Live-line work insulation performance prediction method and system fusing multivariate information

Abstract

The invention provides a multivariate information fused live working insulation performance prediction method and system, and belongs to the technical field of live working safety, and the prediction method is applied to a prediction system comprising a data perception and collection module, a data analysis module, a decision prediction module, a monitoring feedback module and an alarm module. The method specifically comprises the steps of collecting environmental data, stress data and leakage current of the insulation pull rod equipment to identify an operation condition and determine a prediction demand, correspondingly selecting a short-term recursive prediction strategy or a long-term multi-dimensional input and output prediction strategy, predicting the change trend of the leakage current, identifying and predicting an adjustment demand in real time in the prediction process, and predicting the change trend of the leakage current. And performing prediction strategy transition switching correspondingly, updating a prediction result, and performing integrity scoring on the insulation performance of the hot-line work to realize safety early warning of the insulation pull rod equipment. According to the invention, by setting the dual-time scale prediction strategy and the transition switching link, the accuracy of the insulation performance prediction result is improved, and the operation safety of the insulation pull rod equipment is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of live working safety, and in particular to a method and system for predicting insulation performance of live working by integrating multi-dimensional information. Background Art

[0002] Insulating rods, as crucial supporting and insulating components for high-voltage power equipment, primarily support conductors and maintain electrical isolation. Their electrical performance and mechanical strength are directly linked to the safe and stable operation of power systems. During their operational life, insulating rods are exposed to complex environments, including high-voltage electric fields and coupled mechanical loads. Excessive ambient temperature fluctuations can cause structural deformation and degrade the end seals of the insulating rods due to thermal expansion, leading to moisture intrusion and a rapid increase in leakage current. Furthermore, when ambient humidity exceeds the specified limit, a water film forms on the surface of the insulating rods. This moisture penetration degrades the dielectric strength of the epoxy resin matrix, exacerbating the risk of flashover or internal breakdown due to leakage current and partial discharge. Regarding mechanical stress, cyclical load fluctuations can also reduce material life through cumulative fatigue, leading to delamination of the insulating rod laminate interface. Therefore, during the operation of insulating rods, they can experience both short-term abnormalities, such as sudden increases in leakage current, due to extreme conditions such as heavy rain or sudden load increases, and long-term, gradual, cumulative degradation due to material aging and localized degradation. In other words, in complex environments, insulation performance testing of insulated rod equipment actually requires both short-term, high-frequency early warnings and long-term trend identification, depending on the operating conditions. Currently, commonly used insulation testing methods, such as manual periodic testing and offline testing after power outages, are inadequate for the dynamic and complex insulation performance testing requirements of insulated rod equipment due to issues such as poor timeliness and the lack of continuous monitoring.

[0003] In order to solve the above problems, relevant technologies have introduced the long short-term memory model LSTM to realize the insulation performance detection of insulating pull rod equipment. As a deep learning model that can effectively process time series data, LSTM can capture the long-term and short-term dependencies between environmental parameters, load characteristics and historical operation data to realize short-term dynamic prediction and long-term trend prediction of leakage current, which can effectively meet the insulation performance detection needs of insulating pull rod equipment.

[0004] However, in practical applications, short-term predictions rely on high-frequency sampling data as input, focusing on capturing instantaneous fluctuations, while long-term predictions rely on low-frequency trend data and highlight slowly changing patterns. There are significant differences between the two in terms of data time granularity and feature weight distribution. When faced with changing testing requirements and the need to switch prediction strategies, sudden changes in parameters often lead to disconnected prediction results, making it impossible to form a continuous and consistent insulation performance prediction curve. This can lead to the underreporting of short-term sudden faults and the misjudgment of long-term aging trends. Consequently, the insulation performance prediction results for insulating rod equipment are less accurate, making it impossible to accurately identify insulation degradation, and thus, the operational safety of insulating rod equipment cannot be guaranteed. Summary of the Invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art that the long short-term memory model LSTM is used to perform short-term dynamic prediction and long-term trend prediction on leakage current. When the prediction strategy needs to be switched, the prediction results are often disconnected due to sudden changes in parameters, and a continuous and consistent insulation performance prediction curve cannot be formed. The accuracy of the insulation performance prediction results is low, and the operation safety of the insulating rod equipment cannot be guaranteed. A method and system for predicting the insulation performance of live operations that integrates multiple information are provided. Based on a strategy selection mechanism for working condition identification, a short-term recursive prediction strategy or a long-term multi-dimensional input and output prediction strategy is targetedly selected to predict the leakage current change trend, which can effectively meet the insulation performance detection needs of the insulating rod equipment. When the prediction strategy needs to be switched, a transition switching link is set to eliminate the prediction error caused by the sudden change adjustment of parameters, improve the accuracy of the insulation performance prediction results, accurately identify the insulation degradation of the insulating rod equipment, and ensure the operation safety of the insulating rod equipment.

[0006] The purpose of the present invention is achieved through the following technical solutions: The live working insulation performance prediction method integrating multi-information includes: Collect environmental data, stress data and leakage current of insulating rod equipment during live operation; Based on environmental data and stress data, identify the current operating conditions and determine the forecast demand based on the operating conditions; Based on the leakage current, a short-term recursive prediction strategy or a long-term multi-dimensional input-output prediction strategy is selected according to the prediction requirements to predict the leakage current change trend; According to the leakage current change trend prediction results and operating conditions, the prediction adjustment needs are identified in real time, the prediction strategy transition is switched according to the prediction adjustment needs, and the leakage current change trend prediction results are updated; Based on the current leakage current change trend prediction results and the corresponding leakage current prediction strategy, the insulation performance integrity score of live working is evaluated; Provide safety warning for insulating pull rod equipment based on integrity scoring results.

[0007] The strategy selection mechanism based on operating condition identification accurately matches the testing requirements of insulation rod equipment under different operating conditions, and then selects either a short-term recursive prediction strategy or a long-term multi-dimensional input-output prediction strategy to predict leakage current trends. This not only promptly responds to short-term anomalies in equipment insulation performance, but also effectively identifies slowly accumulating insulation degradation trends due to material aging and localized degradation, thus fully covering both short-term and long-term insulation performance testing requirements for insulation rod equipment. A transitional switching process is implemented during the prediction strategy switching process, ensuring a smooth transition between the short-term recursive prediction strategy and the long-term multi-dimensional input-output prediction strategy. This eliminates sudden changes in prediction errors caused by step-like parameter changes, ensures the continuity and consistency of leakage current trend prediction results, and significantly improves the accuracy of insulation performance prediction results. Based on accurate insulation degradation identification results, the insulation integrity of the insulation rod equipment can be precisely quantitatively assessed, enabling timely safety warnings and proactive maintenance measures to effectively avoid flashovers, internal breakdowns, and other faults caused by insulation failure, thereby ensuring the safe operation of the insulation rod equipment.

[0008] Furthermore, the identifying of the current operating condition based on the environmental data and stress data and determining the forecast demand according to the operating condition include: Preprocess the environmental data and stress data, and extract working condition features from the preprocessed environmental data and stress data; Based on the working condition characteristics, the current operating condition is identified in combination with the corresponding working condition classification rules; Obtain the duration and type of the current operating condition to match the corresponding forecast demand.

[0009] Furthermore, based on the leakage current, a short-term recursive prediction strategy or a long-term multi-input and output prediction strategy is selected according to the prediction requirements to predict the leakage current change trend, including: When the working condition type is a short-term extreme working condition and the duration is lower than the preset time threshold, the predicted demand is a short-term high-frequency predicted demand; Select a short-term recursive prediction strategy based on the short-term high-frequency prediction requirements, and set the leakage current prediction model parameters and recursive parameters according to the corresponding operating conditions; Using the collected leakage current as the initial input, the leakage current value at a future time point is predicted by the leakage current prediction model, and the leakage current prediction value is fed back to the leakage current prediction model as a new input; The prediction of leakage current value and new input feedback are repeatedly performed to generate a continuous short-term prediction sequence; Slide through the recursively generated short-term forecast demand with a fixed window size and calculate the trend characteristics within the window; Based on trend characteristics and preset rules, the leakage current change trend prediction results are obtained.

