A method for on-line monitoring and aging early warning of internal insulation state of ring main unit

Abstract

The application discloses a kind of ring network box inside insulation state on-line monitoring and aging early warning method, belong to power equipment state monitoring technical field, by multimodal sensor synchronous acquisition local discharge, temperature and humidity, insulation leakage current, shell vibration and other multi-physical field data, construct synchronous time series state vector;Extract key features to distinguish equipment operating condition, divide normal, condensation risk, abnormal partial discharge, thermal anomaly and composite deterioration scene, and adaptively switch sampling frequency, give consideration to operating energy consumption and fault monitoring sensitivity.Heat, humidity, electricity, force coupling insulation degradation evolution model is constructed, the degradation state is quantified and the insulation remaining life is calculated, and the grading early warning is output combined with the service life numerical value and the change rate.The application realizes quantitative evaluation of insulation degradation, improves early warning lead time by integrating multi-criteria prediction algorithm, effectively reduces false alarm and missed alarm rate by relying on edge processing and cloud parameter iterative optimization model.

Description

Technical Field

[0001] This invention belongs to the field of power equipment condition monitoring technology, specifically a method for online monitoring and aging early warning of the internal insulation status of a ring main unit. Background Technology

[0002] Ring main units (RMUs), as key node equipment in power distribution networks for ring network power supply, load tapping, and fault isolation, play an irreplaceable role in the construction of modern urban power distribution networks, especially in areas requiring high-reliability power supply. The integrity and stability of their internal insulation system are the core foundation for ensuring the continuous and safe operation of the entire distribution network. With the continuous increase in power load density and the increasing complexity of the power grid operating environment, the deterioration of the insulation condition inside RMUs has become increasingly prominent. Statistics show that insulation faults account for more than 40% of all operational faults in RMUs. Unplanned power outages caused by insulation breakdown not only result in huge direct economic losses and high subsequent maintenance costs, but also pose a potential threat to the orderly operation of social production and daily life. Therefore, how to achieve accurate monitoring and scientific early warning of the insulation condition inside RMUs has always been a key research topic in the field of power system reliability.

[0003] In terms of current technical practices, the monitoring methods for the insulation status of ring main units have mainly evolved from offline to online, and from single-parameter to multi-parameter fusion. The initial technical solutions primarily relied on periodic power outage preventative tests, i.e., static testing of parameters such as insulation resistance and AC withstand voltage of the ring main unit within a preset cycle according to relevant power equipment testing procedures. This model played a crucial role in the early stages of power system development, enabling the elimination of some equipment with obvious defects through periodic inspections. However, this time-based offline testing mode has significant limitations. Its testing cycle is typically measured in years, making it difficult to capture sudden and rapidly evolving insulation degradation processes between tests. More importantly, the conditions applied in offline tests often deviate significantly from the complex operating conditions of the equipment during actual operation, failing to accurately reflect the dynamic performance evolution of the equipment under the long-term coupling effects of electric, temperature, humidity, and mechanical stress fields. This results in test results that are often lagging and unsuitable as a basis for precise operation and maintenance.

[0004] With advancements in sensing and communication technologies, single-parameter online monitoring technologies, such as partial discharge monitoring, have gradually gained application. These technologies typically utilize ultra-high frequency sensors to capture partial discharge signals accompanying insulation degradation and combine this with preset thresholds for early warning. While this approach represents a leap from static to dynamic monitoring, it exhibits significant limitations in practical applications. This is because partial discharge is only one of the later-stage characteristics of insulation degradation, and the decline in insulation performance is often the result of the long-term effects of multiple environmental factors. For example, in the early stages before discharge occurs, risks such as condensation, thermal aging, or localized overheating due to poor contact cannot be identified using a single discharge monitoring method. Furthermore, some solutions attempt to incorporate multi-source data such as temperature, humidity, and leakage current for combined monitoring, using linear logic judgment models to assess the degree of moisture absorption. Although this multi-parameter approach expands the data dimensions, its underlying logic still follows a fixed signal processing path, namely, performing feature extraction and threshold comparison at a uniform sampling frequency. This rigid monitoring process leads to unnecessary computational overhead and data redundancy when the equipment is running normally. When faced with rapidly evolving and complex degradation scenarios, the lack of targeted in-depth diagnostic mechanisms often results in insufficient early warning timeliness and an inability to provide a sufficiently early intervention window.

[0005] A thorough analysis of the shortcomings of the existing technologies reveals that their core contradiction lies in the failure to establish a nonlinear coupling relationship between the driving factors of multiple physical fields such as electricity, heat, humidity, and force and the degradation mechanism of insulation materials. Within the enclosed space of the ring main unit, these physical factors do not exist in isolation but rather evolve in an intertwined manner. For example, a high-humidity environment not only reduces the surface resistance of insulation components but also induces weak partial discharges. The thermal effect generated by these discharges accelerates the thermal aging process of organic insulation materials, and the accompanying chemical byproducts, such as ozone, further damage the hydrophobicity of the material surface, thereby expanding the wetted area and ultimately forming a vicious cycle of accelerated aging: moisture absorption, discharge, heat generation, oxidation, and performance loss. Existing technical solutions often treat these physical processes as independent linear superpositions, failing to quantify their inherent coupling acceleration effect. Furthermore, the lack of flexibility in adaptively adjusting the monitoring depth according to real-time operating conditions makes it difficult to achieve a balance between sensitivity and robustness. Furthermore, the prediction of remaining insulation lifetime often relies on simple calculations of single physical quantities, lacking prediction models that integrate multiple mechanisms. This results in early warning information that is insufficient in terms of accuracy and lead time to meet the actual needs of refined operation and maintenance in modern distribution networks. Therefore, how to construct an online monitoring and early warning system that can deeply integrate multiple physical field mechanisms, achieve adaptive monitoring strategies for different scenarios, and possess multi-path lifetime prediction capabilities has become a key technical challenge in improving the asset management level and power supply reliability of distribution networks. Summary of the Invention

[0006] The purpose of this invention is to provide an online monitoring and aging early warning method for the insulation status inside a ring main unit, in order to solve the problems mentioned in the background art, such as the inability to quantify and characterize the nonlinear acceleration effect of multi-physics coupling on insulation degradation, the inability of the monitoring strategy to adaptively adjust according to the operating scenario, and the single dimension and insufficient lead time for remaining lifetime prediction.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for online monitoring and aging early warning of the internal insulation status of a ring main unit includes the following steps: Step 1: Simultaneously collect ultra-high frequency partial discharge signals, ultrasonic partial discharge signals, ambient temperature, relative humidity, leakage current on the surface of insulating components, and vibration acceleration of the shell by deploying a multi-modal sensor group inside the ring network box, and construct an original state vector with a synchronous time scale; Step 2: Extract features from the original state vector and determine the operating conditions of the ring network box as normal operation, condensation risk, partial discharge abnormality, thermal abnormality, or combined deterioration scenario based on the condensation temperature difference index, partial discharge characteristic quantity, and temperature characteristic quantity. Step 3: Based on the identified operating scenario, adaptively switch the sampling frequency of the monitoring system: use the first sampling frequency in normal operating scenarios, and increase the sampling frequency to the second sampling frequency in abnormal operating scenarios, with the second sampling frequency being higher than the first sampling frequency; Step 4: Construct an evolution model of the insulation system's deterioration state that reflects the coupling effects of multiple physical fields such as heat, humidity, electricity, and force. This model determines the current deterioration state variables based on ambient temperature, relative humidity, partial discharge power, and vibration intensity, and calculates the remaining life of the insulation based on the deterioration state variables. Step 5: Real-time calculation of the deterioration state variables and their corresponding remaining lifetimes, and output of graded early warning signals based on the remaining lifetime values ​​and their rate of change.

