Power transmission line insulator surface pollution high-risk component identification method and device after forest fire, electronic equipment and storage medium

Abstract

The invention discloses a power transmission line insulator surface pollution high-risk component identification method and device after forest fire, electronic equipment and a storage medium, and the method comprises the steps: obtaining a Raman spectrum curve of an insulator surface, extracting characteristic peaks, and generating a characteristic peak set; screening candidate characteristic peaks of the suspected high-risk components from the characteristic peak set based on a preset high-risk component peak position frequency shift range, and inputting peak position frequency shifts and peak intensities of the candidate characteristic peaks into a dirty component detection model to generate component information; grouping the candidate characteristic peaks belonging to the same high-risk component to form a high-risk component characteristic peak group; and extracting the peak intensity of each group of characteristic peaks, and calculating the corresponding high-risk component content based on a multiple linear regression model. By implementing the method, the efficiency of identifying the filthy components on the surface of the insulator after the forest fire passes can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of operation state maintenance of power transmission and transformation equipment, and particularly relates to a method, device, electronic device and storage medium for identifying high-risk components on the surface of transmission line insulators after a wildfire. Background Art

[0002] After a wildfire, a large number of fouling particles with complex sources and diverse components often adhere to the surface of transmission line insulators. These fouling components include but are not limited to soluble salts, carbon black, sulfates, and other substances formed by the deposition of combustion residues, ashes, and airborne suspended particles. After these foulings accumulate on the insulator surface, they will significantly change the surface electric field distribution, weaken the insulation layer's resistance to electrical stress, cause local electric field distortion, and increase the leakage current. Once the leakage current exceeds the critical threshold, arc discharge may occur on the insulator surface, leading to accidents such as flashover and tripping, seriously threatening the safe and stable operation of the power grid. In addition, some high-risk fouling components are hygroscopic or conductive, which will exacerbate the contamination degree of the insulator in humid or foggy weather and accelerate the deterioration of its electrical performance. Therefore, after a wildfire, quickly and accurately identifying the high-risk fouling components and their contents on the insulator surface is not only the basis for accurately evaluating the fouling state of the insulator, but also the key link for formulating scientific and reasonable cleaning or replacement strategies and ensuring the power grid to resume operation as soon as possible.

[0003] In the prior art, for the identification of the fouling components on the surface of insulators after a wildfire, there are problems of cumbersome detection processes and low efficiency. The fundamental reason is the lack of a targeted screening mechanism for high-risk components. The current technical means mainly rely on full-component spectral analysis or broad-spectrum chemical detection methods, attempting to comprehensively detect the chemical components and their concentrations of all foulings on the insulator surface. However, these methods do not distinguish the risk levels of the fouling components and often need to spend a lot of time and resources detecting all substances including low-risk components, resulting in redundant detection result information. In fact, not all fouling components will have a significant impact on the insulator performance, and high-risk components such as soluble salts, carbon black, sulfates, etc. are more destructive to the insulation performance. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, electronic device and storage medium for identifying high-risk components on the surface of transmission line insulators after a wildfire. By implementing the present invention, it is possible to identify high-risk components on the surface of transmission line insulators after a wildfire.

[0005] An embodiment of the present invention provides a method for identifying high-risk components on the surface of transmission line insulators after a wildfire, including: obtaining the Raman spectral curve of the fouling on the surface of the transmission line insulator after a wildfire;

[0006] Extract the characteristic peaks of the Raman spectrum curve to generate a set of characteristic peaks; wherein, the characteristic peaks include peak position frequency shift and peak intensity;

[0007] Based on a preset set of peak position frequency shift ranges of high-risk components, screen out the characteristic peaks of suspected high-risk components from the set of characteristic peaks to generate a number of candidate characteristic peaks; wherein, the high-risk components include soluble salts, carbon black, and sulfates;

[0008] Input the peak position frequency shift and peak intensity of the candidate characteristic peaks into a preset fouling component detection model, so that the fouling component detection model generates the component information corresponding to each candidate characteristic peak according to the peak position frequency shift and peak intensity of the candidate characteristic peaks;

[0009] Group the candidate characteristic peaks whose component information belongs to the same high-risk component into one group to generate a number of high-risk component characteristic peak groups;

[0010] For each high-risk component characteristic peak group, extract the peak intensities of the candidate characteristic peaks, and based on the extracted peak intensities, calculate and generate the corresponding high-risk component content based on a preset multiple linear regression model; wherein, one high-risk component corresponds to one multiple linear regression model.