[0010] Furthermore, the method of predicting the leakage current change trend by selecting a short-term recursive prediction strategy or a long-term multi-input and output prediction strategy based on the leakage current according to the prediction requirements may further include: When the working condition type is a long-term stable working condition and the duration is not less than the preset time threshold, the predicted demand is a long-term trend predicted demand; When selecting a long-term multi-dimensional input-output prediction strategy based on long-term trend prediction requirements, set the leakage current prediction model parameters according to the corresponding operating conditions; Construct input data samples based on leakage current and its influencing factors, and output a continuous long-term prediction sequence in combination with the leakage current prediction model; Divide the long-term prediction series and the collected environmental data and stress data according to the time granularity, calculate the trend characteristics of the long-term prediction series in each time period, as well as the trend characteristics of the environmental data and stress data in each time period, and add corresponding trend category labels to them respectively; Based on the trend characteristics and corresponding trend category labels, the causes of the change trend are identified and the leakage current change trend prediction results are obtained.

[0011] Furthermore, the real-time identification and prediction of adjustment requirements based on the leakage current change trend prediction results and operating conditions includes: Identify the current operating conditions and their duration, and obtain the matching degree of the forecast strategy based on the current forecast demand; Obtain the working condition characteristics of the current operating condition, and obtain the working condition adaptability of the prediction strategy based on the leakage current change trend prediction results; According to the actual measured value of the currently collected leakage current, a trend consistency index of the leakage current change trend prediction result is obtained; Identify and predict adjustment needs based on matching, working condition suitability and trend consistency index.

[0012] Furthermore, the prediction strategy transition switching is performed according to the prediction adjustment demand, and the leakage current change trend prediction result is updated, including: Obtain the working condition fluctuation coefficient based on environmental data and stress data, and set the corresponding transition window according to the working condition fluctuation coefficient; In the transition window, the prediction values ​​of the current prediction strategy and the switched target prediction strategy are output synchronously, and the corresponding prediction fusion value is obtained by combining the dynamic weights; When the predicted fusion value and the dynamic weight meet the transition end condition, the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy.

[0013] Furthermore, when the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy, the following steps are also performed: Obtain a transition switching target based on the demand type of the predicted adjustment demand, and select a transition switching direction of the prediction strategy based on the transition switching target; Based on the transition switching direction of the forecast strategy, a corresponding relationship between the trend characteristics of the current forecast strategy and the target strategy is established, and a list of trend characteristics of the target strategy is obtained; Extract the supplementary features of the current prediction strategy and add the supplementary features to the trend feature list of the target strategy; Based on the trend features in the trend feature list, the leakage current change trend prediction result is updated.

[0014] Furthermore, the integrity scoring of the live working insulation performance is performed based on the current leakage current change trend prediction result and the corresponding leakage current prediction strategy, including: Obtain the safe upper limit of leakage current of the insulating rod equipment in the current live working scenario, and obtain the corresponding prediction time window based on the current leakage current prediction strategy; According to the current leakage current change trend prediction result, the leakage current prediction value at each moment in the corresponding prediction time window is obtained; Based on the safety upper limit of leakage current and the predicted leakage current value at each moment in the prediction time window, the integrity score of the insulation performance of live working is obtained.

[0015] A live working insulation performance prediction system integrating multivariate information, used to execute any of the above-mentioned prediction methods, comprises: The data sensing and collection module is installed at the insulating rod equipment and is used to collect environmental data, stress data and leakage current of the insulating rod equipment during live operation; The data analysis module is connected to the data perception and acquisition module and is used to identify operating conditions and determine forecast requirements based on the collected environmental data, stress data, and leakage current; The decision-making prediction module is connected to the data perception and acquisition module and the data analysis module respectively, and is used to select a prediction strategy according to the prediction requirements and predict the leakage current change trend based on the leakage current; The monitoring feedback module is connected to the data analysis module and the decision prediction module respectively, and is used to identify the prediction adjustment needs based on the leakage current change trend prediction results and operating conditions, and control the decision prediction module to transitionally adjust the prediction strategy applied according to the prediction adjustment needs; The alarm module is connected to the decision-making prediction module and is used to score the integrity of the insulation performance of live working according to the leakage current change trend and prediction strategy, and to issue safety warnings for the insulating rod equipment based on the integrity score.

[0016] Furthermore, the data sensing and collection module includes: The temperature and humidity sensor is installed on the middle outer wall of the insulating rod equipment and is used to collect environmental data of the insulating rod equipment during live operation; Stress sensors are installed on the middle surface of the axis of the insulating pull rod equipment and the metal end, and are used to collect stress data of the insulating pull rod equipment during live operation; The leakage current sensor is installed at the end of the insulating pull rod equipment and is used to collect the leakage current of the insulating pull rod equipment during live operation.

[0017] A device for predicting insulation performance of live working by integrating multiple information, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the live working insulation performance prediction method according to any one of the above items.

[0018] A computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method for predicting insulation performance of live working as described in any one of the above.

[0019] The beneficial effects of the present invention are:

[0020] (1) The strategy selection mechanism based on working condition identification can accurately match the detection requirements of the insulating rod equipment under different operating conditions, and then select the short-term recursive prediction strategy or the long-term multi-dimensional input and output prediction strategy to carry out leakage current change trend prediction. It can not only respond to the short-term abnormalities of the equipment insulation performance in a timely manner, but also effectively identify the slowly accumulated insulation performance degradation trends such as material aging and local degradation, thus achieving full coverage of the short-term and long-term needs of the insulation performance detection of the insulating rod equipment. In the process of switching the prediction strategy, a transition switching link is set to make the switching between the short-term recursive prediction strategy and the long-term multi-dimensional input and output prediction strategy present a smooth transition state, eliminating the sudden change of prediction error caused by the step change of parameters, ensuring the continuity and consistency of the leakage current change trend prediction results, and significantly improving the accuracy of the insulation performance prediction results. Based on the accurate insulation degradation identification results, the insulation performance integrity of the insulating rod equipment can be accurately quantitatively evaluated, and then a safety warning can be issued in time, and maintenance measures can be taken in advance to effectively avoid flashover, internal breakdown and other faults caused by insulation performance failure, thereby ensuring the safe operation of the insulating rod equipment; (2) In short-term prediction, a continuous prediction sequence is generated based on recursive feedback, and trend features are extracted by sliding windows. This ensures high-frequency capture of instantaneous fluctuations and avoids the randomness of single prediction values ​​through trend feature analysis, making the identification of abnormal trends such as short-term leakage current surges more reliable. In long-term prediction, input samples are constructed by correlating influencing factors such as environment and stress, and combined with multi-dimensional trend features and cause analysis, the driving mechanism of long-term degradation trends such as material aging is reflected to ensure the reliability of the obtained long-term trend prediction results; (3) Through multi-dimensional evaluation of matching degree, working condition adaptability, and trend consistency index, the adaptability of the prediction strategy is dynamically quantified, providing an accurate basis for strategy switching. During the transition switch, the transition window is set based on the working condition fluctuation coefficient, and the dual strategy prediction values ​​are fused in combination with dynamic weights. At the same time, by establishing a corresponding relationship between feature supplementation and trend features, the continuity and integrity of the change trend during the switching process are ensured, effectively eliminating the prediction fault caused by parameter mutation. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of a process of the present invention;

[0022] Figure 2 This is a structural diagram of an embodiment of the present invention;

[0023] Figure 3 This is a schematic diagram of the location of a data sensing and acquisition module on an insulating pull rod device according to an embodiment of the present invention; Among them: 1. Data perception and acquisition module, 11. Temperature and humidity sensor, 111. Temperature sensor, 112. Humidity sensor, 12. Stress sensor, 121. Tensile load sensor, 122. Shear load sensor, 13. Leakage current sensor, 2. Data analysis module, 3. Decision prediction module, 4. Monitoring feedback module, 5. Alarm module, 6. Insulated pull rod equipment. DETAILED DESCRIPTION

[0024] The present invention will be further described below with reference to the accompanying drawings and examples.

[0025] Example 1: A method for predicting insulation performance of live working by integrating multiple information, such as Figure 1 Shown, including: Collect environmental data, stress data and leakage current of insulating rod equipment during live operation; Based on environmental data and stress data, identify the current operating conditions and determine the forecast demand based on the operating conditions; Based on the leakage current, a short-term recursive prediction strategy or a long-term multi-dimensional input-output prediction strategy is selected according to the prediction requirements to predict the leakage current change trend; According to the leakage current change trend prediction results and operating conditions, the prediction adjustment needs are identified in real time, the prediction strategy transition is switched according to the prediction adjustment needs, and the leakage current change trend prediction results are updated; Based on the current leakage current change trend prediction results and the corresponding leakage current prediction strategy, the insulation performance integrity score of live working is evaluated; Provide safety warning for insulating pull rod equipment based on integrity scoring results.

