Method and system for detecting aging degree of household electric wire cable based on multi-parameter fusion

Abstract

The application provides a household electric wire cable aging degree detection method and system based on multi-parameter fusion, which comprises the following steps: collecting working parameters and environmental parameters of the household electric wire cable to be detected, and constructing an electric wire cable parameter matrix; inputting the electric wire cable parameter matrix into a parameter preprocessing module to eliminate noise interference and complete normalization processing, and obtaining a standardized parameter vector; inputting the standardized parameter vector into a preset aging detection model to predict the aging probability distribution of the electric wire cable, and determining a candidate aging degree set according to the aging probability distribution; the aging detection model is generated based on historical sample data training; for the candidate aging degree set, the possibility of each candidate aging degree is re-evaluated by extracting the time correlation characteristics between multiple parameters, and the prediction confidence value of each candidate aging degree is obtained; and the final aging degree evaluation result is obtained according to the prediction confidence value of each candidate aging degree.

Description

Technical Field

[0001] This invention relates to the field of cable aging detection, and in particular to a method and system for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion. Background Technology

[0002] As the fundamental carriers of power transmission and signal control, wires and cables play a crucial role in household electricity use and other fields. With prolonged use, under the combined effects of electrical, thermal, and mechanical stresses, the insulation and mechanical properties of wires and cables gradually deteriorate, resulting in aging. Severely aged wires and cables are prone to short circuits and leakage, and in serious cases, can even lead to fires, threatening people's lives and property.

[0003] Currently, aging tests on wires and cables are primarily conducted before they leave the factory. Manufacturers use simulation methods such as thermal aging tests and electrical aging tests to accelerate the aging of wires and cables under laboratory conditions to assess the product's service life and reliability. However, in actual use after leaving the factory, wires and cables are affected by a combination of factors, including ambient temperature, humidity, load changes, and laying methods, resulting in a significant difference in their aging process compared to the simulated conditions during factory testing. Existing technologies typically lack effective means to monitor the aging degree of in-service wires and cables; repairs are often only carried out after a fault occurs, by which time economic losses or even personal injury may have already occurred.

[0004] Among existing wire and cable testing technologies, the megohmmeter is a commonly used testing tool. The megohmmeter determines the insulation performance of wires and cables by measuring their insulation resistance, and thus estimates their degree of aging. However, when using a megohmmeter, the power supply to the device under test must be disconnected, and the wires and cables must be short-circuited to ground to ensure measurement safety. This process is cumbersome, and disconnecting the power disrupts normal life, especially in locations requiring continuous power supply, where power outage testing is difficult. Furthermore, the megohmmeter can only measure the single parameter of insulation resistance, making it difficult to comprehensively reflect the aging state of wires and cables. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention provides a method and system for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, so as to improve at least one of the above-mentioned problems.

[0006] This invention provides a method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, comprising:

[0007] S101, Collect the operating parameters and environmental parameters of the household electrical wires and cables to be tested, and construct a parameter matrix for the electrical wires and cables; the operating parameters include current carrying capacity, leakage current, and operating temperature; the environmental parameters include ambient humidity and usage time;

[0008] S102, the wire and cable parameter matrix is ​​input into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector;

[0009] S103, The standardized parameter vector is input into a preset aging detection model to predict the aging probability distribution of wires and cables, and a candidate aging degree set is determined based on the aging probability distribution; the aging detection model is generated based on historical sample data.

[0010] S104, For the candidate aging degree set, the probability of each candidate aging degree is re-evaluated by extracting the time correlation features between multiple parameters, and the predicted confidence value of each candidate aging degree is obtained.

[0011] S105, the final aging assessment result is obtained based on the predicted confidence value of each candidate aging degree.

[0012] Preferably, step S101 specifically includes:

[0013] The current carrying capacity signal of the wires and cables is collected by a current transformer, and the current carrying capacity signal is converted into a digital quantity after being processed by a signal conditioning circuit.

[0014] The leakage current signal of the wire and cable is collected by a leakage current sensor, and the leakage current signal is amplified and converted by an A / D converter to obtain a digital value.

[0015] Operating temperature data is obtained by collecting the insulation temperature or conductor temperature of wires and cables using temperature sensors.

[0016] Ambient humidity data of the environment where wires and cables are laid is collected using a humidity sensor;

[0017] The usage time of wires and cables can be obtained by reading their identification information or querying installation records.

[0018] Preferably, step S102 specifically includes:

[0019] The collected raw parameter data is filtered, and a median filtering algorithm is used to eliminate occasional pulse interference, while a low-pass filtering algorithm is used to smooth data fluctuations.

