Cable aging test system and method based on Internet of Things

Abstract

The invention discloses a cable aging test system and method based on the Internet of Things, and relates to the technical field of cable testing, and the method comprises the steps: collecting cable operation parameters in real time through a circuit breaker node disposed in a cable line, and synchronously monitoring the environment temperature and humidity; wherein the cable operation parameters comprise conductor temperature, leakage current, voltage fluctuation value and load power; the collected cable operation parameters are transmitted to a cloud platform through an Internet of Things communication module; analyzing the aging degree value of the cable insulation layer based on an electrothermal two-factor synergistic aging model; dynamically calculating an ignition risk value of the cable through a fire risk assessment algorithm in combination with the aging degree value; and when the aging degree value or the ignition risk value exceeds a threshold value, generating an early warning report, pushing the early warning report to the terminal equipment, and marking the position of the hidden danger at the same time.

Description

Technical Field

[0001] This application relates to the technical field of cable testing. Specifically, it relates to a cable aging test system and method based on the Internet of Things. Background Art

[0002] During long-term operation, the cable insulation layer ages due to the combined action of electrothermal factors, which can easily lead to potential fire hazards such as short circuits and arc breakdowns. In the prior art, cable aging assessment mainly relies on laboratory accelerated aging tests (such as overcurrent and overvoltage tests), and indirectly evaluates the aging degree by testing parameters such as elongation at break and insulation resistance. However, such methods have at least the following defects:

[0003] Limitations of off-line detection: It is necessary to stop the machine for sampling and sending samples for inspection, and it is impossible to monitor the cable operation status in real time, making it difficult to detect aging hazards in a timely manner; Insufficient evaluation of single factor: Existing models are mostly based on single thermal aging or electrical aging data, without considering the non-linear impact of the combined electrothermal effect on material properties, resulting in low prediction accuracy.

[0004] For the above problems, no effective solution has been proposed yet. Summary of the Invention

[0005] Embodiments of this application provide a cable aging test system and method based on the Internet of Things to solve the above technical problems.

[0006] This application provides a cable aging test system based on the Internet of Things, including:

[0007] A cable data acquisition and monitoring module, configured to collect cable operation parameters in real time through breaker nodes deployed in the cable line, and synchronously monitor the ambient temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value, and load power;

[0008] An Internet of Things transmission module, configured to transmit the collected cable operation parameters to the cloud platform through the Internet of Things communication module;

[0009] An aging degree analysis module, configured to analyze the aging degree value of the cable insulation layer based on the combined electrothermal factor aging model;

[0010] A combustion risk value calculation module, configured to dynamically calculate the combustion risk value of the cable by combining the aging degree value through a fire risk assessment algorithm;

[0011] An early warning module, configured to generate an early warning report and push it to the terminal device when the aging degree value or the combustion risk value exceeds the threshold, and mark the location of the hidden danger at the same time.

[0012] This application provides an Internet of Things-based cable aging test method, including: collecting cable operation parameters in real time through circuit breaker nodes deployed in the cable line, and synchronously monitoring the ambient temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value, and load power;

[0013] Transmit the collected cable operation parameters to the cloud platform through the Internet of Things communication module;

[0014] Analyze the aging degree value of the cable insulation layer based on the electro-thermal dual-factor collaborative aging model;

[0015] Combined with the aging degree value, dynamically calculate the ignition risk value of the cable through the fire risk assessment algorithm;

[0016] When the aging degree value or the ignition risk value exceeds the threshold, generate a warning report and push it to the terminal device, and mark the hidden danger location at the same time.

[0017] Further, the Internet of Things communication module adopts a high-speed power line carrier wired and wireless multi-mode hybrid networking method;

[0018] The input parameters of the electro-thermal dual-factor collaborative aging model include: cable conductor cross-section, material type, and laying method; the current overload frequency, temperature rise rate, and insulation resistance decay curve in the historical operation data;

[0019] The processing logic of the electro-thermal dual-factor collaborative aging model includes the following steps:

[0020] Based on the cable conductor cross-section, material type, and laying method, construct a multi-physical field coupling characteristic matrix, and map the electro-thermal parameters to the material stress distribution function; wherein, the electro-thermal parameters include cable operation parameters and their derived electric field strength distribution and local thermal gradient;

[0021] Perform time series alignment and multi-resolution analysis on the current overload frequency, temperature rise rate, and insulation resistance decay curve in the historical operation data, and extract the characteristics of the aging-sensitive frequency band;

[0022] Adopt a dynamic Bayesian network to establish the probability dependence relationship between the electrical aging factor and the thermal aging factor;

[0023] Fuse the contribution degrees of the electro-thermal dual factors through a non-linear kernel function to generate a cable aging index, and correct the aging rate threshold in combination with the ambient temperature and humidity data.

