Insulator pollution flashover dynamic monitoring and early warning system

Abstract

The embodiment of the invention discloses an insulator pollution flashover dynamic monitoring and early warning system, and relates to a power equipment state monitoring technology, and the system comprises a data collection module which is used for collecting the multi-source monitoring data of the operation state of an insulator, and the multi-source monitoring data comprises an electrical parameter, an environmental parameter and a pollution parameter; the feature processing module is used for performing time-frequency analysis and coupling analysis on the multi-source monitoring data to obtain data after feature processing; the mixed kernel function support vector machine analysis module comprises a kernel function calculation engine, a dynamic weight adjustment unit and a model optimizer, the kernel function calculation engine is used for supporting parallel calculation of a Gaussian kernel function and a polynomial kernel function, the dynamic weight adjustment unit is used for adaptive adjustment based on feature importance, and the model optimizer is used for parameter optimization; the early warning output module is used for outputting pollution flashover early warning information and maintenance decision suggestions according to the analysis result; the problems that a common monitoring system is poor in adaptability to complex environments and low in early warning accuracy can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment state monitoring, and particularly relates to an insulator pollution flashover dynamic monitoring and early warning system. BACKGROUND

[0002] In a power system, insulator pollution flashover is a serious fault that can cause large-scale power outages and equipment damage. In order to effectively prevent and reduce the occurrence of such faults, it is crucial to monitor the operating state of the insulator in real time and to make accurate early warnings.

[0003] Current insulator pollution flashover monitoring systems usually use a single monitoring method, such as periodic manual inspection, single environmental parameter monitoring, or simple leakage current monitoring. In such systems, detecting the leakage current amplitude of the insulator surface is susceptible to environmental noise interference, resulting in a high false positive rate; acoustic and infrared detection are suitable for late-stage warning of partial discharge that has already occurred, and the system prediction is not real-time. In addition, the existing systems lack effective coordination between modules, resulting in low data transmission and processing efficiency, making it difficult to achieve fast and accurate early warning.

[0004] Therefore, there is an urgent need for a new insulator pollution flashover dynamic monitoring and early warning system that can integrate multiple module functions to meet the real-time, accuracy, and comprehensiveness requirements of insulator pollution flashover monitoring. SUMMARY

[0005] The main purpose of the present application is to provide an insulator pollution flashover dynamic monitoring and early warning system to solve the problem of poor adaptability to complex environments and low early warning accuracy of traditional monitoring systems.

[0006] To achieve the above-mentioned purpose, the present application provides an insulator pollution flashover dynamic monitoring and early warning system in the first aspect, comprising a data acquisition module, a feature processing module, a hybrid kernel function support vector machine analysis module, and an early warning output module, wherein: The data acquisition module is used to acquire multi-source monitoring data of the operating state of the insulator in real time, and the multi-source monitoring data includes electrical parameters, environmental parameters, and contamination parameters. The feature processing module is used to perform time-frequency analysis and coupling analysis on the multi-source monitoring data to obtain data after feature processing. The mixed kernel function support vector machine analysis module is configured to construct a mixed kernel function support vector machine model, the mixed kernel function support vector machine model adopts a mixed kernel function of a Gaussian kernel and a polynomial kernel, and dynamically optimizes a kernel function proportion through an adaptive weight adjustment mechanism; wherein the mixed kernel function support vector machine analysis module comprises a kernel function calculation engine, a dynamic weight adjustment unit, and a model optimizer; the kernel function calculation engine is configured to support parallel calculation of a Gaussian kernel function and a polynomial kernel function; the dynamic weight adjustment unit is configured to be adaptively adjusted based on feature importance; and the model optimizer is configured to optimize parameters. The early warning output module is configured to output pollution flashover early warning information and maintenance decision suggestions according to the analysis result.

[0007] Optionally, the feature processing module comprises a time-frequency analysis unit and a coupling analysis unit. The time-frequency analysis unit is configured to implement parallel fast Fourier transform processing based on an FPGA. The coupling analysis unit is configured to calculate an environment-pollution dynamic correlation matrix.

[0008] Optionally, the time-frequency analysis unit is implemented by using an Xilinx Zynq-7020 SoC.

