A power distribution room switch cabinet partial discharge on-line monitoring and early warning system

Abstract

This invention relates to the field of switchgear discharge monitoring, specifically to an online monitoring and early warning system for partial discharge in switchgear in a power distribution room. The invention determines the authenticity of discharge in original signals based on the correlation coefficients of their modal components within a set historical time window. If a genuine discharge exists in an original signal, the entropy value of each pulse data of its dominant discharge component is obtained, the pulse stability index is analyzed, the discharge timing characteristic type is determined based on the recursive change rate, and the discharge intensity level is analyzed. A set of candidate state change vectors is constructed based on historical discharge intensity levels, and the predicted probability of each discharge intensity level in the future window is analyzed. The predicted probability is corrected based on the discharge timing characteristic type, and the discharge intensity level in the future window is predicted. When the discharge intensity level reaches a point requiring an early warning, the discharge area is identified based on the location of the corresponding acquisition device and the signal strength, and a corresponding early warning is issued, thus improving monitoring efficiency.

Description

Technical Field

[0001] This invention relates to the field of switchgear discharge monitoring, and specifically to an online monitoring and early warning system for partial discharge in switchgear in a power distribution room. Background Technology

[0002] High-voltage switchgear is widely used in power distribution systems, and its operational reliability directly determines the safety, stability, and continuity of power supply in the power distribution network. Partial discharge is the most significant cause of insulation defects in switchgear. Currently, online partial discharge monitoring has gradually replaced traditional periodic offline detection during power outages, becoming the mainstream technology, enabling non-intrusive and uninterrupted signal acquisition. However, in practical applications in complex power distribution rooms, existing monitoring systems generally suffer from low reliability in extracting true discharge signals, insufficient early warning efficiency, and significant false alarms and missed alarms, making it difficult to meet the high reliability, high precision, and intelligent monitoring requirements of smart grids.

[0003] Existing technologies, such as Chinese Patent Publication No. CN118054552B, disclose a method and alarm system for intelligent monitoring of the status of a power distribution room. This invention collects partial discharge, audio and environmental data through partial discharge sensors and environmental sensors, performs anomaly detection, optimizes the hyperparameters of the model using an overband algorithm, and analyzes the discharge status of the switchgear based on the target detection data to achieve real-time monitoring of the status of the power distribution room.

[0004] However, the existing technology has the following problems: 1. The existing technology directly inputs the raw signals collected by the sensor into the model for training and detection, without effectively separating the noise in the raw signal from the real discharge pulse. In essence, it still relies on a fixed noise threshold or preprocessing of the collected environmental noise, which is difficult to adapt to the complex dynamic noise environment of the power distribution room, resulting in low reliability of real discharge signal extraction and unreliable subsequent early warning results.

[0005] 2. Existing technologies use static characteristic parameters such as discharge amplitude and frequency based on a single time window as model inputs. They do not take into account the dynamic evolution process of effectively capturing the discharge pulse sequence, and cannot distinguish between smoothly evolving discharges and drastically fluctuating random discharges. When the discharge behavior exhibits random fluctuation characteristics, the model recognition results are unstable and prone to false or missed warnings. Maintenance personnel find it difficult to accurately judge the trend of equipment insulation degradation. Summary of the Invention

[0006] This invention aims to address the shortcomings of existing technologies by providing an online monitoring and early warning system for partial discharge in switchgear of power distribution rooms. The invention determines the authenticity of discharges through signal decomposition and modal component correlation, then identifies the discharge timing characteristic type based on the pulse stability index and recursive rate of change, and adaptively corrects the probability prediction results according to this type, achieving accurate identification and differentiated prediction of partial discharges in complex noise environments.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: The present invention provides an online monitoring and early warning system for partial discharge in switchgear of a power distribution room, comprising: a discharge authenticity identification module, a feature type determination module, an intensity level analysis module, and an early warning module. The connection relationship between the modules is as follows: the discharge authenticity identification module is connected to the feature type determination module, and the intensity level analysis module is connected to both the feature type determination module and the early warning module.

[0008] The discharge authenticity identification module decomposes each original signal within a set historical time window to obtain each modal component, and judges the authenticity of the discharge of the original signal based on the correlation coefficient of each modal component.

[0009] The feature type determination module, if a certain original signal has a real discharge, obtains the entropy value of each pulse data of its dominant discharge component, analyzes the pulse stability index, and obtains the recursive change rate of the historical stability index to determine the discharge timing feature type.