[0011] Further, after obtaining the Raman spectrum curve of the surface fouling of the transmission line insulator after a wildfire, it further includes:

[0012] Based on the Savitzky-Golay filtering algorithm, perform denoising processing on the Raman spectrum curve to generate a denoised Raman spectrum curve;

[0013] Based on the polynomial function fitting algorithm, perform baseline correction on the denoised Raman spectrum curve to generate a baseline-corrected Raman spectrum curve;

[0014] Based on the Min-Max normalization algorithm, perform normalization processing on the baseline-corrected Raman spectrum curve to generate a normalized Raman spectrum curve;

[0015] Update the Raman spectrum curve of the surface fouling of the transmission line insulator after a wildfire according to the normalized Raman spectrum curve.

[0016] Further, the extracting the characteristic peaks of the Raman spectrum curve to generate a set of characteristic peaks includes:

[0017] Calculate the first derivative and the second derivative of the Raman spectrum curve, and determine the Raman frequency shift corresponding to the zero-crossing point of the first derivative and the minimum value of the second derivative as the peak position frequency shift of the characteristic peak;

[0018] Extract the corresponding Raman scattered light intensity from the Raman spectrum curve according to the peak position frequency shift of the characteristic peak;

[0019] Determine the intensity of the Raman scattered light as the peak intensity of the corresponding characteristic peak, and generate a number of characteristic peaks according to the peak position frequency shift and the peak intensity of the characteristic peak;

[0020] Combine the several characteristic peaks to generate a characteristic peak set.

[0021] Further, based on a preset set of peak position frequency shift ranges of high-risk components, screen out the characteristic peaks of suspected high-risk components from the characteristic peak set to generate a number of candidate characteristic peaks, including:

[0022] Extract the peak position frequency shift range of each high-risk component in the preset set of peak position frequency shift ranges of high-risk components;

[0023] Traverse each characteristic peak in the characteristic peak set, and mark the characteristic peak whose peak position frequency shift falls within the peak position frequency shift range of any high-risk component as a candidate characteristic peak.

[0024] Further, the training of the fouling component detection model includes:

[0025] Obtain a number of training characteristic peaks for training and the component labels corresponding to the training characteristic peaks; wherein, the component labels contain the component information corresponding to the training characteristic peaks;

[0026] Construct a number of training samples according to the training characteristic peaks and the component labels corresponding to the training characteristic peaks;

[0027] Input each training sample into the fouling component detection model in turn to train the fouling component detection model until the preset number of training times is reached; wherein, when the fouling component detection model receives each training sample, it outputs the predicted component information corresponding to the training sample; according to the predicted component information and the corresponding component label, calculate the loss function value through the loss function; update the fouling component detection model according to the loss function value.

[0028] Further, the construction of the multiple linear regression model includes:

[0029] Obtain the Raman spectrum curve of each standard sample in the standard sample set; wherein, the standard sample set contains standard samples covering various high-risk components with different contents;

[0030] Extract the characteristic peak intensity corresponding to the high-risk component in the Raman spectrum curve of each standard sample to generate a characteristic peak intensity data set;

[0031] Group the characteristic peak intensities belonging to the same high-risk component in the characteristic peak intensity data set to generate a number of characteristic peak intensity groups;

[0032] For each set of characteristic peak intensities, using the characteristic peak intensity as the independent variable and the known high-risk component content of the standard sample as the dependent variable, the linear regression coefficients are fitted by the least squares method to establish a multiple linear regression model for each high-risk component.

[0033] Further, the multiple linear regression model is specifically:

[0034] C = a0 + a1I1 + a2I2 +... + a n I n

[0035] where C is the high-risk component content; I1, I2,..., I n are the characteristic peak intensities; a0, a1,..., a n are the linear regression coefficients.

[0036] Based on the above method embodiment, the present invention correspondingly provides an apparatus embodiment.

[0037] An embodiment of the present invention provides a device for identifying high-risk components on the surface of transmission line insulators after a wildfire, including: a Raman spectral curve generation module, a characteristic peak set generation module, a candidate characteristic peak generation module, a component information solving module, a high-risk component characteristic peak group generation module, and a component content solving module;

[0038] The Raman spectral curve generation module is used to obtain the Raman spectral curve of the surface contamination of the transmission line insulator after the wildfire;

[0039] The characteristic peak set generation module is used to extract the characteristic peaks of the Raman spectral curve and generate a characteristic peak set; where the characteristic peaks include peak position frequency shift and peak intensity;

[0040] The candidate characteristic peak generation module is used to screen out the characteristic peaks of suspected high-risk components from the characteristic peak set based on a preset set of high-risk component peak position frequency shift ranges, and generate a number of candidate characteristic peaks; where the high-risk components include soluble salts, carbon black, and sulfates;

[0041] The component information solving module is used to input the peak position frequency shift and peak intensity of the candidate characteristic peaks into a preset contamination component detection model, so that the contamination component detection model generates the component information corresponding to each candidate characteristic peak according to the peak position frequency shift and peak intensity of the candidate characteristic peaks;

[0042] The high-risk component characteristic peak group generation module is used to group the candidate characteristic peaks with component information belonging to the same high-risk component into one group, and generate a number of high-risk component characteristic peak groups;

[0043] The component content solving module is used to extract the peak intensities of each candidate characteristic peak for each high-risk component characteristic peak group, and calculate and generate the corresponding high-risk component content based on the extracted peak intensities and a preset multiple linear regression model; where one high-risk component corresponds to one multiple linear regression model.

