Nondestructive testing method, device and system for silicone rubber cable accessory

Abstract

The invention belongs to the technical field of cable accessory detection, and particularly relates to a nondestructive testing method, device and system for a silicone rubber cable accessory, and the method comprises the steps: obtaining infrared thermal imaging detection data; acquiring ultraviolet imaging detection data, constructing an infrared-ultraviolet correlation model of the silicone rubber cable accessory, and matching the infrared-ultraviolet correlation level of the silicone rubber cable accessory; determining surface layer defect data of the silicone rubber cable accessory; collecting appearance detection data of the silicone rubber cable accessory to obtain an appearance detection signal of the silicone rubber cable accessory; outputting the appearance detection grade of the silicone rubber cable accessory based on the BP neural network model; using a deep belief network model to output a silicone rubber cable accessory electrical performance classification result; matching the silicone rubber cable accessory nondestructive testing comprehensive score, and judging whether the silicone rubber cable accessory is qualified or not. According to the method, the limitation of a traditional single detection method is overcome, potential safety hazards can be found more accurately, and a powerful guarantee is provided for safe operation of a power system.

Description

Technical Field

[0001] The present invention belongs to the technical field of cable accessory detection, and in particular relates to a nondestructive detection method, device and system for silicone rubber cable accessories. Background Art

[0002] As the demand for electricity continues to grow, the scale of power grid construction continues to expand. As a key component in the power transmission system, the use of silicone rubber cable accessories is also increasing. These cable accessories usually operate in complex environments such as outdoors and underground. They are affected by temperature changes, humidity, mechanical stress, electromagnetic interference and other factors for a long time, and are prone to aging, insulation performance degradation, partial discharge and other problems. If these problems are not discovered and handled in time, they may cause cable failures and affect power supply. Therefore, ensuring the quality and performance of silicone rubber cable accessories is crucial to ensuring the safe and stable operation of the power grid.

[0003] However, there are still some deficiencies in the current research on non-destructive testing of silicone rubber cable accessories. Traditional cable accessory testing methods such as sampling dissection are destructive testing, which can only be performed before the installation of cable accessories or after retirement. It is impossible to perform real-time testing during their operation, and it will cause damage to the cable accessories, affecting their normal use. In addition, some conventional electrical testing methods, such as insulation resistance testing and DC withstand voltage testing, can reflect the electrical performance of cable accessories to a certain extent, but the detection sensitivity for some local defects and early faults is low, and it is difficult to accurately detect potential safety hazards. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a method, device and system for non-destructive testing of silicone rubber cable accessories, which comprehensively evaluate the status of silicone rubber cable accessories from multiple dimensions such as thermal characteristics, corona discharge characteristics, surface defects, appearance and electrical performance, overcoming the limitations of traditional single detection methods, and can more accurately discover potential safety hazards, providing strong protection for the safe operation of the power system.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A nondestructive testing method for silicone rubber cable accessories comprises the following steps: Step 1: Perform infrared thermal imaging detection on silicone rubber cable accessories to obtain infrared thermal imaging detection data; Step 2: Perform ultraviolet imaging detection on silicone rubber cable accessories to obtain ultraviolet imaging detection data, combine with infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor for silicone rubber cable accessories, and match the infrared-ultraviolet correlation level for silicone rubber cable accessories; Step 3: Use ultrasonic surface wave detection and optical coherence tomography detection to determine the surface layer defect data of the silicone rubber cable accessories; Step 4: Collect the appearance inspection data of the silicone rubber cable accessories, and combine the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories; Step 5: Based on the BP neural network model and in combination with the silicone rubber cable accessory surface detection signal, output the silicone rubber cable accessory surface detection level; Step 6: Collect electrical performance test data of silicone rubber cable accessories, obtain electrical performance factors of silicone rubber cable accessories, and use a deep belief network model to output electrical performance classification results of silicone rubber cable accessories; Step 7. Based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, match the comprehensive score of non-destructive testing of silicone rubber cable accessories to determine whether the silicone rubber cable accessories are qualified.

[0006] Preferably, in step 1, the process of obtaining infrared thermal imaging detection data is: Conduct infrared thermal imaging inspection on silicone rubber cable accessories, and use infrared thermal imager to detect the temperature distribution on the surface of silicone rubber cable accessories; Obtain infrared thermal imaging test data, including surface temperature differences of silicone rubber cable accessories , Maximum surface temperature of silicone rubber cable accessories , Thermal contrast of the highest surface temperature area of ​​silicone rubber cable accessories .

[0007] Preferably, in the step 2, an infrared-ultraviolet correlation model of silicone rubber cable accessories is constructed, a correlation factor of silicone rubber cable accessories is output, and an infrared-ultraviolet correlation level of silicone rubber cable accessories is matched. The specific process is as follows: Perform UV imaging inspection on silicone rubber cable accessories and obtain UV imaging inspection data, including UV photon counting , discharge intensity , discharge area ; Construct an infrared-ultraviolet correlation model for silicone rubber cable accessories, and output the correlation factor of silicone rubber cable accessories based on ultraviolet imaging detection data and infrared thermal imaging detection data. The correlation factor of silicone rubber cable accessories serves as the analysis basis for matching the infrared-ultraviolet correlation level of silicone rubber cable accessories. The infrared-ultraviolet correlation model of silicone rubber cable accessories is expressed as: ; In the formula, is the silicone rubber cable accessory correlation factor, Infrared analysis factor for silicone rubber cable accessories, UV analysis factor for silicone rubber cable accessories, For setting The weight factor of For setting The weight factor of , e is a natural constant; Obtain the silicone rubber cable accessory association factor and the silicone rubber cable accessory infrared-ultraviolet association level mapping table pre-stored in the database, and find the matching silicone rubber cable accessory infrared-ultraviolet association level according to the silicone rubber cable accessory association factor by searching the mapping table.