[0010] The intensity level analysis module analyzes the discharge intensity level based on the pulse repetition rate and amplitude sum of squares within a set historical time window. It constructs a set of candidate state change vectors based on each historical discharge intensity level and analyzes the predicted probability of each discharge intensity level occurring in the future window.

[0011] The early warning module corrects the predicted probability of each discharge intensity level based on the discharge timing characteristics and predicts the discharge intensity level of the future window. When the discharge intensity level reaches the required level, it identifies the discharge area based on the location of the corresponding acquisition device and the signal strength and issues the corresponding warning.

[0012] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention decomposes each original signal within a set historical time window to obtain each modal component, and judges the authenticity of the discharge of the original signal based on the correlation coefficient of each modal component. It has strong anti-interference ability, is suitable for complex electromagnetic environment in power distribution room, and improves the reliability of real signal extraction.

[0013] (2) This invention obtains the entropy value of each pulse data of the dominant component of the discharge from the original signal with real discharge, analyzes the pulse stability index, obtains the recursive change rate of the historical stability index to determine the discharge time sequence characteristic type, realizes quantitative analysis of discharge time sequence stability, distinguishes between stable discharge and random discharge types, and reduces the probability of subsequent false warnings and missed warnings.

[0014] (3) This invention analyzes the discharge intensity level based on the pulse repetition rate and amplitude sum of squares within a set historical time window, constructs a set of candidate state change vectors based on each historical discharge intensity level, introduces a basic minimum probability to correct the statistically obtained initial occurrence probability, analyzes the predicted occurrence probability of each discharge intensity level in the future window, improves the scientific nature of the level classification, realizes the probabilistic prediction of the discharge intensity level, and improves the prediction robustness.

[0015] (4) Based on the discharge timing characteristic type, the present invention corrects the predicted probability of each discharge intensity level and predicts the discharge intensity level of the future window. When the discharge intensity level reaches the level that requires an early warning, the discharge area is identified based on the location of the corresponding acquisition device and the signal strength and a corresponding early warning is issued, thereby improving the accuracy of the final probability analysis and realizing the rapid location of the discharge fault area of ​​the switchgear. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the system module connections of the present invention;

[0018] Figure 2 This is a schematic flowchart of the method steps for predicting the occurrence probability of each discharge intensity level in the future window in this invention.

[0019] Figure 3 This is a flowchart illustrating the steps of the method for obtaining the final probability of occurrence in this invention. Detailed Implementation

[0020] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0021] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0022] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0023] This invention determines the authenticity of discharges in original signals based on the correlation coefficients of each modal component within a set historical time window. If a real discharge exists in an original signal, the entropy value of each pulse data of its dominant discharge component is obtained, the pulse stability index is analyzed, the discharge timing characteristic type is determined based on the recursive change rate, and the discharge intensity level is analyzed. A set of candidate state change vectors is constructed based on historical discharge intensity levels, the predicted probability of each discharge intensity level in the future window is analyzed, the predicted probability of occurrence is corrected based on the discharge timing characteristic type, and the discharge intensity level in the future window is predicted. When the discharge intensity level reaches the required warning level, the discharge area is identified based on the location of the corresponding acquisition device and the signal strength, and a corresponding warning is issued, thereby improving monitoring efficiency.

[0024] Please see Figure 1 As shown, this invention provides an online monitoring and early warning system for partial discharge in switchgear of a power distribution room, comprising: a discharge authenticity identification module, a characteristic type determination module, an intensity level analysis module, and an early warning module. The connection relationships between the modules are as follows: the discharge authenticity identification module is connected to the characteristic type determination module, and the intensity level analysis module is connected to both the characteristic type determination module and the early warning module.

[0025] In this embodiment, at least one ultra-high frequency sensor is deployed in each switch cabinet in the power distribution room, for example, installed on the side wall of the cable compartment.

[0026] The discharge authenticity identification module determines the authenticity of the original signal discharge.

[0027] Considering the complex electromagnetic interference in power distribution rooms, the original partial discharge signal contains a large amount of noise and interference components. Directly using a fixed threshold for denoising is insufficient to effectively distinguish between genuine discharge pulses and interference signals, easily leading to missed detections of genuine discharges or misjudgments of noise. Furthermore, a single-mode signal cannot fully reflect the characteristics of the discharge waveform. It is necessary to decompose the original signal into multiple mode components and perform correlation matching with a standard reference discharge pulse waveform to accurately identify genuine discharges from complex signals. Therefore, it is necessary to determine the authenticity of the discharge in the original signal based on the correlation coefficients between each mode component and the reference waveform sequence, thereby improving the reliability of genuine discharge signal identification.