[0044] Based on the above method item embodiments, the present invention correspondingly provides electronic device item embodiments.

[0045] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it can implement the method for identifying high-risk components on the surface of transmission line insulators after a wildfire as described in any one of the above method item embodiments.

[0046] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments.

[0047] An embodiment of the present invention provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, it can implement the method for identifying high-risk components on the surface of transmission line insulators after a wildfire as described in any one of the above method item embodiments.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] An embodiment of the present invention provides a method, device, electronic device, and storage medium for identifying high-risk components on the surface of transmission line insulators after a wildfire. The method obtains the Raman spectrum curve of the insulator surface and extracts spectral characteristic peaks, including peak position frequency shift and peak intensity; then, based on a preset high-risk component peak position frequency shift range, candidate characteristic peaks suspected of being high-risk components are screened out from the set of characteristic peaks; subsequently, the peak position frequency shift and peak intensity of the candidate characteristic peaks are input into a pollution component detection model to identify the component information corresponding to the candidate characteristic peaks, and the candidate characteristic peaks belonging to the same high-risk component are grouped; the peak intensities of each characteristic peak are extracted, and the corresponding high-risk component content is calculated based on a multiple linear regression model.

[0050] Through the pre-screening mechanism of high-risk components based on the characteristic peaks of Raman spectroscopy, combined with the preset peak position frequency shift range set, the characteristic peaks of key harmful components such as soluble salts, carbon black, and sulfates are quickly locked, effectively avoiding the inefficiency of full-component analysis and the interference of redundant data. By introducing a screening mechanism for high-risk components before component identification, the detection process no longer blindly covers all fouling components, but preferentially locks components such as soluble salts, carbon black, and sulfates that have a significant impact on insulation performance, thus reducing the ineffective detection of low-risk components and simplifying the analysis process. In addition, by further using the fouling component detection model and the multiple linear regression algorithm, precise qualitative and quantitative analysis of high-risk components is achieved for the screened characteristic peak groups, improving the efficiency and accuracy of the post-disaster insulator fouling state assessment, thereby ensuring the safe and stable operation of the transmission line. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 FIG. is a schematic flowchart of a method for identifying high-risk components on the surface of insulators of a transmission line after a wildfire according to an embodiment of the present invention.

[0052] Figure 2 FIG. is a schematic structural diagram of a device for identifying high-risk components on the surface of insulators of a transmission line after a wildfire according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0054] As Figure 1 shown, an embodiment of the present invention provides a method for identifying high-risk components on the surface of insulators of a transmission line after a wildfire, which at least includes the following steps:

[0055] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0056] As Figure 1 shown, an embodiment of the present invention provides a method for identifying high-risk components on the surface of insulators of a transmission line after a wildfire, which at least includes the following steps:

[0057] Step S1: Obtain the Raman spectral curve of the surface contamination of the transmission line insulator after a wildfire;

[0058] Specifically, when obtaining the Raman spectral curve of the surface contamination of the transmission line insulator after a wildfire, a non-destructive sampling method is required to collect the surface contamination sample to avoid introducing secondary contamination or damaging the insulator structure. The sample is scanned by a Raman spectrometer in the near-infrared band (such as 785 nm or 1064 nm), and the laser power and integration time are dynamically adjusted to balance the signal suppression effect of highly absorbent substances such as carbon black. At the same time, for the fluorescence background interference caused by the coexistence of salts and carbon black in the mixed contamination, baseline correction and denoising algorithms are used to eliminate the masking of broad peaks and ensure the effective capture of the characteristic peak signals. To improve the data reliability, the same sample is scanned multiple times and the average spectral curve is superimposed to suppress random noise interference. Finally, spectral data reflecting the characteristics of the contamination components is obtained, providing a basis for the accurate identification and quantitative analysis of high-risk components.