[0024] Further, the electro-thermal dual-factor collaborative aging model is constructed in the following way:

[0025] Establish a cable accelerated aging test database, including the mechanical properties, insulation resistance, and thermal stability data of the cable under different current levels and temperature and humidity conditions;

[0026] The transient features in the electrothermal parameters are extracted by using the wavelet multi-scale decomposition algorithm, and the activation energy of material decomposition under the action of thermal stress is calculated by combining with the Arrhenius equation;

[0027] Based on the Weibull distribution, an electric breakdown probability model is constructed, and the transient features and steady-state aging data are fused through a Bayesian probability graph model to generate an electrothermal double-factor coupling coefficient;

[0028] The electrothermal double-factor coupling coefficient is nonlinearly weighted by using the random forest regression algorithm, an aging rate prediction function is established, and the model deviation is dynamically corrected through an online adversarial training mechanism.

[0029] Furthermore, the construction process of the Bayesian probability graph model specifically includes:

[0030] Taking the transient features as the observed nodes and the cable aging index as the latent variable nodes, a directed acyclic graph is constructed;

[0031] The Markov chain Monte Carlo sampling method is used to infer the node conditional probability distribution in combination with the aging test data;

[0032] The environmental temperature and humidity are introduced as covariate nodes to dynamically adjust the posterior probability weights of the electrothermal double factors;

[0033] The variational inference algorithm is used to compress the model complexity to achieve real-time inference at the edge computing device end.

[0034] Furthermore, the fire risk assessment algorithm includes:

[0035] According to the glowing wire ignition test data, a critical ignition temperature mapping table of the cable is established. The critical ignition temperature mapping table associates the non-linear relationship between the aging degree value and the critical ignition temperature, and the temperature prediction values in the missing interval of the glowing wire ignition test data are filled by using the fractal interpolation algorithm;

[0036] Based on the results of the arc shock test, a breakdown probability model of the cable insulation layer is constructed, and the hidden Markov model is used to dynamically predict the number of breakdowns that the cable can withstand at different aging stages. The input of the breakdown probability model includes the real-time leakage current spectrum characteristics and the voltage fluctuation peak value;

[0037] The aging index time series data and the breakdown probability curve are aligned by using the dynamic time warping algorithm to generate the evaluation result of the ignition risk value.

[0038] Furthermore, the breakdown probability model is constructed in the following way:

[0039] The high-frequency oscillation waveform and energy accumulation characteristics of the breakdown event are extracted from the arc shock test, and the sparse autoencoder is used to compress the high-dimensional data and extract the breakdown precursor pattern;

[0040] Based on the hidden Markov model, define the cable aging state as a hidden variable and the breakdown event as an observed variable, and decode the most likely aging state sequence corresponding to the real-time leakage current data through the Viterbi algorithm;

[0041] Combined with the environmental temperature and humidity data, adopt a multi-task reinforcement learning algorithm to dynamically optimize the breakdown probability threshold; wherein, the reward function of the multi-task reinforcement learning algorithm is generated by the matching degree between the historical breakdown event and the aging index.

[0042] Furthermore, the calculation of the ignition risk value further includes:

[0043] Introduce the test data of the cone calorimeter, analyze the ignition time and the peak value of the heat release rate of the cable under 30 kW of heat radiation, and extract the key combustion modes from the heat release curve by non-negative matrix factorization;

[0044] Fuse the key combustion modes with the time-frequency domain features of the real-time leakage current data to construct a multi-modal risk feature vector;

[0045] Adopt the deep deterministic policy gradient algorithm to dynamically correct the breakdown probability, and its action space is defined as the breakdown probability adjustment amount, and the state space includes the aging index, environmental temperature and humidity, and combustion mode similarity;

[0046] Through an online meta-learning framework, adapt the risk assessment strategies for different cable models and laying scenarios.

[0047] Furthermore, the warning report includes the cable aging level, remaining life prediction, and recommended maintenance measures, and displays the heat map of multi-dimensional perception data through a visual interface.

[0048] Furthermore, the breaker node integrates a high-speed ADC sampling circuit and a DSP signal processing unit, and the acquisition accuracy meets the requirements of conductor temperature ±2°C and leakage current ±1 mA.