[0008] According to the above technical solution, in step two, the condensation temperature difference index is the difference between the ambient temperature and the dew point temperature. The dew point temperature is calculated based on the ambient temperature, relative humidity and ambient atmospheric pressure. When the condensation temperature difference index is lower than the first preset threshold and the leakage current growth rate exceeds the second preset threshold, it is determined to be a condensation risk scenario.

[0009] According to the above technical solution, in step two, the partial discharge characteristic quantity includes the apparent discharge quantity; when the apparent discharge quantity exceeds the third preset threshold and the amplitude of the ultrasonic partial discharge signal shows a trend of increasing, it is determined to be an abnormal partial discharge scenario.

[0010] According to the above technical solution, in step two, the temperature characteristic quantity includes the ambient temperature value and its temperature rise slope; when the ambient temperature exceeds the fourth preset threshold or the temperature rise slope exceeds the fifth preset threshold, it is determined to be a thermal anomaly scenario.

[0011] According to the above technical solution, in step two, when the judgment criteria of at least two of the three scenarios—condensation risk scenario, partial discharge abnormal scenario, and thermal abnormal scenario—are met simultaneously, it is determined to be a composite deterioration scenario.

[0012] According to the above technical solution, in step three, the first sampling frequency is determined based on the update requirements of the discharge statistics within the power frequency cycle, and the second sampling frequency meets the Nyquist sampling requirements for the rising edge of the pulse in the UHF partial discharge signal.

[0013] According to the above technical solution, in step four, the insulation system degradation state evolution model adopts an integral form to express the evolution of degradation state variables over time. The integral kernel function includes a temperature exponent term reflecting the thermal activation effect, a humidity exponent term reflecting the change in surface conductivity caused by humidity, an electro-induced degradation term reflecting the erosion of accumulated discharge energy, and a vibration intensity term reflecting mechanical vibration stress fatigue. The integral kernel function also includes a self-acceleration term that is positively correlated with the degradation state variables.

[0014] Based on the above technical solution, the evolution model of the insulation system's deterioration state is represented by the following integral equation:

[0015] in: The dimensionless degradation state variable of the insulation system at time t, with a range of values. ; This represents the initial degradation state value. This indicates the critical degradation state value for insulation failure; The degradation rate constant, representing the intrinsic properties of the insulating material, has dimensions of . ; This represents a multiphysics coupling acceleration function, which is dimensionless. This represents the self-accelerating function in the deteriorated state, and its dimension is dimensionless. The variable 't' represents the integration time variable, in seconds (s); 't' represents the current time, in seconds (s). Indicates the integration time. The instantaneous value of relative humidity, Indicates the integration time. The instantaneous value of the thermodynamic temperature. Indicates the integration time. The instantaneous value of partial discharge power, Representing the time of integration Instantaneous value of effective vibration intensity Indicates the integration time. The instantaneous value of the deteriorated state variable.

[0016] According to the above technical solution, in step four, the percentage of remaining insulation life... This is determined through the following mapping relationship:

[0017] in, represents the critical degradation state value for insulation failure, calibrated through accelerated aging tests, with a typical value of 1.0; p represents the lifetime mapping shape factor, which is dimensionless and used to adjust the nonlinearity of the lifetime decay curve.

[0018] According to the above technical solution, step five also includes: the edge computing gateway generates an early warning message containing the abnormal scenario type, the dominant degradation factor and the predicted remaining lifetime, and sends the early warning message to the operation and maintenance management system through the wireless communication module.

[0019] Compared with the prior art, the present invention has the following beneficial effects: This invention constructs a comprehensive degradation factor model, incorporating four types of driving factors—electricity, heat, humidity, and force—into a unified nonlinear mathematical framework. This model not only characterizes the influence of each individual physical quantity on insulation but also precisely describes the accelerating effects across physical fields, such as high humidity-induced discharge, accelerated heat generation from discharge, and thermal aging that damages hydrophobicity. This transforms the evaluation of insulation status from empirical qualitative to physical quantitative, significantly improving the scientific rigor of the assessment.

[0020] This invention abandons the traditional rigid linear signal processing path and drives the dynamic adjustment of sampling frequency and computation depth through scene discrimination logic. This strategy greatly reduces the power consumption and communication bandwidth pressure of the edge computing terminal during normal device operation, and can immediately switch to a full-bandwidth, high-precision deep diagnostic mode when an anomaly occurs, ensuring the complete capture of rapidly evolving fault characteristics and balancing real-time monitoring with sensitivity.

[0021] By running a multi-path lifetime prediction model in parallel and incorporating Dempster-Shafer evidence theory for multi-criteria fusion, this invention overcomes the limitations of extrapolating from a single index. The system can dynamically adjust weights according to environmental changes, identifying potential evolution trends in the early stages of insulation degradation (such as the moisture stage or weak discharge stage). Compared with traditional threshold alarms, the early warning lead time is increased by more than 40%, providing sufficient decision-making time for condition-based maintenance.

[0022] Through real-time processing at the edge and asynchronous updates of model parameters in the cloud, the system possesses adaptive capabilities to complex environments and the operating characteristics of different batches of devices. As operational data accumulates, the cloud-based expert database can continuously refine the coupled model and prediction weights, enabling the accuracy of the early warning algorithm to continuously improve with the extension of operating time, effectively reducing false alarm and false negative rates. Attached Figure Description

[0023] Figure 1This is a flowchart illustrating the early warning method of the present invention. Figure 2 This is a flowchart of the processing steps of the early warning method of the present invention; Figure 3 This is a schematic diagram of the ring network box early warning module of the present invention. Detailed Implementation

[0024] 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.