[0059] In a preferred embodiment, after obtaining the Raman spectral curve of the surface contamination of the transmission line insulator after a wildfire, it further includes:

[0060] Based on the Savitzky-Golay filtering algorithm, denoise the Raman spectral curve to generate a denoised Raman spectral curve;

[0061] Based on the polynomial function fitting algorithm, perform baseline correction on the denoised Raman spectral curve to generate a baseline-corrected Raman spectral curve;

[0062] Based on the Min-Max normalization algorithm, normalize the baseline-corrected Raman spectral curve to generate a normalized Raman spectral curve;

[0063] According to the normalized Raman spectral curve, update the Raman spectral curve of the surface contamination of the transmission line insulator after a wildfire.

[0064] Step S2: Extract the characteristic peaks of the Raman spectral curve to generate a set of characteristic peaks; where the characteristic peaks include peak position frequency shift and peak intensity;

[0065] In a preferred embodiment, the extraction of the characteristic peaks of the Raman spectral curve to generate a set of characteristic peaks includes:

[0066] Calculate the first derivative and the second derivative of the Raman spectral curve, and determine the peak position frequency shift of the characteristic peak as the Raman frequency shift corresponding to the zero-crossing point of the first derivative and the minimum value of the second derivative;

[0067] Extract the corresponding Raman scattering light intensity from the Raman spectral curve according to the peak position frequency shift of the characteristic peak;

[0068] Determine the Raman scattering light intensity as the peak intensity of the corresponding characteristic peak, and generate a number of characteristic peaks according to the peak position frequency shift and the peak intensity of the characteristic peak;

[0069] Combine the several characteristic peaks to generate a characteristic peak set.

[0070] Specifically, perform denoising processing on the Raman spectrum curve based on the Savitzky-Golay filtering algorithm, smooth the high-frequency noise through local polynomial fitting and retain the characteristic peak morphology, and suppress the random noise interference caused by the scattering of carbon black and particulate matter in the dirt after the wildfire; on this basis, use the polynomial function fitting algorithm to perform baseline correction on the denoised spectrum curve, iteratively fit the baseline shape and deduct the interference of the carbon black broad peak and fluorescence background, eliminate the masking effect of baseline drift on the characteristic peak intensity, and generate a baseline-corrected Raman spectrum curve; subsequently, map the intensity of the baseline-corrected spectrum curve to the [0,1] interval based on the Min-Max normalization algorithm, eliminate the difference in the light intensity dimension caused by uneven sample thickness or instrument fluctuation, and generate comparable normalized spectral data; finally, update the Raman spectrum curve of the dirt on the surface of the transmission line insulator after the wildfire according to the normalization processing result, and provide a standardized data basis for the extraction and quantitative analysis of the characteristic peaks of high-risk components.

[0071] Step S3: Based on a preset set of peak position frequency shift ranges of high-risk components, screen out the characteristic peaks of suspected high-risk components from the characteristic peak set to generate a number of candidate characteristic peaks; among them, the high-risk components include soluble salts, carbon black, and sulfates;

[0072] In a preferred embodiment, the screening out the characteristic peaks of suspected high-risk components from the characteristic peak set based on a preset set of peak position frequency shift ranges of high-risk components to generate a number of candidate characteristic peaks includes:

[0073] Extract the peak position frequency shift range of each high-risk component in the preset set of peak position frequency shift ranges of high-risk components;

[0074] Traverse each characteristic peak in the characteristic peak set, and mark the characteristic peak whose peak position frequency shift falls within the peak position frequency shift range of any high-risk component as a candidate characteristic peak.

[0075] It should be noted here that based on the preset high-risk component peak position frequency shift range set, the characteristic peak frequency shift intervals corresponding to target components such as soluble salts, carbon black, and sulfates are extracted. By traversing each characteristic peak in the characteristic peak set, the peak position frequency shift value is matched with the preset range one by one, and the characteristic peaks that meet the high-risk component characteristic peak frequency shift range are screened out and marked as candidate characteristic peaks, so as to focus on the key pollutants that significantly harm the insulation performance, effectively exclude the interference of non-key components and background noise in the complex pollution after the wildfire, realize the rapid preliminary screening of high-risk components, and provide a high-confidence candidate characteristic peak set for subsequent accurate identification and quantitative analysis.

[0076] The high-risk component peak position frequency shift range set is a database established based on the Raman spectral characteristics of typical pollutants (such as soluble salts, carbon black, sulfates, etc.) on the surface of transmission line insulators after the wildfire, and contains the characteristic peak frequency shift intervals of each high-risk component. For example: the typical characteristic peaks of soluble salts (such as NaCl, CaCl2) are located at 220 - 240 cm -1 (corresponding to lattice vibration) and 340 - 360 cm -1 (molecular vibration); the characteristic peaks of carbon black are concentrated at 1330 - 1370 cm -1 (disordered carbon structure D peak) and 1570 - 1590 cm -1 (graphitized carbon G peak); the characteristic peaks of sulfates (such as CaSO4) are distributed at 970 - 990 cm -1 (symmetric stretching vibration of sulfate radical). This range set is determined by fusing experimental calibration and literature data, covering the typical spectral peak positions of key components in the pollution after the wildfire, and allowing an instrument measurement error and peak position offset tolerance of ±10 - 20 cm -1 to quickly screen out the characteristic peaks matching high-risk components from the complex spectrum, improving the detection efficiency and reliability.