[0049] Based on the embodiments provided in this application, by deploying breaker nodes in the cable line, it is possible to collect cable operation parameters in real time, such as conductor temperature, leakage current, voltage fluctuation value, and load power, and simultaneously monitor the ambient temperature and humidity. This solves the limitations of offline detection in the prior art, eliminates the need for shutdown sampling and inspection, and can promptly detect potential cable aging hazards. Analyzing the aging degree of the cable insulation layer based on the electro-thermal dual-factor collaborative aging model takes into account the non-linear impact of the electro-thermal collaborative effect on material properties, overcomes the problem of low prediction accuracy caused by existing models mostly based on single thermal aging or electrical aging data, and improves the accuracy of assessment. Combining the aging degree value, the ignition risk value of the cable is dynamically calculated through a fire risk assessment algorithm. When the aging degree value or the ignition risk value exceeds the threshold, an early warning report can be generated in a timely manner and pushed to the terminal device, while marking the location of the hidden danger, facilitating maintenance personnel to take measures in a timely manner and effectively preventing potential fire hazards. Using the Internet of Things communication module to transmit the collected data to the cloud platform realizes remote monitoring and management of the data, improves the intelligent level of cable maintenance, and reduces the workload and cost of manual inspection. Through real-time monitoring and early warning, problems can be detected and processed in a timely manner before serious failures are caused by cable aging, effectively reducing the fire risks such as short circuits and arc breakdowns caused by cable aging, and ensuring the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of this application, and the illustrative embodiments and descriptions thereof are used to explain this application without unduly limiting this application. In the drawings:

[0051] Figure 1 FIG. is a structural diagram of an optional Internet of Things-based cable aging test system according to an embodiment of this application;

[0052] Figure 2 FIG. is a flowchart of an optional Internet of Things-based cable aging test method according to an embodiment of this application;

[0053] Figure 3 FIG. is a flowchart of another optional Internet of Things-based cable aging test method according to an embodiment of this application.

[0054] The realization of the objectives, functional features, and advantages of the present invention will be further described in conjunction with the embodiments with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

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

[0056] Optionally, as Figure 1 shown, the present application provides an Internet of Things-based cable aging test system, including:

[0057] A cable data acquisition and monitoring module 101, configured to collect cable operation parameters in real time through breaker nodes deployed in the cable line, and synchronously monitor the ambient temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value, and load power;

[0058] An Internet of Things transmission module 102, configured to transmit the collected cable operation parameters to the cloud platform through an Internet of Things communication module;

[0059] An aging degree analysis module 103, configured to analyze the aging degree value of the cable insulation layer based on an electrothermal dual-factor collaborative aging model;

[0060] A combustion risk value calculation module 104, configured to dynamically calculate the combustion risk value of the cable by combining the aging degree value through a fire risk assessment algorithm;

[0061] An early warning module 105, configured to generate an early warning report and push it to the terminal device when the aging degree value or the combustion risk value exceeds the threshold, and mark the hidden danger location at the same time.

[0062] Optionally, as Figure 2 shown, the present application provides an Internet of Things-based cable aging test method, including:

[0063] S201, collect cable operation parameters in real time through breaker nodes deployed in the cable line, and synchronously monitor the ambient temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value, and load power;

[0064] S202, transmit the collected cable operation parameters to the cloud platform through an Internet of Things communication module;

[0065] S203, analyze the aging degree value of the cable insulation layer based on an electrothermal dual-factor collaborative aging model;

[0066] S204, dynamically calculate the combustion risk value of the cable by combining the aging degree value through a fire risk assessment algorithm;

[0067] S205. When the aging degree value or the ignition risk value exceeds the threshold, generate a warning report and push it to the terminal device, and mark the hidden danger location at the same time.

[0068] In the related art, there is a lack of dynamic monitoring of fire hazard parameters such as the critical temperature of cable ignition and the number of times of arc shock resistance, making it difficult to evaluate the fire risk during actual operation. And the monitoring means rely on manual inspection or single sensors, and multi-parameter fusion perception cannot be achieved.

[0069] Based on the embodiments provided in the present application, by deploying breaker nodes in the cable line, cable operation parameters such as conductor temperature, leakage current, voltage fluctuation value, and load power can be collected in real time, and the ambient temperature and humidity can be monitored synchronously. It solves the limitations of offline detection in the prior art, eliminates the need for shutdown sampling and inspection, and can timely detect cable aging hidden dangers. Analyze the aging degree of the cable insulation layer based on the electro-thermal dual-factor collaborative aging model, considering the non-linear influence of the electro-thermal collaborative effect on material properties, overcoming the problem of low prediction accuracy caused by existing models mostly based on single thermal aging or electrical aging data, and improving the accuracy of evaluation. Combine the aging degree value and dynamically calculate the ignition risk value of the cable through the fire risk assessment algorithm. When the aging degree value or the ignition risk value exceeds the threshold, a warning report can be generated in time and pushed to the terminal device, and the hidden danger location is marked at the same time, facilitating the maintenance personnel to take measures in time and effectively preventing fire hazards. Use the Internet of Things communication module to transmit the collected data to the cloud platform, realizing remote monitoring and management of the data, improving the intelligent level of cable maintenance, and reducing the workload and cost of manual inspection. Through real-time monitoring and warning, problems can be detected and processed in time before serious failures are caused by cable aging, effectively reducing the fire risks such as short circuits and arc breakdowns caused by cable aging, and ensuring the safe and stable operation of the power system.