[0008] Preferably, in step 3, the process of determining the surface layer defect data of the silicone rubber cable accessories is: Ultrasonic surface waves are emitted to the surface of silicone rubber cable accessories, and optical coherence tomography is performed. The principle of low-coherence light interference is used to perform micron-level resolution tomography on the surface of silicone rubber cable accessories to obtain surface layer defect data of silicone rubber cable accessories. Silicone rubber cable accessories surface layer defect data including the maximum defect depth of silicone rubber cable accessories surface layer , Number of pores on the surface of silicone rubber cable accessories , Number of bubbles on the surface of silicone rubber cable accessories .

[0009] Preferably, in step 4, the process of obtaining the silicone rubber cable accessory surface detection signal is: Collect appearance inspection data of silicone rubber cable accessories, including the number of cracks on the appearance of silicone rubber cable accessories , Appearance corrosion area of ​​silicone rubber cable accessories , Total area of ​​impurities embedded in the appearance of silicone rubber cable accessories ; Based on the appearance inspection data of silicone rubber cable accessories and the surface layer defect data of silicone rubber cable accessories, the appearance inspection signal of silicone rubber cable accessories is obtained through comprehensive analysis. The appearance inspection signal of silicone rubber cable accessories is used as the analysis basis for outputting the appearance inspection grade of silicone rubber cable accessories. The calculation method of the surface detection signal of the silicone rubber cable accessories is: ; In the formula, It is the detection signal of the appearance of the silicone rubber cable accessories. It is the appearance detection signal of silicone rubber cable accessories. is the surface detection signal of the silicone rubber cable accessories, A is the set The weight factor of The weight factor of , e is a natural constant.

[0010] Preferably, in step 5, the method of outputting the appearance inspection grade of the silicone rubber cable accessories is: Obtain the BP neural network model training data set, normalize the input variables and output variables, and map them to the [0,1] interval; Determine that the number of neurons in the input layer of the BP neural network model is equal to the number of input variables, and the input variable is the surface detection signal of the silicone rubber cable accessories in the training data set, which is 1; The number of neurons in the output layer is equal to the number of categories of the silicone rubber cable accessories appearance detection level. The number of categories of the silicone rubber cable accessories appearance detection level in the training data set is 3, which are 1, 2, and 3, representing high, medium, and low levels, respectively. Unique hot encoding is used to convert them into numerical vectors, with high levels represented as [1,0,0], medium levels represented as [0,1,0], and low levels represented as [0,0,1]. The number of neurons in the hidden layer is h; Randomly initialize the connection weights from the input layer to the hidden layer and from the hidden layer to the output layer, as well as the thresholds of the hidden layer and the output layer; Input the surface detection signal of the silicone rubber cable accessory in the preprocessed training data set; Use the Sigmoid activation function to calculate the output of the sth neuron in the hidden layer : ; In the formula, is the connection weight from the input layer to the sth neuron in the hidden layer, is the threshold of the sth neuron in the hidden layer, s is the neuron number of the hidden layer, s=1,2,3...h; Use the Sigmoid activation function to calculate the output of the tth neuron in the output layer : ; In the formula, is the connection weight from the sth neuron in the hidden layer to the tth neuron in the output layer, is the threshold of the tth neuron in the output layer, t is the neuron number of the output layer, t=1,2,3; The error for the tth neuron in the output layer is , and its calculation formula is: ; In the formula, is the expected output of the tth neuron in the output layer, is the actual output of the tth neuron in the output layer; The gradient descent method is used to update the connection weights and thresholds until the network error is less than the accuracy threshold stored in the database or the maximum number of iterations is reached; Input the pre-processed current silicone rubber cable accessory appearance detection signal into the trained BP neural network model; Get the output vector of the output layer; According to the output vector, the maximum membership principle is adopted to determine the appearance inspection level of silicone rubber cable accessories.

[0011] Preferably, in step 6, the process of outputting the electrical performance classification results of the silicone rubber cable accessories is: Collect electrical performance test data of silicone rubber cable accessories, including insulation resistance of silicone rubber cable accessories , Dielectric loss factor of silicone rubber cable accessories , Silicone rubber cable accessories conductor resistance ; Based on the electrical performance test data of silicone rubber cable accessories, the electrical performance factor of silicone rubber cable accessories is obtained through comprehensive analysis. The electrical performance factor of silicone rubber cable accessories is used as the analysis basis for outputting the electrical performance classification results of silicone rubber cable accessories. The formula for obtaining the electrical performance factor of silicone rubber cable accessories is: ; In the formula, is the electrical performance factor of silicone rubber cable accessories, and e is a natural constant; The electrical performance factor of the silicone rubber cable accessories is normalized, and the preprocessed electrical performance factor of the silicone rubber cable accessories is input into the trained deep belief network model. The probability of each category is calculated by passing through each layer of restricted Boltzmann machine and softmax classifier in turn: For each layer of restricted Boltzmann machine, based on , calculate the activation value of the hidden layer of the l-th layer restricted Boltzmann machine ,in is the electrical performance factor of the pre-treated silicone rubber cable accessories, is the weight of the l-th layer restricted Boltzmann machine, is the hidden layer bias of the l-th layer restricted Boltzmann machine, is the sigmoid function, , x represents the input of the sigmoid function; The hidden layer activation value of the lth layer of restricted Boltzmann machine is used as the input of the next layer of restricted Boltzmann machine. If it is the last layer of restricted Boltzmann machine, the output of the last layer of restricted Boltzmann machine is As the input of the softmax classifier, calculate , where is the output of the softmax classifier, is the softmax classifier weight, is the softmax classifier bias of the softmax classifier; The output of the softmax classifier is converted into a probability distribution through the softmax function; According to the output probability vector, the category with the highest probability is selected as the final electrical performance classification result.