[0020] The filtered parameter data is normalized to convert various parameters into standardized values ​​in the range of 0 to 1.

[0021] Preferably, step S103 specifically includes:

[0022] The standardized parameter vector is input into the aging detection model, which is constructed using the following mathematical relationships:

[0023] D = w1*I load + w2*I leak + w3*T+ w4*H+ w5*t

[0024] Among them, I load I represents the actual measured current carrying capacity. leak The actual measured leakage current value represents the rated current carrying capacity of the wire and cable; T represents the actual measured operating temperature value; H represents the actual measured ambient relative humidity value; and t represents the length of time since the wire and cable were put into use.

[0025] D represents the aging index, with a value ranging from 0 to 1. The closer the D value is to 1, the more severe the aging. w1, w2, w3, w4, and w5 are preset weight coefficients, which are determined through training with a large number of samples and satisfy w1 + w2 + w3 + w4 + w5 = 1.

[0026] Based on the aging degree index D, a threshold determination strategy is used to determine the candidate aging degree set; wherein, when D is greater than or equal to the first preset threshold D1, the wire and cable is determined to be in a severely aged state; when D is greater than or equal to the second preset threshold D2 and less than D1, the wire and cable is determined to be in a moderately aged state; when D is less than D2, the wire and cable is determined to be in a slightly aged state; the value range of D1 is 0.7 to 0.8, and the value range of D2 is 0.4 to 0.5.

[0027] Preferably, the weight coefficient training process of the aging detection model includes:

[0028] Collect several wire and cable samples, each sample containing its actual current carrying capacity, leakage current, operating temperature, ambient humidity, and service time parameters during operation;

[0029] Professional technicians assess the actual aging level of each sample and generate sample label values.

[0030] Linear regression or neural network algorithms are used to train the sample data, and the weight coefficients are optimized to minimize the mean square error between the model output value and the actual aging degree assessment value.

[0031] Preferably, step S104 specifically includes:

[0032] For each candidate aging level, the time correlation features between the parameters in its corresponding standardized parameter vector are extracted; among them, a multi-dimensional time correlation feature vector is constructed by calculating the correlation coefficient between current carrying capacity and leakage current, and the interaction feature values ​​between operating temperature and usage time.

[0033] The multidimensional time-related feature vector is input into the confidence prediction network to obtain the predicted confidence value for each candidate aging level. The confidence prediction network consists of three fully connected layers, each containing 128 neurons. The ReLU activation function is used, and the output layer uses the Sigmoid activation function. The calculation process of the confidence prediction network is defined as follows:

[0034] Confidence=σ(W3⋅ReLU(W2⋅ReLU(W1⋅F+b1)+b2)+b3);

[0035] Where F represents the multidimensional temporal correlation feature vector, W1, W2, W3 and b1, b2, b3 are the learnable parameters of the network, σ represents the Sigmoid activation function, and Confidence represents the prediction confidence value.

[0036] Preferably, in step S105, the Soft-NMS algorithm is used to suppress redundant results; wherein:

[0037] First, sort all candidate aging levels according to their corresponding confidence scores, and calculate the IoU overlap between the candidate aging level with the highest confidence score and all other candidate aging levels.

[0038] Then, the confidence scores of candidate aging degrees with IoU overlap higher than the preset overlap threshold are attenuated. This operation is repeated for all remaining candidate aging degrees to obtain the final reordered set of candidate aging degrees, where N is the number of final aging degree evaluation results used for output.

[0039] Preferably, step S105 specifically includes:

[0040] Adaptive weights are used to weight and fuse different types of parameters along the parameter dimension to align the differences in importance of each parameter in the evaluation;

[0041] The fused feature vectors are dimensionally transformed, and the feature dimensions are mapped to the aging degree category space through a fully connected layer;

[0042] The probability distribution of each aging level category is output using the Softmax activation function to estimate the aging level of wires and cables. The calculation process for aging level classification is defined as: P = softmax(W⋅F processed +b), where F processedLet W be the processed feature vector, b be the weight matrix, b be the bias vector, Softmax be the normalized exponential function, and P be the probability distribution vector for each aging level.

[0043] Preferably, for the aging detection model, three loss functions are used to optimize its learning parameters, including aging degree regression loss, confidence prediction loss, and aging level classification loss; the aging degree regression loss uses the mean squared error loss function, which is expressed as: Lregression=N∑(D predi -D truei ) 2 D predi Let D be the predicted aging index for the i-th sample. truei Let N be the true aging index of the i-th sample, and N be the number of training samples;

[0044] The confidence prediction loss is defined as: Lconfidence = −M∑[y i ⋅log(σ i )+(1−y i )⋅log(1−σ i )], where y i Let σ be the true label of the i-th sample. i Let M be the prediction confidence score of the i-th sample, and M be the number of samples used for training.