[0028] Based on this, the specific implementation steps of the discharge authenticity identification module include: S11, decomposing each original signal within a set historical time window to obtain each modal component. In a specific embodiment of the present invention, Empirical Mode Decomposition (EMD) is used for decomposition. Specifically, firstly, the upper and lower envelopes of the original signal are fitted, the local mean is calculated, and the mean is subtracted from the original signal to obtain a component prototype. This prototype is iterated repeatedly until an intrinsic mode function component satisfies that the number of extrema and zeros is equal or at most differs by 1; the local mean of the upper and lower envelopes is 0, and this intrinsic mode function is stripped from the original signal. The above process is repeated for the remaining signal until the residual signal cannot be further decomposed. Empirical Mode Decomposition (EMD) and the fitted envelope are existing technologies, and will not be described in detail in this invention.

[0029] S12. Determine the authenticity of the discharge of the original signal based on the correlation coefficient of each modal component.

[0030] The specific content of the method for judging the authenticity of the discharge of the original signal includes: S121, extracting reference discharge pulse waveforms of various types from the background database of the switch cabinet partial discharge monitoring and early warning system, obtaining the amplitude of each acquisition point based on the acquisition frequency of each modal component, and forming a reference waveform sequence of various types.

[0031] The method for obtaining the reference discharge pulse waveform includes: acquiring signals from known historical discharge events of the switchgear; calculating the similarity between two discharge signals of the same discharge type; calculating the average similarity between each discharge signal and other signals; and recording the discharge signal with the highest average similarity to other signals as the reference discharge signal for the corresponding discharge type.

[0032] S122. Using the width of the reference discharge pulse waveform as the sliding window width, select each component segment on each modal component with a set step size, and obtain the amplitude of each acquisition point in each component segment to form a corresponding component segment sequence. Specifically, each modal component refers to each modal component within a set historical time window, where the set historical time window refers to the past 1 minute, and the implementer can also set other time lengths according to the actual situation. In addition, in a specific embodiment of the present invention, the set step size is 1 second. The length of each component segment sequence is the same as that of the reference waveform sequence.

[0033] S123. Calculate the Pierre correlation coefficient between each component segment sequence of each modal component and each type of reference waveform sequence, and record the maximum value as the correlation coefficient between the corresponding modal component and the reference waveform sequence.

[0034] The specific calculation method for the Peer correlation coefficient is as follows: .

[0035] in This represents the amplitude of the i-th acquisition point in the component segment sequence. This represents the average amplitude at each sampling point of the component segment sequence. This represents the amplitude of the i-th acquisition point in the reference waveform sequence. The mean amplitude of each acquisition point in the reference waveform sequence is represented by i = 1, 2, 3, ..., n. The formula for calculating the Pierre correlation coefficient is existing technology and will not be elaborated upon here.

[0036] S124. If the maximum value of the correlation coefficients of each modal component of a certain original signal is greater than the set correlation threshold, it is determined that the original signal has a real discharge, and the current discharge type is determined based on the type of the corresponding reference waveform; otherwise, it is determined that there is no real discharge.

[0037] In this embodiment, the correlation threshold can be set by obtaining the minimum similarity between two discharge signals according to the above-mentioned reference discharge pulse waveform acquisition method, and setting it as the correlation threshold. In this embodiment, it is set to 0.6, but the implementer can also set other specific values ​​according to the actual situation.

[0038] This invention decomposes each original signal within a set historical time window to obtain each modal component, and judges the authenticity of the discharge of the original signal based on the correlation coefficient of each modal component. It has strong anti-interference ability, is suitable for complex electromagnetic environments in power distribution rooms, and improves the reliability of real signal extraction.

[0039] The feature type determination module analyzes the pulse stability index to determine the discharge timing feature type.

[0040] Considering the dynamic fluctuations in the pulse amplitude and time interval of partial discharge, relying solely on single-window pulse characteristics cannot reflect the evolution of the discharge time series and makes it difficult to distinguish between stable and random discharge time series characteristics. Furthermore, the stability of the pulse amplitude and the regularity of the time interval jointly determine the stability of the discharge time series, requiring joint quantification using the amplitude stability coefficient and permutation entropy. In addition, the long-term evolution trend of the discharge time series needs to be characterized by a recursive rate of change to provide reliable characteristic data for subsequent predictions. Therefore, it is necessary to calculate the pulse stability index and determine the discharge time series characteristic type based on the recursive rate of change of the historical stability index.