[0077] Step S4: Input the peak position frequency shift and peak intensity of the candidate characteristic peaks into a preset pollution component detection model, so that the pollution component detection model generates the component information corresponding to each candidate characteristic peak according to the peak position frequency shift and peak intensity of the candidate characteristic peaks;

[0078] Specifically, input the peak position frequency shift and peak intensity of the candidate characteristic peaks into a pre-trained fouling component detection model. This model conducts multi-dimensional analysis on each candidate characteristic peak by fusing the matching degree of the peak position frequency shift and the weight relationship of the peak intensity, combined with the context correlation information of adjacent characteristic peaks (such as adjacent peak position intervals, intensity ratios, etc.), to generate the probability distribution results of their corresponding components (such as the attribution probabilities of NaCl, carbon black, or CaSO4); the model is based on a convolutional neural network and an attention mechanism, dynamically learning the separation characteristics of overlapping peaks (such as the D peak of carbon black and the organic pollutant peak) and the suppression logic of interference peaks (such as silicate miscellaneous peaks) in the mixed fouling after a wildfire, and finally outputting the high-risk component types and confidence levels of the attribution of each candidate characteristic peak. If there are multi-peak associations for the same component (such as the double-peak characteristics of NaCl), the determination reliability is improved through cross-peak collaborative verification, so as to achieve accurate identification and classification of high-risk components under a complex spectral background.

[0079] In a preferred embodiment, the training of the fouling component detection model includes:

[0080] Obtain a number of training characteristic peaks for training and the component labels corresponding to the training characteristic peaks; wherein, the component labels contain the component information corresponding to the training characteristic peaks;

[0081] Construct a number of training samples according to the training characteristic peaks and the component labels corresponding to the training characteristic peaks;

[0082] Input each training sample into the fouling component detection model in turn to train the fouling component detection model until the preset number of training times is reached; wherein, when the fouling component detection model receives each training sample, it outputs the predicted component information corresponding to the training sample; according to the predicted component information and the corresponding component label, calculate the loss function value through a loss function; update the fouling component detection model according to the loss function value.

[0083] It should be noted here that when obtaining a number of training characteristic peaks for training and their corresponding component labels, it is necessary to construct training samples of the "input-output" mapping relationship based on the spectral characteristic peaks of high-risk components (including peak position frequency shift and peak intensity parameters) verified in the standard sample set or historical detection data. Each training characteristic peak's frequency shift-intensity combination and its corresponding component label (such as soluble salts, carbon black, sulfates, etc.) are used to construct the training samples. By sequentially inputting the training samples into the fouling component detection model, the model is driven to generate predicted component information based on the frequency shift matching degree and intensity weight relationship of the input characteristic peaks, and the loss value is calculated based on the difference between the prediction result and the true component label (such as the cross-entropy loss function). The backpropagation algorithm is used in combination with the gradient descent optimizer to dynamically adjust the model parameters (such as convolutional kernel weights, fully connected layer bias terms), iteratively updating the model's analytical ability for overlapping peaks, weak peaks, and interference peaks until the preset training times or loss convergence threshold is reached. Finally, the model can accurately determine the type of high-risk component to which the candidate characteristic peak belongs based on the frequency shift-intensity characteristics of the candidate characteristic peaks, realizing the automatic analysis and classification of multi-component mixed spectra in complex fouling scenarios after wildfires.

[0084] Step S5: Group the candidate characteristic peaks whose component information belongs to the same high-risk component to generate several high-risk component characteristic peak groups;

[0085] Specifically, according to the candidate characteristic peak component information output by the fouling component detection model, the candidate characteristic peaks belonging to the same high-risk component (such as soluble salts, carbon black, or sulfates) are clustered and grouped according to their component labels to generate several high-risk component characteristic peak groups. Through component label consistency verification and cross-peak logic verification (such as multiple characteristic peaks of the same component need to meet the preset frequency shift interval and intensity ratio relationship), candidate characteristic peaks with misjudgment or low confidence by the model are excluded to ensure that the component attribution of the characteristic peaks within each group is consistent and conforms to the actual spectral characteristic law. For the multi-peak association characteristics that may exist for the same high-risk component (such as soluble salts containing main peaks and secondary peaks), multi-peak combinations that meet the co-verification conditions are retained to improve the robustness of subsequent quantitative analysis. Finally, a set of characteristic peak groups corresponding one-to-one with high-risk components is formed, providing standardized input data for the content calculation of the multiple linear regression model.