[0026] In order to achieve accurate detection and prediction of the insulation performance of insulating rod equipment, in the data collection link, considering that the insulation performance degradation of insulating rod equipment is closely related to the environment, stress and its own operating status, it is necessary to synchronously collect multi-dimensional data during its live operation.

[0027] Environmental data covers key parameters such as temperature and humidity, capturing the impact of temperature and humidity fluctuations on water film formation and moisture penetration on equipment surfaces. Stress data focuses on the cyclical changes and transient impacts of loads, allowing analysis of the effects of mechanical stress on material fatigue accumulation and interfacial delamination. Leakage current data directly reflects the real-time status of equipment insulation performance and is a core indicator for determining the degree of insulation degradation. Therefore, during live operations on insulated pull rod equipment, these three types of data are continuously collected to provide complete, real-time basic data support for subsequent operating condition identification, prediction strategy selection, and performance evaluation.

[0028] Based on the collected environmental data and stress data, further identification of operating conditions and determination of prediction needs are carried out.

[0029] Specifically, identifying the current operating condition based on environmental data and stress data, and determining the forecast demand according to the operating condition, includes: Preprocess the environmental data and stress data, and extract working condition features from the preprocessed environmental data and stress data; Based on the working condition characteristics, the current operating condition is identified in combination with the corresponding working condition classification rules; Obtain the duration and type of the current operating condition to match the corresponding forecast demand.

[0030] Considering the complex operating environment of insulating rod equipment, environmental and stress data often contain noise, missing data, or abnormal fluctuations. Directly using raw data for operating condition identification can lead to inaccurate identification results, which in turn affects the determination of subsequent forecast requirements. Therefore, environmental and stress data are preprocessed, including outlier removal and data standardization, to eliminate noise interference.

[0031] Moreover, since the characteristics of equipment insulation degradation vary greatly under different working conditions, only by accurately extracting the working condition characteristics, identifying the working conditions in combination with classification rules, and matching the prediction requirements according to the duration and type of the working conditions can the applicability of subsequent prediction strategies be ensured to meet the different needs of short-term high-frequency warnings and long-term trend identification.

[0032] Therefore, working condition feature extraction is performed based on the preprocessed environmental data and stress data to mine key information in the data that can reflect the essence of the working conditions, such as temperature change rate, humidity peak, stress fluctuation period, etc.

[0033] After obtaining the operating condition characteristics, the current operating condition is accurately identified using pre-set operating condition classification rules. Specifically, the operating condition classification rules are constructed based on the degradation mechanism of the insulating pull rod equipment and historical operating data. For example, based on historical operating data, a short-term extreme operating condition is defined as humidity > 90% for a duration > 1 hour, with a peak load exceeding 1.2 times the rated value. A long-term stable operating condition is defined as a temperature fluctuation < 5°C / day, humidity < 60%, and a load fluctuation frequency < 0.1Hz.

[0034] During the identification process, the type of working condition is determined by calculating the matching degree between the current working condition characteristics and the rules of each category, such as the number of matching characteristic parameters and the degree of deviation.

[0035] After identifying the current operating condition type, it is necessary to further count its duration and combine the two to determine the corresponding forecast demand.

[0036] For short-term extreme conditions, if the duration is short, that is, below the preset time threshold, and the characteristic parameters show a sudden rise and fall trend, the forecast demand focuses on short-term high-frequency warnings. To meet the short-term high-frequency forecast demand, the leakage current prediction model must be able to capture future instantaneous changes in leakage current to trigger an emergency response in a timely manner. If the short-term extreme condition lasts for a long time and the characteristic parameters remain at a high level of fluctuation, such as continuous heavy rain causing humidity to exceed the standard for a long time, in addition to short-term warnings, the acceleration of the condition on long-term degradation must also be taken into account, and a preliminary assessment of subsequent long-term trends must be included in the forecast demand.

[0037] For long-term stable operating conditions, if the duration is long (i.e., no less than a preset time threshold) and the characteristic parameters do not fluctuate significantly, the forecasting requirement will primarily focus on identifying long-term trends. To achieve this, the leakage current prediction model must output leakage current trends over longer periods, such as weeks or months, to assess the cumulative effects of material aging. If slow, gradual changes in characteristic parameters occur during long-term stable operating conditions, the weighting of these gradual changes in the long-term trend forecast must be strengthened to ensure timely identification of potential signs of accelerated degradation.

[0038] And the above preset time threshold can be set according to needs.

[0039] By combining the demand matching mechanism of working condition type and duration, the selection of prediction strategy is more in line with the actual operating status of the equipment, effectively avoiding the problem of insufficient accuracy caused by a single prediction strategy.

[0040] After determining the relevant prediction requirements, the corresponding prediction strategy can be further selected to carry out the corresponding leakage current change trend prediction through the corresponding leakage current prediction model.

[0041] This embodiment uses a long short-term memory (LSTM) network as a modeling tool to establish a corresponding leakage current prediction model, effectively capturing long-term dependencies and trends in time series. Specifically, data such as the magnitude and frequency of tensile and shear loads on insulating pull rods during actual operation are used as primary input features. Historical environmental variables such as temperature and humidity are also incorporated to construct the input data sequence. The leakage current prediction model is trained using supervised learning to learn the evolutionary trends of leakage current under different loads and environmental conditions, enabling it to predict future leakage current changes. Furthermore, an attention module is introduced into the LSTM backbone architecture to enhance the model's response accuracy to sudden events and critical time periods, improving the weighting of critical moments such as sudden load changes or sharp increases in humidity, thereby improving prediction accuracy. The leakage current prediction model uses mean squared error (MSE) as the primary loss function during training, and a weighted penalty mechanism is introduced for periods of sudden increases, ensuring more accurate predictions at key leakage risk points. The trained leakage current prediction model is deployed on an embedded edge computing platform, enabling on-site online computing and real-time feedback.

[0042] In terms of prediction strategies, the leakage current prediction model adopts two prediction modes. One is recursive prediction, in which the leakage current prediction model only predicts the leakage current at one point in the future and uses the output feedback as input for the next prediction. This can be used for continuous prediction in short time steps. The other is a long-term multi-input and output prediction strategy, which uses historical data from a period of time as input and outputs leakage current prediction values ​​for multiple future time steps at once. This is suitable for medium- and long-term trend analysis and whole-segment performance evaluation.

[0043] For short-term, high-frequency demands under extreme operating conditions, such as sudden bursts of humidity and excessive peak loads, a recursive prediction strategy is preferred. Under these conditions, leakage current is susceptible to transient environmental and stress fluctuations, requiring high-frequency, short-term granularity to capture sudden increases. The single-step output and feedback iteration characteristics of recursive prediction enable real-time responses to subtle changes at each point in time. Especially when short-term extreme conditions are short-lived and feature parameters rise and fall suddenly, recursive prediction can capture sudden changes in leakage current through continuously generated short-term sequences, providing a rapid warning window for emergency response.

[0044] Under such operating conditions, the leakage current is based on a short-term recursive prediction strategy selected according to the prediction requirements to predict the leakage current change trend, including: When the working condition type is a short-term extreme working condition and the duration is lower than the preset time threshold, the predicted demand is a short-term high-frequency predicted demand; Select a short-term recursive prediction strategy based on the short-term high-frequency prediction requirements, and set the leakage current prediction model parameters and recursive parameters according to the corresponding operating conditions; Using the collected leakage current as the initial input, the leakage current value at a future time point is predicted by the leakage current prediction model, and the leakage current prediction value is fed back to the leakage current prediction model as a new input; The prediction of leakage current value and new input feedback are repeatedly performed to generate a continuous short-term prediction sequence; Slide through the recursively generated short-term forecast demand with a fixed window size and calculate the trend characteristics within the window; Based on trend characteristics and preset rules, the leakage current change trend prediction results are obtained.