[0025] Example 1 like Figure 1 As shown, a method for online monitoring and aging early warning of the internal insulation status of a ring main unit includes the following steps: Step 1: Synchronous Acquisition and Timescale Construction of Multidimensional Physical Quantities. A multimodal sensor array is deployed within the ring network enclosure at key insulation points and in the surrounding environment. This array includes: a UHF partial discharge sensor, an ultrasonic partial discharge sensor, a high-precision temperature and humidity sensor, a leakage current transformer, and a mechanical vibration sensor. The synchronous trigger pulses of the edge computing gateway are used to achieve microsecond-level synchronous acquisition of data from each sensor, and nanosecond-level timescale alignment is achieved for the UHF partial discharge signal channel. This constructs an original state vector encompassing four dimensions: electrical, thermal, humidity, and force. Its expression is:

[0026] in: Represents the original state vector at time t; Indicates a high-frequency partial discharge signal; This indicates an ultrasonic partial discharge signal; Indicates the ambient temperature in Celsius; Relative humidity is expressed as a dimensionless percentage from 0 to 100. Indicates the leakage current on the surface of the insulating component; This indicates the vibration acceleration of the shell.

[0027] Furthermore, the ultra-high frequency partial discharge sensor has a detection bandwidth of 300MHz to 1.5GHz, used to capture discharge pulses inside and on the surface of insulation; the ultrasonic partial discharge sensor has a center frequency of 40kHz, used to detect mechanical wave signals caused by insulation defects; the leakage current transformer adopts a zero-flux current transformer with a range of 0 to 10mA and a resolution better than... .

[0028] The necessity of constructing the aforementioned synchronized timescale lies in the fact that the correspondence between partial discharge pulses and voltage phases determines the accuracy of discharge type identification, while the impact of environmental humidity fluctuations on surface leakage current exhibits a second-level hysteresis effect, and the correlation between mechanical vibration and partial discharge pulses manifests as a microsecond-level transient response. By aligning the timescales to nanosecond levels, causal chain tracing of different physical quantities along the time axis can be achieved. For example, when a sub-microsecond-level correspondence is detected between a UHF discharge pulse train and the peak of shell vibration acceleration, it can be determined that partial discharge has triggered the propagation of microcracks in the solid insulating material, thus transforming the coupled analysis of electrical and mechanical dimensions from qualitative speculation to quantitative correlation.

[0029] Step 2: Adaptive scene discrimination based on feature energy mapping. The original state vectors acquired in Step 1 are preprocessed to extract time-domain, frequency-domain, and statistical features. A scene discrimination function is then established. Based on real-time feature vectors, the operating conditions of the ring main unit are divided into five scenarios: normal operation, condensation risk, partial discharge anomaly, thermal anomaly, and combined degradation. Scenario discrimination function. The following logic is adopted: First, calculate the condensation temperature difference index, which is expressed as follows:

[0030] in: This indicates the temperature difference index of condensation. Indicates the ambient temperature in Celsius, as defined in step one; Indicates based on ambient temperature Relative humidity (H) and ambient atmospheric pressure The dew point temperature value is calculated using a dew point temperature estimation model.

[0031] When the condensation temperature difference index Below the preset first threshold And leakage current The growth rate exceeded the second preset threshold. At that time, it was determined to be a condensation risk scenario; Secondly, for Pulse feature extraction is performed, and the discharge repetition frequency is calculated. Apparent discharge quantity q and discharge energy If the apparent discharge quantity q exceeds the third preset threshold And accompanied by An increase in amplitude indicates an abnormal partial discharge scenario; Furthermore, when Exceeding the preset operating temperature upper limit and the fourth preset threshold Or its temperature rise slope exceeds the fifth preset threshold. At that time, it was determined to be a thermal anomaly scenario; Finally, when at least two of the above anomaly criteria are met simultaneously, it is determined to be a composite degradation scenario.

[0032] The physical basis for the above scenario classification is as follows: The condensation risk scenario corresponds to the stage where surface resistivity decreases exponentially due to moisture adsorption, and its dominant physical process is the abrupt change in conductivity at the solid-gas interface; the partial discharge anomaly scenario corresponds to the local electric field inside or on the surface of the insulator exceeding the breakdown field strength threshold, and its dominant physical process is the development of collisional ionization and electron avalanche; the thermal anomaly scenario corresponds to the thermal activation process where intensified molecular chain segment motion leads to an increase in free volume; and the composite degradation scenario corresponds to the occurrence of positive feedback coupling between at least two degradation mechanisms. Scenario discrimination function. Essentially, it is a phase space partitioning mapping of the current thermodynamic and electrodynamic states of the insulation system.

[0033] Step 3: Adaptive adjustment of dynamic sampling frequency and algorithm complexity. Based on the operating scenario identified in Step 2, the sampling depth of the monitoring system is switched in real time. Under normal operating conditions, the system executes a low-energy monitoring mode, reducing the ultra-high frequency sampling frequency to [missing information]. It only performs basic statistical calculations; when a scenario is identified as condensation risk, abnormal partial discharge, or thermal anomaly, the system automatically switches to a specialized diagnostic mode, increasing the sampling frequency to [specific parameters]. The system also enables a deep feature extraction algorithm; in complex degradation scenarios, the system enables a full-dimensional deep analysis mode and performs online solution of the multi-physics coupling model.

[0034] The physical constraints for sampling frequency switching are: The Nyquist sampling theorem should be satisfied to ensure complete capture of the rising edge of the UHF partial discharge pulse. Considering that the upper limit of the detection bandwidth of the UHF sensor is 1.5 GHz, to ensure that the signal components within this frequency band are not distorted, It should be no less than 2 × 1.5 GHz = 3 GS / s; The lower limit is determined based on the update requirements of discharge statistics within the power frequency cycle, and its lower limit is determined by the statistical stability of the discharge repetition frequency. Through derivation, when... When the sampling rate is 10MS / s, it can guarantee that no less than 200 sampling points can be obtained within a 20ms power frequency period, which is sufficient to meet the accuracy requirements of envelope extraction and basic statistical feature calculation.

[0035] Step four: Construction and calculation of an integrated prediction model for multi-physics coupled degradation mechanism and remaining lifetime. This invention abandons the conventional approach of independently evaluating and simply superimposing thermal, electrical, humidity, and mechanical factors. Instead, it constructs a comprehensive degradation-lifetime coupled model that reflects the continuous evolution of the insulation system from its initial state to its final failure state under the nonlinear coupling of multiple physics fields. This model integrates degradation degree evaluation and remaining lifetime prediction into a unified mathematical expression:

[0036] in: The dimensionless degradation state variable of the insulation system at time t, with a range of values. ; This represents the initial degradation state value, calibrated through factory testing or historical data, with a typical range of 0 ≤ ≤0.1; The degradation rate constant, representing the intrinsic properties of the insulating material, has dimensions of . ; This represents a multiphysics coupling acceleration function, which is dimensionless. This represents the self-accelerating function in the deteriorated state, and its dimension is dimensionless. The variable 't' represents the integration time variable, in seconds (s); 't' represents the current time, in seconds (s). Indicates the integration time. The instantaneous value of relative humidity, Indicates the integration time. The instantaneous value of the thermodynamic temperature. Indicates the integration time. The instantaneous value of partial discharge power, Representing the time of integration Instantaneous value of effective vibration intensity Indicates the integration time. The instantaneous value of the deteriorated state variable.