[0086] Step S6: For each high-risk component characteristic peak group, extract the peak intensity of each candidate characteristic peak, and based on the extracted peak intensity, calculate and generate the corresponding high-risk component content according to a preset multiple linear regression model; where one high-risk component corresponds to one multiple linear regression model.

[0087] Specifically, for each group of characteristic peaks of high-risk components, the peak intensities of all candidate characteristic peaks within the group are extracted as independent variables and input into the preset multiple linear regression model corresponding to the component. The peak intensities of each characteristic peak are weighted and summed through the regression coefficients (weights) and intercept terms built into the model. Combining the multi-peak synergistic effect and the non-linear correction factor, the quantitative content value of the corresponding high-risk component is comprehensively calculated. Among them, the multiple linear regression model for each high-risk component is independently trained based on its unique spectral peak response law. For example, the carbon black model focuses on the ratio and absolute intensity of the D peak and G peak intensities, while the soluble salt model utilizes the intensity superposition relationship between the main peak and the secondary peak. Through multi-dimensional parameter fusion, the influence of background interference and peak coupling effect in complex contamination after wildfire is eliminated. Finally, the normalized and calibrated quantitative results are output, and the rationality of the predicted content value is verified (such as non-negative constraint, content upper limit threshold filtering) to ensure that the detection results conform to the actual physical meaning, and high-precision content calculation based on multi-peak joint analysis is achieved.

[0088] In a preferred embodiment, the construction of the multiple linear regression model includes:

[0089] Obtain the Raman spectral curves of each standard sample in the standard sample set; wherein, the standard sample set contains standard samples covering various high-risk components with different contents;

[0090] Extract the characteristic peak intensities corresponding to the high-risk components in the Raman spectral curves of each standard sample to generate a characteristic peak intensity data set;

[0091] Group the characteristic peak intensities belonging to the same high-risk component in the characteristic peak intensity data set to generate several characteristic peak intensity groups;

[0092] For each characteristic peak intensity group, with the characteristic peak intensity as the independent variable and the known high-risk component content of the standard sample as the dependent variable, the linear regression coefficients are fitted by the least squares method to establish the multiple linear regression models corresponding to each high-risk component.

[0093] It should be noted here that when obtaining the Raman spectral curves of each standard sample in the standard sample set, it is necessary to pre-prepare standard samples covering high-risk components such as soluble salts, carbon black, and sulfates with different content gradients (such as gradient samples containing 0.1% - 10% NaCl), collect their spectral data through a Raman spectrometer, and perform baseline correction and normalization processing to ensure the consistency of data quality; extract the characteristic peak intensities corresponding to each high-risk component from the preprocessed spectral curves (such as the main peak and secondary peak intensities of soluble salts, and the D peak and G peak intensities of carbon black), generate a characteristic peak intensity data set containing multi-component and multi-peak type characteristics; group the characteristic peak intensities of the same high-risk component according to the component labels (such as the carbon black group contains D peak and G peak intensity data), forming a characteristic peak intensity group corresponding one-to-one to the target component; for each characteristic peak intensity group, using all the characteristic peak intensities in the group as independent variables and the known components of the corresponding standard sample as the dependent variable, fit the linear relationship between each intensity variable and the content by the least squares method, calculate the regression coefficient and the intercept term, establish a multiple linear regression model such as "carbon black content = 0.12 × D peak intensity + 0.08 × G peak intensity + 0.05", and verify the model prediction error through an independent validation set (such as R 2 > 0.95), to ensure the adaptability of the model to multi-peak interference and content non-linear response in the complex polluted scenario after wildfires.

[0094] In a preferred embodiment, the multiple linear regression model is specifically:

[0095] C = a0 + a1I1 + a2I2 +... + a n I n

[0096] where C is the content of the high-risk component; I1, I2,..., I n are the characteristic peak intensities; a0, a1,..., a n are the linear regression coefficients.

[0097] Based on the above method item embodiment, the present invention correspondingly provides a device item embodiment.