[0045] During the parameter setting phase, the leakage current prediction model parameters and recursive parameters must be specifically configured based on the characteristics of the short-term extreme operating conditions corresponding to the forecast requirements. For example, when the operating condition is a sudden increase in humidity due to heavy rain and accompanied by a short-term load shock, the model parameters need to increase the weight coefficient of the humidity factor and increase the number of hidden units in the LSTM layer of the leakage current prediction model to enhance the ability to fit high-frequency fluctuating data. The recursive parameters, including the recursive step size and the maximum number of recursions, must be set according to the timeliness requirements of the early warning response. At the same time, the upper and lower thresholds of the predicted value should be set based on the historical maximum leakage current value under the operating condition to avoid prediction drift caused by extreme data disturbances.

[0046] After completing the relevant parameter settings, the collected leakage current is used as the initial input, and the leakage current value at a future time point is predicted through the leakage current prediction model. The leakage current prediction value is fed back to the leakage current prediction model as a new input, and so on to complete the iterative feedback process.

[0047] Trend feature calculations utilize a fixed window sliding traversal approach, with the window size set based on the fluctuation period of short-term extreme operating conditions. During the sliding process, trend features within each window are calculated, including the average growth rate of leakage current, peak frequency, and fluctuation variance. These features effectively reflect the variation patterns of a specific time period.

[0048] Based on the calculated trend characteristics, the system generates corresponding leakage current trend prediction results in combination with preset rules. These rules are set in conjunction with the degradation threshold of the insulation rod equipment. For example, if the average growth rate within the window is greater than 100% and the peak value is ≥30μA, it is considered a high-risk sudden increase trend. If the average growth rate is between 50% and 100% and the fluctuation variance is stable, it is considered a controllable growth trend.

[0049] By matching the calculated trend features with the rules, the final output prediction results not only include the specific leakage current numerical sequence, but also the trend type label and key time nodes.

[0050] This condition-adaptive parameter configuration and iterative feedback mechanism enables the prediction sequence to accurately track transient leakage current fluctuations under short-term extreme operating conditions, effectively reducing corresponding prediction errors. Trend feature extraction using a sliding window avoids the randomness of single prediction values, and aggregate analysis of local trends effectively improves the accuracy of trend judgment, effectively avoiding the risk of flashover in insulating rod equipment. The generated continuous prediction sequence and trend table also provide fine-grained feature support for subsequent strategy transitions, ensuring a smooth transition from short-term warning to long-term trend analysis.

[0051] For long-term trend prediction needs under long-term stable operating conditions, such as scenarios with gentle fluctuations in temperature and humidity and stable loads, a long-term multi-input and output prediction strategy is suitable. This strategy uses historical data segments as input and outputs multiple time-step results at once, which can cover the slow degradation process caused by material aging. When characteristic parameters in long-term stable operating conditions change slowly, the multi-input and output model can also integrate corresponding parameters, such as the long-term trend characteristics of temperature, humidity, and load, to reflect the cumulative impact of these factors on leakage current in the prediction results, providing a quantitative basis for long-term life assessment over the entire period.

[0052] Under such operating conditions, the leakage current-based, long-term multi-input and output prediction strategy is selected according to the prediction requirements to predict the leakage current change trend, including: When the working condition type is a long-term stable working condition and the duration is not less than the preset time threshold, the predicted demand is a long-term trend predicted demand; When selecting a long-term multi-dimensional input-output prediction strategy based on long-term trend prediction requirements, set the leakage current prediction model parameters according to the corresponding operating conditions; Construct input data samples based on leakage current and its influencing factors, and output a continuous long-term prediction sequence in combination with the leakage current prediction model; Divide the long-term prediction series and the collected environmental data and stress data according to the time granularity, calculate the trend characteristics of the long-term prediction series in each time period, as well as the trend characteristics of the environmental data and stress data in each time period, and add corresponding trend category labels to them respectively; Based on the trend characteristics and corresponding trend category labels, the causes of the change trend are identified and the leakage current change trend prediction results are obtained.

[0053] During the corresponding parameter setting phase, the leakage current prediction model parameters must be specifically configured to suit the characteristics of long-term stable operating conditions. Since the degradation of insulating tie rod equipment in long-term stable operating conditions is primarily caused by material aging, slowly accumulated environmental impacts, and stable mechanical stress, the leakage current prediction model parameters should prioritize the ability to fit low-frequency trends. For example, in the LSTM leakage current prediction model, the learning rate can be appropriately reduced to avoid overfitting high-frequency noise, the time step can be increased to cover longer-term variations, and the number of hidden layer neurons can be adjusted to improve the ability to model the synergistic effects of multiple factors. Furthermore, for different long-term stable operating condition subtypes, such as high-temperature and dry conditions and normal temperature and humidity conditions, the weighting of environmental factors should be differentiated. For example, the weight of the temperature parameter should be increased in high-temperature and dry conditions, while the weight of the humidity parameter should be appropriately increased in normal temperature and humidity conditions. This ensures that the leakage current prediction model accurately adapts to the degradation driving mechanisms of specific operating conditions.

[0054] After completing the corresponding parameter settings, the multi-dimensional data on leakage current and its influencing factors must be integrated to construct an input data sample. These influencing factors include environmental data such as daily average temperature and average humidity, as well as stress data such as daily average load and load fluctuation. This data is time-aligned with the leakage current data for the corresponding time period to construct an input data sample containing multiple features. The leakage current prediction model then performs predictions based on the input data sample, outputting a continuous long-term prediction sequence.

[0055] This multi-factor fusion of input samples and model parameter settings adapted to long-term operating conditions can enable the long-term prediction sequence to accurately reflect the slowly changing trend of leakage current. Compared with the prediction model driven by a single factor, it can effectively reduce the corresponding prediction error.

[0056] After outputting a continuous long-term prediction sequence, the long-term prediction sequence and the collected environmental data and stress data are divided according to the preset time granularity, such as weeks. For each time period, the trend characteristics of the long-term prediction sequence are calculated, such as the weekly average growth rate of the leakage current, the difference between the weekly maximum and minimum values, the slope of the trend line, etc. For the environmental data, the average temperature change rate and the average humidity fluctuation range of each time period are calculated. For the stress data, the average load fluctuation amplitude and the load peak occurrence frequency of each time period are calculated. Subsequently, corresponding trend category labels are added to these trend characteristics. For example, the leakage current trend characteristics can be marked as slowly rising, basically stable, slightly declining, etc. The environmental data trend characteristics can be marked as slowly rising temperature, stable humidity, etc. The stress data trend characteristics can be marked as gentle load fluctuation, slight increase in load mean, etc.

[0057] Based on the trend characteristics of each time period and the corresponding trend category labels, correlation analysis is used to accurately determine the cause of the leakage current trend. For example, if the leakage current trend characteristic of a certain period is a slow increase, the corresponding environmental data trend characteristic is a slow increase in humidity, and the stress data trend characteristic is a gentle load fluctuation, combined with the preset cause analysis rules, such as the long-term slow increase in humidity that easily leads to moisture penetration, causing a slow increase in leakage current, it can be determined that the main cause of the leakage current increase in this period is the long-term cumulative effect of humidity. Alternatively, if the leakage current is slowly increasing and the stress data trend characteristic is a slight increase in the load mean, while the environmental data trend is stable, it can be determined that the main cause is the accumulation of material fatigue under the long-term action of mechanical stress.

[0058] Through detailed trend characteristics and cause analysis, it can also provide sufficient feature support for the smooth transition between short-term prediction strategies and long-term prediction strategies, ensuring the coherence and consistency between different prediction strategies, and further improving the overall reliability of insulation performance testing of insulating pull rod equipment.

[0059] During the actual operation of insulated pull rod equipment, operating conditions are not static but dynamically shift between short-term extreme conditions and long-term stable conditions. For example, a sudden rainstorm may cause the equipment to instantly switch from long-term stable normal temperature and humidity conditions to short-term extreme high humidity shock conditions. Once the rainstorm ends, the equipment will gradually return to its long-term stable state. Alternatively, during long-term stable operation, the changing seasons may cause the environmental parameters to slowly change, causing the operating conditions to gradually transition from high temperature and dry conditions to normal temperature and high humidity.

[0060] While the aforementioned short-term recursive prediction strategy and long-term multi-input / output prediction strategy can respectively adapt to the prediction needs under short-term extreme operating conditions and long-term stable operating conditions, if a single strategy is still used during the dynamic transition between operating conditions, the compatibility between the strategy and the operating conditions will decrease, leading to reduced prediction accuracy. For example, when the equipment transitions from short-term extreme operating conditions to long-term stable operating conditions, the short-term recursive prediction strategy, due to its excessive focus on high-frequency fluctuations, has difficulty capturing the long-term trend after the operating conditions stabilize. Meanwhile, the long-term multi-input / output prediction strategy, due to its slow response to transient changes, cannot promptly track leakage current fluctuations during the transition phase.