[0070] Further, the Internet of Things communication module adopts a high-speed power line carrier wired and wireless multi-mode hybrid networking method;

[0071] The input parameters of the electro-thermal dual-factor collaborative aging model include: cable conductor cross-section, material type, and laying method; the current overload frequency, temperature rise rate, and insulation resistance decay curve in the historical operation data.

[0072] As Figure 3 shown, the processing logic of the electro-thermal dual-factor collaborative aging model includes the following steps:

[0073] S301. Based on the cable conductor cross-section, material type, and laying method, construct a multi-physical field coupling characteristic matrix, and map the electro-thermal parameters to a material stress distribution function; among them, the electro-thermal parameters include cable operation parameters and the derived electric field strength distribution and local thermal gradient;

[0074] Among them, the electric field strength distribution is calculated from the voltage fluctuation value and the conductor cross-section, reflecting the local electric field distortion of the cable; the local thermal gradient is deduced from the conductor temperature and the laying method (such as the heat dissipation condition), quantifying the thermal stress distribution;

[0075] S302. Perform time series alignment and multi-resolution analysis on the current overload frequency, temperature rise rate, and insulation resistance decay curve in the historical operation data, and extract the characteristics of the aging-sensitive frequency band;

[0076] S303. Use a dynamic Bayesian network to establish the probabilistic dependence relationship between the electrical aging factor and the thermal aging factor; among them, the electrical aging factor includes the electric field distortion rate and the partial discharge energy; the thermal aging factor includes the oxidation rate and the pyrolysis activation energy;

[0077] S304. Generate the cable aging index by fusing the contribution degrees of the electro-thermal dual factors through a non-linear kernel function, and correct the aging rate threshold in combination with the environmental temperature and humidity data.

[0078] Based on the embodiments provided in this application, by constructing a multi-physical field coupling feature matrix, the electro-thermal parameters are mapped into a material stress distribution function, accurately quantifying the local aging hot spots of the cable. Use a dynamic Bayesian network to establish the probabilistic dependence relationship between the electrical aging factor (electric field distortion rate) and the thermal aging factor (oxidation rate), and capture the non-linear interaction effect of electro-thermal synergistic aging.

[0079] Furthermore, the electro-thermal dual-factor synergistic aging model is constructed in the following manner:

[0080] Establish a cable accelerated aging test database, including the mechanical properties, insulation resistance, and thermal stability data of the cable under different current levels, temperature and humidity conditions;

[0081] Use the wavelet multi-scale decomposition algorithm to extract the transient characteristics in the electro-thermal parameters, and calculate the material decomposition activation energy under the action of thermal stress in combination with the Arrhenius equation; among them, the transient characteristics in the electro-thermal parameters include current spikes and temperature rise mutations;

[0082] Based on the Weibull distribution, construct an electric breakdown probability model, and fuse the transient characteristics and steady-state aging data through a Bayesian probability graph model to generate an electro-thermal dual-factor coupling coefficient;

[0083] Use the random forest regression algorithm to perform non-linear weighting on the electro-thermal dual-factor coupling coefficient, establish an aging rate prediction function, and dynamically correct the model deviation through an online adversarial training mechanism.

[0084] In the embodiments of this application,

[0085]

[0086] Among them, Ψ agingis the cable aging index, representing the comprehensive aging degree; t0 is the initial monitoring time point (s); t' is the integration time variable (s), representing the historical time window; t is the monitoring time point; represents the electro-thermal coupling response function; the input is the real-time electric field intensity distribution E (V / m) and the local heat gradient / m); the output is the material stress response (MPa); Δσ(t') is the insulation resistance decay rate (% / h), determined by the time-varying function fitted from historical data; α and β are dynamic weight coefficients, adjusted through an online adversarial training mechanism.

[0087] Among them, the value of α can be determined through the following steps: Real-time monitor the voltage fluctuation value and conductor temperature distribution of the cable, and calculate the real-time electric field intensity (unit: V / m) and local heat gradient (unit: °C / m) in combination with the conductor cross-section parameters; Obtain the maximum allowable electric field intensity (such as the insulation material breakdown threshold) and maximum allowable heat gradient (such as the material thermal deformation threshold) of this model from the cable design specifications; Dynamically calculate the value of α, and the formula is: The closer the value of α is to 1, the closer the electro-thermal synergistic effect is to the design limit, and early warning should be given preferentially.

[0088] The value of β can be determined through the following steps: Fit the insulation resistance decay rate curve through historical data, and calculate the current decay rate (% / h) in real time; Monitor the environmental temperature and humidity (humidity RH%, temperature T °C), and calculate the environmental correction coefficient; Judge whether the current decay rate exceeds the preset threshold. If it exceeds, amplify the value of β through the hyperbolic tangent function, otherwise keep the reference value. When it is humid and hot or the insulation deterioration accelerates, the value of β increases significantly, highlighting the risk of aging mutation.