[0012] Preferably, in step seven, the process of judging whether the silicone rubber cable accessories are qualified is: The infrared-ultraviolet correlation grade of silicone rubber cable accessories, the appearance inspection grade of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories are stored as specified tags; Obtain a mapping table of designated tags and comprehensive scores of nondestructive testing of silicone rubber cable accessories pre-stored in the database, and find a matching comprehensive score of nondestructive testing of silicone rubber cable accessories according to the designated tags by searching the mapping table; Compare the comprehensive score of the nondestructive testing of silicone rubber cable accessories with the defined score of the nondestructive testing of silicone rubber cable accessories stored in the database; If the comprehensive score of the non-destructive testing of the silicone rubber cable accessories is not lower than the defined score of the non-destructive testing of the silicone rubber cable accessories, the silicone rubber cable accessories corresponding to the comprehensive score of the non-destructive testing of the silicone rubber cable accessories are qualified; If the comprehensive score of non-destructive testing of silicone rubber cable accessories is lower than the definition score of non-destructive testing of silicone rubber cable accessories, the silicone rubber cable accessories corresponding to the comprehensive score of non-destructive testing of silicone rubber cable accessories are unqualified, and an early warning reminder for replacement of silicone rubber cable accessories is issued.

[0013] A nondestructive testing device for silicone rubber cable accessories comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor. The above method is implemented by executing the computer program by the processor.

[0014] A nondestructive testing system for silicone rubber cable accessories, used to implement the above method, comprising: Infrared thermal imaging detection module, used to perform infrared thermal imaging detection on silicone rubber cable accessories and obtain infrared thermal imaging detection data; The infrared-ultraviolet correlation model analysis module is used to perform ultraviolet imaging detection on silicone rubber cable accessories, obtain ultraviolet imaging detection data, combine infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor of silicone rubber cable accessories, and match the infrared-ultraviolet correlation level of silicone rubber cable accessories; A surface layer defect data acquisition module is used to determine the surface layer defect data of silicone rubber cable accessories by using ultrasonic surface wave detection and optical coherence tomography detection; The cable accessories appearance inspection module is used to collect the appearance inspection data of the silicone rubber cable accessories, and combine the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories; The cable accessory appearance detection module is used to output the silicone rubber cable accessory appearance detection level based on the BP neural network model and the silicone rubber cable accessory appearance detection signal; The cable accessories electrical performance classification module is used to collect the electrical performance test data of silicone rubber cable accessories, obtain the electrical performance factors of silicone rubber cable accessories, and use the deep belief network model to output the electrical performance classification results of silicone rubber cable accessories; The cable accessories qualification judgment module is used to match the comprehensive score of non-destructive testing of silicone rubber cable accessories based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, to judge whether the silicone rubber cable accessories are qualified.

[0015] The present invention has the following beneficial effects: The present invention can obtain information on silicone rubber cable accessories from different angles by comprehensively using multiple technologies such as infrared thermal imaging detection, ultraviolet imaging detection, ultrasonic surface wave detection, optical coherence tomography detection, appearance detection and electrical performance detection, covering multiple dimensions such as thermal characteristics, corona discharge characteristics, surface layer defects, appearance conditions and electrical performance, avoiding the limitations of a single detection method, and can fully understand the status of cable accessories. It provides a standardized basis for quality control and operation and maintenance of cable accessories, which helps to improve the management level and operation efficiency of the power system.

[0016] The present invention can comprehensively utilize the complementary information of the two detection data by constructing an infrared-ultraviolet correlation model, more accurately reflect the actual situation of cable accessories, and the output correlation factor and correlation level can locate the problem more accurately. At the same time, the application of BP neural network model and deep confidence network model can efficiently process and analyze the corresponding detection data, improve the accuracy and reliability of the detection results, and reduce the possibility of misjudgment and missed judgment. The data obtained by different detection methods are fused and analyzed, such as combining surface layer defect data and appearance detection data to obtain appearance detection signals, and then based on this, the appearance detection level evaluation is performed, which can fully explore the internal connection between the data and make the final detection result more in line with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of the steps of the method of the present invention; Figure 2 It is a module connection diagram of the system of the present invention. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0019] Embodiment 1: like Figure 1 As shown, a nondestructive testing method for silicone rubber cable accessories includes: Step 1: Perform infrared thermal imaging detection on silicone rubber cable accessories to obtain infrared thermal imaging detection data.

[0020] The specific acquisition process is: perform infrared thermal imaging detection on silicone rubber cable accessories, use infrared thermal imager to detect the temperature distribution on the surface of silicone rubber cable accessories; obtain infrared thermal imaging detection data, including the temperature difference on the surface of silicone rubber cable accessories , Maximum surface temperature of silicone rubber cable accessories , Thermal contrast of the highest surface temperature area of ​​silicone rubber cable accessories .

[0021] By detecting the surface temperature distribution and obtaining data such as surface temperature difference and maximum temperature, abnormal heating caused by poor contact and insulation degradation of cable accessories can be detected, and timely warnings can be given at the early stage of the fault to avoid the expansion of the fault. Data such as thermal contrast in the area with the highest surface temperature can help accurately locate the specific location where problems may exist in cable accessories, and can intuitively reflect the operating status of silicone rubber cable accessories, providing a quantitative basis for evaluating their health status.

[0022] Step 2: Perform ultraviolet imaging detection on silicone rubber cable accessories to obtain ultraviolet imaging detection data, combine it with infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor of silicone rubber cable accessories, and match the infrared-ultraviolet correlation level of silicone rubber cable accessories.

[0023] The specific process is: perform UV imaging detection on silicone rubber cable accessories and obtain UV imaging detection data, including UV photon counting , discharge intensity , discharge area ; Construct an infrared-ultraviolet correlation model for silicone rubber cable accessories, and output the correlation factor of silicone rubber cable accessories based on ultraviolet imaging detection data and infrared thermal imaging detection data. The correlation factor of silicone rubber cable accessories serves as the analysis basis for matching the infrared-ultraviolet correlation level of silicone rubber cable accessories; The infrared-ultraviolet correlation model of silicone rubber cable accessories is expressed as: ; In the formula, is the silicone rubber cable accessory correlation factor, Infrared analysis factor for silicone rubber cable accessories, UV analysis factor for silicone rubber cable accessories, For setting The weight factor of For setting The weight factor of , e is a natural constant; lg represents the logarithm with base 10.