[0061] Therefore, in order to ensure that the accuracy and continuity of the prediction can be maintained when the working conditions change dynamically, based on the leakage current change trend prediction results and real-time operating conditions, the prediction adjustment needs are identified in real time, and then it is determined whether the prediction strategy needs to be switched to make up for the limitations of a single prediction strategy during working condition conversion, ensuring that the prediction results always match the actual operating status of the equipment, and providing reliable support for the insulation performance evaluation and safety warning of insulating pull rod equipment.

[0062] The step of identifying and predicting adjustment requirements in real time based on the leakage current change trend prediction results and operating conditions includes: Identify the current operating conditions and their duration, and obtain the matching degree of the forecast strategy based on the current forecast demand; Obtain the working condition characteristics of the current operating condition, and obtain the working condition adaptability of the prediction strategy based on the leakage current change trend prediction results; According to the actual measured value of the currently collected leakage current, a trend consistency index of the leakage current change trend prediction result is obtained; Identify and predict adjustment needs based on matching, working condition suitability and trend consistency index.

[0063] Based on the need for forecast strategy adaptability during dynamic operating condition transitions, the applicability of the current forecast strategy is quantitatively assessed using multi-dimensional indicators to accurately identify forecast adjustment needs. Specifically, the forecast strategy's matching degree, operating condition adaptability, and trend consistency index are used as applicability evaluation indicators.

[0064] The matching degree calculation of the prediction strategy requires a comprehensive assessment of the current operating conditions, their duration, and the current forecast demand. First, the type and duration of the current operating condition are determined. Then, the degree of adaptation is calculated using the preset matching scoring rules, comparing it to the current forecast demand.

[0065] The matching score rule may be: matching score = working condition type adaptation score × duration correction coefficient, where the working condition type adaptation is divided into short-term extreme working conditions and short-term recursive prediction strategy, and long-term stable working conditions and long-term multi-input and output prediction strategy adaptation. The basic score is 90, otherwise, the basic score is 40. The duration correction parameter is set according to the degree to which the working condition duration deviates from the typical adaptation duration. The greater the deviation, the smaller the duration correction coefficient.

[0066] For the calculation of the working condition adaptability of the prediction strategy, it is necessary to focus on the degree of fit between the working condition characteristics of the current operating condition and the leakage current change trend prediction results. First, extract the key characteristics of the current working condition, such as the humidity surge amplitude and load impact frequency under short-term extreme working conditions, and the temperature fluctuation range and load mean change rate under long-term stable working conditions. Then analyze whether the leakage current change trend prediction results can accurately reflect the impact of these characteristics. For example, in the short-term extreme working condition of humidity surge, if the prediction results show that the leakage current shows a slow growth trend, which is inconsistent with the working condition characteristics that humidity surge should cause the leakage current to rise rapidly, then the working condition adaptability is low. If the growth trend of the prediction results is positively correlated with the humidity surge amplitude, then the working condition adaptability is high.

[0067] To calculate the trend consistency index of the prediction strategy, the currently collected leakage current measured value is used as a benchmark to measure the degree of consistency between the prediction result and the actual change. First, by comparing the predicted leakage current change trend curve with the measured curve, the deviation rate of the two at the same time node is calculated. Then, combined with the deviation rate mean, maximum deviation value and other indicators of the entire monitoring period, the trend consistency index is obtained according to the consistency scoring formula. The consistency scoring formula can be: consistency score = 1-deviation rate mean.

[0068] Based on the evaluation results of the above three indicators, preset decision rules are used to identify forecast adjustment needs. For example, if the matching degree is less than 60 points, the working condition adaptability is low, and the trend consistency index is less than 75%, it is determined to be an urgent adjustment need, and the forecast strategy must be switched immediately. If two of these indicators are not up to standard, it is determined to be a routine adjustment need, and the strategy must be switched within a certain period of time. If only one indicator is not up to standard, it is determined to be an observation adjustment need, and subsequent changes in working conditions and forecast results will be continuously monitored.

[0069] Through the above-mentioned multi-dimensional quantitative evaluation, accurate identification of forecast adjustment needs is achieved, avoiding the blindness of strategy switching. Through clear indicator thresholds and decision-making rules, errors in human judgment are reduced, and the accuracy of identifying adjustment needs is improved, laying a solid foundation for the smooth transition switching of subsequent forecast strategies, ensuring that the leakage current change trend prediction is always highly consistent with the equipment operating status.

[0070] After identifying the need for prediction adjustment through matching degree, operating condition adaptability and trend consistency index, it means that the current prediction strategy can no longer accurately adapt to the operating conditions of the equipment. If it is not adjusted in time, it may cause deviations in the prediction results of the leakage current change trend, affecting the accurate judgment of the insulation performance of the insulating pull rod equipment.

[0071] To ensure the continuity and accuracy of the prediction, a transitional switch in the prediction strategy is necessary based on the identified prediction adjustment needs. This transitional switch is not a simple strategy replacement, but rather a smooth transition mechanism between the two prediction strategies to address dynamic changes in operating conditions. Through a reasonable transitional switch, sudden changes in prediction errors caused by step-like parameter changes are eliminated, ensuring the continuity and consistency of the leakage current trend prediction results. This in turn updates the leakage current trend prediction results, ensuring that they always accurately reflect the insulation performance status of the equipment, providing reliable data support for safety assessment and early warning of insulated pull rod equipment.

[0072] Specifically, the prediction strategy transition switching according to the prediction adjustment demand and updating the leakage current change trend prediction result include: Obtain the working condition fluctuation coefficient based on environmental data and stress data, and set the corresponding transition window according to the working condition fluctuation coefficient; In the transition window, the prediction values ​​of the current prediction strategy and the switched target prediction strategy are output synchronously, and the corresponding prediction fusion value is obtained by combining the dynamic weights; When the predicted fusion value and the dynamic weight meet the transition end condition, the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy.

[0073] After identifying the forecast adjustment needs, in order to achieve a smooth transition switch of the forecast strategy and update the leakage current change trend forecast results, it is necessary to control the transition process through the operating condition fluctuation coefficient.

[0074] The operating condition fluctuation coefficient is derived based on the dynamic changes in environmental and stress data, and is used to quantitatively assess the severity of operating condition changes. The calculation uses the instantaneous rate of change of temperature and humidity, and the standard deviation of load fluctuations, as core parameters. After normalization, the operating condition fluctuation coefficient is calculated using a weighted summation formula. A higher coefficient indicates a more drastic operating condition change. The duration of the transition window is then determined based on the magnitude of the operating condition fluctuation coefficient. A larger operating condition fluctuation coefficient indicates more drastic changes in environmental and stress parameters. In this case, the adaptability gap between the two prediction strategies will increase dramatically. For example, the short-term recursive strategy can capture transient fluctuations but struggles to adapt to the upcoming stable trend. The target strategy, with its long-term multi-input and output strategy, adapts to new operating conditions but requires a longer time to learn the patterns following drastic changes. Therefore, a longer transition window is necessary. By slowly adjusting the dynamic weights, the predictions of the two strategies converge, gradually offsetting the prediction bias caused by drastic fluctuations. Therefore, a larger fluctuation coefficient increases the duration of the transition window, ensuring that the transition process matches the speed of operating condition changes.

[0075] Within the transition window, a dual-strategy parallel output and dynamic weight fusion mechanism is adopted. Specifically, the current prediction strategy and the target prediction strategy are run synchronously to generate leakage current prediction values ​​at corresponding time points. The allocation of dynamic weights is based on the real-time changes in the operating condition fluctuation coefficient. At the initial moment, the current prediction strategy weight is 1 and the target strategy weight is 0. As the window progresses, the current prediction strategy weight decreases in a linear attenuation manner, and the target strategy weight increases in a linear increasing manner. The attenuation or increasing rate is determined by the window duration. The prediction fusion value is the sum of the product of the prediction values ​​of the two strategies and the corresponding weights. It not only retains the adaptability of the current strategy to the previous operating conditions, but also gradually introduces the target strategy's ability to fit the new operating conditions, avoiding prediction mutations caused by switching a single strategy.