[0089] Based on the embodiments provided in this application, transient features such as current spikes and temperature rise mutations are extracted through the wavelet multi-scale decomposition algorithm to capture the sudden signals of accelerated aging; The random forest regression algorithm is used to non-linearly weight the electro-thermal coupling coefficient, and the deviation is dynamically corrected in combination with online adversarial training to improve the robustness of the aging rate prediction.

[0090] Furthermore, the construction process of the Bayesian probability graph model specifically includes:

[0091] Construct a directed acyclic graph with transient features as observation nodes and the cable aging index as hidden variable nodes;

[0092] Adopt the Markov chain Monte Carlo sampling method to infer the node conditional probability distribution in combination with the aging test data;

[0093] Introduce the environmental temperature and humidity as covariate nodes to dynamically adjust the posterior probability weights of the electro-thermal dual factors;

[0094] Compress the model complexity through the variational inference algorithm to achieve real-time inference on the edge computing device side.

[0095] Based on the embodiments provided in this application, infer the node conditional probability distribution through the Markov chain Monte Carlo sampling method, and accurately quantify the implicit correlation between the electrothermal factor and the aging index. Use the variational inference algorithm to compress the model complexity, realize real-time inference of aging assessment on the edge device side, and reduce the dependence on the cloud.

[0096] Furthermore, the fire risk assessment algorithm includes:

[0097] According to the glowing wire ignition test data, establish a critical ignition temperature mapping table for the cable. The critical ignition temperature mapping table associates the non-linear relationship between the aging degree value and the critical ignition temperature, and fills the temperature prediction values in the missing interval of the glowing wire ignition test data through the fractal interpolation algorithm;

[0098] Based on the results of the arc shock test, construct a breakdown probability model for the cable insulation layer, and use the hidden Markov model to dynamically predict the number of breakdowns that the cable can withstand at different aging stages. The inputs of the breakdown probability model include the real-time leakage current spectrum characteristics and the peak voltage fluctuation;

[0099] Align the aging index time series data and the breakdown probability curve through the dynamic time warping algorithm to generate the evaluation result of the ignition risk value.

[0100] Based on the embodiments provided in this application, fill the critical temperature prediction values in the missing interval of the glowing wire test data through the fractal interpolation algorithm to solve the model generalization problem under sparse data. Use the dynamic time warping algorithm to align the aging index and the breakdown probability curve to quantify the dynamic correlation between the aging process and the breakdown event.

[0101] Furthermore, the breakdown probability model is constructed in the following way:

[0102] Extract the high-frequency oscillation waveform and energy accumulation characteristics of the breakdown event from the arc shock test, and use the sparse autoencoder to compress the high-dimensional data and extract the breakdown precursor pattern;

[0103] Based on the hidden Markov model, define the cable aging state as a hidden variable, such as healthy, slightly aged, severely aged, and the breakdown event as an observable variable, and decode the most likely aging state sequence corresponding to the real-time leakage current data through the Viterbi algorithm;

[0104] Combine the environmental temperature and humidity data, and use the multi-task reinforcement learning algorithm (MTRL) to dynamically optimize the breakdown probability threshold; among them, the reward function of the multi-task reinforcement learning algorithm is generated by the matching degree between the historical breakdown event and the aging index.

[0105] In the embodiments of this application,

[0106]

[0107] Among them, P breakdown is the breakdown probability, ranging from 0 to 1, and is used to predict the risk of cable insulation failure in real time; SAE(S arc(k) ) is the k-th high-frequency oscillation waveform feature extracted by the sparse autoencoder; K is the feature dimension of the high-frequency oscillation waveform; ω k is the feature weight, optimized by multi-task reinforcement learning; θ HMM (s t ) is the aging state threshold output by the hidden Markov model; s t is the current hidden state, healthy or aging; γ is the non-linear scaling factor, dynamically corrected by the environmental temperature and humidity.

[0108] Among them, the value of ω k is determined based on the following steps: Extract the high-frequency oscillation waveform before breakdown from the arc impact test, compress the data through the sparse autoencoder, and extract the key waveform features (such as energy peak, oscillation frequency); analyze the frequency-domain energy distribution of the leakage current, and calculate the energy proportion of each frequency component; dynamically allocate the weight ω k , and the rule is: the weight of the low-frequency component (such as below 100 Hz) is high (the arc precursor is mostly low-frequency oscillation); the weight of the high-frequency noise (such as above 1 kHz) approaches 0; combined with the effective value of the real-time leakage current, automatically adjust the frequency attenuation coefficient. The role of ω k is to focus on the physical features sensitive to breakdown and suppress the interference of invalid noise.