[0009] Optionally, the early warning output module comprises: A hierarchical early warning indicator configured to display multi-level early warning information. A maintenance decision generator configured to construct a disposal suggestion library based on case reasoning; the maintenance decision generator comprises a knowledge graph engine configured to store historical pollution flashover case data. A data communication interface configured to support a communication protocol.

[0010] Optionally, the data acquisition module comprises a leakage current monitoring device, a meteorological monitoring unit, and a pollution detection device. The leakage current monitoring device is configured to acquire electrical parameters, the electrical parameters comprising a leakage current time-domain waveform. The meteorological monitoring unit is configured to acquire environmental parameters, the environmental parameters comprising environmental temperature, relative humidity, dew point temperature, wind speed, precipitation intensity, and salt mist concentration. The pollution detection device is configured to acquire pollution parameters, the pollution parameters comprising salt density, ash density, and hydrophobicity grade.

[0011] Optionally, the leakage current monitoring device has a leakage current measurement range of 10 uA to 100 mA and a leakage current measurement accuracy of 1%.

[0012] Optionally, the dynamic weight adjustment unit comprises a temperature compensation circuit, and has a working temperature range of -40°C to +85°C.

[0013] Optionally, the hybrid kernel function support vector machine analysis module supports OTA remote upgrade and supports dynamically changing kernel function combination methods.

[0014] Optionally, the form of the hybrid kernel function is: K_hybrid = α·K_RBF + (1-α)·K_poly wherein α∈(0,1) is a hybrid weight coefficient, K_RBF is a Gaussian kernel function: exp(-γ||x_i - x_j||²), K_poly is a polynomial kernel function: (x_i·x_j + C)^d, C∈[0.1,100] is a penalty factor, the Gaussian kernel function kernel parameter: γ=1 / (n_features·X.var()), the polynomial kernel degree d=3.

[0015] Optionally, the dynamic weight adjustment unit is specifically configured to implement the adaptive adjustment by using the following adjustment strategy: α = 0.5 + 0.2·tanh(β·(f_e - f_p)) wherein α is the hybrid weight coefficient, f_e is an environmental feature importance score, f_p is a pollution feature importance score, β is an adjustment sensitivity coefficient, the tanh function is a smooth S-shaped nonlinear conversion, and the output range is (-1, 1).

[0016] The application provides an insulator pollution flashover dynamic monitoring and early warning system, comprising a data acquisition module, a feature processing module, a hybrid kernel function support vector machine analysis module, and an early warning output module, wherein: the data acquisition module is configured to acquire multi-source monitoring data of an insulator operating state in real time, the multi-source monitoring data comprising electrical parameters, environmental parameters, and pollution parameters; the feature processing module is configured to perform time-frequency analysis and coupling analysis on the multi-source monitoring data to obtain data after feature processing; the hybrid kernel function support vector machine analysis module is configured to construct a hybrid kernel function support vector machine model, the hybrid kernel function support vector machine model adopting a hybrid kernel function of a Gaussian kernel and a polynomial kernel and dynamically optimizing the kernel function proportion through an adaptive weight adjustment mechanism; wherein: the hybrid kernel function support vector machine analysis module comprises a kernel function calculation engine, a dynamic weight adjustment unit, and a model optimizer; the kernel function calculation engine is configured to support parallel calculation of a Gaussian kernel function and a polynomial kernel function; the dynamic weight adjustment unit is configured to perform adaptive adjustment based on feature importance; the model optimizer is configured to perform parameter optimization; and the early warning output module is configured to output pollution flashover early warning information and maintenance decision suggestions according to the analysis result; the beneficial effects are specifically embodied in: Through the integration of multiple functional modules, efficient collaborative work between modules is realized, and the overall performance and reliability of the system are improved. Through the collaborative work of multiple modules, the system can automatically monitor the running state of the insulator in real time and accurately throughout the whole process, and timely output early warning information and maintenance decision suggestions; In addition, in the system, a hybrid kernel function support vector machine model is adopted, which combines the advantages of Gaussian kernel function and polynomial kernel function, captures local and global features of data, and significantly improves the prediction accuracy of the model for insulator pollution flashover risk. Through the online updating mechanism, the system dynamically adjusts the model parameters according to the environmental changes, ensuring that the model always maintains high adaptability and stability during long-term operation. This dynamic optimization mechanism not only improves the reliability of the system, but also reduces false positives caused by environmental changes, enhances the practicality of the system, and provides a strong guarantee for the safe operation of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Among them: Figure 1 A structure diagram of an insulator pollution flashover dynamic monitoring and early warning system provided by an embodiment of the present application; Figure 2 A flowchart of an insulator pollution flashover dynamic monitoring and early warning method provided by an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make those skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0020] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0022] The embodiments of this application are described below with reference to the accompanying drawings.