[0076] The transition end condition must meet the dual requirements of the predicted fusion value and the dynamic weight. Specifically, when the fluctuation amplitude of the fusion value at three consecutive time points is less than the preset threshold and the target strategy weight reaches 1 and remains stable, the transition end condition is determined to be met, the transition window is terminated, and the target prediction strategy is officially switched to predict the leakage current change trend. If the conditions are not met within the window duration, the window is automatically extended to ensure the transition process is fully completed.

[0077] This approach, which dynamically adjusts the transition window based on operating condition fluctuations, precisely matches the strategy switching rhythm with the speed of operating condition changes. During drastic operating condition changes, the window is extended to smooth the transition, while during gentle changes, the window is shortened to improve response efficiency. Furthermore, a dynamic weight fusion mechanism effectively eliminates prediction gaps during strategy switching, achieving a seamless transition between short-term recursive and long-term multi-input and output strategies, ensuring both the continuity and accuracy of leakage current trend predictions.

[0078] In addition, although the transition process has achieved smooth prediction connection through the fusion mechanism after terminating the transition window and switching to the target prediction strategy, the target strategy may ignore the key details captured by the current strategy in the old working conditions due to its focus on the trend characteristics of the new working conditions, such as the potential impact of instantaneous fluctuation characteristics under short-term extreme working conditions on long-term trends. In addition, the short-term strategy focuses on high-frequency fluctuation characteristics, while the long-term strategy focuses on low-frequency trend characteristics. There are differences in the trend feature dimensions of the two strategies. Direct switching may lead to incomplete feature transfer, affecting the target strategy's comprehensive judgment of the leakage current change trend. Therefore, it is necessary to establish feature correspondences and supplement key features to ensure that the target strategy can integrate the feature information of the old and new working conditions to improve the accuracy and completeness of the prediction results.

[0079] Specifically, when the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy, the following steps are also performed: Obtain a transition switching target based on the demand type of the predicted adjustment demand, and select a transition switching direction of the prediction strategy based on the transition switching target; Based on the transition switching direction of the forecast strategy, a corresponding relationship between the trend characteristics of the current forecast strategy and the target strategy is established, and a list of trend characteristics of the target strategy is obtained; Extract the supplementary features of the current prediction strategy and add the supplementary features to the trend feature list of the target strategy; Based on the trend features in the trend feature list, the leakage current change trend prediction result is updated.

[0080] First, the transition target is determined based on the demand type for forecast adjustment, which then determines the transition direction for the forecast strategy. For example, when the demand type is switching from short-term extreme conditions to long-term stable conditions, the transition target is the long-term multi-input and output forecast strategy, and the transition direction is from short-term to long-term. When the demand type is switching from long-term stable conditions to short-term extreme conditions, the transition target is the short-term recursive forecast strategy, and the transition direction is from long-term to short-term. Clarifying the transition direction provides guidance for establishing subsequent feature mapping relationships.

[0081] Based on the switching direction, determine the trend feature dimensions that the current prediction strategy and the target strategy focus on to establish a feature correspondence. Taking the short-term to long-term switching as an example, the trend features of the short-term prediction strategy include the instantaneous growth rate of leakage current, high-frequency fluctuation peaks, etc., and the trend features of the long-term prediction strategy include the weekly average growth rate, low-frequency trend slope, etc. Through the corresponding relationship, it is clear that the instantaneous growth rate corresponds to the initial growth momentum in the long-term trend, and the high-frequency peak corresponds to the fluctuation risk point in the long-term trend. Subsequently, the trend features that have been identified by the target strategy under the new working conditions are extracted, such as the temperature cumulative impact characteristics and load mean change characteristics in the long-term stable working conditions, to form the initial trend feature list of the target strategy. This trend feature list records the characteristic parameters of historical data.

[0082] Supplementary features that complement the target strategy are then extracted from the historical prediction results of the current prediction strategy. If the current strategy is short-term recursive, these supplementary features may include the maximum leakage current value under short-term extreme operating conditions and the duration of the leakage current surge. These features can reflect the degree of material damage under extreme conditions and affect the long-term degradation rate. If the current strategy is long-term multi-input and output, these supplementary features may include the periodic small fluctuations hidden in the long-term trend, providing a basis for judging transient anomalies in the short-term strategy.

[0083] These supplementary features are mapped to the feature dimensions of the target strategy according to the corresponding relationship, and added to the trend feature list of the target strategy, so that the trend feature list covers the key features of both new and old working conditions.

[0084] The target strategy then uses the supplemented trend feature list as input to retrain or adjust the leakage current prediction model parameters. For example, after incorporating the long-term supplementary features, the short-term strategy adjusts the threshold for transient anomalies based on a cyclical fluctuation benchmark to avoid misinterpreting normal fluctuations in the long-term trend as sudden anomalies. Based on this, the adjusted leakage current prediction model outputs an updated prediction result.

[0085] By clarifying the correspondence between switching direction and features as described above, the targeted and accurate transmission of trend features is ensured, and feature loss or mismatch is avoided. The addition of supplementary features can also make the trend feature list of the target strategy more comprehensive, improve the adaptability of the leakage current prediction model to complex working conditions, ensure the continuity and integrity of the changing trend during the switching process, and effectively eliminate the prediction gap caused by parameter mutations.

[0086] Based on the prediction results of the leakage current change trend of the insulating pull rod equipment, in order to quantitatively evaluate the insulation status, an insulation integrity scoring mechanism is further introduced to quantitatively express the leakage current trend and insulation performance degradation risk in real time.

[0087] Specifically, the integrity scoring of the live working insulation performance is performed based on the current leakage current change trend prediction result and the corresponding leakage current prediction strategy, including: Obtain the safe upper limit of leakage current of the insulating rod equipment in the current live working scenario, and obtain the corresponding prediction time window based on the current leakage current prediction strategy; According to the current leakage current change trend prediction result, the leakage current prediction value at each moment in the corresponding prediction time window is obtained; Based on the safety upper limit of leakage current and the predicted leakage current value at each moment in the prediction time window, the integrity score of the insulation performance of live working is obtained.

[0088] The scoring formula for the above completeness score is: ; in, is the set safety upper limit of leakage current; is the leakage current prediction value at the i-th moment; H is the prediction time window length; S is the score value.

[0089] In the formula It is used to measure the degree of deviation of the predicted value at time i relative to the upper limit value. The lower the score S, the higher the leakage risk and the worse the integrity.

[0090] This integrity scoring is performed in real time during the prediction process, and when it is identified that the score value S is lower than the set safety threshold, a corresponding insulating pull rod equipment safety warning will be issued.

[0091] Example 2: The embodiment of the present invention provides a live working insulation performance prediction system integrating multiple information, such as Figure 2 As shown, the method that can be used in real-time embodiment 1 includes: The data sensing and acquisition module 1 is provided at the insulating rod device 6 and is used to collect environmental data, stress data and leakage current of the insulating rod device during live operation; Data analysis module 2, connected to the data perception and acquisition module, is used to identify operating conditions and determine forecast requirements based on the collected environmental data, stress data, and leakage current; The decision prediction module 3 is connected to the data perception and acquisition module and the data analysis module respectively, and is used to select a prediction strategy according to the prediction requirements and predict the leakage current change trend according to the leakage current; The monitoring feedback module 4 is connected to the data analysis module and the decision prediction module respectively, and is used to identify the prediction adjustment needs based on the leakage current change trend prediction results and the operating conditions, and control the decision prediction module to adjust the prediction strategy applied in the transition according to the prediction adjustment needs; The alarm module 5 is connected to the decision prediction module and is used to score the integrity of the insulation performance of live working according to the leakage current change trend and prediction strategy, and to issue a safety warning for the insulation rod equipment based on the integrity score.

[0092] The data sensing and acquisition module includes: The temperature and humidity sensor 11 is provided on the outer wall of the middle portion of the insulating rod device and is used to collect environmental data of the insulating rod device during live operation; The stress sensor 12 is provided on the middle surface of the axis of the insulating pull rod equipment and the metal end, and is used to collect stress data of the insulating pull rod equipment during live operation; The leakage current sensor 13 is provided at the end of the pull rod of the insulating pull rod device and is used to collect the leakage current of the insulating pull rod device during live operation.