[0109] Suppose there are 3 high-frequency oscillation waveform features (K = 3), corresponding to different frequency components respectively:

[0110] Feature 1: Low-frequency component (such as 50 Hz), with a relatively high energy peak and a relatively low oscillation frequency.

[0111] Feature 2: Medium-frequency component (such as 500 Hz), with a moderate energy peak and a medium oscillation frequency.

[0112] Feature 3: High-frequency component (such as 2 kHz), with a relatively low energy peak and a relatively high oscillation frequency.

[0113] According to the rule, the weight of the low-frequency component is high, and the weight of the high-frequency noise approaches 0. Combined with the effective value of the real-time leakage current, automatically adjust the frequency attenuation coefficient. Suppose the current effective value of the leakage current is low and the frequency attenuation coefficient is small.

[0114] Optimized by multi-task reinforcement learning, the following feature weights are obtained:

[0115] ω_1 = 0.6 (the weight of the low-frequency component is high)

[0116] ω_2 = 0.3 (the weight of the medium-frequency component is moderate)

[0117] ω_3 = 0.1 (the weight of the high-frequency component approaches 0)

[0118] These weight values focus on the low-frequency physical features sensitive to breakdown and suppress high-frequency noise interference. In practical applications, the weights will be dynamically adjusted according to real-time data to adapt to different working environments and cable states.

[0119] Based on the embodiments provided in this application, the high-frequency oscillation waveform data is compressed by a sparse autoencoder to accurately identify the transient features of breakdown precursors. The breakdown probability threshold is dynamically optimized by multi-task reinforcement learning to achieve adaptive early warning under complex working conditions.

[0120] Furthermore, the calculation of the ignition risk value further includes:

[0121] Introduce the test data of the cone calorimeter, analyze the ignition time and the peak value of the heat release rate of the cable under 30 kW of heat radiation, and extract the key combustion modes from the heat release curve by non-negative matrix factorization (NMF);

[0122] Fuse the key combustion modes with the time-frequency domain features of the real-time leakage current data to construct a multi-modal risk feature vector; the time-frequency domain features include wavelet packet energy entropy;

[0123] In the embodiments of this application,

[0124] F risk = W NMF × C heat + U DWT × J leak (f)

[0125] where F risk is the multi-modal risk feature vector used to describe the fire risk; W NMF is the combustion mode basis matrix generated by non-negative matrix factorization; C heat is the main component of the heat release curve (MJ / m 2 ) measured by the cone calorimeter; U DWT is the time-frequency domain projection matrix of the discrete wavelet transform; J leak (f) is the frequency domain energy distribution (A / Hz) of the leakage current, extracted by wavelet packet transform; f is the frequency domain component (Hz) of the leakage current.

[0126] The deep deterministic policy gradient algorithm (DDPG) is used to dynamically correct the breakdown probability, and its action space is defined as the breakdown probability adjustment amount, and the state space includes the aging index, environmental temperature and humidity, and combustion mode similarity;

[0127] Through the online meta-learning framework, the risk assessment strategies for different cable models and laying scenarios are adapted.

[0128] In the embodiments of the present application,

[0129]

[0130] where π DDPG is the optimal action policy (breakdown probability adjustment amount) generated by deep deterministic policy gradient; ρ π is the probability density function of the state distribution under policy π; R(s, a) is the immediate reward function, with the input being the state space variable s (such as aging index, environmental temperature and humidity) and the action space variable a (such as probability adjustment amount); Meta(s, φ cable ) is the scenario adaptation reward output by the online meta-learning framework; φ cable is the cable model feature encoding (such as the hash value of cross-section and material type); λ is the meta-learning weight coefficient.

[0131] Among them, the value of λ is determined based on the following steps: Feature encoding is performed on the cable model (such as cross-section size, material flame retardant grade, laying method) to generate a unique feature vector; Calculate the cosine similarity between the current cable model and the historical reference model, and if the similarity is high, the value of λ increases; Count the available data volume of the current scenario, and the less the data, the smaller the value of λ (to prevent overfitting); Finally, the λ value is normalized to the [0, 1] interval through the Sigmoid function.

[0132] For example, there are three cable models, and their feature vectors are respectively:

[0133] Cable model A: 2, 3, 1, indicating a cross-section size of 2, a material flame retardant grade of 3, and a laying method of 1.

[0134] Cable model B: 1, 2, 0, indicating a cross-section size of 1, a material flame retardant grade of 2, and a laying method of 0.

[0135] Cable model C: 3, 4, 2, indicating a cross-section size of 3, a material flame retardant grade of 4, and a laying method of 2.