[0041] Based on this, the specific implementation steps of the feature type determination module are as follows: S21. If a certain original signal has a real discharge, then obtain the entropy value of each pulse data of its dominant discharge component and analyze the pulse stability index. The specific implementation steps include: S211. Record the mode component corresponding to the maximum correlation coefficient in the original signal as the dominant mode component, obtain the amplitude of each pulse in the dominant mode component, calculate the average value and standard deviation of each pulse amplitude, and record the ratio of the average value to the standard deviation as the amplitude stability coefficient.

[0042] S212. Obtain the occurrence time points of each pulse in the dominant mode component, calculate the time interval between adjacent pulses based on the occurrence time points of each pulse, construct an interval sequence, and calculate the permutation entropy of the interval sequence. In a specific embodiment of the present invention, when calculating the permutation entropy, the embedding dimension m=4 and the delay time τ=1 are set. The phase space of the interval sequence is reconstructed according to these parameters, and then the permutation probability distribution of each reconstructed component is calculated to finally obtain the permutation entropy value. The implementer can also adjust the above parameters appropriately according to the signal complexity. The calculation of the permutation entropy of the sequence is a prior art known in the art and will not be described in detail in the present invention. In addition, the range of the permutation entropy is [0, ln(m!)].

[0043] S213. Normalize the permutation entropy and amplitude stability coefficient respectively, and record the product of the normalized amplitude stability coefficient and permutation entropy as the pulse stability index with a set historical time window.

[0044] Specifically, the ratio of the permutation entropy to its maximum value is denoted as the normalized permutation entropy, which is normalized based on the maximum and minimum values ​​of the amplitude stability coefficient within each historical time window using the min-max normalization method.

[0045] S22. Obtain the recursive change rate of the historical stability index to determine the discharge timing characteristic type. The specific implementation steps are as follows: S221. Obtain the pulse stability index for each historical time window within a set historical time period, construct a pulse stability index sequence, and calculate the corresponding stability index change rate based on adjacent pulse stability indices to obtain the recursive change rate of the historical stability index. In a specific embodiment of this invention, the historical time period is set to the last hour; however, the implementer can set other specific values ​​according to actual conditions.

[0046] S222. Calculate the average and standard deviation of the recursive rate of change of the historical stability index, and record the ratio of the average to the standard deviation as the discharge characteristic coefficient.

[0047] S223. If the discharge characteristic coefficient is greater than the set characteristic coefficient threshold, the current discharge time sequence characteristic type is determined to be a stable discharge type; otherwise, it is determined to be a random discharge type. The larger the discharge characteristic coefficient, the more stable the recursive change rate of its historical stability index. In this embodiment, based on the original signals of each historical time window without partial discharge, the discharge characteristic coefficient of the original signal is calculated, and the average value and standard deviation of the discharge characteristic coefficient of the original signal of each historical time window are obtained. The sum of the average value and three times the standard deviation is recorded as the characteristic coefficient threshold.

[0048] This invention obtains the entropy value of each pulse data of the dominant component of the original signal with real discharge, analyzes the pulse stability index, obtains the recursive change rate of the historical stability index to determine the discharge time sequence characteristic type, realizes quantitative analysis of discharge time sequence stability, distinguishes between stable discharge and random discharge types, and reduces the probability of subsequent false alarms and missed alarms.

[0049] The intensity level analysis module analyzes the discharge intensity level, constructs a set of candidate state change vectors, and analyzes the predicted probability of each discharge intensity level occurring in the future window.

[0050] Considering that the severity of discharge hazards is related not only to pulse amplitude but also to the number of pulses per unit time, a single indicator cannot comprehensively characterize the discharge intensity. Furthermore, directly relying on statistical probabilities based on historical state changes can result in a zero probability of a discharge intensity level not occurring, which contradicts actual discharge development patterns and can easily lead to prediction bias. Therefore, it is necessary to establish a baseline probability guarantee for any unoccurred discharge intensity levels to ensure prediction completeness and robustness. Thus, it is necessary to combine pulse repetition rate and amplitude sum of squares to classify discharge intensity levels, and calculate the predicted probability of each discharge intensity level occurring in the future window based on historical state change vectors and the baseline probability guarantee mechanism.

[0051] Based on this, the specific content of the intensity level analysis module includes: S31, analyzing the discharge intensity level based on the pulse repetition rate and amplitude sum of squares within a set historical time window. Its specific implementation steps include: S311, obtaining the number of pulses within the set historical time window, and recording the ratio of this number to the duration of the set historical time window as the pulse repetition rate.