[0023] Figure 1 This is a schematic diagram of the structure of an insulator flashover dynamic monitoring and early warning system provided in an embodiment of this application. Figure 1 As shown, the insulator flashover dynamic monitoring and early warning system 100 includes a data acquisition module 110, a feature processing module 120, a hybrid kernel function support vector machine analysis module 130, and an early warning output module 140; wherein: The aforementioned data acquisition module 110 is used to acquire multi-source monitoring data of the insulator's operating status in real time. The multi-source monitoring data includes electrical parameters, environmental parameters, and pollution parameters. The aforementioned feature processing module 120 is used to perform time-frequency analysis and coupling analysis on the aforementioned multi-source monitoring data to obtain the feature-processed data; The aforementioned hybrid kernel function support vector machine analysis module 130 is used to construct a hybrid kernel function support vector machine model. This model employs a hybrid kernel function combining Gaussian and polynomial kernels, and dynamically optimizes the kernel function ratio through an adaptive weight adjustment mechanism. The module 130 includes a kernel function calculation engine 131, a dynamic weight adjustment unit 132, and a model optimizer 133. The kernel function calculation engine 131 supports parallel computation of Gaussian and polynomial kernel functions. The dynamic weight adjustment unit 132 is used for adaptive adjustment based on feature importance. The model optimizer 133 is used for parameter optimization. The aforementioned early warning output module 140 is used to output pollution flashover early warning information and maintenance decision suggestions based on the above analysis results.

[0024] The insulator pollution flashover dynamic monitoring and early warning system 100 in the embodiment of the application can collect multi-source monitoring data of the running state of the insulator in real time, including electrical parameters, environmental parameters and pollution parameters; perform feature processing on the multi-source monitoring data, construct a space-time feature matrix through time-frequency analysis and coupling analysis, and calculate an environment-pollution dynamic coupling factor; adopt a hybrid kernel function support vector machine model of a Gaussian kernel and a polynomial kernel, dynamically optimize the proportion of the kernel function through an adaptive weight adjustment mechanism, and realize multi-level pollution flashover early warning based on the model output. The system can accelerate feature extraction by using FPGA, and integrates anti-interference design and self-diagnosis functions. The system can solve the problems of poor adaptability to complex environments and low early warning accuracy of traditional monitoring methods.

[0025] Optionally, the data acquisition module 110 includes a leakage current monitoring device 111, a meteorological monitoring unit 112 and a pollution detection device 113. The leakage current monitoring device 111 is configured to collect electrical parameters, and the electrical parameters include a leakage current time-domain waveform. The meteorological monitoring unit 112 is configured to collect environmental parameters, and the environmental parameters include environmental temperature, relative humidity, dew point temperature, wind speed, precipitation intensity and salt mist concentration. The pollution detection device 113 is configured to collect pollution parameters, and the pollution parameters include salt density, ash density and hydrophobicity grade.

[0026] Further optionally, the leakage current monitoring device has a leakage current measurement range of 10 uA to 100 mA and a leakage current measurement accuracy of 1%.