[0136] Calculate the cosine similarity between cable model A and B: First, multiply the corresponding elements of the two vectors and sum them to get the numerator part: 2 multiplied by 1 plus 3 multiplied by 2 plus 1 multiplied by 0 equals 2 + 6 + 0 = 8. Then, calculate the modulus lengths of the two vectors respectively. The modulus length of cable A is the square root of the sum of the squares of each element, that is, the square of 2 plus the square of 3 plus the square of 1 equals 4 + 9 + 1 = 14, and the square root is √14; The modulus length of cable B is the square of 1 plus the square of 2 plus the square of 0 equals 1 + 4 + 0 = 5, and the square root is √5. Finally, the cosine similarity is the numerator 8 divided by the product of the two modulus lengths √14 multiplied by √5, approximately equal to 8 divided by 8.306, and the result is approximately 0.963.

[0137] Assume that the available data volume in the current scenario is 50 samples, and the data volume is small. Therefore, the value of λ can be small. The value of λ can be calculated by combining the similarity and the data volume: assume that the weight of the similarity is 0.7 and the weight of the data volume is 0.3. Normalize the data volume to the interval [0, 1]. For example, when the data volume is 50, it is normalized to 0.5. Then the calculation of the λ value is: λ = Sigmoid(0.7×0.963 + 0.3×0.5) = Sigmoid(0.674 + 0.15) = Sigmoid(0.824) ≈ 0.695.

[0138] Therefore, in this case, the value of λ is approximately 0.695. That is to say, the specific value of λ can be calculated in a similar way according to the specific cable model and the data volume situation.

[0139] Based on the embodiments provided in this application, the key combustion modes in the heat release curve are extracted by non - negative matrix factorization, and are fused with the time - frequency characteristics of the leakage current to quantify the physical - electrical coupling effect of the fire risk. An online meta - learning framework is used to dynamically generate risk assessment strategies to achieve minute - level model adaptation for different cable models and laying scenarios.

[0140] Furthermore, the early warning report includes the cable aging level, the prediction of the remaining life, and the recommended maintenance measures, and displays the heat map of multi - dimensional perception data through a visual interface.

[0141] Furthermore, the breaker node integrates a high - speed ADC sampling circuit and a DSP signal processing unit, and the acquisition accuracy meets ±2℃ for the conductor temperature and ±1mA for the leakage current.

[0142] It should be noted that in this application, the embodiments implemented on the side of the cable aging test system based on the Internet of Things can be referred to each other with the embodiments implemented on the side of the cable aging test method based on the Internet of Things, and this application will not elaborate one by one.

[0143] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. All equivalent structural or equivalent process transformations made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, are equally included in the patent protection scope of the present invention.

Claims

1. An Internet of Things-based cable aging test system, characterized in that, Including: A cable data acquisition and monitoring module, which is used to collect cable operation parameters in real time through breaker nodes deployed in the cable line, and synchronously monitor the environmental temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value and load power; An Internet of Things transmission module, which is used to transmit the collected cable operation parameters to the cloud platform through the Internet of Things communication module; An aging degree analysis module, which is used to analyze the aging degree value of the cable insulation layer based on the electro-thermal dual-factor collaborative aging model; A combustion risk value calculation module, which is used to dynamically calculate the combustion risk value of the cable by combining the aging degree value through a fire risk assessment algorithm; An early warning module, which is used to generate an early warning report and push it to the terminal device when the aging degree value or the combustion risk value exceeds the threshold, and mark the hidden danger location at the same time.

2. A cable aging test method based on the Internet of Things, characterized in that, Including; Collect cable operation parameters in real time through breaker nodes deployed in the cable line, and synchronously monitor the environmental temperature and humidity; wherein, the cable operation parameters include conductor temperature, leakage current, voltage fluctuation value and load power; Transmit the collected cable operation parameters to the cloud platform through the Internet of Things communication module; Analyze the aging degree value of the cable insulation layer based on the electro-thermal dual-factor collaborative aging model; Dynamically calculate the combustion risk value of the cable by combining the aging degree value through a fire risk assessment algorithm; When the aging degree value or the combustion risk value exceeds the threshold, generate an early warning report and push it to the terminal device, and mark the hidden danger location at the same time.

3. The method for testing cable aging based on the Internet of Things according to claim 2, wherein, The Internet of Things communication module adopts a high-speed power line carrier wired and wireless multi-mode hybrid networking method; The input parameters of the electro-thermal dual-factor collaborative aging model include: cable conductor cross-section, material type and laying method; the frequency of current overload, temperature rise rate and insulation resistance decay curve in historical operation data; The processing logic of the electro-thermal dual-factor collaborative aging model includes the following steps: Based on the cable conductor cross-section, material type and laying method, construct a multi-physical field coupling characteristic matrix, and map the electro-thermal parameters to a material stress distribution function; wherein, the electro-thermal parameters include cable operation parameters and their derived electric field intensity distribution and local thermal gradient; Perform time series alignment and multi-resolution analysis on the frequency of current overload, temperature rise rate and insulation resistance decay curve in historical operation data, and extract the characteristics of aging-sensitive frequency bands; Adopt a dynamic Bayesian network to establish the probability dependence relationship between the electrical aging factor and the thermal aging factor; Fuse the contribution degrees of the electro-thermal dual factors through a non-linear kernel function to generate a cable aging index, and correct the aging rate threshold in combination with environmental temperature and humidity data.