[0093] As a key mechanical support and insulating element in high-voltage electrical equipment, the surface of the insulating rod equipment is easily affected by pollutants such as dust, moisture, and salt spray during long-term operation, forming a conductive channel along the surface, which in turn causes leakage current and even causes flashover failure. Before these phenomena occur, the leakage current often first flows along the rod end area of ​​the insulating rod equipment, which is close to the conductive connection end, has a significant electric field edge effect, and has high local electrical stress. At this time, the rod end of the insulating rod equipment becomes an important channel for the initial or convergence of leakage current. From the perspective of structure and maintenance, the space in the rod end area is relatively open, which is also convenient for the installation and subsequent maintenance of sensors, and will not cause mechanical damage to the insulating structure body, which helps to ensure the long-term safe operation of the equipment. Therefore, this embodiment specifically arranges leakage current sensors at the rod ends of the insulating rod equipment, which can achieve early detection and continuous monitoring of leakage current.

[0094] The temperature and humidity sensors specifically include a temperature sensor 111 and a humidity sensor 112. The central area of ​​the insulating rod device body is a sensitive area with a relatively stable thermal field and thermal coupling, which can represent the overall temperature change trend of the rod. This location avoids high-voltage conductor terminals and metal connectors, helps to reduce electric field interference, improve temperature measurement accuracy, and at the same time ensure electrical isolation between the sensor and the device. Therefore, in this embodiment, the temperature sensor is installed on the central outer wall of the insulating rod device. Through this arrangement, its thermal aging trend can be effectively monitored, providing key input for subsequent leakage current prediction.

[0095] For the humidity sensor, its monitoring point is set on the middle outer wall close to the ambient temperature monitoring point. This place has good local humidity response capability and can form a thermal and humidity coupling monitoring point with the temperature sensor, which is convenient for analyzing the impact of humidity and temperature on leakage current under the synergistic effect, and improving the accuracy of predicting the risk of leakage current mutation.

[0096] The stress sensor mainly includes a tensile load sensor 121 for collecting the tensile stress of the insulating pull rod equipment and a shear load sensor 122 for collecting the shear stress of the insulating pull rod equipment. Among them, the axial direction area of ​​the middle surface of the insulating pull rod equipment is the main stress concentration area under the action of the tensile load, which can represent the stress state of the pull rod of the entire insulating pull rod equipment during operation. Therefore, the measuring point of the stress sensor for monitoring the tensile stress is set on the axial middle surface of the insulating pull rod equipment. In addition, the cylindrical surface area of ​​the insulating pull rod equipment close to the metal connection end is a significant area of ​​stress concentration and interface coupling effect, and is a precursor site for initial damage such as debonding and slippage. Therefore, the measuring point of the stress sensor for monitoring the shear stress is further set at the metal end of the insulating pull rod equipment, specifically, it can be selected within the cylindrical surface area of ​​the metal end.

[0097] The installation positions of the sensors in the data sensing and acquisition module on the insulating rod equipment are as follows: Figure 3 shown.

[0098] In this embodiment, the current mutual inductance sensor with a ring-shaped closed structure is a leakage current sensor, and is fixedly installed in an embracing manner. During installation, the surface of the rod end of the insulating pull rod device must be thoroughly cleaned to remove oil, moisture and conductive pollutants, ensuring that the leakage current sensor fits tightly with the surface of the insulator to avoid the formation of new discharge hazards. Secondly, the coil is wrapped around the position of the rod end of the insulating pull rod device closest to the conductive connector to form a complete closed induction loop, and the leakage current sensor is reinforced with a heat shrink fixing ring to ensure that it does not loosen or shift during long-term operation. Subsequently, a double-layer heat shrink tubing is used to encapsulate and protect the coil body to improve its moisture-proof, dust-proof and UV-resistant properties. The signal lead wire should use a shielded twisted pair cable, which is led out from the end to the non-electric field area and reasonably wired to avoid crossing with the high-voltage part. Finally, it is connected to the data analysis module to transmit the collected data.

[0099] For the temperature and humidity sensor, first select a flat area 3-5 cm from the center outer wall of the insulating rod. Thoroughly clean this area with anhydrous ethanol to remove any oxide layer and ensure a clean contact surface for the sensor. To ensure a good fit, apply a layer of thermally conductive adhesive to the bottom of the temperature sensor to improve the response speed and accuracy of temperature acquisition, ensuring it does not conduct electricity and damage the insulation. Next, secure the temperature and humidity sensor to the center outer wall of the insulating rod using a high-grade, high-temperature-resistant polyimide tape. This prevents mechanical loosening or movement due to thermal expansion and contraction. For humidity measurement, since the insulating rod surface is not easily hygroscopic, a hydrophilic film is applied to the humidity sensor to enhance moisture sensing. 3.5 mm ventilation slots are provided on both sides to prevent moisture accumulation and strong wind interference. When laying the leads of the temperature and humidity sensor, use double-shielded wire for the signal line and keep it away from high-voltage cables. The terminal is silver-plated and tin-enameled, then sealed with three layers of heat shrink tubing, and ferrite rings are installed to suppress high-frequency interference.

[0100] For the stress sensor, a strain gauge is used as the tensile load sensor. It is installed on a cylindrical surface within ±5 cm of the midpoint of the connecting piece at each end of the insulated tie rod to collect tensile stress data. Before installation, the target surface should be carefully sanded with sandpaper until the insulating substrate is evenly and smoothly exposed, and then degreased with anhydrous ethanol. The tensile load sensor is arranged unidirectionally along the axis of the insulated tie rod. The strain gauge is glued with a high-strength insulating modified epoxy resin adhesive and cured under constant pressure for at least 24 hours to ensure a tight fit between the strain gauge and the substrate. For the tensile load sensor leads, double-shielded signal cables are used and kept away from high-voltage cables. The terminals are silver-plated, tin-plated, and sealed with three layers of heat shrink tubing. Ferrite rings are also installed to suppress high-frequency interference. After installation, zero-point calibration and static load testing should be performed to verify the linearity and repeatability of the tensile load sensor response, ensuring stable and reliable axial tensile stress detection capabilities.

[0101] A shear stress sensor is used as the shear load sensor and is installed approximately 20 to 30 cm from the metal connection of the insulated tie rod, near the cylindrical surface, to collect shear stress data. Before installation, the target surface must be carefully sanded with sandpaper until the insulating substrate is exposed, followed by degreasing with anhydrous ethanol. The shear load sensors are installed in pairs, affixed to symmetrical positions on either side of the insulated tie rod at a ±45° angle, forming a strain gauge structure to accurately extract shear stress. High-strength insulating adhesive, modified epoxy resin, is used for adhesion and cured under constant pressure for at least 24 hours to ensure a tight bond between the strain gauge and the substrate. For shear load sensor wiring, double-shielded signal cables are used and kept away from high-voltage cables. The terminals are silver-plated, tin-coated, and sealed with three layers of heat shrink tubing. Ferrite rings are also installed to suppress high-frequency interference. After installation, zero point calibration and static loading test should be carried out to verify the response linearity and repeatability of the shear load sensor and ensure that it has stable and reliable shear stress detection capabilities.

[0102] The data analysis module, decision prediction module and monitoring feedback module can use high-performance calculators, servers, microcontrollers and other data analysis and processing devices, and carry relevant algorithms for corresponding operating condition identification, forecast demand matching, forecast strategy selection, leakage current change trend prediction, forecast adjustment demand identification and forecast strategy transition switching. It can realize efficient and accurate prediction of leakage current change trend through the collected environmental data, stress data and leakage current.

[0103] The alarm module can also utilize high-performance data analysis and processing devices such as calculators, servers, and microprocessors. It incorporates a corresponding integrity scoring formula, which determines the integrity score of the insulated rod equipment based on the leakage current trend prediction results obtained by the decision-making and prediction module and the specific prediction strategy selected. The alarm module also incorporates a corresponding safety threshold, which can trigger a dynamic response based on the safety threshold. When the safety threshold is exceeded, the relevant integrity score and the data applied to it can be directly connected to the operation and maintenance monitoring backend, visually displaying the integrity score of the insulated rod equipment through the relevant software interface. When the safety threshold is lower than the safety threshold, the corresponding graded alarm is activated, triggering a yellow or red alert based on the score difference below the safety threshold. If the score difference below the safety threshold is low, a yellow alert is triggered. The yellow alert generates a maintenance work order with GIS positioning and pushes it to the operation and maintenance personnel's mobile terminal. This data is also fed back to the decision-making and prediction module to initiate incremental training of the relevant leakage current prediction model to optimize prediction accuracy. If the score difference below the safety threshold is high, the corresponding hard contact interlock tripping of the insulating pull rod equipment will be directly executed to cut off the circuit, and the sound and light alarm will be triggered synchronously, and the relevant time axis will be recorded.