[0027] In an optional embodiment, the hybrid kernel function has the following form: K_hybrid = α·K_RBF + (1-α)·K_poly wherein a is a hybrid weight coefficient in the range of 0 to 1, K_RBF is a Gaussian kernel function: exp(-γ||x_i - x_j||²), K_poly is a polynomial kernel function: (x_i·x_j + C)^d, C is a penalty factor in the range of 0.1 to 100, the Gaussian kernel function kernel parameter is: γ=1 / (n_features·X.var()), and the polynomial kernel degree d=3. Optionally, the dynamic weight adjustment unit 132 is specifically configured to realize the adaptive adjustment by using the following adjustment strategy: α = 0.5 + 0.2·tanh(β·(f_e - f_p)) Wherein, a is the above-mentioned mixing weight coefficient, f e is the environmental feature importance score, f p is the contamination feature importance score, β is the adjustment sensitivity coefficient, the tanh function is a smooth S-shaped nonlinear conversion, and the output range is (-1, 1).

[0028] In an optional embodiment, the dynamic weight adjustment unit 132 comprises a temperature compensation circuit, and the working temperature range is -40℃ to +85℃.

[0029] The hybrid kernel function support vector machine analysis module 130 in the embodiment of the application can support OTA remote upgrading and support dynamic replacement of kernel function combination modes.

[0030] Optionally, the feature processing module 120 comprises a time-frequency analysis unit and a coupling analysis unit. The time-frequency analysis unit is configured to implement parallel fast Fourier transform processing based on FPGA. The coupling analysis unit is configured to calculate an environment-contamination dynamic correlation matrix.

[0031] In an optional embodiment, the early warning output module 140 comprises: The hierarchical early warning indicator 141 is configured to display multi-level early warning information. The maintenance decision generator 142 is configured to construct a case-based reasoning-based treatment suggestion library.The maintenance decision generator comprises a knowledge graph engine configured to store historical pollution flashover case data. The data communication interface 143 is configured to support communication protocols.

[0032] Optionally, the early warning output module is integrated with an audible light alarm, and the sound pressure level is greater than or equal to 90 dB (@1 m).

[0033] Based on the description of the foregoing device embodiments, the embodiment of the application further provides an insulator pollution flashover dynamic monitoring and early warning method, which can be implemented based on an insulator pollution flashover dynamic monitoring and early warning system 100 as shown in the following. Figure 1

[0034] Figure 2 A flowchart of an insulator pollution flashover dynamic monitoring and early warning system provided by the embodiment of the application is shown in the following. Figure 2 The method comprises the following steps.201. Real-time collection of multi-source monitoring data of the running state of the insulator, wherein the multi-source monitoring data comprises electrical parameters, environmental parameters and contamination parameters.

[0035] The execution subject of the method in the embodiment of the application can be an insulator pollution flashover dynamic monitoring and early warning system, which can be implemented on an electronic device in actual application.

[0036] Specifically, the multi-source monitoring data of the insulator operating state can be collected in real time by the data collection module, i.e., periodically collecting various types of monitoring data of the insulator operating state. Optionally, the electrical parameters can be collected by the leakage current monitoring device, the environmental parameters can be collected by the weather monitoring unit, and the contamination parameters can be collected by the contamination detection device. The electrical parameters include leakage current time-domain waveform; the environmental parameters include environmental temperature, relative humidity, dew point temperature, wind speed, precipitation intensity, and salt mist concentration; and the contamination parameters include salt density, ash density, and hydrophobicity grade.

[0037] In an optional embodiment, the system can include a preprocessing module, and the method can further include a corresponding preprocessing step, such as an abnormal data cleaning step: An abnormal detection algorithm based on isolated forest is used to eliminate data points deviating from normal distribution by more than 3σ. The condition for determining abnormality is:

[0038] wherein: μ and σ is the mean and standard deviation of the abnormal score of the historical normal data; is the abnormal score of the new sample x j is the average path length of the sample in all isolation numbers, and c(N) is a normalization factor.

[0039] 202. Time-frequency analysis and coupling analysis are performed on the multi-source monitoring data to obtain data after feature processing.

[0040] For feature fusion of multi-source monitoring data, feature extraction is first needed.

