in operation. Each of the estimated current state, voltage state, power state, power frequency, vibration amplitude, and partial discharge characteristic state is vector-encoded to obtain a state vector of the estimated state value. The state vector... for: in, It is a state vector. It is a voltage state estimate. It is a current state estimate. It is the power state estimate. It is an estimate of the power frequency state. It is an estimated value of the vibration amplitude state. It is an estimate of the characteristic state of partial discharge. It is a time marker.

5. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, The state vectors are concatenated to obtain the state matrix of the state estimate. The state matrix is ​​then subjected to eigenvalue decomposition to obtain the intrinsic mode components of the state estimate. The state vector is eigenvalued by performing feature decomposition according to a preset empirical mode algorithm to generate the intrinsic mode components of the state estimate, wherein the preset empirical mode algorithm is: in, It is a state vector. It is the initial signal for the k-th decomposition. It is the residual after the (k-1)th decomposition. It is the i-th intrinsic mode component after the eigenvalue decomposition of the state vector. The mean function after the i-th sieve iteration in the k-th decomposition. It is obtained by cubic spline interpolation. The mean of the upper and lower envelopes, It is the standard deviation, which measures the stability of the signal between two iterations. It is the total number of the intrinsic modal components. It is the screening stop threshold.

6. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, When the eigenvalue decomposition module performs filtering on the intrinsic mode components to obtain the key intrinsic mode components of the state estimate, it includes: High-frequency noise is removed from the intrinsic mode components based on preset kurtosis and energy percentage to obtain the key intrinsic mode components of the state estimate. The preset kurtosis and energy percentage are calculated using the following formulas: in, It is the kth The steepness, It is the key intrinsic mode component after the kth screening. It is the expected value. represent The mean, It is the standard deviation, used to eliminate the influence of dimensions, making kurtosis a dimensionless quantity. It is the kth The proportion of energy, It is the state vector. It is a time marker.

7. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, When the feature optimization module performs wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate, it includes: The optimal basis of the wavelet packet decomposition is obtained by minimizing the cost function and the key intrinsic mode components according to the preset entropy criterion, wherein the entropy criterion minimization cost function is: in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the first Layer Individual belt position wavelet coefficients at It is coefficient energy. Total energy of the sub-band; Based on comparing the entropy values ​​of all wavelet bases generated by wavelet packet decomposition, the base that minimizes the preset entropy criterion cost function is selected. The maximum number of wavelet packet decomposition layers is obtained based on a preset layering algorithm and the key intrinsic mode components, wherein the preset layering algorithm is: in, It's the sampling rate. Decomposition layer number, It is the highest effective frequency. It is the optimal number of decomposition levels selected dynamically; The high-frequency fault sub-band of the key intrinsic mode components is generated according to the preset wavelet packet recursive decomposition formula and the key intrinsic mode components, wherein the preset wavelet packet recursive decomposition formula is: in, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter. It is the nth wavelet basis function, representing the wavelet position. It is a normalization factor that ensures consistent wavelet energy across different scales. Control the scaling and translation operations of wavelet compression, stretching, and position shift. These are the coefficients of a low-pass filter, used to extract the low-frequency components of the signal. These are high-pass filter coefficients, used to extract the high-frequency components of a signal. It is the first Layer The wavelet coefficients at position m in the sub-band It is the position index of the current layer. It is a time marker. It is the first Layer, First Sub-band, translation parameters are: wavelet basis functions in time The value at that location, It is the first Layer, First The child is in position wavelet coefficients at the location, It is the first Layer, First The child is in position The wavelet coefficients at that point.

8. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, When the joint feature generation module calculates the energy proportion of the key intrinsic mode components and the energy entropy of the high-frequency fault sub-band and performs vector concatenation to obtain the joint feature matrix of the state estimate, it includes: The energy proportion of the key intrinsic mode components of the state estimate is calculated based on a preset energy proportion formula and the key intrinsic mode components, wherein the preset energy proportion formula is: in, It is the key intrinsic mode component of the i-th state estimate. It is the total energy. It is the first Energy values ​​of key intrinsic mode components; The energy entropy of the high-frequency fault sub-band of the state estimate is calculated based on the preset energy entropy formula and the high-frequency fault sub-band, wherein the preset energy entropy formula is: in, It is the sum of coefficient energy. It is the first Layer Individual belt position wavelet coefficients at the location, It is the number of decomposition levels. It is a sub-band index. It is the wavelet function translation parameter; The joint feature matrix of the state estimate is obtained by vector concatenation of the energy proportion of the key intrinsic mode components and the energy entropy of the high-frequency fault sub-band.

9. The online monitoring system for power distribution switchgear as described in claim 1, characterized in that, After performing the comparison of the joint feature matrix of the operating status of the power distribution switchgear with the preset joint feature matrix of the normal status of the power distribution switchgear, the fault alarm module further includes: A prediction model for the operating status of the power distribution switchgear is established based on the joint feature matrix of the operating status of the power distribution switchgear to predict the future operating status and probability of failure of the equipment.

10. A method for online monitoring of power distribution switchgear, characterized in that, The method includes: S1. Based on the pre-collected operating status data, generate an estimated state value of the power distribution switchgear in the operating state, and perform vector encoding on the estimated state value to obtain the state vector of the estimated state value; S2. Concatenate the state vectors to obtain the state matrix of the state estimate, perform eigenvalue decomposition on the state matrix to obtain the intrinsic mode components of the state estimate, and filter the intrinsic mode components to obtain the key intrinsic mode components of the state estimate. S3. Perform wavelet packet decomposition on the key intrinsic mode components to obtain the high-frequency fault sub-band of the state estimate; S4. Calculate the energy entropy of the high-frequency fault sub-band and the energy proportion of the key intrinsic mode components, and concatenate the vectors to obtain the joint feature matrix of the state estimate; S5. Compare the joint feature matrix with the baseline feature matrix of the preset state, and determine whether there is an abnormality in the operating state based on the comparison result. When the power distribution switchgear is in an abnormal operating state, an alarm is triggered to notify the maintenance personnel.

Images