[0052] S312. Obtain the sum of the squares of the amplitudes of each pulse and record it as the amplitude sum of squares. The ratio of this sum to the maximum historical amplitude sum of squares is recorded as the discharge level. The maximum historical amplitude sum of squares is, for example, the maximum value among the amplitude sums of squares of each historical time window within the past 24 hours.

[0053] S313. The product of the discharge level and the pulse repetition rate is recorded as the discharge intensity. The discharge intensity level of the set historical time window is determined according to the preset discharge intensity range corresponding to each discharge intensity level. In a specific embodiment of the present invention, the discharge intensity range corresponding to each discharge intensity level is as follows: [0, 0.2] is low level, (0.2, 0.6] is medium level, and (0.6, 1] is high level. The implementer may also set other types of discharge intensity levels and corresponding discharge intensity ranges according to the actual situation.

[0054] S32. Construct a set of candidate state change vectors based on historical discharge intensity levels. The specific implementation steps include: S321. Obtain the discharge intensity level of each historical time window within a set historical time period, construct a discharge intensity level sequence in chronological order, and sequentially obtain the state change vectors using the level of each historical time window in the sequence as the starting state and the discharge intensity level of adjacent historical time windows as the final state. For example, if the discharge intensity level sequence is [Medium, Low, High, High, Medium, High, Medium], then the state change vectors are (Medium, Low), (Low, High), (High, High), (High, Medium), (Medium, High), (High, Medium).

[0055] S322. Select the state change vectors from each state change vector that correspond to the same initial state and discharge intensity level as the set historical time window, and form a set of candidate state change vectors. For example, if the discharge intensity level of the set historical time window is high, then select (high, high), (high, medium), and (high, medium) to form a set of candidate state change vectors.

[0056] S33. Analyze the predicted probability of occurrence of each discharge intensity level within the future window. For example... Figure 2 As shown, the specific implementation steps are as follows: S331, based on the discharge intensity level corresponding to the final state of each state change vector in the candidate state change vector set, count the number of each discharge intensity level. In a preferred embodiment of the present invention, for example, the number of discharge intensity levels corresponding to the final state in the candidate state change vector set that are medium is 2, and the number that are high is 1.

[0057] S332. The ratio of the number of each discharge intensity level to the total number of candidate state change vectors is denoted as the initial occurrence probability of the corresponding discharge intensity level.

[0058] S333. Determine whether the discharge intensity level corresponding to the final state includes all discharge intensity levels. If it includes all discharge intensity levels, then record the initial occurrence probability of each discharge intensity level as the predicted occurrence probability of each discharge intensity level in the future window.

[0059] S334. If not all discharge intensity levels are included, the excluded discharge intensity levels are recorded as other discharge intensity levels, and the predicted occurrence probability of other discharge intensity levels in the future window is recorded as the preset basic minimum probability. For example, if the initial occurrence probability of high-level discharge intensity levels corresponding to the final state is 70%, the initial occurrence probability of medium-level discharge intensity levels is 30%, and low-level discharge intensity levels are not included, then low-level discharge intensity levels are recorded as other discharge intensity levels. It is considered that when the discharge intensity level corresponding to the final state in the candidate state change vector set does not include all discharge intensity levels, directly recording the initial occurrence probability of the missing level as zero will result in the predicted occurrence probability of that level in the future window being zero. However, the lack of historical statistics does not mean that the actual probability of future occurrence is zero, which may lead to incorrect discharge intensity level prediction and missed warnings. In this embodiment, the basic minimum probability is set to 20%. Implementers can also set other values ​​according to specific circumstances, but it should be less than 1 / k, where k is the number of discharge intensity level types.

[0060] S335. Obtain the remaining probability after deducting the basic minimum probability. The product of the initial occurrence probability and the remaining probability for each discharge intensity level is recorded as the predicted occurrence probability for each discharge intensity level in the future window. The remaining probability refers to 1 minus the sum of the basic minimum probabilities for other discharge intensity levels, for example, 80% in this embodiment. Therefore, the predicted occurrence probabilities for each discharge intensity level are: low level: 20%, medium level: 24%, high level: 56%.

[0061] This invention analyzes discharge intensity levels based on pulse repetition rate and amplitude sum of squares within a set historical time window, constructs a set of candidate state change vectors based on each historical discharge intensity level, introduces a baseline probability to correct the statistically obtained initial occurrence probability, and analyzes the predicted occurrence probability of each discharge intensity level in the future window. This improves the scientific nature of level classification, realizes probabilistic prediction of discharge intensity levels, and enhances prediction robustness.