[0041] In an optional embodiment, the spatio-temporal feature fusion processing includes time-domain feature extraction, frequency-domain feature extraction, and environment-contamination coupling feature calculation. Specifically, it can include: (1) Time-domain feature extraction: the time-frequency analysis unit can calculate the sliding window mean, peak-to-peak value, and waveform factor of the leakage current; (2) Frequency-domain feature extraction: the time-frequency analysis unit can obtain the energy distribution of the 0-1 kHz frequency band of the leakage current through fast Fourier transform; (3) Environment-contamination coupling feature calculation: the coupling analysis unit establishes a dynamic correlation matrix of temperature, humidity, salt density, and ash density (environment-contamination dynamic correlation matrix).

[0042] Further optionally, the calculation method of the dynamic correlation matrix is: ​​

[0043] where ESDD is salt density (mg / cm2), NSDD is non-soluble deposit density (mg / cm2), T is temperature (℃), RH is relative humidity (%), Δt is time difference (h), and λ is attenuation coefficient; kT is temperature influence weight factor:

[0044] where T is current temperature (℃), Tbase is base temperature, and τT is temperature sensitivity coefficient; kRH is humidity influence weight factor:

[0045] where RH is relative humidity (%), RH50 is critical humidity for pollution flashover, and ΔRH is transition interval width; kC is contamination influence factor:

[0046] where a is salt density weight, and b is non-linear coefficient of non-soluble deposit density.

[0047] 203、constructing a hybrid kernel support vector machine model, the hybrid kernel support vector machine model adopts a hybrid kernel function of a Gaussian kernel and a polynomial kernel, and dynamically optimizes a kernel function proportion through an adaptive weight adjustment mechanism.

[0048] The hybrid kernel support vector machine model in the embodiments of the present application adopts a hybrid kernel function of a Gaussian kernel and a polynomial kernel, and specifically can support parallel calculation of a Gaussian kernel function and a polynomial kernel function through a kernel function calculation engine; performs adaptive adjustment based on feature importance through a dynamic weight adjustment unit, and can perform parameter optimization through a model optimizer.

[0049] The Gaussian kernel (Gaussian Kernel) and the polynomial kernel (Polynomial Kernel) mentioned in the embodiments of the present application are two commonly used kernel functions in support vector machines (SVM). They each have different characteristics and applicable scenarios. The two kernel functions are combined together in the embodiments of the present application to form a hybrid kernel function, which can integrate their advantages and improve the flexibility and adaptability of the model.

[0050] In an optional implementation, the form of the hybrid kernel function is: K_hybrid = α·K_RBF + (1-α)·K_poly wherein a is a mixing weight coefficient in (0, 1), K RBF is a Gaussian kernel function: exp(-γ||x i - x j|| 2), K poly is a polynomial kernel function: (x i ·x j + C) d, C is a penalty factor in [0.1, 100], the Gaussian kernel function kernel parameter: γ = 1 / (n features·X.var()), and the polynomial kernel degree d = 3.

[0051] In an optional embodiment, the adjustment strategy in the dynamic weight adjustment unit is: a = 0.5 + 0.2·tanh(β·(f e - f p)) wherein a is a mixing weight coefficient, also referred to as a dynamic weight coefficient; f e is an environmental feature importance score, f p is a pollution feature importance score, β is an adjustment sensitivity coefficient, and tanh function is a smooth S-shaped nonlinear conversion with an output range of (-1, 1).

[0052] The mixing weight coefficient (a) in the embodiment of the application is an important parameter in the mixed kernel function support vector machine model, used to dynamically balance the contribution of the Gaussian kernel function and the polynomial kernel function. Through the dynamic weight adjustment unit, the model can better adapt to different data features and environmental changes, thereby improving the accuracy and reliability of the prediction.

[0053] Under different environmental conditions (such as different meteorological conditions, pollution levels, etc.), the feature importance of the data may change. The mixing weight coefficient a can automatically adjust according to these changes, so that the model can maintain high prediction performance under different conditions.

[0054] 204. Outputting a pollution flashover warning information and a maintenance decision suggestion according to the above analysis result.

[0055] Specifically, the warning output module can train the support vector machine model using the historical data set, divide different prediction levels according to the output value of the decision function, and trigger the pollution flashover warning according to the prediction level.

[0056] Optionally, the above implementation of the pollution flashover warning comprises: dividing into three levels of safe, attention, and dangerous according to the output value of the above mixed kernel function support vector machine model; issuing a warning information when the prediction is dangerous for consecutive N sampling periods, wherein N is an integer greater than 1.