[0037] Multiphysics Coupling Acceleration Function The expression is:

[0038] Where: T represents the real-time thermodynamic temperature, and T = +273.15, of which The ambient temperature in Celsius is denoted as 1 in step one; H represents the relative humidity, expressed as a dimensionless percentage from 0 to 100, as defined in step one. This represents the partial discharge power, expressed in watts (W). This represents the effective value of the measured vibration intensity, in m / s². 2 ; The activation energy of the material is expressed in J / mol; R represents the molar gas constant. denoted by , representing the humidity coupling coefficient, dimensionless; denoted by n, representing the humidity influence index, dimensionless. Represents the electro-degradation coupling coefficient, in units of ,and To accumulate discharge energy; This represents the stress sensitivity coefficient, with units of (s). 2 / m); The threshold vibration intensity represents the accelerating effect of mechanical stress on degradation, measured in m / s². 2 R represents the molar gas constant.

[0039] Degraded state self-acceleration function The expression is:

[0040] in: represents the self-acceleration coefficient, which is dimensionless; m represents the nonlinear exponent, which is dimensionless. Should meet , This represents a dimensionless deterioration state variable.

[0041] Percentage of remaining insulation life Deteriorated state variables This is determined through the following mapping relationship:

[0042] in: represents the critical degradation state value for insulation failure, calibrated through accelerated aging tests, with a typical value of 1.0; p represents the lifetime mapping shape factor, which is dimensionless and used to adjust the nonlinearity of the lifetime decay curve.

[0043] The physical essence of the above-mentioned integrated model lies in unifying the two separate steps of degradation assessment and remaining lifetime prediction in traditional methods into a continuous state evolution process. The product terms in the function correspond to the nonlinear coupling of four acceleration mechanisms: thermal activation, surface conductivity increase caused by humidity, cumulative discharge energy erosion, and mechanical vibration stress fatigue. According to the chemical reaction rate theory, when multiple acceleration factors coexist, their contributions to the total reaction rate are exponentially additive, so the product form is used to express their synergistic effect. The function characterizes the positive feedback features of the degradation process: as microcracks propagate or carbonization channels form, local electric field distortion intensifies, moisture penetration paths increase, and the degradation rate under the same external stress further increases. This model, through... The integral solution of the evolution trajectory can simultaneously output the current degree of degradation and the time prediction to reach the failure threshold, fundamentally avoiding information separation and error propagation between discrete models.

[0044] As a preferred embodiment of the present invention, the coupling coefficient , , and self-acceleration coefficient The results were obtained through accelerated aging tests. Specifically, in a controlled laboratory environment, combined electrical, thermal, humidity, and vibration stresses of different gradients were applied to the insulation components of ring main units of the same specifications. The insulation breakdown time and the evolution trajectory of each physical quantity during the process were recorded, and the values ​​of each coefficient were obtained by fitting using nonlinear regression analysis. As an example, for epoxy resin insulation components, calibration was performed under test conditions of a reference temperature of 80℃ and a relative humidity of 90%. =0.015, , , =2.0, =0.05, m=1.5. The above parameter values ​​can be adjusted according to the actual type of insulation material.

[0045] The above parameter values ​​are derived based on the optimal fit of the electro-thermal-humid combined aging test data of epoxy resin materials. Among them, n=2.0 corresponds to the surface conductivity being approximately proportional to the square of the relative humidity when the surface moisture adsorption is in the initial stage of multilayer adsorption; m=1.5 reflects the nonlinear relationship between the stress intensity factor at the crack tip and the crack length after the microcrack propagation enters the stable propagation stage. =0.05 indicates that the self-accelerating effect contributes about 5%-15% to the total degradation rate in the middle stage of degradation, which is consistent with the three-stage law of latent-development-outburst in the aging process of insulation materials.

[0046] Step 5: Real-time calculation and multi-level early warning based on the integrated model of remaining lifetime. The edge computing gateway, within each evaluation cycle, utilizes current and historical data... H , Measured value, via T= After converting +273.15 to thermodynamic temperature, the integral equation in step four is solved online numerically to obtain the current deterioration state. and the corresponding percentage of remaining lifespan. .according to The values ​​and rate of change The system generates a level four warning: when >50% and When the value is below the stable threshold, output green normal status information; When 30% < ≤50% or When a trend of increase is observed, a yellow alert will be issued, and a task instruction to strengthen inspections will be automatically issued. When 10% < When the percentage is ≤30%, an orange abnormality warning will be issued, and the expert diagnostic system will be activated for auxiliary analysis. when ≤10% or When the number of cases increases exponentially, a red critical alarm will be triggered, and the protection device will be activated or a power outage for maintenance will be recommended.

[0047] The above-mentioned four-level early warning thresholds are set based on the typical nonlinear characteristics of the performance degradation process of insulating materials. When When the value is greater than 50%, it corresponds to the latent stage, where molecular chain breakage exists in the form of isolated point defects; the 30%-50% range corresponds to the vicinity of the microcrack connectivity seepage threshold; the 10%-30% range corresponds to the crack propagation entering the instability stage; when it is less than 10%, the remaining insulation strength can no longer guarantee to withstand a single overvoltage impact. It provides instantaneous acceleration information for the degradation process, and... The absolute values ​​complement each other, and the early warning logic links the two together with an OR relationship, taking into account both the accuracy and timeliness of the state assessment.

[0048] In a preferred embodiment of the present invention, in step one, the installation of the multimodal sensor group adopts a non-invasive structure, wherein the ultra-high frequency partial discharge sensor is fixed to the signal window on the inner wall of the ring network box by magnetic attraction, and the high-precision temperature and humidity sensor is set 20cm to 30cm above the bus connection point inside the ring network box to capture environmental parameters of the core hot spot area.

[0049] The installation height of the temperature and humidity sensor is determined based on the following: Under natural convection conditions, the thickness of the thermal plume boundary layer above the hot spot on the busbar is related to the Rayleigh number. It is estimated that, with a busbar temperature rise of 50K and a characteristic length of 0.1m, the fully developed thermal plume zone is located within 15cm to 40cm above the heat source. Setting the sensor at 20cm to 30cm ensures that the measured temperature and humidity represent the actual airflow state across the surface of the insulating component, rather than non-representative parameters from the cold wall boundary layer or local dead zones.