[0024] Obtain the silicone rubber cable accessory association factor and the silicone rubber cable accessory infrared-ultraviolet association level mapping table pre-stored in the database, and find the matching silicone rubber cable accessory infrared-ultraviolet association level according to the silicone rubber cable accessory association factor by searching the mapping table.

[0025] Combining infrared thermal imaging and ultraviolet imaging detection data, comprehensive analysis from the two aspects of thermal characteristics and corona discharge characteristics can more comprehensively and accurately reflect the operating status of cable accessories, avoid the one-sidedness of a single detection method, and improve the accuracy of the assessment. By building a model to calculate the correlation factor and match the correlation level, the status of cable accessories is quantitatively graded, making it easier to take targeted measures, such as giving priority to accessories with lower levels and potential risks.

[0026] Ultraviolet imaging detection can detect corona discharge, and correlation analysis with infrared thermal imaging detection data can help to uncover potential problems such as temperature anomalies caused by discharge, provide early warning of possible faults, and reduce the probability of accidents.

[0027] Step 3: Use ultrasonic surface wave detection and optical coherence tomography detection to determine the surface defect data of silicone rubber cable accessories.

[0028] The specific determination process is: emit ultrasonic surface waves to the surface of the silicone rubber cable accessories, and perform optical coherence tomography detection. Using the principle of low-coherence light interference, the surface of the silicone rubber cable accessories is tomographically imaged with micron-level resolution to obtain the surface layer defect data of the silicone rubber cable accessories; the surface layer defect data of the silicone rubber cable accessories includes the maximum defect depth of the surface layer of the silicone rubber cable accessories , Number of pores on the surface of silicone rubber cable accessories , Number of bubbles on the surface of silicone rubber cable accessories .

[0029] By utilizing the low-coherence light interference principle of optical coherence tomography, tomography with micron-level resolution can be achieved. Combined with ultrasonic surface wave detection, it can accurately detect subtle defects in the surface layer of silicone rubber cable accessories, such as tiny pores and bubbles. By detecting surface layer defects, problems can be discovered in time before serious failures occur in cable accessories, and repair or replacement measures can be taken in advance to avoid serious consequences such as degradation of insulation performance and electrical failures due to the gradual development of surface layer defects, thereby ensuring the stable operation of the power system.

[0030] Step 4: Collect the appearance inspection data of the silicone rubber cable accessories, and combine it with the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories.

[0031] The specific process is: Collect the appearance inspection data of silicone rubber cable accessories, including the number of cracks on the appearance of silicone rubber cable accessories , Appearance corrosion area of ​​silicone rubber cable accessories , Total area of ​​impurities embedded in the appearance of silicone rubber cable accessories Based on the appearance inspection data of silicone rubber cable accessories and the surface defect data of silicone rubber cable accessories, a comprehensive analysis is performed to obtain the appearance inspection signal of silicone rubber cable accessories. The appearance inspection signal of silicone rubber cable accessories is used as the analysis basis for outputting the appearance inspection grade of silicone rubber cable accessories. The calculation method of the surface detection signal of the silicone rubber cable accessories is: ; In the formula, It is the detection signal of the appearance of the silicone rubber cable accessories. It is the appearance detection signal of silicone rubber cable accessories. is the surface detection signal of the silicone rubber cable accessories, A is the set The weight factor of The weight factor of ; ln represents the logarithm with base e.

[0032] Comprehensively considering the appearance data such as cracks, corrosion area, impurity embedding, as well as the depth of defects, pores and bubble counts on the surface layer, can fully understand the appearance damage of cable accessories, avoiding focusing only on the local area and ignoring other potential problems. Quantifying the appearance condition into specific values, so that the appearance status of different cable accessories has a unified measurement standard, and accessories with more serious problems can be quickly screened out.

[0033] The appearance inspection signal is used as the analysis basis for the appearance inspection level, which is helpful for the hierarchical management of cable accessories. According to the classification results, maintenance resources can be reasonably arranged, and cable accessories with lower levels and more prominent appearance problems can be given priority, so as to improve maintenance efficiency and ensure the safe operation of the power system.

[0034] Step 5: Based on the BP neural network model and combined with the silicone rubber cable accessory surface detection signal, output the silicone rubber cable accessory surface detection level.

[0035] The specific output method is as follows: obtain the BP neural network model training data set, normalize the input variables and output variables, and map them to the [0,1] interval; determine that the number of neurons in the input layer of the BP neural network model is equal to the number of input variables, and the input variable is the silicone rubber cable accessory appearance detection signal in the training data set, which is 1; the number of neurons in the output layer is equal to the number of categories of silicone rubber cable accessory appearance detection level, and the number of categories of silicone rubber cable accessory appearance detection level in the training data set is 3, which are 1, 2, and 3, representing high, medium, and low, respectively, and use unique hot encoding to convert them into numerical vectors, with high level represented as [1,0,0], medium level represented as [0,1,0], and low level represented as [0,0,1]; the number of neurons in the hidden layer is h; randomly initialize the connection weights from the input layer to the hidden layer and from the hidden layer to the output layer, as well as the thresholds of the hidden layer and the output layer; input the silicone rubber cable accessory appearance detection signal in the preprocessed training data set; Use the Sigmoid activation function to calculate the output of the sth neuron in the hidden layer : ; In the formula, is the connection weight from the input layer to the sth neuron in the hidden layer, is the threshold value of the sth neuron in the hidden layer (bias value, used to adjust the output of the activation function of the neuron), s is the neuron number of the hidden layer, s=1,2,3...h; Use the Sigmoid activation function to calculate the output of the tth neuron in the output layer : ; In the formula, is the connection weight from the sth neuron in the hidden layer to the tth neuron in the output layer, is the threshold of the tth neuron in the output layer, t is the neuron number of the output layer, t=1,2,3; The error for the tth neuron in the output layer is , and its calculation formula is: ; In the formula, is the expected output of the tth neuron in the output layer, is the actual output of the tth neuron in the output layer; The gradient descent method is used to update the connection weights and thresholds until the network error is less than the accuracy threshold stored in the database or the maximum number of iterations is reached; the preprocessed current silicone rubber cable accessory surface detection signal is input into the trained BP neural network model; the output vector of the output layer is obtained; based on the output vector, the maximum membership principle is used to determine the silicone rubber cable accessory surface detection level.