[0057] For example, if N = 3, the prediction levels are divided into level I (safe), level II (attention), and level III (dangerous), and a warning information is issued when the prediction is level III for consecutive 3 sampling periods.

[0058] The early warning output module can display multi-level early warning information through a hierarchical early warning indicator. The maintenance decision generator can construct a case-based reasoning disposal suggestion library, which includes a knowledge graph engine for storing historical pollution flashover case data. The early warning output module can further include a data communication interface for supporting communication protocols such as the IEC61850 protocol.

[0059] The form of the early warning information is not limited.

[0060] In an optional embodiment, in the model construction and parameter optimization link, a differentiated penalty factor C can be set for different voltage grade insulators, for example: 110kV and below: C [1, 10]; 220kV: C [10, 50]; 500kV and above: C [50, 100].

[0061] The penalty factor C in the embodiments of the present application is used to control the complexity of the model and the degree of punishment for misclassification. By selecting an appropriate C value, the performance of the model can be optimized to adapt to the operating characteristics of different voltage grades, thereby improving the accuracy and reliability of the insulator pollution flashover dynamic monitoring and early warning.

[0062] Further optionally, the decision function output of the hybrid kernel function support vector machine in the embodiments of the present application is appended with an uncertainty estimate: uncertainty = 1 / (1 + exp(-|f(x)|)) The additional uncertainty estimate in the embodiments of the present application is used to evaluate the confidence of the model in the prediction result. This uncertainty estimate can help users better understand the prediction reliability of the model, so as to make more intelligent decisions in actual applications.

[0063] Correspondingly, the early warning output in the embodiments of the present application can output early warning signals according to the risk level and the uncertainty estimate value, and display early warning information of different levels through a hierarchical early warning indicator.

[0064] Optionally, the hybrid kernel function support vector machine model described above can also be updated online, specifically including: An incremental learning mechanism of a sliding time window is established, and model retraining is triggered when the environmental mutation index ΔE is greater than a threshold value; The calculation method of the above-mentioned environmental mutation index is: ΔE = ||E_t - E_{t-1}||_2 / σE Where E_t is the current environmental feature vector, and σE is the standard deviation of historical environmental data.

[0065] Specifically, the model can be updated online in this application, and the model parameters can be dynamically adjusted according to environmental changes to ensure the adaptability and accuracy of the model. Among them, a sliding time window (for example, the length of 7 days) incremental learning mechanism can be established, and the model retraining is triggered when the environmental mutation index ΔE is greater than the threshold value.

[0066] In an embodiment, the specific steps of online updating can be as follows: 1. Data acquisition and feature extraction: Real-time acquisition of multi-source monitoring data of insulator operating state, and feature extraction, construction of space-time feature matrix and environment-dirty dynamic coupling factor.

[0067] 2. Environmental mutation detection: Calculate the change between the current environmental feature vector E_t and the environmental feature vector E_{t-1} at the previous time point, and get the environmental mutation index ΔE: ΔE = ||E_t - E_{t-1}||_2 / σE Where σE is the standard deviation of historical environmental data.

[0068] 3. Trigger update mechanism: If the environmental mutation index ΔE exceeds the preset threshold value, trigger the online updating mechanism of the model. This indicates that the current environment has changed significantly, and the model parameters need to be updated to adapt to the new environmental conditions.

[0069] 4. Model retraining: Use the latest monitoring data and historical data to retrain the hybrid kernel function support vector machine model. In the retraining process, optimize the model parameters (such as penalty factor C, kernel function parameter γ and α, etc.) to ensure the adaptability of the model to the current data.

[0070] 5. Model updating and deployment: Deploy the updated model parameters to the monitoring and early warning system to replace the old model parameters. The updated model will be used for subsequent pollution flashover risk assessment and early warning.

[0071] The current pollution flashover disaster risk assessment method has imperfect dynamic response mechanism to sudden weather events, cannot adjust the risk assessment threshold in real time, and has insufficient prediction accuracy.