[0050] In a preferred embodiment of the present invention, in step two, when extracting the features of the ultra-high frequency discharge signal, the variational mode decomposition (VMD) algorithm is used to decompose the original pulse sequence into a finite number of intrinsic mode components. By calculating the center frequency and energy distribution of each component, the discharge type is identified as tip discharge, floating discharge, or surface discharge. The number of mode decompositions K can be determined using the center frequency observation method: initially, K=2 is preset, and the K value is gradually increased for decomposition. When the difference in center frequency between adjacent mode components is less than a preset frequency resolution (e.g., 50kHz), it is considered that over-decomposition has occurred, and the previous K value is taken as the optimal number of decompositions.

[0051] The physical basis for using VMD decomposition to identify partial discharge types lies in the fact that different discharge types have different equivalent excitation source impedance characteristics and transmission path frequency responses, thus exhibiting characteristic energy distributions in the frequency domain. Point discharge energy is concentrated in the high-frequency band; the proportion of low-frequency components in the spectrum of surface discharge increases significantly; and suspended discharge exhibits narrowband enhancement in specific frequency bands. The center frequency difference threshold of 50kHz is an empirically optimal value determined after comprehensively considering the bandwidth of ultra-high frequency sensors and the typical partial discharge pulse spectrum width to avoid mode aliasing.

[0052] In a preferred embodiment of the present invention, in step three, the sampling frequency... The speed was set to 3.2 G / s to obtain complete discharge waveform details; The speed was set to 10 MS / s, retaining only envelope information and basic statistical features.

[0053] The 3.2GS / s sampling rate corresponds to a Nyquist frequency of 1.6GHz, fully covering the effective detection bandwidth of UHF sensors from 300MHz to 1.5GHz, ensuring that the highest frequency signal components are not distorted. The 10MS / s sampling rate corresponds to a sampling interval of 100ns. Combined with peak hold and mean detection techniques, the power frequency phase distribution spectrum of the discharge amplitude can be accurately reconstructed, meeting the trend monitoring requirements under normal operating conditions.

[0054] In a preferred embodiment of the present invention, during the calculation process in step four, the partial discharge power Defined as the product of a single discharge energy and the repetition frequency, its expression is:

[0055] in: This represents the partial discharge power, expressed in watts (W). This represents the energy of a single discharge, expressed in J. This indicates the discharge repetition frequency, measured in Hz.

[0056] RMS value of vibration intensity The calculation is performed using the composite value of the root mean square values ​​of the triaxial acceleration signals.

[0057] As a preferred embodiment of the present invention, the high-precision temperature and humidity sensor adopts a digital bus interface, with a temperature measurement accuracy of ±0.1℃, a relative humidity measurement accuracy of ±1.5%RH, and a response time of less than 5s, ensuring a rapid response to the risk of condensation.

[0058] The necessity of a response time of less than 5 seconds lies in the fact that the condensation process inside the ring main unit caused by diurnal temperature differences or sudden load changes can result in a dew point temperature change rate of up to 2℃ / min. If the sensor response time is too long, the calculated value of the condensation temperature difference index will lag behind the actual thermodynamic state, leading to missed detection or delayed alarms for condensation risks.

[0059] In a preferred embodiment of the present invention, the mechanical vibration sensor is a triaxial accelerometer with a frequency response range of 10Hz to 10kHz, used to monitor the mechanical structural stability of the ring main unit under switching operation or external vibration interference.

[0060] The lower limit of the frequency response range of 10Hz is set to filter out steady-state gravitational acceleration components and low-frequency environmental disturbances, while the upper limit of 10kHz covers the main detectable frequency band of partial discharge-induced ultrasonic waves propagating in solid media and the effective frequency band of the impact response spectrum generated by switching operations.

[0061] Example 2 In the specific engineering implementation process, the sensor network is deployed first. Monitoring nodes are arranged in the cable compartment, circuit breaker compartment, and busbar compartment of the ring main unit. (Refer to...) Figure 1 The system architecture shown comprises a sensing layer consisting of an ultra-high frequency partial discharge sensor, an ultrasonic partial discharge sensor, a high-precision temperature and humidity sensor, a leakage current transformer, and a mechanical vibration sensor, all deployed inside the ring main unit. The ultra-high frequency sensor is installed near the insulating bushing to maximize the signal-to-noise ratio of the partial discharge signal. The zero-flux current transformer is sleeved on the surge arrester grounding wire or the cable shield grounding wire, and the shielding ensures that only the weak leakage current from the insulation surface is extracted.

[0062] The data acquisition layer utilizes an embedded microprocessor as its core, which possesses multi-channel synchronous AD acquisition capabilities. After system startup, the microprocessor continuously runs the scene recognition engine from step two. When a surge in ambient humidity causes the dew point temperature to approach the internal temperature of the ring network enclosure, the recognition engine calculates the condensation temperature difference index. Immediately identify the condensation risk scenario. At this point, the edge computing gateway increases the leakage current sampling frequency from once per minute to hundreds of times per second, focusing on analyzing the harmonic and pulse components of the leakage current to determine whether continuous tracking has occurred. Specifically, FFT analysis is performed on the leakage current signal to extract the amplitude of the 50Hz fundamental frequency and the 3rd, 5th, and 7th harmonics. When the harmonic content exceeds 15% of the total effective value, the surface tracking process is determined to have started, triggering a yellow alert.

[0063] The 15% threshold for leakage current harmonic content is based on the following: the leakage current waveform of a linear insulation system should be a pure sine wave with a harmonic distortion rate close to zero. When localized carbonized conductive channels appear on the surface, the current-voltage characteristics become nonlinear, and the third harmonic component increases significantly. Experimental data shows that in the early stages of carbonization on the epoxy resin surface, the third harmonic amplitude accounts for approximately 5%-8% of the total effective value; when the carbonized channels are about to connect, this proportion jumps to 15%-20%. Therefore, the 15% threshold provides sufficient advance warning between the onset of nonlinearity and complete failure.

[0064] In the scenario of partial discharge anomalies, the system invokes the VMD algorithm to deconstruct the UHF signal. If the discharge spectrum is identified as conforming to the characteristics of tip discharge, the system adjusts the electro-induced degradation coupling coefficient when calculating the integrated degradation-lifetime coupling model in step four. Perform temperature correction, specifically by adjusting the original temperature setting. Multiplying by the temperature correction factor, its expression is:

[0065] in: This represents the temperature correction factor, which is dimensionless. This represents the temperature sensitivity coefficient of a material, measured in Kelvin (K). -1 Typical values ​​range from 0.01 to 0.05K. -1 T represents the real-time thermodynamic temperature, with units of 1000°C. ; This indicates the reference temperature, which is the thermodynamic temperature value taken during the calibration test.

[0066] After acquiring the measurement sequences of various physical quantities, the lifetime prediction layer uses recursive least squares to update the integral trajectory of the integrated degradation-lifetime coupling model in real time. The prediction engine not only provides the expected value of the remaining lifetime but also gives the confidence interval of the prediction result based on historical statistical distribution. When the calculated percentage of remaining lifetime reaches the warning threshold, the edge computing unit sends a structured warning message to the operation and maintenance management system through the wireless communication module. The message content includes: the type of abnormal scenario, the dominant degradation factor, the predicted remaining lifetime, and the recommended operation and maintenance measures.