[0045] The aging level classification loss uses the cross-entropy loss function, which is expressed as: Lclassification=−K*∑q ij ⋅log(p ij ), where q ij Let p be the true probability distribution of the i-th sample belonging to the j-th aging level. ij Let K be the predicted probability distribution, and K be the sample size.

[0046] This invention also provides a device for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, comprising:

[0047] The parameter acquisition unit is used to collect the operating parameters and environmental parameters of the household electrical wires and cables to be tested, and to construct a parameter matrix for the electrical wires and cables. The operating parameters include current carrying capacity, leakage current, and operating temperature. The environmental parameters include ambient humidity and usage time.

[0048] The normalization unit is used to input the wire and cable parameter matrix into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector.

[0049] An aging detection unit is used to input the standardized parameter vector into a preset aging detection model to predict the aging probability distribution of wires and cables, and to determine a candidate aging degree set based on the aging probability distribution; the aging detection model is generated based on historical sample data.

[0050] The confidence calculation unit is used to re-evaluate the probability of each candidate aging degree by extracting the time correlation features between multiple parameters from the candidate aging degree set, and obtain the predicted confidence value of each candidate aging degree.

[0051] An aging assessment unit is used to obtain the final aging assessment result based on the predicted confidence value of each candidate aging level.

[0052] Compared with the prior art, the present invention has at least the following advantages:

[0053] 1) This invention overcomes the limitations of single-parameter detection. By simultaneously collecting parameters from five dimensions—current carrying capacity, leakage current, operating temperature, ambient humidity, and usage time—it constructs a multi-parameter fusion parameter matrix for wires and cables, which can more comprehensively reflect the aging state of wires and cables.

[0054] 2) This invention designs and implements an effective aging detection model. By fusing multiple parameters through a preset weighted calculation formula, it outputs a quantitative aging index, thereby realizing a quantitative assessment of the aging degree of wires and cables and solving the problem of coarse detection results in the prior art.

[0055] 3) This invention designs and implements a reasonable confidence prediction network. It extracts the time correlation features between multiple parameters for confidence evaluation and uses a non-maximum suppression algorithm to remove redundant results, which effectively improves the accuracy and reliability of aging degree detection. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating the method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion provided in the first embodiment of the present invention.

[0058] Figure 2 This is a schematic diagram of the structure of a household electrical wire and cable aging detection system based on multi-parameter fusion provided in the second embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1 This invention provides a method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, which includes the following steps:

[0062] S101, Collect the operating parameters and environmental parameters of the household electrical wires and cables to be tested, and construct a parameter matrix for the electrical wires and cables; the operating parameters include current carrying capacity, leakage current and operating temperature; the environmental parameters include ambient humidity and usage time.

[0063] In this embodiment of the invention, the operating parameters and environmental parameters of the wire and cable to be tested are first collected to construct a wire and cable parameter matrix. The operating parameters include current carrying capacity, leakage current, and operating temperature; the environmental parameters include ambient humidity and usage time.

[0064] In this embodiment, in order to comprehensively acquire multi-dimensional parameter information reflecting the aging state of wires and cables, the parameter acquisition module consists of the following five independent sensor units:

[0065] (1) Current carrying capacity acquisition unit

[0066] The current-carrying capacity signal of a wire or cable is acquired using a current transformer. The current transformer employs a through-core structure, with its iron core fitted onto the conductor of the wire or cable under test to sense the alternating current in the conductor. The secondary output of the current transformer is processed by a signal conditioning circuit and converted into a digital signal; where I... load I represents the actual measured current carrying capacity. rated This indicates the rated current carrying capacity of the wire or cable. The signal conditioning circuit includes a sampling resistor, an operational amplifier, and a low-pass filter. It amplifies the weak current signal and filters out high-frequency interference before sending it to the A / D converter to be converted into a digital value for subsequent processing.

[0067] (2) Leakage current acquisition unit

[0068] The leakage current signal of the wires and cables is collected by a leakage current sensor. The leakage current sensor adopts a zero-sequence current transformer structure with a small core opening, facilitating installation on a DIN rail inside the distribution box. The output of the leakage current sensor is amplified and converted to a digital value by an A / D converter; where I... leak This represents the actual measured leakage current value. With the amplifier circuit gain set to 100, it can detect leakage current changes at the 0.5mA level, sensitively reflecting even minor damage to the insulation layer of wires and cables.