[0072] Through real-time monitoring, intelligent analysis and dynamic adjustment, the application embodiment provides an efficient, accurate and adaptable insulator pollution flashover dynamic monitoring and early warning system, which not only improves the accuracy and reliability of early warning, but also ensures the long-term stability of the system through the online updating mechanism, and provides a strong guarantee for the safe operation of the power system.

[0073] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment system can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer readable storage medium. When the program is executed, it can include the method steps in the above-mentioned embodiments.

[0074] The technical features of the above embodiments can be combined in any way. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, but as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the description.

[0075] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the patent scope of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An insulator pollution flashover dynamic monitoring and early warning system, characterized in that, The device comprises a data acquisition module, a feature processing module, a hybrid kernel function support vector machine analysis module and an early warning output module. The data acquisition module is configured to acquire multi-source monitoring data of the operating state of the insulator in real time, and the multi-source monitoring data comprises electrical parameters, environmental parameters and contamination parameters. The feature processing module is configured to perform time-frequency analysis and coupling analysis on the multi-source monitoring data to obtain processed data. The hybrid kernel function support vector machine analysis module is configured to construct a hybrid kernel function support vector machine model, wherein the model adopts a hybrid kernel function of a Gaussian kernel and a polynomial kernel and dynamically optimizes the proportion of the kernel function through an adaptive weight adjustment mechanism. The early warning output module is configured to output pollution flashover early warning information and maintenance decision suggestions based on the analysis result.

2. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The feature processing module comprises a time-frequency analysis unit and a coupling analysis unit. The time-frequency analysis unit is configured to perform parallel fast Fourier transform processing based on FPGA. The coupling analysis unit is configured to calculate an environment-contamination dynamic correlation matrix.

3. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 2, characterized in that, The time-frequency analysis unit is implemented by using an Xilinx Zynq-7020 SoC.

4. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The early warning output module comprises: a hierarchical early warning indicator configured to display multi-level early warning information; a maintenance decision generator configured to construct a case-based reasoning disposal suggestion library, wherein the maintenance decision generator comprises a knowledge graph engine configured to store historical pollution flashover case data; and a data communication interface configured to support a communication protocol.

5. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The data acquisition module comprises a leakage current monitoring device, a meteorological monitoring unit and a contamination detection device. The leakage current monitoring device is configured to acquire electrical parameters, and the electrical parameters comprise a leakage current time-domain waveform. The meteorological monitoring unit is configured to acquire environmental parameters, and the environmental parameters comprise environmental temperature, relative humidity, dew point temperature, wind speed, precipitation intensity and salt mist concentration. The contamination detection device is configured to acquire contamination parameters, and the contamination parameters comprise salt density, ash density and hydrophobicity grade.

6. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 5, characterized in that, The leakage current monitoring device has a leakage current measurement range of 10 uA to 100 mA and a leakage current measurement accuracy of 1%.

7. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The dynamic weight adjustment unit comprises a temperature compensation circuit, and has a working temperature range of -40 DEG C to +85 DEG C.

8. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The hybrid kernel function support vector machine analysis module supports OTA remote upgrading and dynamic replacement of kernel function combination modes.

9. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 1, characterized in that, The hybrid kernel function has the form of K_hybrid = α·K_RBF + (1-α)·K_poly. ​ Wherein, alpha is a mixing weight coefficient in (0, 1), K_RBF is a Gaussian kernel function: exp(-gamma||x_i - x_j||^2), K_poly is a polynomial kernel function: (x_i*x_j + C)^d, C is a penalty factor in [0.1, 100], the Gaussian kernel function kernel parameter: gamma = 1 / (n_features*X.var()), the polynomial kernel degree d = 3.

10. The pollution flashover dynamic monitoring and early warning system for insulators according to claim 9, characterized in that, The dynamic weight adjustment unit is specifically configured to implement the adaptive adjustment by adopting the following adjustment strategy: alpha = 0.5 + 0.2*tanh(beta*(f_e - f_p)) Wherein, alpha is the mixing weight coefficient, f_e is the environmental feature importance score, f_p is the pollution feature importance score, beta is the adjustment sensitivity coefficient, the tanh function is a smooth S-shaped nonlinear conversion, and the output range is (-1, 1).

Images