[0062] The early warning module predicts the discharge intensity level of the future window, identifies the discharge area, and issues corresponding warnings.

[0063] Considering the significant differences in the development patterns of stable and random discharges, directly using a uniform prediction probability will lead to low prediction accuracy in random discharge scenarios. Therefore, it is necessary to adaptively adjust the prediction probability based on the discharge time-series characteristics. Furthermore, simply issuing an alert when a warning level is reached is insufficient for operational needs; precise location of the discharge area requires combining signal strength and location information from multiple acquisition devices. Therefore, it is necessary to adjust the prediction probability based on the discharge time-series characteristics to determine the future discharge intensity level, and then identify the discharge area and issue a corresponding alert based on the location and signal strength of the acquisition devices when the warning threshold is reached.

[0064] The specific content of the aforementioned early warning module includes: S41, correcting the predicted occurrence probability of each discharge intensity level based on the discharge timing characteristic type and predicting the discharge intensity level of the future window. Its specific implementation steps include: S411, based on the predicted occurrence probability of each discharge intensity level in the future window, correcting the predicted occurrence probability in conjunction with the discharge timing characteristic type to obtain the final occurrence probability of each discharge intensity level in the future window. For example... Figure 3 As shown, the specific implementation steps are as follows: W1. Obtain the predicted occurrence probability of each discharge intensity level in the future window. If the discharge timing characteristic type is a stable discharge type, then the predicted occurrence probability of each discharge intensity level is recorded as the final occurrence probability.

[0065] W2. If the discharge timing characteristic type is a random discharge type, then the equal probability of each discharge intensity level is obtained based on the number of discharge intensity level types. For example, in this embodiment, the number of discharge intensity level types is 3, so the equal probability is 1 / 3 = 33.3%.

[0066] W3. Based on the discharge characteristic coefficient, set the weight coefficient for the predicted occurrence probability, and sum the predicted occurrence probability of each discharge intensity level corresponding to the future window with the equal probability to obtain the final occurrence probability after correction for the corresponding intensity level.

[0067] The method for setting the weighting coefficients of the predicted probability is as follows: .

[0068] in The weighting coefficients represent the probability of the predicted occurrence. This represents the discharge characteristic coefficient. This formula references the sigmoid function, so that the larger the discharge characteristic coefficient is, the closer the weight coefficient of the predicted probability is to 1, and the smaller the discharge characteristic coefficient is, the closer the weight coefficient of the predicted probability is to 0.

[0069] Furthermore, the sum of the weighting coefficient for the predicted probability and the weighting coefficient for equal probability is 1, that is, the weighting coefficient for equal probability is... .

[0070] S412. The discharge intensity level corresponding to the maximum probability of eventual occurrence is taken as the discharge intensity level of the future window.

[0071] S42. When the discharge intensity level reaches a point requiring an early warning, the discharge area is identified based on the location of the corresponding acquisition device and the signal strength, and a corresponding early warning is issued. In this embodiment, when the discharge intensity level reaches the medium level, it is determined that an early warning needs to be issued.

[0072] It should be explained that the method for identifying the discharge region includes: S421, based on the dominant mode components of each original signal, filtering from the dominant mode components corresponding to the original components acquired by all acquisition devices to see if there are acquisition devices with similar waveform signals; if so, they are recorded as associated devices. It should be explained that the method for obtaining acquisition devices with similar waveform signals is as follows: calculating the Pearson correlation coefficient between the dominant mode components of each acquisition device, and determining devices with correlation coefficients greater than a set similarity threshold as having similar waveform signals. In this embodiment, the similarity threshold is exemplarily set to 0.8, but implementers can also set other specific values ​​according to actual conditions.

[0073] S422. If there are associated devices for a certain original signal acquisition device, form an associated device group with the acquisition device corresponding to the original signal, and obtain the signal strength and installation position of the dominant mode component of each acquisition device in the associated device group.

[0074] S423. Based on the preset propagation distance corresponding to the signal strength, construct the expected discharge location range with the installation location of each associated device in the associated device group as the center and the preset corresponding propagation distance as the radius. The intersection of each discharge location range is used as the discharge location. Specifically, the device with the strongest signal strength in the group is selected as the reference device. According to the logarithmic distance path loss model, the distance ratio between any i-th device and the reference device satisfies: .