4. The method for testing cable aging based on the Internet of Things according to claim 2, characterized in that The electro-thermal dual-factor collaborative aging model is constructed in the following way: Establish a cable accelerated aging test database, which contains the mechanical properties, insulation resistance and thermal stability data of the cable under different current levels and temperature and humidity conditions; Adopt the wavelet multi-scale decomposition algorithm to extract the transient characteristics in the electro-thermal parameters, and combine the Arrhenius equation to calculate the material decomposition activation energy under the action of thermal stress; Construct an electric breakdown probability model based on the Weibull distribution, fuse transient characteristics and steady-state aging data through a Bayesian probability graph model, and generate an electro-thermal double-factor coupling coefficient; Use the random forest regression algorithm to perform non-linear weighting on the electro-thermal double-factor coupling coefficient, establish an aging rate prediction function, and dynamically correct the model deviation through an online adversarial training mechanism.

5. The method for testing cable aging based on the Internet of Things according to claim 4, wherein, The construction process of the Bayesian probability graph model specifically includes: Construct a directed acyclic graph with the transient characteristics as the observation nodes and the cable aging index as the hidden variable nodes; Adopt the Markov chain Monte Carlo sampling method to infer the node conditional probability distribution in combination with the aging test data; Introduce environmental temperature and humidity as covariate nodes to dynamically adjust the posterior probability weights of the electro-thermal double factors; Compress the model complexity through the variational inference algorithm to achieve real-time inference on the edge computing device side.

6. The method for testing cable aging based on the Internet of Things according to claim 3, characterized in that The fire risk assessment algorithm includes: Based on the glow wire ignition test data, establish a critical ignition temperature mapping table for the cable. The critical ignition temperature mapping table associates the non-linear relationship between the aging degree value and the critical ignition temperature, and fills the temperature prediction values in the missing interval of the glow wire ignition test data through the fractal interpolation algorithm; Based on the results of the arc shock test, construct a breakdown probability model for the cable insulation layer, and use the hidden Markov model to dynamically predict the number of breakdown resistances of the cable at different aging stages. The input of the breakdown probability model includes the real-time leakage current spectrum characteristics and the voltage fluctuation peak value; Align the aging index time series data and the breakdown probability curve through the dynamic time warping algorithm to generate the evaluation result of the ignition risk value.

7. The method for testing cable aging based on the Internet of Things according to claim 6, wherein The breakdown probability model is constructed in the following way: Extract the high-frequency oscillation waveform and energy accumulation characteristics of the breakdown event from the arc shock test, and use the sparse autoencoder to compress the high-dimensional data and extract the breakdown precursor pattern; Based on the hidden Markov model, define the cable aging state as the hidden variable and the breakdown event as the observation variable, and decode the most likely aging state sequence corresponding to the real-time leakage current data through the Viterbi algorithm; Combined with the environmental temperature and humidity data, use the multi-task reinforcement learning algorithm to dynamically optimize the breakdown probability threshold; among them, the reward function of the multi-task reinforcement learning algorithm is generated by the matching degree between the historical breakdown event and the aging index.

8. The cable aging test method based on the Internet of Things according to claim 7, wherein The calculation of the ignition risk value further includes: Introduce the cone calorimeter test data, analyze the ignition time and the peak value of the heat release rate of the cable under 30kW heat radiation, and extract the key combustion mode from the heat release curve by non-negative matrix factorization; Fuse the key combustion mode with the time-frequency domain characteristics of the real-time leakage current data to construct a multi-modal risk feature vector; Adopt the deep deterministic policy gradient algorithm to dynamically correct the breakdown probability, and its action space is defined as the breakdown probability adjustment amount, and the state space includes the aging index, environmental temperature and humidity, and combustion mode similarity; Through the online meta-learning framework, adapt the risk assessment strategy for different cable models and laying scenarios.

9. The method for testing cable aging based on the Internet of Things according to claim 2, wherein, The warning report includes the cable aging level, remaining life prediction, and recommended maintenance measures, and displays the heat map of multi-dimensional perception data through a visual interface.

10. The Internet of Things-based cable aging test method according to claim 2, wherein The breaker node integrates a high-speed ADC sampling circuit and a DSP signal processing unit, and the acquisition accuracy meets ±2°C for conductor temperature and ±1 mA for leakage current.

Images