[0069] (3) Temperature acquisition unit

[0070] Operating temperature data is obtained by collecting the insulation or conductor temperature of wires and cables using temperature sensors. The temperature sensors employ a dual-redundancy design using both an NTC thermistor and an infrared temperature sensor. The NTC thermistor is fixed to the surface of the cable insulation and makes close contact with the cable sheath via thermally conductive silicone grease to accurately sense the insulation temperature. The infrared temperature sensor is installed on the inner wall of the distribution box facing the cable for non-contact measurement of the cable surface temperature. Here, T represents the actual measured operating temperature value. The measurement results from both temperature sensors can be cross-checked, improving the reliability of the temperature data.

[0071] (4) Humidity acquisition unit

[0072] A humidity sensor collects ambient humidity data of the cable laying environment. The humidity sensor uses a capacitive humidity-sensitive element, and its sensing membrane is made of polymer material, capable of sensing changes in ambient relative humidity; where H represents the actual measured ambient relative humidity value. The humidity sensor has a measurement range of 10%RH to 95%RH, an accuracy of ±2%RH, and is installed inside the distribution box near the cable laying area to accurately reflect the ambient humidity of the cable operating environment.

[0073] (5) Use time acquisition unit

[0074] The usage time of electrical wires and cables can be obtained by reading their identification information or querying installation records. In this embodiment, each electrical wire and cable to be tested is equipped with a QR code or RFID tag during installation. The tag stores information such as the cable's specifications, rated voltage, conductor cross-section, and installation date; where 'tt' represents the length of time the cable has been in use. Maintenance personnel can obtain the cable's usage time by scanning the QR code or reading the RFID tag using a handheld terminal. If the cable has no identification information, the installation date can be obtained by consulting the as-built drawings of the power distribution system or the equipment ledger.

[0075] The five independent sensor units mentioned above collect parameters such as current carrying capacity, leakage current, operating temperature, ambient humidity, and usage time, respectively, to construct a parameter matrix for the wire and cable, providing a data foundation for subsequent aging assessment.

[0076] S102, the wire and cable parameter matrix is ​​input into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector.

[0077] In this embodiment, the wire and cable parameter matrix can be input into the parameter preprocessing module to eliminate noise interference and complete normalization processing to obtain a standardized parameter vector.

[0078] In this embodiment, the parameter preprocessing module includes two parts: a filtering processing unit and a normalization processing unit.

[0079] (1) Filtering

[0080] The collected raw parameter data is filtered. A median filter algorithm is used to eliminate occasional pulse interference, and a low-pass filter algorithm is used to smooth data fluctuations. The median filter algorithm takes the median of five consecutive sample values ​​as the filter output to eliminate occasional pulse interference; the low-pass filter algorithm uses a first-order inertial filter algorithm to smooth the data and eliminate high-frequency components. The mathematical expression for the filtering process is:

[0081] Y k =α•X k +(1−α)•Y k−1

[0082] Among them, Y k Let X be the output value of the k-th filter. k k is the kth sample value, Y k−1 The output value of the previous filter is α, which is the filter coefficient, ranging from 0.1 to 0.3. In this embodiment, α is set to 0.2.

[0083] (2) Normalization

[0084] The filtered parameter data is then normalized, converting all parameters into standardized values ​​within the range of 0 to 1. The mathematical expression for this normalization process is:

[0085] I load_norm =I load / I rated

[0086] I leak_norm = I leak / I leak_max

[0087] T norm= T / T max

[0088] H norm =H / 100

[0089] t norm =t / t service

[0090] Among them, I load_norm I leak_norm T norm H norm t norm To normalize the current carrying capacity, leakage current, operating temperature, ambient humidity, and usage time, I leak_max To preset the maximum leakage current threshold, it is set to 100mA in this embodiment. max The maximum allowable operating temperature for wires and cables is 70℃ for PVC insulated cables and 90℃ for cross-linked polyethylene insulated cables. service The expected service life of the wires and cables is set to 30 years in this embodiment.

[0091] After processing by the parameter preprocessing module, a standardized parameter vector is obtained. This standardized parameter vector includes five dimensions: normalized current carrying capacity, normalized leakage current, normalized operating temperature, normalized ambient humidity, and normalized usage time, providing a unified input format for the subsequent aging test model.

[0092] S103, the standardized parameter vector is input into a preset aging detection model to predict the aging probability distribution of wires and cables, and the candidate aging degree set is determined according to the aging probability distribution; the aging detection model is generated based on historical sample data.

[0093] In this embodiment, the standardized parameter vector is input into a preset aging detection model to predict the aging probability distribution of wires and cables, and a candidate aging degree set is determined based on the aging probability distribution.