[0075] in For the signal strength of the reference device, This represents the signal strength of the i-th device. Here, n is the path loss exponent, which is taken as n=2.5 in this embodiment for a power distribution room environment. This ratio is independent of the absolute strength of the transmitting source. If the propagation distance of the preset reference device is R, then the propagation distance of the i-th device is... The position coordinates of each associated device are used to establish a sphere equation and solve for the intersection point. Here, R is the unknown quantity to be solved. The installation position coordinates of each associated device are taken as the center, and... Establish a system of equations for the sphere with respect to the radius, and solve for the coordinates of the intersection point and the unknown R.

[0076] S424. If there is no associated device for a certain original signal acquisition device, then the discharge area is determined with a preset length as the radius and the original signal acquisition device as the center. The preset length can be determined based on the maximum value of the original signal strength and the original signal strength of the device based on signal attenuation.

[0077] This invention corrects the predicted probability of each discharge intensity level based on the discharge timing characteristic type and predicts the discharge intensity level of the future window. When the discharge intensity level reaches the level that requires an early warning, the discharge area is identified based on the location of the corresponding acquisition device and the signal strength, and a corresponding early warning is issued, thereby improving the accuracy of the final probability analysis and enabling rapid location of the discharge fault area of ​​the switchgear.

[0078] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0079] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0080] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0082] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A partial discharge online monitoring and early warning system for switchgear in a power distribution room, characterized in that, include: The discharge authenticity identification module decomposes each original signal within a set historical time window to obtain each mode component, and judges the authenticity of the discharge of the original signal based on the correlation coefficient of each mode component. The feature type determination module, if a certain original signal has a real discharge, obtains the entropy value of each pulse data of its dominant discharge component, analyzes the pulse stability index, and obtains the recursive change rate of the historical stability index to determine the discharge time sequence feature type. The intensity level analysis module analyzes the discharge intensity level based on the pulse repetition rate and amplitude sum of squares within a set historical time window. It constructs a set of candidate state change vectors based on each historical discharge intensity level and analyzes the predicted probability of each discharge intensity level occurring in the future window. The early warning module corrects the predicted probability of each discharge intensity level based on the discharge timing characteristic type and predicts the discharge intensity level of the future window. When the discharge intensity level reaches the point where an early warning needs to be issued, it identifies the discharge area based on the location of the corresponding acquisition device and the signal strength and issues a corresponding early warning. The specific content of the method for determining the authenticity of the discharge of the original signal includes: The reference discharge pulse waveforms of various types are extracted from the background database of the switch cabinet partial discharge monitoring and early warning system. The amplitude of each acquisition point is obtained based on the acquisition frequency of each modal component, and a series of reference waveforms of various types are formed. Using the width of the reference discharge pulse waveform as the width of the sliding window, each component segment is selected by sliding with a set step size on each modal component, and the amplitude of each acquisition point in each component segment is obtained to form the corresponding component segment sequence. Calculate the Pierre correlation coefficient between each component segment sequence of each modal component and each type of reference waveform sequence, and record the maximum value as the correlation coefficient between the corresponding modal component and the reference waveform sequence; If the maximum value of the correlation coefficients of each modal component of a certain original signal is greater than the set correlation threshold, it is determined that the original signal has a real discharge, and the current discharge type is determined based on the type of the corresponding reference waveform; otherwise, it is determined that there is no real discharge. The analysis method for the pulse stability index includes: The modal component corresponding to the maximum correlation coefficient in the original signal is denoted as the dominant modal component. The amplitude of each pulse in the dominant modal component is obtained, and the average value and standard deviation of the amplitude of each pulse are calculated. The ratio of the average value to the standard deviation is denoted as the amplitude stability coefficient. Obtain the occurrence time of each pulse in the dominant mode component, calculate the time interval between adjacent pulses based on the occurrence time of each pulse, construct an interval sequence, and calculate the permutation entropy of the interval sequence; The permutation entropy and amplitude stability coefficient are normalized respectively, and the product of the normalized amplitude stability coefficient and permutation entropy is denoted as the pulse stability index with a set historical time window. The analysis method for the discharge intensity level includes: Obtain the number of pulses within a set historical time window, and record the ratio of the number of pulses to the duration of the set historical time window as the pulse repetition rate; The sum of the squares of the amplitudes of each pulse is recorded as the sum of squares of amplitudes, and the ratio of this sum to the maximum historical sum of squares of amplitudes is recorded as the degree of discharge. The product of the discharge level and the pulse repetition rate is recorded as the discharge intensity. The discharge intensity level of the set historical time window is determined according to the preset discharge intensity range corresponding to each discharge intensity level. The method for constructing the set of candidate state change vectors includes: Obtain the discharge intensity level of each historical time window within a set historical time period, construct a discharge intensity level sequence in chronological order, and obtain the change vector of each state by taking the level of each historical time window in the sequence as the starting state and the discharge intensity level of each adjacent historical time window as the final state. From each state change vector, select those state change vectors that correspond to the same initial state and discharge intensity level as the set historical time window, and form a set of candidate state change vectors.