[0036] BP neural network has powerful nonlinear mapping capabilities and can automatically learn the relationship between complex features in appearance detection signals and appearance detection levels. It does not require manual extraction of complex features and can efficiently process large amounts of complex data information.

[0037] By learning and training a large amount of training data, the connection weights and thresholds of the network are continuously adjusted, so that the model can accurately determine the corresponding detection level based on the external detection signal, which is more accurate and consistent than manual evaluation. The use of BP neural network can realize automatic external detection level evaluation, reduce manual intervention, improve evaluation efficiency, and provide strong support for the intelligent operation and maintenance of cable accessories, so as to facilitate the timely detection of cable accessories with external problems and ensure the stable operation of the power system.

[0038] Step 6. Collect electrical performance test data of silicone rubber cable accessories, obtain electrical performance factors of silicone rubber cable accessories, and use a deep belief network model (Deep Belief Networks) to output electrical performance classification results of silicone rubber cable accessories.

[0039] The specific output process is: Collect the electrical performance test data of silicone rubber cable accessories, including the insulation resistance of silicone rubber cable accessories , Dielectric loss factor of silicone rubber cable accessories , Silicone rubber cable accessories conductor resistance ; Based on the electrical performance test data of silicone rubber cable accessories, the electrical performance factor of silicone rubber cable accessories is obtained through comprehensive analysis. The electrical performance factor of silicone rubber cable accessories is used as the analysis basis for outputting the electrical performance classification results of silicone rubber cable accessories; The formula for obtaining the electrical performance factor of silicone rubber cable accessories is: ; In the formula, Electrical performance factor of silicone rubber cable accessories; The electrical performance factor of the silicone rubber cable accessories is normalized, and the preprocessed electrical performance factor of the silicone rubber cable accessories is input into the trained deep belief network model. The probability of each category is calculated by passing through each layer of restricted Boltzmann machine and softmax classifier in turn: For each layer of restricted Boltzmann machine, based on , calculate the activation value of the hidden layer of the l-th layer restricted Boltzmann machine ,in is the electrical performance factor of the pre-treated silicone rubber cable accessories, is the weight of the l-th layer restricted Boltzmann machine, is the hidden layer bias of the l-th layer restricted Boltzmann machine, is the sigmoid function, , x represents the input of the sigmoid function; The hidden layer activation value of the lth layer of restricted Boltzmann machine is used as the input of the next layer of restricted Boltzmann machine. If it is the last layer of restricted Boltzmann machine, the output of the last layer of restricted Boltzmann machine is As the input of the softmax classifier, calculate , where is the output of the softmax classifier, is the softmax classifier weight, The softmax classifier is biased for the softmax classifier; the output of the softmax classifier is converted into a probability distribution through the softmax function; according to the output probability vector, the category with the largest probability is selected as the final electrical performance classification result.

[0040] Taking into account the insulation resistance, dielectric loss factor, conductor resistance and other electrical performance test data, the electrical performance factor is calculated to fully reflect the electrical performance of cable accessories and avoid the one-sidedness of a single indicator evaluation.

[0041] The deep belief network model is composed of multiple restricted Boltzmann machine layers (each restricted Boltzmann machine layer is a separate restricted Boltzmann machine) stacked together. It has powerful unsupervised learning and feature extraction capabilities, can automatically learn the complex characteristics and inherent laws in electrical performance factors, and explore the potential connections between data.

[0042] After layer-by-layer processing and probability calculation of the softmax classifier, the electrical performance of cable accessories can be classified more accurately, providing a reliable basis for judging the electrical health status of cable accessories, helping to promptly detect accessories with poor electrical performance and reduce the risk of electrical failures.

[0043] Step 7. Based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, match the comprehensive score of non-destructive testing of silicone rubber cable accessories to determine whether the silicone rubber cable accessories are qualified.

[0044] The specific process is as follows: the infrared-ultraviolet correlation grade of silicone rubber cable accessories, the appearance inspection grade of silicone rubber cable accessories and the classification results of electrical performance of silicone rubber cable accessories are stored as designated tags; a mapping table of designated tags and comprehensive scores of nondestructive testing of silicone rubber cable accessories pre-stored in the database is obtained, and the matching comprehensive scores of nondestructive testing of silicone rubber cable accessories are found according to the designated tags by searching the mapping table; the comprehensive scores of nondestructive testing of silicone rubber cable accessories are compared with the defined scores of nondestructive testing of silicone rubber cable accessories stored in the database; If the comprehensive score of non-destructive testing of silicone rubber cable accessories is not lower than the definition score of non-destructive testing of silicone rubber cable accessories, then the silicone rubber cable accessories corresponding to the comprehensive score of non-destructive testing of silicone rubber cable accessories are qualified; if the comprehensive score of non-destructive testing of silicone rubber cable accessories is lower than the definition score of non-destructive testing of silicone rubber cable accessories, then the silicone rubber cable accessories corresponding to the comprehensive score of non-destructive testing of silicone rubber cable accessories are unqualified, and an early warning reminder for replacement of silicone rubber cable accessories is issued.

[0045] Taking into comprehensive consideration the infrared-ultraviolet correlation level, appearance inspection level and electrical performance classification results, the silicone rubber cable accessories are evaluated from multiple dimensions such as thermal characteristics, appearance status and electrical performance, avoiding the limitations of single-dimensional judgment and reflecting the actual condition of the cable accessories more comprehensively and accurately.

[0046] By storing the specified tags, using the pre-stored mapping table to find the matching comprehensive score, and then comparing it with the defined score, a standardized judgment process is formed. This method makes the judgment process clear and objective, and different testers can reach consistent conclusions, which improves the reliability and comparability of the test results.