[0104] Embodiment 3: The embodiment of the present invention provides a device for predicting insulation performance of live working by integrating multiple information, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of any one of the methods in the first embodiment.

[0105] Embodiment 4: The embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of any one of the methods in Embodiment 1 are implemented.

[0106] The embodiment described above is only a preferred solution of the present invention and does not limit the present invention in any form. Other variations and modifications are possible without exceeding the technical solution described in the claims.

Claims

1. A method for predicting insulation performance of live working by integrating multivariate information, characterized in that: include: Collect environmental data, stress data and leakage current of insulating rod equipment during live operation; Based on environmental data and stress data, identify the current operating conditions and determine the forecast demand based on the operating conditions; Based on the leakage current, a short-term recursive prediction strategy or a long-term multi-dimensional input-output prediction strategy is selected according to the prediction requirements to predict the leakage current change trend; According to the leakage current change trend prediction results and operating conditions, the prediction adjustment needs are identified in real time, the prediction strategy transition is switched according to the prediction adjustment needs, and the leakage current change trend prediction results are updated; Based on the current leakage current change trend prediction results and the corresponding leakage current prediction strategy, the insulation performance integrity score of live working is evaluated; Provide safety warning for insulating pull rod equipment based on integrity scoring results.

2. The method for predicting insulation performance of live working by integrating multivariate information according to claim 1 is characterized in that: The process of identifying the current operating condition based on the environmental data and stress data and determining the forecast demand according to the operating condition includes: Preprocess the environmental data and stress data, and extract working condition features from the preprocessed environmental data and stress data; Based on the working condition characteristics, the current operating condition is identified in combination with the corresponding working condition classification rules; Obtain the duration and type of the current operating condition to match the corresponding forecast demand.

3. The method for predicting insulation performance of live working by integrating multivariate information according to claim 2, characterized in that: The method of predicting the leakage current change trend by selecting a short-term recursive prediction strategy or a long-term multi-input and output prediction strategy based on the leakage current according to the prediction requirements includes: When the working condition type is a short-term extreme working condition and the duration is lower than the preset time threshold, the predicted demand is a short-term high-frequency predicted demand; Select a short-term recursive prediction strategy based on the short-term high-frequency prediction requirements, and set the leakage current prediction model parameters and recursive parameters according to the corresponding operating conditions; Using the collected leakage current as the initial input, the leakage current value at a future time point is predicted by the leakage current prediction model, and the leakage current prediction value is fed back to the leakage current prediction model as a new input; The prediction of leakage current value and new input feedback are repeatedly performed to generate a continuous short-term prediction sequence; Slide through the recursively generated short-term forecast demand with a fixed window size and calculate the trend characteristics within the window; Based on trend characteristics and preset rules, the leakage current change trend prediction results are obtained.

4. The method for predicting insulation performance of live working by integrating multivariate information according to claim 2, characterized in that: The method of predicting the leakage current change trend by selecting a short-term recursive prediction strategy or a long-term multi-input and output prediction strategy based on the leakage current according to the prediction requirements may further include: When the working condition type is a long-term stable working condition and the duration is not less than the preset time threshold, the predicted demand is a long-term trend predicted demand; When selecting a long-term multi-dimensional input-output prediction strategy based on long-term trend prediction requirements, set the leakage current prediction model parameters according to the corresponding operating conditions; Construct input data samples based on leakage current and its influencing factors, and output a continuous long-term prediction sequence in combination with the leakage current prediction model; Divide the long-term prediction series and the collected environmental data and stress data according to the time granularity, calculate the trend characteristics of the long-term prediction series in each time period, as well as the trend characteristics of the environmental data and stress data in each time period, and add corresponding trend category labels to them respectively; Based on the trend characteristics and corresponding trend category labels, the causes of the change trend are identified and the leakage current change trend prediction results are obtained.

5. The method for predicting insulation performance of live working by integrating multivariate information according to claim 1, characterized in that: The real-time identification and prediction of adjustment requirements based on the leakage current change trend prediction results and operating conditions includes: Identify the current operating conditions and their duration, and obtain the matching degree of the forecast strategy based on the current forecast demand; Obtain the working condition characteristics of the current operating condition, and obtain the working condition adaptability of the prediction strategy based on the leakage current change trend prediction results; According to the actual measured value of the currently collected leakage current, a trend consistency index of the leakage current change trend prediction result is obtained; Identify and predict adjustment needs based on matching, working condition suitability and trend consistency index.

6. The method for predicting insulation performance of live working by integrating multivariate information according to claim 1, characterized in that: The prediction strategy transition switching according to the prediction adjustment requirements and updating the leakage current change trend prediction result include: Obtain the working condition fluctuation coefficient based on environmental data and stress data, and set the corresponding transition window according to the working condition fluctuation coefficient; In the transition window, the prediction values ​​of the current prediction strategy and the switched target prediction strategy are output synchronously, and the corresponding prediction fusion value is obtained by combining the dynamic weights; When the predicted fusion value and the dynamic weight meet the transition end condition, the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy.

7. The method for predicting insulation performance of live working by integrating multivariate information according to claim 6, characterized in that: When the transition window is terminated and the leakage current change trend is predicted according to the target prediction strategy, the following steps are also performed: Obtain a transition switching target based on the demand type of the predicted adjustment demand, and select a transition switching direction of the prediction strategy based on the transition switching target; Based on the transition switching direction of the forecast strategy, a corresponding relationship between the trend characteristics of the current forecast strategy and the target strategy is established, and a list of trend characteristics of the target strategy is obtained; Extract the supplementary features of the current prediction strategy and add the supplementary features to the trend feature list of the target strategy; Based on the trend features in the trend feature list, the leakage current change trend prediction result is updated.

8. The method for predicting insulation performance of live working by integrating multivariate information according to claim 1, characterized in that: The integrity scoring of the insulation performance of live working is performed based on the current leakage current change trend prediction result and the corresponding leakage current prediction strategy, including: Obtain the safe upper limit of leakage current of the insulating rod equipment in the current live working scenario, and obtain the corresponding prediction time window based on the current leakage current prediction strategy; According to the current leakage current change trend prediction result, the leakage current prediction value at each moment in the corresponding prediction time window is obtained; Based on the safety upper limit of leakage current and the predicted leakage current value at each moment in the prediction time window, the integrity score of the insulation performance of live working is obtained.

9. A live working insulation performance prediction system integrating multivariate information, used to execute the prediction method according to any one of claims 1 to 8, characterized in that: include: The data sensing and collection module is installed at the insulating rod equipment and is used to collect environmental data, stress data and leakage current of the insulating rod equipment during live operation; The data analysis module is connected to the data perception and acquisition module and is used to identify operating conditions and determine forecast requirements based on the collected environmental data, stress data, and leakage current; The decision-making prediction module is connected to the data perception and acquisition module and the data analysis module respectively, and is used to select a prediction strategy according to the prediction requirements and predict the leakage current change trend based on the leakage current; The monitoring feedback module is connected to the data analysis module and the decision prediction module respectively, and is used to identify the prediction adjustment needs based on the leakage current change trend prediction results and operating conditions, and control the decision prediction module to transitionally adjust the prediction strategy applied according to the prediction adjustment needs; The alarm module is connected to the decision-making prediction module and is used to score the integrity of the insulation performance of live working according to the leakage current change trend and prediction strategy, and to issue safety warnings for the insulating rod equipment based on the integrity score.

10. The live working insulation performance prediction system integrating multi-element information according to claim 9 is characterized in that: The data perception and acquisition module includes: The temperature and humidity sensor is installed on the middle outer wall of the insulating rod equipment and is used to collect environmental data of the insulating rod equipment during live operation; Stress sensors are installed on the middle surface of the axis of the insulating pull rod equipment and the metal end, and are used to collect stress data of the insulating pull rod equipment during live operation; The leakage current sensor is installed at the end of the insulating pull rod equipment and is used to collect the leakage current of the insulating pull rod equipment during live operation.

11. A device for predicting insulation performance of live working by integrating multiple information, characterized in that: including processors and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method for predicting insulation performance of live working according to any one of claims 1 to 8.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting insulation performance of live working according to any one of claims 1 to 8 are implemented.

Images