[0094] The aging detection model is implemented using a weighted summation model, and its mathematical formula is as follows:

[0095] D = w1*I load + w2*I leak + w3*T+ w4*H+ w5*t

[0096] Wherein, D represents the aging degree index, with a value ranging from 0 to 1. The closer the D value is to 1, the more severe the aging degree. w1, w2, w3, w4, and w5 are preset weight coefficients, each of which is determined through training with a large number of samples and satisfies w1 + w2 + w3 + w4 + w5 = 1.

[0097] (1) Weight coefficient training process

[0098] In this embodiment, the weight coefficient training process of the aging detection model includes: collecting several wire and cable samples, each sample containing parameters such as actual current carrying capacity, leakage current, operating temperature, ambient humidity, and usage time; having professional technicians evaluate the actual aging degree of each sample and generate sample label values; and using a linear regression algorithm or a neural network algorithm to train the sample data and optimize the weight coefficients so that the mean square error between the model output value and the actual aging degree evaluation value is minimized.

[0099] Based on extensive sample training, the recommended weighting coefficients in this embodiment are: w1=0.15, w2=0.30, w3=0.20, w4=0.10, and w5=0.25. Among these, leakage current has the highest weight because it is a direct indicator of cable insulation degradation; service time has the second highest weight because service time is a fundamental factor in cable aging; operating temperature and current carrying capacity have moderate weights because overload operation and overheating accelerate cable aging; and ambient humidity has a relatively low weight because its impact on aging is relatively indirect.

[0100] (2) Determination of candidate aging degree set

[0101] In this embodiment, a threshold determination strategy is used to determine the candidate aging degree set based on the aging degree index D. The specific determination rule is as follows:

[0102] When D is greater than or equal to the first preset threshold D1, the wire and cable are determined to be in a severely aged state; when D is greater than or equal to the second preset threshold D2 and less than D1, the wire and cable are determined to be in a moderately aged state; when D is less than D2 and less than D1, the wire and cable are determined to be in a slightly aged state.

[0103] In this embodiment, specifically, the value range of D1 is 0.7 to 0.8; and the value range of D2 is 0.4 to 0.5. Through threshold determination, the continuous aging degree index DD is discretized into three candidate aging degree levels, forming a candidate aging degree set.

[0104] S104, For the candidate aging degree set, the probability of each candidate aging degree is re-evaluated by extracting the time correlation features between multiple parameters, and a predicted confidence value for each candidate aging degree is obtained.

[0105] In this embodiment, the probability of each candidate aging degree is re-evaluated by extracting the temporal correlation features between multiple parameters from the candidate aging degree set, thereby obtaining the predicted confidence value of each candidate aging degree.

[0106] A confidence prediction network was designed to achieve this function. This network can further optimize the determination results of candidate aging degree by extracting and evaluating the temporal correlation features between multiple parameters.

[0107] (1) Extraction of temporal correlation features

[0108] For each candidate aging level, extract the temporal correlation features between parameters in its corresponding standardized parameter vector. The specific implementation is as follows:

[0109] A multidimensional time-related feature vector is constructed by calculating the correlation coefficient between current carrying capacity and leakage current, and the interaction feature value between operating temperature and usage time. The correlation coefficient reflects the degree of linear correlation between changes in current carrying capacity and leakage current; the higher the correlation coefficient, the more significant the impact of changes in current carrying capacity on leakage current. The interaction feature value reflects the accelerated aging effect of operating temperature accumulating over usage time; the higher the interaction feature value, the more obvious the accelerating effect of temperature on aging.

[0110] (2) Confidence prediction network

[0111] The multidimensional time-related feature vector is input into the confidence prediction network to obtain the predicted confidence value for each candidate aging level. The confidence prediction network consists of three fully connected layers, each containing 128 neurons. The ReLU activation function is used, and the output layer uses the Sigmoid activation function.

[0112] The calculation process of the confidence prediction network is defined as follows:

[0113] Confidence=σ(W3⋅ReLU(W2⋅ReLU(W1⋅F+b1)+b2)+b3)

[0114] Where F represents the multidimensional temporal correlation feature vector, W1, W2, W3 and b1, b2, b3 are the learnable parameters of the network, σ represents the Sigmoid activation function, and Confidence represents the prediction confidence value.

[0115] The output value of the confidence prediction network ranges from 0 to 1. A higher confidence level indicates a higher reliability of the candidate aging degree determination. By introducing the confidence prediction network, obviously erroneous candidate aging degrees can be effectively screened out, improving the overall evaluation accuracy.

[0116] S105, the final aging assessment result is obtained based on the predicted confidence value of each candidate aging level.