[0047] For unqualified cable accessories whose comprehensive scores are lower than the defined scores, timely issuance of replacement warning prompts will help operation and maintenance personnel take timely measures to avoid power accidents caused by cable accessory failures and ensure the safe and stable operation of the power system. It will also help to reasonably arrange maintenance resources and improve maintenance efficiency.

[0048] In a specific example, the temperature difference on the surface of the silicone rubber cable accessory The maximum surface temperature of the silicone rubber cable accessories is 5.2 degrees Celsius. The thermal contrast ratio of the highest surface temperature area of ​​the silicone rubber cable accessories is 68 degrees Celsius. 0.85, UV photon count 1200, discharge intensity 45mW / cm², discharge area 2.5cm², set The weight factor is 0.6, and the The weight factor is 0.4, and the calculated correlation factor of silicone rubber cable accessories is 80.0914. According to the preset mapping table, the infrared-ultraviolet correlation level of the matched silicone rubber cable accessories is excellent.

[0049] Maximum defect depth of the surface layer of silicone rubber cable accessories 0.03mm, number of pores on the surface of silicone rubber cable accessories 12. Number of bubbles on the surface of silicone rubber cable accessories 5, the number of cracks on the appearance of silicone rubber cable accessories 3. Appearance corrosion area of ​​silicone rubber cable accessories The total area of ​​impurities embedded in the appearance of silicone rubber cable accessories is 0.8cm² 0.2cm², set The weight factor is 0.7, and the The weight factor is 0.3, the calculated silicone rubber cable accessory surface detection signal is 0.62, the BP neural network output vector is [0.85, 0.12, 0.03], and the silicone rubber cable accessory surface detection level is determined to be advanced according to the maximum membership principle.

[0050] Silicone Rubber Cable Accessories Insulation Resistance Dielectric loss factor for 5000 ohm, silicone rubber cable accessories 0.02, silicone rubber cable accessories conductor resistance The electrical performance factor of the silicone rubber cable accessories is 0.05 ohms, and the calculated electrical performance factor is 37.8049. According to the probability vector [0.92, 0.07, 0.01] output by the deep belief network model, the category with the largest probability is selected as the final electrical performance classification result of 0.92, and the corresponding category is excellent.

[0051] The infrared-ultraviolet correlation grade of silicone rubber cable accessories, the appearance inspection grade of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories are stored as specified labels. The label combination is: A+Advanced+Excellent. The comprehensive score of non-destructive testing of the matched silicone rubber cable accessories is 95 points, which is not less than the defined score of 80 points for non-destructive testing of silicone rubber cable accessories. The silicone rubber cable accessories corresponding to the comprehensive score of non-destructive testing of the silicone rubber cable accessories are qualified.

[0052] Embodiment 2: Based on Example 1, this example defines the training process of the deep belief network model, specifically introducing an adversarial training mechanism to enhance the robustness of the model, specifically including: Construct a generative adversarial network architecture, where the generator receives a random noise vector and generates simulated electrical performance factor data, and the discriminator is modified from the original deep belief network; In the pre-training stage, the discriminator is supervised and trained using a real electrical performance factor dataset to enable it to have basic classification capabilities; In the adversarial training phase, the generator and discriminator are updated alternately: the generator optimizes the distribution matching degree between the generated data and the real data through gradient backpropagation, and the discriminator simultaneously learns to identify the real data and the generated data and maintains the classification accuracy; After freezing the generator parameters, fine-tune the discriminator, focusing on strengthening the classification decision boundary of boundary samples; Finally, the optimized discriminator is deployed for electrical performance classification, and its hidden layer adds an attention mechanism to dynamically adjust the feature weights.

[0053] Embodiment 3: A nondestructive testing device for silicone rubber cable accessories includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The method described in Example 1 or Example 2 is implemented by executing the computer program by the processor.

[0054] Embodiment 4: like Figure 2 As shown, a nondestructive testing system for silicone rubber cable accessories is used in the method described in Example 1 or Example 2, comprising: Infrared thermal imaging detection module, used to perform infrared thermal imaging detection on silicone rubber cable accessories and obtain infrared thermal imaging detection data; The infrared-ultraviolet correlation model analysis module is used to perform ultraviolet imaging detection on silicone rubber cable accessories, obtain ultraviolet imaging detection data, combine infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor of silicone rubber cable accessories, and match the infrared-ultraviolet correlation level of silicone rubber cable accessories; A surface layer defect data acquisition module is used to determine the surface layer defect data of silicone rubber cable accessories by using ultrasonic surface wave detection and optical coherence tomography detection; The cable accessories appearance inspection module is used to collect the appearance inspection data of the silicone rubber cable accessories, and combine the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories; The cable accessory appearance detection module is used to output the silicone rubber cable accessory appearance detection level based on the BP neural network model and the silicone rubber cable accessory appearance detection signal; The cable accessories electrical performance classification module is used to collect the electrical performance test data of silicone rubber cable accessories, obtain the electrical performance factors of silicone rubber cable accessories, and use the deep belief network model to output the electrical performance classification results of silicone rubber cable accessories; The cable accessories qualification judgment module is used to match the comprehensive score of non-destructive testing of silicone rubber cable accessories based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, to judge whether the silicone rubber cable accessories are qualified.