[0067] Reference Figure 2 The method flow shown, in step 101 of the multidimensional physical quantity synchronous acquisition and synchronous time scale construction, is as follows: Figure 3As shown, this invention incorporates a multi-mode sensor array in key insulated areas such as the cable compartment, busbar compartment, and mechanism compartment within the ring main unit. The ultra-high frequency partial discharge sensor employs a wideband disc antenna structure, installed at the signal observation window on the inner wall of the ring main unit. Its detection frequency band covers 300MHz to 1.5GHz, with a dynamic range better than 70dB. The ultrasonic partial discharge sensor uses a piezoelectric ceramic transducer, with a center frequency calibrated at 40kHz and a sensitivity of not less than -65dB (0dB=1V / μ The high-precision temperature and humidity sensor uses an integrated digital sensing element. The temperature measurement range is -40℃ to +125℃ with an accuracy of ±0.1℃, and the relative humidity measurement range is 0%RH to 100%RH with an accuracy of ±1.5%RH. The response time is less than 5 seconds. It is installed in the convection zone 20cm to 30cm above the busbar connection point inside the ring mains enclosure. The leakage current transformer is a zero-flux current transformer with a range of 0 to 10mA and a resolution better than 1. μ A. The mechanical vibration sensor is a triaxial MEMS accelerometer with a frequency response range of 10Hz to 10kHz and a full-scale range of ±16g.

[0068] To ensure the consistency of correlation between different physical quantities over time, the FPGA module inside the edge computing gateway generates a high-precision synchronization trigger pulse. This pulse is distributed to the interface circuits of each sensor via a synchronization bus, achieving phase locking of the sampling clock for each channel. The data collected by each sensor is encapsulated into a raw state vector containing a microsecond-level synchronization timescale. By aligning the time stamps, the system can accurately analyze the instantaneous correspondence between partial discharge pulses and voltage phase, ambient humidity fluctuations, and sudden changes in leakage current.

[0069] The phase-locking accuracy of the synchronous trigger pulse determines the temporal resolution of multiphysics causal analysis. In this embodiment, the rising edge jitter of the synchronous pulse generated by the FPGA is less than 10 ns, and the difference in the analog-to-digital conversion start-up time of each sensor channel is controlled within ±50 ns. This level of accuracy is sufficient to distinguish the time difference between the arrival of partial discharge electromagnetic waves and ultrasonic waves at the sensors, thereby providing a reliable time reference for the acoustic-electric joint localization of the discharge source.

[0070] Furthermore, in the adaptive discrimination step 102 based on feature energy mapping, the edge computing gateway performs sliding window preprocessing on the acquired original state vector. For The wavelet hard threshold denoising algorithm was used to remove background electromagnetic interference, and the discharge repetition frequency in each power frequency cycle was calculated. Apparent discharge quantity q and cumulative discharge energy .against The fundamental component and harmonic amplitudes are extracted using Fast Fourier Transform. A scene discrimination function is then established. The operating conditions are divided into five categories: normal operation, condensation risk, partial discharge abnormality, thermal abnormality, and combined deterioration.

[0071] In one specific embodiment, the logic for determining condensation risk scenarios relies on the condensation temperature difference index. The calculation is performed when the difference between the ambient temperature and the dew point temperature decreases to a first preset threshold. (For example, within 2℃) and leakage current is observed at this temperature. The growth rate exceeded the second preset threshold for three consecutive cycles. When the condensation rate reaches 5% / min (for example), the system automatically determines that the current situation is a condensation risk scenario. In the determination of partial discharge abnormal scenarios, if... The apparent discharge quantity q in the feature exceeds the third preset threshold. If the amplitude of the ultrasonic partial discharge sensor is observed to be increasing (e.g., 50 pC), and a trend of increasing amplitude is simultaneously observed, then it is determined to be an abnormal partial discharge. The measured bus hotspot temperature exceeded the fourth preset threshold. (For example (or the temperature rise slope exceeds the fifth preset threshold) If the temperature is 10℃ / h, it is considered a thermal anomaly scenario. When the judgment logic triggers multiple of the above indicators simultaneously, the system enters the complex degradation scenario judgment.

[0072] First preset threshold The 2℃ threshold setting takes into account actual measurement errors and the non-uniformity of the microscopic state of the insulation surface. Limited by sensor accuracy and spatial temperature gradient, setting the threshold to 2℃ ensures sensitivity in detecting condensation risks while avoiding frequent false alarms due to measurement noise. Second preset threshold. The setting of 5% / min is based on a dynamic model of the evolution of surface conductivity over time during water film formation, which can effectively capture the early stage of the condensation process.

[0073] In step 103, which involves adaptively adjusting the dynamic sampling frequency and algorithm complexity, the edge computing gateway adjusts the system's resource configuration in real time based on the determination result of step 102. Under normal operating conditions, the system executes a low-power monitoring mode, reducing the sampling frequency of the ultra-high frequency partial discharge channel to [a lower value]. (e.g., 10MS / s). Once the system detects an abnormal scenario, it immediately switches to a dedicated diagnostic mode, increasing the sampling frequency to [missing value]. (For example, 3.2 GS / s) to obtain nanosecond-level discharge waveform details, supporting subsequent VMD algorithm deconstruction. In complex degradation scenarios, the system enables a full-dimensional deep analysis mode, not only maintaining a high sampling rate but also initiating online iterative calculation of the integrated degradation-lifetime coupling model in step four.

[0074] In step 104 of the integrated prediction model construction and solution for multiphysics coupling degradation mechanism and remaining lifetime, the edge computing gateway performs the numerical solution of the integral equation in step four. This is achieved by real-time data acquisition... H , Convert to the corresponding physical quantity (where T= +273.15), substitute into the coupling acceleration function And combined with the deterioration state of the previous cycle The current degradation state is updated recursively using numerical integration methods. Then, the remaining lifespan percentage can be obtained through the mapping relationship. In a preferred embodiment of the present invention, the coupling coefficient and self-acceleration coefficient in the model are obtained through accelerated aging tests and stored in the non-volatile memory of the edge computing gateway, and the parameters can be fine-tuned according to the actual material batches during operation and maintenance.

[0075] In step 105 of the real-time calculation and multi-level early warning of remaining lifetime based on the integrated model, the system calculates the remaining lifetime based on the results obtained from the calculation. and Execute the level four early warning logic. When >50% and When the condition is stable, the output is green, indicating a normal state. When 30% < ≤50% or A yellow alert will be issued when a trend of increase is observed. When 10% < When the percentage is ≤50%, an orange alert will be issued. ≤10% or When there is an exponential surge, a red critical alarm will be issued.