[0117] In this embodiment, based on the predicted confidence value of each candidate aging degree, redundant candidate aging degrees are removed by non-maximum suppression to obtain the final aging degree evaluation result.

[0118] In this embodiment, the Soft-NMS algorithm is used to suppress redundant results. Compared with the traditional NMS algorithm, this algorithm can handle the overlap between adjacent candidate aging levels more smoothly and avoid the problem of unstable evaluation results caused by threshold abrupt changes.

[0119] (1) Implementation of Soft-NMS algorithm

[0120] The specific implementation method is as follows: First, sort all candidate aging levels according to their corresponding confidence scores, and calculate the IoU overlap between the candidate aging level with the highest confidence score and all other candidate aging levels; then, perform a decay process on the confidence scores of candidate aging levels with IoU overlap higher than a preset overlap threshold, and repeat this operation on all other candidate aging levels to obtain the final reordered set of candidate aging levels, where N is the number of final aging level evaluation results used for output.

[0121] The decay function of the Soft-NMS algorithm is a Gaussian decay function, and its mathematical expression is:

[0122] score′=score*e −(IoU^2) / σ

[0123] Wherein, score is the original confidence score of the candidate aging degree, IoU is the overlap between the current highest confidence candidate and the candidate, and σ is the decay parameter, which is taken as 0.5 in this embodiment.

[0124] The preset overlap threshold is set to 0.5. When the IoU between two candidate aging levels is greater than 0.5, they are considered to have a redundant relationship and need to be suppressed. After processing by the Soft-NMS algorithm, the candidate aging levels with the highest confidence are retained as the final aging level evaluation results.

[0125] (2) Classification of aging levels

[0126] Adaptive weights are used to weight and fuse different types of parameters along the parameter dimension to align the differences in importance of each parameter in the evaluation; the fused feature vector is dimensionally transformed and mapped to the aging degree category space through a fully connected layer; the probability distribution of each aging degree category is output through the Softmax activation function to estimate the aging level of the wire and cable.

[0127] The calculation process for aging level classification is defined as follows:

[0128] P=softmax(W⋅F processed +b), where F processedLet W be the processed feature vector, b be the weight matrix, b be the bias vector, Softmax be the normalized exponential function, and P be the probability distribution vector for each aging level.

[0129] In this embodiment, the aging level is divided into three levels: mild aging, moderate aging, and severe aging. Each level corresponds to a probability value, and the level with the highest probability is the final judgment result.

[0130] (3) Loss function optimization

[0131] For the aging detection model, three loss functions are used to optimize its learning parameters, including aging degree regression loss, confidence prediction loss, and aging level classification loss.

[0132] The regression loss for aging degree is expressed using the mean squared error loss function, which is as follows:

[0133] Its expression is: Lregression = N∑(D predi -D truei ) 2 D predi Let D be the predicted aging index for the i-th sample. truei Let N be the true aging index of the i-th sample, and N be the number of training samples;

[0134] The confidence prediction loss is defined as: Lconfidence = −M∑[y i ⋅log(σ i )+(1−y i )⋅log(1−σ i )], where y i Let σ be the true label of the i-th sample. i Let M be the prediction confidence score of the i-th sample, and M be the number of samples used for training.

[0135] The aging level classification loss uses the cross-entropy loss function, which is expressed as: Lclassification=−K*∑q ij ⋅log(p ij ), where q ij Let p be the true probability distribution of the i-th sample belonging to the j-th aging level. ij Let K be the predicted probability distribution, and K be the sample size.

[0136] By jointly optimizing the three loss functions mentioned above, the aging detection model can simultaneously possess the ability to accurately predict the aging degree index, provide a reliable confidence assessment, and correctly classify the aging level.

[0137] Compared with the prior art, the present invention has at least the following advantages:

[0138] 1) This invention overcomes the limitations of single-parameter detection. By simultaneously collecting parameters from five dimensions—current carrying capacity, leakage current, operating temperature, ambient humidity, and usage time—it constructs a multi-parameter fusion parameter matrix for wires and cables, which can more comprehensively reflect the aging state of wires and cables.

[0139] 2) This invention designs and implements an effective aging detection model. By fusing multiple parameters through a preset weighted calculation formula, it outputs a quantitative aging index, thereby realizing a quantitative assessment of the aging degree of wires and cables and solving the problem of coarse detection results in the prior art.

[0140] 3) This invention designs and implements a reasonable confidence prediction network. It extracts the time correlation features between multiple parameters for confidence evaluation and uses a non-maximum suppression algorithm to remove redundant results, which effectively improves the accuracy and reliability of aging degree detection.