Claims

1. A nondestructive testing method for silicone rubber cable accessories, characterized in that: The following steps are involved: Step 1: Perform infrared thermal imaging detection on silicone rubber cable accessories to obtain infrared thermal imaging detection data; Step 2: Perform ultraviolet imaging detection on silicone rubber cable accessories to obtain ultraviolet imaging detection data, combine with infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor for silicone rubber cable accessories, and match the infrared-ultraviolet correlation level for silicone rubber cable accessories; Step 3: Use ultrasonic surface wave detection and optical coherence tomography detection to determine the surface layer defect data of the silicone rubber cable accessories; Step 4: Collect the appearance inspection data of the silicone rubber cable accessories, and combine the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories; Step 5: Based on the BP neural network model and in combination with the silicone rubber cable accessory surface detection signal, output the silicone rubber cable accessory surface detection level; Step 6: Collect electrical performance test data of silicone rubber cable accessories, obtain electrical performance factors of silicone rubber cable accessories, and use a deep belief network model to output electrical performance classification results of silicone rubber cable accessories; Step 7. Based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, match the comprehensive score of non-destructive testing of silicone rubber cable accessories to determine whether the silicone rubber cable accessories are qualified.

2. A nondestructive testing method for silicone rubber cable accessories according to claim 1, characterized in that: In the step 1, the process of obtaining infrared thermal imaging detection data is as follows: Conduct infrared thermal imaging inspection on silicone rubber cable accessories, and use infrared thermal imager to detect the temperature distribution on the surface of silicone rubber cable accessories; Obtain infrared thermal imaging test data, including surface temperature differences of silicone rubber cable accessories , Maximum surface temperature of silicone rubber cable accessories , Thermal contrast of the highest surface temperature area of ​​silicone rubber cable accessories .

3. A nondestructive testing method for silicone rubber cable accessories according to claim 2, characterized in that: In the step 2, an infrared-ultraviolet correlation model of silicone rubber cable accessories is constructed, a correlation factor of silicone rubber cable accessories is output, and an infrared-ultraviolet correlation level of silicone rubber cable accessories is matched. The specific process is as follows: Perform UV imaging inspection on silicone rubber cable accessories and obtain UV imaging inspection data, including UV photon counting , discharge intensity , discharge area ; Construct an infrared-ultraviolet correlation model for silicone rubber cable accessories, and output the correlation factor of silicone rubber cable accessories based on ultraviolet imaging detection data and infrared thermal imaging detection data. The correlation factor of silicone rubber cable accessories serves as the analysis basis for matching the infrared-ultraviolet correlation level of silicone rubber cable accessories. The infrared-ultraviolet correlation model of silicone rubber cable accessories is expressed as: ； In the formula, is the silicone rubber cable accessory correlation factor, Infrared analysis factor for silicone rubber cable accessories, UV analysis factor for silicone rubber cable accessories, For setting The weight factor of For setting The weight factor of , e is a natural constant; Obtain the silicone rubber cable accessory association factor and the silicone rubber cable accessory infrared-ultraviolet association level mapping table pre-stored in the database, and find the matching silicone rubber cable accessory infrared-ultraviolet association level according to the silicone rubber cable accessory association factor by searching the mapping table.

4. A nondestructive testing method for silicone rubber cable accessories according to claim 1, characterized in that: In the step 3, the process of determining the surface layer defect data of the silicone rubber cable accessories is as follows: Ultrasonic surface waves are emitted to the surface of silicone rubber cable accessories, and optical coherence tomography is performed. The principle of low-coherence light interference is used to perform micron-level resolution tomography on the surface of silicone rubber cable accessories to obtain surface layer defect data of silicone rubber cable accessories. Silicone rubber cable accessories surface layer defect data including the maximum defect depth of silicone rubber cable accessories surface layer , Number of pores on the surface of silicone rubber cable accessories , Number of bubbles on the surface of silicone rubber cable accessories .

5. A nondestructive testing method for silicone rubber cable accessories according to claim 4, characterized in that: In the step 4, the process of obtaining the silicone rubber cable accessory surface detection signal is as follows: Collect appearance inspection data of silicone rubber cable accessories, including the number of cracks on the appearance of silicone rubber cable accessories , Appearance corrosion area of ​​silicone rubber cable accessories , Total area of ​​impurities embedded in the appearance of silicone rubber cable accessories ; Based on the appearance inspection data of silicone rubber cable accessories and the surface layer defect data of silicone rubber cable accessories, the appearance inspection signal of silicone rubber cable accessories is obtained through comprehensive analysis. The appearance inspection signal of silicone rubber cable accessories is used as the analysis basis for outputting the appearance inspection grade of silicone rubber cable accessories. The calculation method of the surface detection signal of the silicone rubber cable accessories is: ； In the formula, It is the detection signal of the appearance of the silicone rubber cable accessories. It is the appearance detection signal of silicone rubber cable accessories. is the surface detection signal of the silicone rubber cable accessories, A is the set The weight factor of The weight factor of , e is a natural constant.

6. A nondestructive testing method for silicone rubber cable accessories according to claim 5, characterized in that: In the step 5, the method of outputting the appearance inspection grade of the silicone rubber cable accessories is: Obtain the BP neural network model training data set, normalize the input variables and output variables, and map them to the [0,1] interval; Determine that the number of neurons in the input layer of the BP neural network model is equal to the number of input variables, and the input variable is the surface detection signal of the silicone rubber cable accessories in the training data set, which is 1; The number of neurons in the output layer is equal to the number of categories of the silicone rubber cable accessories appearance detection level. The number of categories of the silicone rubber cable accessories appearance detection level in the training data set is 3, which are 1, 2, and 3, representing high, medium, and low levels, respectively. Unique hot encoding is used to convert them into numerical vectors, with high levels represented as [1,0,0], medium levels represented as [0,1,0], and low levels represented as [0,0,1]. The number of neurons in the hidden layer is h; Randomly initialize the connection weights from the input layer to the hidden layer and from the hidden layer to the output layer, as well as the thresholds of the hidden layer and the output layer; Input the surface detection signal of the silicone rubber cable accessory in the preprocessed training data set; Use the Sigmoid activation function to calculate the output of the sth neuron in the hidden layer : ； In the formula, is the connection weight from the input layer to the sth neuron in the hidden layer, is the threshold of the sth neuron in the hidden layer, s is the neuron number of the hidden layer, s=1,2,3...h; Use the Sigmoid activation function to calculate the output of the tth neuron in the output layer : ； In the formula, is the connection weight from the sth neuron in the hidden layer to the tth neuron in the output layer, is the threshold of the tth neuron in the output layer, t is the neuron number of the output layer, t=1,2,3; The error for the tth neuron in the output layer is , and its calculation formula is: ； In the formula, is the expected output of the tth neuron in the output layer, is the actual output of the tth neuron in the output layer; The gradient descent method is used to update the connection weights and thresholds until the network error is less than the accuracy threshold stored in the database or the maximum number of iterations is reached; Input the pre-processed current silicone rubber cable accessory appearance detection signal into the trained BP neural network model; Get the output vector of the output layer; According to the output vector, the maximum membership principle is adopted to determine the appearance inspection level of silicone rubber cable accessories.