[0098] As Figure 2 shown, an embodiment of the present invention provides a device for identifying high-risk components on the surface of insulators of transmission lines after a wildfire, including: a Raman spectral curve generation module, a characteristic peak set generation module, a candidate characteristic peak generation module, a component information solving module, a high-risk component characteristic peak group generation module, and a component content solving module;

[0099] The Raman spectral curve generation module is used to obtain the Raman spectral curve of the dirt on the surface of the insulator of the transmission line after the wildfire;

[0100] The feature peak set generation module is configured to extract the feature peaks of the Raman spectrum curve and generate a feature peak set; wherein, the feature peaks include peak position frequency shift and peak intensity;

[0101] The candidate feature peak generation module is configured to screen out the feature peaks of suspected high-risk components from the feature peak set based on a preset high-risk component peak position frequency shift range set, and generate a number of candidate feature peaks; wherein, the high-risk components include soluble salts, carbon black, and sulfates;

[0102] The component information solving module is configured to input the peak position frequency shift and peak intensity of the candidate feature peaks into a preset pollution component detection model, so that the pollution component detection model generates the component information corresponding to each candidate feature peak according to the peak position frequency shift and peak intensity of the candidate feature peaks;

[0103] The high-risk component feature peak group generation module is configured to group the candidate feature peaks whose component information belongs to the same high-risk component into one group, and generate a number of high-risk component feature peak groups;

[0104] The component content solving module is configured to, for each high-risk component feature peak group, extract the peak intensities of the candidate feature peaks, and calculate and generate the corresponding high-risk component content based on the extracted peak intensities and a preset multiple linear regression model; wherein, one high-risk component corresponds to one multiple linear regression model.

[0105] It should be noted that the embodiments of the devices described above correspond to the above embodiments of the present invention, and can implement the method for identifying high-risk components on the surface of transmission line insulators after a wildfire in any one of the above embodiments of the present invention. In addition, the embodiments of the above devices are merely illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationships between the modules indicate that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement without creative efforts.

[0106] Based on the above method embodiment of the present invention, an embodiment of an electronic device is correspondingly provided.

[0107] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the method for identifying high-risk components on the surface of transmission line insulators after a wildfire in any one of the present invention, or when the processor executes the computer program, it implements the functions of each module in the above device embodiments.

[0108] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0109] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0110] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, and connects various parts of the entire terminal device through various interfaces and circuits.

[0111] The memory can be used to store the computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and invoking the data stored in the memory, the processor realizes various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash device, or other volatile solid-state storage devices.

[0112] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments;

[0113] Another embodiment of the present invention provides a storage medium, the storage medium includes a stored computer program, wherein, when the computer program runs, it controls the device where the storage medium is located to execute the above-mentioned method for identifying high-risk components on the surface of transmission line insulators after a wildfire in any one of the present invention.

[0114] Among them, the above storage medium is a computer-readable storage medium, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0115] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of this application. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0116] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for identifying high-risk contamination components on the surface of transmission line insulators after a wildfire, characterized in that: include: Obtain the Raman spectrum curve of the surface contamination of the transmission line insulator after a wildfire; Extracting characteristic peaks of the Raman spectrum curve to generate a characteristic peak set; wherein the characteristic peaks include peak position frequency shift and peak intensity; Based on the preset high-risk component peak position frequency shift range set, characteristic peaks of suspected high-risk components are screened out from the characteristic peak set to generate a number of candidate characteristic peaks; wherein the high-risk components include soluble salts, carbon black and sulfate; Inputting the peak frequency shift and peak intensity of the candidate characteristic peaks into a preset pollution component detection model, so that the pollution component detection model generates component information corresponding to each candidate characteristic peak according to the peak frequency shift and peak intensity of the candidate characteristic peaks; The candidate characteristic peaks whose component information belongs to the same high-risk component are grouped together to generate several high-risk component characteristic peak groups; For each high-risk component characteristic peak group, the peak intensity of each candidate characteristic peak is extracted, and the corresponding high-risk component content is calculated and generated based on the extracted peak intensity and a preset multivariate linear regression model; wherein one high-risk component corresponds to one multivariate linear regression model.

2. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 1, characterized in that: After obtaining the Raman spectrum curve of the surface contamination of the transmission line insulator after the wildfire, the method further includes: Based on the Savitzky-Golay filtering algorithm, the Raman spectrum curve is subjected to denoising processing to generate a denoised Raman spectrum curve; Based on a polynomial function fitting algorithm, baseline correction is performed on the denoised Raman spectrum curve to generate a baseline-corrected Raman spectrum curve; Based on the Min-Max normalization algorithm, the baseline-corrected Raman spectrum curve is normalized to generate a normalized Raman spectrum curve; According to the normalized Raman spectrum curve, the Raman spectrum curve of the surface contamination of the transmission line insulator after the wildfire is updated.

3. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 2, characterized in that: The step of extracting characteristic peaks of the Raman spectrum curve to generate a characteristic peak set includes: Calculating the first-order derivative and the second-order derivative of the Raman spectrum curve, and determining the Raman frequency shift corresponding to the zero-crossing point of the first-order derivative and the minimum value of the second-order derivative as the peak position frequency shift of the characteristic peak; The corresponding Raman scattered light intensity is extracted from the Raman spectrum curve according to the peak position frequency shift of the characteristic peak; Determining the intensity of the Raman scattered light as the peak intensity of the corresponding characteristic peak, and generating a plurality of characteristic peaks according to the peak position frequency shift of the characteristic peak and the peak intensity of the characteristic peak; The several characteristic peaks are combined to generate a characteristic peak set.

4. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 3, characterized in that: The method of screening out characteristic peaks of suspected high-risk components from the characteristic peak set based on the preset high-risk component peak position frequency shift range set to generate a number of candidate characteristic peaks includes: Extracting the peak frequency shift range of each high-risk component in the preset high-risk component peak frequency shift range set; Each characteristic peak in the characteristic peak set is traversed, and a characteristic peak whose peak position frequency shift falls into the peak position frequency shift range of any high-risk component is recorded as a candidate characteristic peak.

5. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 4, characterized in that: The training of the pollution component detection model includes: Acquire a number of training characteristic peaks for training and component labels corresponding to the training characteristic peaks; wherein the component labels include component information corresponding to the training characteristic peaks; Constructing a number of training samples according to the training characteristic peaks and the component labels corresponding to the training characteristic peaks; Input each training sample into the contaminant component detection model in turn, and train the contaminant component detection model until a preset number of training times is reached; wherein, the contaminant component detection model outputs predicted component information corresponding to the training sample each time it receives a training sample; calculates a loss function value through a loss function based on the predicted component information and the corresponding component label; and updates the contaminant component detection model based on the loss function value.

6. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 5, characterized in that: The construction of the multiple linear regression model includes: Obtaining a Raman spectrum curve of each standard sample in a standard sample set; wherein the standard sample set includes standard samples covering various high-risk components at different contents; Extracting the characteristic peak intensity corresponding to the high-risk component in the Raman spectrum curve of each standard sample to generate a characteristic peak intensity data set; Grouping the characteristic peak intensities belonging to the same high-risk component in the characteristic peak intensity data set into one group to generate a plurality of characteristic peak intensity groups; For each characteristic peak intensity group, the characteristic peak intensity was used as the independent variable, and the known high-risk component content of the standard sample was used as the dependent variable. The linear regression coefficients were fitted by the least squares method to establish a multivariate linear regression model corresponding to each high-risk component.

7. The method for identifying high-risk contamination components on the surface of transmission line insulators after wildfire according to claim 6, characterized in that: The multiple linear regression model is as follows: C=a0+a1I1+a2I2+…+a n I n Among them, C is the content of high-risk ingredients; I1, I2, ..., I n is the characteristic peak intensity; a1, a2, ..., a n is the linear regression coefficient.

8. A device for identifying high-risk contamination components on the surface of transmission line insulators after wildfire, characterized in that: include: Raman spectrum curve generation module, characteristic peak set generation module, candidate characteristic peak generation module, component information solution module, high-risk component characteristic peak group generation module and component content solution module; The Raman spectrum curve generation module is used to obtain the Raman spectrum curve of the surface contamination of the transmission line insulator after the wildfire; The characteristic peak set generation module is used to extract characteristic peaks of the Raman spectrum curve and generate a characteristic peak set; wherein the characteristic peaks include peak position frequency shift and peak intensity; The candidate characteristic peak generation module is used to screen out characteristic peaks of suspected high-risk components from the characteristic peak set based on a preset high-risk component peak position frequency shift range set, and generate a number of candidate characteristic peaks; wherein the high-risk components include soluble salts, carbon black and sulfate; The component information solving module is used to input the peak frequency shift and peak intensity of the candidate characteristic peaks into a preset pollution component detection model, so that the pollution component detection model generates the component information corresponding to each candidate characteristic peak according to the peak frequency shift and peak intensity of the candidate characteristic peaks; The high-risk component characteristic peak group generating module is used to group candidate characteristic peaks whose component information belongs to the same high-risk component into one group to generate a plurality of high-risk component characteristic peak groups; The component content solving module is used to extract the peak intensity of each candidate characteristic peak for each high-risk component characteristic peak group, and calculate and generate the corresponding high-risk component content based on the extracted peak intensity and a preset multivariate linear regression model; wherein one high-risk component corresponds to one multivariate linear regression model.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for identifying high-risk components of contamination on the surface of insulators of a transmission line after a wildfire as described in any one of claims 1 to 7 can be implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it can implement the method for identifying high-risk components of contamination on the surface of transmission line insulators after a wildfire as described in any one of claims 1 to 7.

Images