2. The partial discharge online monitoring and early warning system for switchgear in a power distribution room according to claim 1, characterized in that, The current method for determining the discharge timing characteristic type includes: Obtain the pulse stability index for each historical time window within a set historical time period, construct a pulse stability index sequence, and calculate the rate of change of the corresponding stability index based on the adjacent pulse stability index to obtain the recursive rate of change of the historical stability index. Calculate the average and standard deviation of the recursive rate of change of the historical stability index, and record the ratio of the average to the standard deviation as the discharge characteristic coefficient. If the discharge characteristic coefficient is greater than the set characteristic coefficient threshold, the current discharge timing characteristic type is determined to be a stable discharge type; otherwise, it is determined to be a random discharge type.

3. The online monitoring and early warning system for partial discharge of switchgear in a power distribution room according to claim 1, characterized in that, The analysis method for predicting the occurrence probability of each discharge intensity level in the future window includes: Based on the discharge intensity level corresponding to the final state of each state change vector in the set of candidate state change vectors, the number of each discharge intensity level is counted. The ratio of the number of each discharge intensity level to the total number of candidate state change vectors is denoted as the initial probability of occurrence of the corresponding discharge intensity level. Determine whether the discharge intensity level corresponding to the final state includes all discharge intensity levels. If it includes all discharge intensity levels, then record the initial occurrence probability of each discharge intensity level as the predicted occurrence probability of each discharge intensity level in the future window. If not all discharge intensity levels are included, the discharge intensity levels not included are recorded as other discharge intensity levels, and the predicted probability of other discharge intensity levels in the future window is recorded as the preset basic minimum probability. Obtain the remaining probability after deducting the basic minimum probability, and record the product of the initial probability of occurrence of each discharge intensity level and the remaining probability as the predicted probability of occurrence of each discharge intensity level in the future window.

4. The online monitoring and early warning system for partial discharge of switchgear in a power distribution room according to claim 1, characterized in that, The analysis method for the discharge intensity level of the future window includes: Based on the predicted probability of occurrence of each discharge intensity level in the future window, the predicted probability of occurrence is corrected by combining the discharge time series feature type, so as to obtain the final probability of occurrence of each discharge intensity level in the future window. The discharge intensity level corresponding to the maximum probability of eventual occurrence is used as the discharge intensity level for the future window.

5. The partial discharge online monitoring and early warning system for switchgear in a power distribution room according to claim 4, characterized in that, The method for obtaining the final occurrence probability includes: Obtain the predicted probability of occurrence for each discharge intensity level in the future window. If the discharge timing characteristic type is a stable discharge type, then the predicted probability of occurrence for each discharge intensity level is recorded as the final probability of occurrence. If the discharge timing characteristic type is a random discharge type, then the equal probability of each discharge intensity level is obtained based on the number of discharge intensity level types. The weighting coefficients for the predicted occurrence probability are set based on the discharge characteristic coefficients. The predicted occurrence probability of each discharge intensity level in the future window is weighted and summed with the equal probability to obtain the final occurrence probability after correction for the corresponding intensity level.

6. The online monitoring and early warning system for partial discharge of switchgear in a power distribution room according to claim 1, characterized in that, The method for identifying the discharge region includes: Based on the dominant mode components of each original signal, from the dominant mode components corresponding to the original components acquired by all acquisition devices, we screen for acquisition devices with similar waveform signals. If they exist, they are recorded as associated devices. If a certain original signal has associated devices, it is combined with the original signal acquisition device to form an associated device group, and the signal strength and installation position of the dominant mode component of each acquisition device in the associated device group are obtained. Based on the signal strength, a corresponding propagation distance is preset. The installation location of each associated device in the associated device group is taken as the center and the preset corresponding propagation distance is taken as the radius to construct the expected discharge location range. The intersection of each discharge location range is obtained by solving the problem. If there is no associated device for a certain original signal acquisition device, the discharge area is determined with the original signal acquisition device as the center and a preset length as the radius.

Images