7. A nondestructive testing method for silicone rubber cable accessories according to claim 1, characterized in that: In step 6, the process of outputting the electrical performance classification results of the silicone rubber cable accessories is as follows: Collect electrical performance test data of silicone rubber cable accessories, including insulation resistance of silicone rubber cable accessories , Dielectric loss factor of silicone rubber cable accessories , Conductor resistance of silicone rubber cable accessories ; Based on the electrical performance test data of silicone rubber cable accessories, the electrical performance factor of silicone rubber cable accessories is obtained through comprehensive analysis. The electrical performance factor of silicone rubber cable accessories is used as the analysis basis for outputting the electrical performance classification results of silicone rubber cable accessories. The formula for obtaining the electrical performance factor of silicone rubber cable accessories is: ； In the formula, is the electrical performance factor of silicone rubber cable accessories, and e is a natural constant; The electrical performance factor of the silicone rubber cable accessories is normalized, and the preprocessed electrical performance factor of the silicone rubber cable accessories is input into the trained deep belief network model. The probability of each category is calculated by passing through each layer of restricted Boltzmann machine and softmax classifier in turn: For each layer of restricted Boltzmann machine, based on , calculate the activation value of the hidden layer of the l-th layer restricted Boltzmann machine ,in is the electrical performance factor of the pre-treated silicone rubber cable accessories, is the weight of the l-th layer restricted Boltzmann machine, is the hidden layer bias of the l-th layer restricted Boltzmann machine, is the sigmoid function, , x represents the input of the sigmoid function; The hidden layer activation value of the lth layer of restricted Boltzmann machine is used as the input of the next layer of restricted Boltzmann machine. If it is the last layer of restricted Boltzmann machine, the output of the last layer of restricted Boltzmann machine is As the input of the softmax classifier, calculate , where is the output of the softmax classifier, is the softmax classifier weight, is the softmax classifier bias of the softmax classifier; The output of the softmax classifier is converted into a probability distribution through the softmax function; According to the output probability vector, the category with the highest probability is selected as the final electrical performance classification result.

8. The nondestructive testing method for silicone rubber cable accessories according to claim 1 is characterized in that: In step 7, the process of judging whether the silicone rubber cable accessories are qualified is as follows: The infrared-ultraviolet correlation grade of silicone rubber cable accessories, the appearance inspection grade of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories are stored as specified tags; Obtain a mapping table of designated tags and comprehensive scores of nondestructive testing of silicone rubber cable accessories pre-stored in the database, and find a matching comprehensive score of nondestructive testing of silicone rubber cable accessories according to the designated tags by searching the mapping table; Compare the comprehensive score of the nondestructive testing of silicone rubber cable accessories with the defined score of the nondestructive testing of silicone rubber cable accessories stored in the database; If the comprehensive score of the non-destructive testing of the silicone rubber cable accessories is not lower than the defined score of the non-destructive testing of the silicone rubber cable accessories, the silicone rubber cable accessories corresponding to the comprehensive score of the non-destructive testing of the silicone rubber cable accessories are qualified; If the comprehensive score of non-destructive testing of silicone rubber cable accessories is lower than the definition score of non-destructive testing of silicone rubber cable accessories, the silicone rubber cable accessories corresponding to the comprehensive score of non-destructive testing of silicone rubber cable accessories are unqualified, and an early warning reminder for replacement of silicone rubber cable accessories is issued.

9. A nondestructive testing device for silicone rubber cable accessories, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, and the method according to any one of claims 1 to 8 is implemented by executing the computer program by the processor.

10. A nondestructive testing system for silicone rubber cable accessories, used to implement the method described in any one of claims 1 to 8, characterized in that: include: Infrared thermal imaging detection module, used to perform infrared thermal imaging detection on silicone rubber cable accessories and obtain infrared thermal imaging detection data; The infrared-ultraviolet correlation model analysis module is used to perform ultraviolet imaging detection on silicone rubber cable accessories, obtain ultraviolet imaging detection data, combine infrared thermal imaging detection data, build an infrared-ultraviolet correlation model for silicone rubber cable accessories, output the correlation factor of silicone rubber cable accessories, and match the infrared-ultraviolet correlation level of silicone rubber cable accessories; A surface layer defect data acquisition module is used to determine the surface layer defect data of silicone rubber cable accessories by using ultrasonic surface wave detection and optical coherence tomography detection; The cable accessories appearance inspection module is used to collect the appearance inspection data of the silicone rubber cable accessories, and combine the surface layer defect data of the silicone rubber cable accessories to obtain the appearance inspection signal of the silicone rubber cable accessories; The cable accessory appearance detection module is used to output the silicone rubber cable accessory appearance detection level based on the BP neural network model and the silicone rubber cable accessory appearance detection signal; The cable accessories electrical performance classification module is used to collect the electrical performance test data of silicone rubber cable accessories, obtain the electrical performance factors of silicone rubber cable accessories, and use the deep belief network model to output the electrical performance classification results of silicone rubber cable accessories; The cable accessories qualification judgment module is used to match the comprehensive score of non-destructive testing of silicone rubber cable accessories based on the infrared-ultraviolet correlation level of silicone rubber cable accessories, the appearance inspection level of silicone rubber cable accessories and the electrical performance classification results of silicone rubber cable accessories, to judge whether the silicone rubber cable accessories are qualified.

Images