[0076] Furthermore, the present invention also includes a data feedback and cloud-based expert database closed-loop learning step 106. The edge computing gateway encrypts and uploads the feature vector, scene judgment result, and degradation state evolution trajectory of each early warning cycle to the cloud server through a communication channel. The cloud server uses deep learning algorithms to train on massive historical samples, continuously corrects the threshold in the scene judgment logic and the coupling coefficient in the integrated model, and periodically sends the optimized parameters to the edge to achieve continuous algorithm evolution.

[0077] At the cloud-based collaborative level, the closed-loop learning mechanism implemented in step 106 operates as follows: when multiple ring main units distributed in different geographical areas simultaneously transmit similar degradation evolution pattern data, the cloud server automatically optimizes the coupling coefficient and self-acceleration coefficient in the global parameter library. Through this collective intelligence perception and big data analysis, the system can adapt to ring main units of different brands and insulation material processes, greatly reducing the calibration workload in the initial stage of system deployment.

[0078] The core logic of this invention lies in abandoning the limitation of treating each physical quantity in isolation in traditional monitoring, and instead using an integrated coupled model to uniformly describe the continuous degradation process of the insulation system under the action of multiple physical fields. In practical engineering applications, the real-time kernel inside the edge computing gateway adopts a double-buffered task scheduling mechanism to ensure that high-frequency sampling tasks and complex integral equation solving tasks do not conflict with resources. In terms of hardware circuit design, considering the complex electromagnetic environment inside the ring network box, the signal conditioning circuits of all sensors have undergone rigorous electromagnetic compatibility design, the signal transmission lines use double-shielded twisted cables, and multi-level surge protection is set at the gateway entrance.

[0079] In summary, this invention, by deploying a multimodal sensor network, constructs a microsecond-level synchronous acquisition mechanism, utilizes a scenario-adaptive driving sampling strategy, and based on a unique integrated prediction model of degradation and lifespan using coupled electrical, thermal, humidity, and mechanical multi-physics fields, achieves high-precision online monitoring and aging early warning of the insulation status inside the ring main unit. This invention not only senses the current insulation health in real time but also accurately predicts the remaining lifespan based on a nonlinear coupling mechanism, providing scientific data support for condition-based maintenance of power systems and significantly reducing the risk of power outages caused by sudden insulation faults.

[0080] Example 3 Based on Example 1, this embodiment proposes a deeply optimized degradation evolution model to address the spatial location differences of partial discharge and the environmental temperature measurement hysteresis effect.

[0081] In the actual operation of ring main units, partial discharge can occur on the surface of insulating components or inside the insulating body. Air gap discharges inside the body are often far from the heat dissipation surface and located in high electric field stress zones, posing a much higher risk of insulation breakdown than surface corona discharges. Meanwhile, although high-precision temperature and humidity sensors can accurately capture the state of the ambient medium, under sudden load increases, the ambient temperature measured by the sensors cannot transiently reflect the thermal stress generated by the drastic temperature rise of the main circuit conductor on the internal insulation due to the limitations of the heat conduction physical process.

[0082] To further quantify the destructive power of partial discharge, this embodiment adds a load current transformer, such as a high-frequency Rogowski coil, connected to the main circuit cable in the multimodal sensor group for synchronously acquiring the main circuit load current. The configuration of the remaining sensors is the same as in Embodiment 1. Simultaneously, the edge computing gateway introduces a combined acoustic-electric spatial positioning mechanism and an electrothermal feedforward compensation strategy to optimize the integrated degradation-lifetime coupling model in step four in the following two aspects.

[0083] Spatial Depth Weighted Correction Based on Acoustic-Electronic Time Delay Difference: Monitoring System Acquires UHF Partial Discharge Signals in Parallel With ultrasonic partial discharge signal The transient envelopes of the two signals are extracted using Hilbert transform, and a cross-correlation algorithm is employed to accurately locate the starting edge of the envelope waveform. The arrival time of the UHF envelope is then obtained based on the starting edge time node. With time of arrival of the ultrasonic envelope Calculate the acoustic-electric time delay difference:

[0084] in: Indicates the difference in acoustic and electrical time delay; Indicates the arrival time of the envelope of the ultra-high frequency partial discharge signal; This indicates the arrival time of the ultrasonic partial discharge signal envelope excited by the same discharge source.

[0085] Because electromagnetic waves travel at extremely high speeds This can be approximated as the moment of discharge occurrence. The equivalent acoustic path depth of the partial discharge source relative to the ultrasonic sensor is calculated based on the acoustic-electric time delay difference:

[0086] in: Indicates the equivalent acoustic path depth; This indicates the equivalent propagation speed of ultrasound in an insulating medium; the typical calibration value for epoxy resin insulators is 2600 m / s.

[0087] Get path depth Then, a spatial depth weighting factor is constructed. :

[0088] in: Indicates the spatial depth weighting factor; This represents the sensitivity coefficient to internal defects in the material, which is dimensionless and set to 0.8 in this embodiment. The characteristic physical thickness of the insulating component is indicated in meters (m), and in this embodiment it is set to 0.05m.

[0089] Then, the spatial depth weighting factor is used to adjust the initial electro-induced degradation coupling coefficient. Dynamic correction is performed to obtain the corrected electro-degradation coupling coefficient:

[0090] in This represents the initial electro-degradation coupling coefficient calibrated through accelerated aging tests. Replace the multiphysics coupling acceleration function described in step four. In This allows for differentiated degradation assessment based on spatial depth. When the acoustic-electric time delay difference is extremely small and the discharge source is located in a high-field-strength concentration region deep within the insulation... Approaching the maximum value of 1.8, the rate of electrical degradation increases significantly; when the discharge source is located on the surface of the insulator or at external hardware... Larger Approaching version 1.0, the system maintains a normal evaluation rate.

[0091] Electrothermal feedforward compensation based on load current: To eliminate the time lag error caused by heat transfer in ambient temperature monitoring, the multiphysics coupling acceleration function described in step four... The real-time thermodynamic temperature T is introduced into the electrothermal feedforward compensation term. Specifically, the instantaneous effective value of the main circuit load current is acquired using the load current transformer. And calculate the real-time thermodynamic temperature using the following formula:

[0092] in: This indicates the ambient temperature measured by a high-precision temperature and humidity sensor. express The instantaneous effective value of the main circuit load current at any given moment; This represents the equivalent Joule heat transfer coefficient related to the conductor material and heat dissipation structure of the ring main unit, expressed in units of... The typical value is 1.2 × 10. -6 ; This represents the integral sliding time window, in seconds. In this embodiment, it is set to 900 seconds, corresponding to the typical thermal conduction characteristic time constant of the insulating surface.