[0141] Please see Figure 2 The second embodiment of the present invention also provides a multi-parameter fusion-based aging detection system for household electrical wires and cables, comprising:

[0142] The parameter acquisition unit 210 is used to acquire the operating parameters and environmental parameters of the household electrical wires and cables to be tested, and to construct a parameter matrix for the electrical wires and cables; the operating parameters include current carrying capacity, leakage current and operating temperature; the environmental parameters include ambient humidity and usage time.

[0143] The normalization unit 220 is used to input the wire and cable parameter matrix into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector.

[0144] The aging detection unit 230 is used to input the standardized parameter vector into a preset aging detection model to predict the aging probability distribution of wires and cables, and to determine a candidate aging degree set based on the aging probability distribution; the aging detection model is generated based on historical sample data.

[0145] The confidence calculation unit 240 is used to re-evaluate the probability of each candidate aging degree by extracting the time correlation features between multiple parameters from the candidate aging degree set, and obtain the predicted confidence value of each candidate aging degree.

[0146] Aging assessment unit 250 is used to obtain the final aging assessment result based on the predicted confidence value of each candidate aging degree.

[0147] The third embodiment of the present invention also provides a multi-parameter fusion-based aging detection system for household electrical wires and cables, which includes a memory and a processor. The memory stores a computer program, which can be executed by the processor to realize the multi-parameter fusion-based aging detection method for household electrical wires and cables as described above.

[0148] In the several embodiments provided in this invention, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0149] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0150] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further restrictions, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0151] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, characterized in that, include: S101: Collect the operating and environmental parameters of the household electrical wires and cables to be tested, and construct the electrical wire and cable parameter matrix; The operating parameters include current carrying capacity, leakage current, and operating temperature; the environmental parameters include ambient humidity and usage time. S102, the wire and cable parameter matrix is ​​input into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector; S103, The standardized parameter vector is input into a preset aging detection model to predict the aging probability distribution of wires and cables, and a candidate aging degree set is determined based on the aging probability distribution; the aging detection model is generated based on historical sample data. S104, For the candidate aging degree set, the probability of each candidate aging degree is re-evaluated by extracting the time correlation features between multiple parameters, and the predicted confidence value of each candidate aging degree is obtained. S105, the final aging assessment result is obtained based on the predicted confidence value of each candidate aging degree.

2. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 1, characterized in that, Step S101 specifically includes: The current carrying capacity signal of the wires and cables is collected by a current transformer, and the current carrying capacity signal is converted into a digital quantity after being processed by a signal conditioning circuit. The leakage current signal of the wire and cable is collected by a leakage current sensor, and the leakage current signal is amplified and converted by an A / D converter to obtain a digital value. Operating temperature data is obtained by collecting the insulation temperature or conductor temperature of wires and cables using temperature sensors. Ambient humidity data of the environment where wires and cables are laid is collected using a humidity sensor; The usage time of wires and cables can be obtained by reading their identification information or querying installation records.

3. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 2, characterized in that, Step S102 specifically includes: The collected raw parameter data is filtered, and a median filtering algorithm is used to eliminate occasional pulse interference, while a low-pass filtering algorithm is used to smooth data fluctuations. The filtered parameter data is normalized to convert various parameters into standardized values ​​in the range of 0 to 1.

4. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 3, characterized in that, Step S103 specifically includes: The standardized parameter vector is input into the aging detection model, which is constructed using the following mathematical relationships: D = w1*I load + w2*I leak + w3*T+ w4*H+ w5*t; Among them, I load I represents the actual measured current carrying capacity. leak The actual measured leakage current value represents the rated current carrying capacity of the wire and cable; T represents the actual measured operating temperature value; H represents the actual measured ambient relative humidity value; and t represents the length of time since the wire and cable were put into use. D represents the aging degree index, with a value ranging from 0 to 1. The closer the D value is to 1, the more severe the aging degree. w1, w2, w3, w4, and w5 are preset weight coefficients, which are determined through training with a large number of samples and satisfy w1 + w2 + w3 + w4 + w5 = 1. Based on the aging degree index D, a threshold determination strategy is used to determine the candidate aging degree set; wherein, when D is greater than or equal to the first preset threshold D1, the wire and cable is determined to be in a severely aged state; when D is greater than or equal to the second preset threshold D2 and less than D1, the wire and cable is determined to be in a moderately aged state; when D is less than D2, the wire and cable is determined to be in a slightly aged state; the value range of D1 is 0.7 to 0.8, and the value range of D2 is 0.4 to 0.

5.

5. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 4, characterized in that, The training process for the weight coefficients of the aging detection model includes: Collect several wire and cable samples, each sample containing its actual current carrying capacity, leakage current, operating temperature, ambient humidity, and service time parameters during operation; Professional technicians assess the actual aging level of each sample and generate sample label values. Linear regression or neural network algorithms are used to train the sample data, and the weight coefficients are optimized to minimize the mean square error between the model output value and the actual aging degree assessment value.

6. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 4, characterized in that, Step S104 specifically includes: For each candidate aging level, the time correlation features between the parameters in its corresponding standardized parameter vector are extracted; among them, a multi-dimensional time correlation feature vector is constructed by calculating the correlation coefficient between current carrying capacity and leakage current, and the interaction feature values ​​between operating temperature and usage time. The multidimensional time-related feature vector is input into the confidence prediction network to obtain the predicted confidence value for each candidate aging level. The confidence prediction network consists of three fully connected layers, each containing 128 neurons. The ReLU activation function is used, and the output layer uses the Sigmoid activation function. The calculation process of the confidence prediction network is defined as follows: Confidence=σ(W3⋅ReLU(W2⋅ReLU(W1⋅F+b1)+b2)+b3); Where F represents the multidimensional temporal correlation feature vector, W1, W2, W3 and b1, b2, b3 are the learnable parameters of the network, σ represents the Sigmoid activation function, and Confidence represents the prediction confidence value.

7. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 6, characterized in that, In step S105, the Soft-NMS algorithm is used to suppress redundant results; wherein: First, sort all candidate aging levels according to their corresponding confidence scores, and calculate the IoU overlap between the candidate aging level with the highest confidence score and all other candidate aging levels. Then, the confidence scores of candidate aging degrees with IoU overlap higher than the preset overlap threshold are attenuated. This operation is repeated for all remaining candidate aging degrees to obtain the final reordered set of candidate aging degrees, where N is the number of final aging degree evaluation results used for output.

8. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 1, characterized in that, Step S105 specifically includes: Adaptive weights are used to weight and fuse different types of parameters along the parameter dimension to align the differences in importance of each parameter in the evaluation; The fused feature vectors are dimensionally transformed, and the feature dimensions are mapped to the aging degree category space through a fully connected layer; The probability distribution of each aging level category is output using the Softmax activation function to estimate the aging level of wires and cables. The calculation process for aging level classification is defined as: P = softmax(W⋅F processed +b), where F processed Let W be the processed feature vector, b be the weight matrix, b be the bias vector, Softmax be the normalized exponential function, and P be the probability distribution vector for each aging level.

9. The method for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion according to claim 1, characterized in that, For the aging detection model, three loss functions are used to optimize its learning parameters, including aging degree regression loss, confidence prediction loss, and aging level classification loss; the aging degree regression loss uses the mean squared error loss function, which is expressed as: Lregression=N∑(D predi -D truei ) 2 D predi Let D be the predicted aging index for the i-th sample. truei Let N be the true aging index of the i-th sample, and N be the number of training samples; The confidence prediction loss is defined as: Lconfidence = −M∑[y i ⋅log(σ i )+(1−y i )⋅log(1−σ i )], where y i Let σ be the true label of the i-th sample. i Let M be the prediction confidence score of the i-th sample, and M be the number of samples used for training. The aging level classification loss uses the cross-entropy loss function, which is expressed as: Lclassification=−K*∑q ij ⋅log(p ij ), where q ij Let p be the true probability distribution of the i-th sample belonging to the j-th aging level. ij Let K be the predicted probability distribution, and K be the sample size.

10. A system for detecting the aging degree of household electrical wires and cables based on multi-parameter fusion, characterized in that, include: The parameter acquisition unit is used to collect the operating and environmental parameters of the household electrical wires and cables to be tested, and to construct the parameter matrix of the electrical wires and cables. The operating parameters include current carrying capacity, leakage current, and operating temperature; the environmental parameters include ambient humidity and usage time. The normalization unit is used to input the wire and cable parameter matrix into the parameter preprocessing module to eliminate noise interference and complete the normalization process to obtain a standardized parameter vector. An aging detection unit is used to input the standardized parameter vector into a preset aging detection model to predict the aging probability distribution of wires and cables, and to determine a candidate aging degree set based on the aging probability distribution; the aging detection model is generated based on historical sample data. The confidence calculation unit is used to re-evaluate the probability of each candidate aging degree by extracting the time correlation features between multiple parameters from the candidate aging degree set, and obtain the predicted confidence value of each candidate aging degree. An aging assessment unit is used to obtain the final aging assessment result based on the predicted confidence value of each candidate aging level.

Images