[0093] This electrothermal feedforward compensation mechanism directly incorporates the Joule heating effect of the load current into the thermodynamic temperature calculation, bypassing the heat conduction delay, and ensuring that the aging early warning model can predict the accelerated insulation thermal activation effect caused by the heat surge during sudden large loads or fault ride-throughs in the distribution network.

[0094] To verify the engineering performance of the optimized model, a comparative verification was conducted in a 10kV ring main unit operating in a high-load industrial park. During the experiment, internal micro-air gap defects were artificially created, and fluctuating loads were applied. Using the basic model (i.e., the model described in Example 1) without incorporating acoustic-electric time delay depth weighting and electrothermal feedforward compensation as a benchmark, the warning response times of the two models under different operating conditions were recorded. The experimental results are as follows: Under surface creepage and stable load conditions, the basic model's early warning time was 124.5 hours, while the optimized model's was 120.2 hours, representing a warning time lead of approximately 3.5%. Under internal air gap and stable load conditions, the basic model's early warning time was 115.8 hours, while the optimized model's was 82.4 hours, representing a warning time lead of approximately 28.8%. Under surface creepage and drastically fluctuating load conditions, the basic model's early warning time was 108.3 hours, while the optimized model's was 88.5 hours, representing a warning time lead of approximately 18.3%. Under internal air gap and drastically fluctuating load conditions, the basic model's early warning time was 98.6 hours, while the optimized model's was 55.2 hours, representing a warning time lead of approximately 44.0%.

[0095] It is evident that by introducing acoustic-electric time delay depth weighting and load current thermal feedforward compensation, the monitoring system can more sensitively capture the destructive power of rapidly increasing multi-physics coupling when facing high-risk internal air gap discharges and severe load fluctuations. The advance warning time is increased by up to 44.0%, effectively extending the time window for maintenance intervention. This optimization scheme further enriches the adaptability of the technical solution of this invention under different operating scenarios.

[0096] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "include," "contain," 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.

[0097] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for online monitoring and aging early warning of the internal insulation status of a ring main unit, characterized in that: Includes the following steps: Step 1: Simultaneously collect ultra-high frequency partial discharge signals, ultrasonic partial discharge signals, ambient temperature, relative humidity, leakage current on the surface of insulating components, and vibration acceleration of the shell by deploying a multi-modal sensor group inside the ring network box, and construct an original state vector with a synchronous time scale; Step 2: Extract features from the original state vector and determine the operating conditions of the ring network box as normal operation, condensation risk, partial discharge abnormality, thermal abnormality, or combined deterioration scenario based on the condensation temperature difference index, partial discharge characteristic quantity, and temperature characteristic quantity. Step 3: Based on the identified operating scenario, adaptively switch the sampling frequency of the monitoring system: use the first sampling frequency in normal operating scenarios, and increase the sampling frequency to the second sampling frequency in abnormal operating scenarios, with the second sampling frequency being higher than the first sampling frequency; Step four involves constructing an evolution model of the insulation system's degradation state, reflecting the coupling effects of multiple physical fields including heat, humidity, electricity, and force. This model determines the current degradation state variables based on ambient temperature, relative humidity, partial discharge power, and vibration intensity, and calculates the remaining insulation lifetime and percentage of remaining insulation lifetime based on these variables. This is determined through the following mapping relationship: in, represents the critical degradation state value for insulation failure, calibrated through accelerated aging tests, with a typical value of 1.0; p represents the lifetime mapping shape factor, dimensionless, used to adjust the nonlinearity of the lifetime decay curve. This represents the dimensionless degradation state variable of the insulation system at time t. This represents the initial degradation state value; Step 5: Real-time calculation of the deterioration state variables and their corresponding remaining lifetimes, and output of graded early warning signals based on the remaining lifetime values ​​and their rate of change.

2. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 1, characterized in that: In step two, the condensation temperature difference index is the difference between the ambient temperature and the dew point temperature. The dew point temperature is calculated based on the ambient temperature, relative humidity, and ambient atmospheric pressure. When the condensation temperature difference index is lower than the first preset threshold and the leakage current growth rate exceeds the second preset threshold, it is determined to be a condensation risk scenario.

3. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 2, characterized in that: In step two, the partial discharge characteristic quantities include the apparent discharge quantity; when the apparent discharge quantity exceeds the third preset threshold and the amplitude of the ultrasonic partial discharge signal shows a trend of increasing, it is determined to be an abnormal partial discharge scenario.

4. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 3, characterized in that: In step two, the temperature characteristic quantities include the ambient temperature value and its temperature rise slope; when the ambient temperature exceeds the fourth preset threshold or the temperature rise slope exceeds the fifth preset threshold, it is determined to be a thermal anomaly scenario.

5. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 4, characterized in that: In step two, when the criteria for judging at least two of the three scenarios—condensation risk scenario, partial discharge abnormal scenario, and thermal abnormal scenario—are met simultaneously, it is judged as a composite degradation scenario.

6. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 1, characterized in that: In step three, the first sampling frequency is determined based on the update requirements of the discharge statistics within the power frequency cycle, and the second sampling frequency meets the Nyquist sampling requirements for the rising edge of the pulse in the UHF partial discharge signal.

7. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 1, characterized in that: In step four, the insulation system degradation state evolution model uses an integral form to express the evolution of degradation state variables over time. The integral kernel function includes a temperature exponential term reflecting the thermal activation effect, a humidity exponential term reflecting the change in surface conductivity caused by humidity, an electro-induced degradation term reflecting the erosion of accumulated discharge energy, and a vibration intensity term reflecting mechanical vibration stress fatigue. The integral kernel function also includes a self-acceleration term that is positively correlated with the degradation state variables.

8. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 1, characterized in that: The evolution model of the degradation state of the insulation system is represented by the following integral equation: in: The dimensionless degradation state variable of the insulation system at time t, with a range of values. ; This represents the initial degradation state value. This indicates the critical degradation state value for insulation failure; The degradation rate constant, representing the intrinsic properties of the insulating material, has dimensions of . ; This represents a multiphysics coupling acceleration function, which is dimensionless. This represents the self-accelerating function in the deteriorated state, and its dimension is dimensionless. The variable 't' represents the integration time variable, in seconds (s); 't' represents the current time, in seconds (s). Indicates the integration time. The instantaneous value of relative humidity, Indicates the integration time. The instantaneous value of the thermodynamic temperature. Indicates the integration time. The instantaneous value of partial discharge power, Indicates the integration time. The instantaneous effective value of vibration intensity Indicates the integration time. The instantaneous value of the deteriorated state variable.

9. The method for online monitoring and aging early warning of the internal insulation status of a ring main unit according to claim 1, characterized in that: Step five also includes: the edge computing gateway generating an early warning message containing the type of abnormal scenario, the dominant degradation factor, and the predicted remaining lifetime, and sending the early warning message to the operation and maintenance management system through the wireless communication module.

Images