A switchgear internal temperature mapping transient inversion method

Abstract

The application provides a switch cabinet internal temperature mapping transient inversion method, comprising the following steps: step one, obtaining switch cabinet internal and external temperature data streams to obtain a model training group and a model verification group; step two, setting temperature calculation tracing time periods, time period division segment numbers, initial weights and pre-designed calculation accuracies for each data stream in the model training group and the model verification group formed in step one to obtain training samples; step three, inputting the four groups of variables in the training samples obtained in step two into a regression algorithm for training to obtain a convergence formula; step four, when the algorithm is completely converged and the accuracy meets the requirements, using the obtained convergence formula to test the non-training sample accuracy interval, if the requirements are met, directly giving a temperature full value formula at the current time, and if the requirements are not met, performing feedback adjustment and resetting the length of the tracing time period and the segment number. The application can eliminate the error and time lag of the current time inversion technology and realizes real-time temperature inversion.

Description

Technical Field

[0001] This invention relates to the field of temperature inversion inside switchgear, specifically a transient inversion method for temperature mapping inside switchgear. Background Technology

[0002] Overheating is a significant factor contributing to switchgear failures. Prolonged overheating can lead to the burning and contamination of contacts and nearby insulation materials, ultimately resulting in insulation faults such as internal short circuits and arcing. This seriously threatens the safety of maintenance personnel and may cause localized or widespread power outages. Therefore, understanding the heating characteristics of switchgear and its influencing factors, and implementing heating status monitoring, is crucial for intelligent switchgear operation and maintenance, reducing overheating failures, ensuring the safety of maintenance personnel, and improving power supply reliability.

[0003] Thermal failures in switchgear are often caused by loose conductive circuits, poor connections, or excessive contact resistance due to surface oxidation. These thermal defects are mostly found in high-voltage, high-current primary circuits. Directly measuring hot spot temperatures has many drawbacks, such as electromagnetic compatibility issues with sensors at high potentials leading to inaccurate measurements, and the potential to damage insulation structures. Therefore, industrial applications often use surface temperature measurement for online monitoring of switchgear thermal failures. However, heat transfer from internal hot spots to the cabinet surface via convection and radiation takes time, which can allow the thermal failure to escalate to a level that threatens equipment safety. Therefore, a technology that can provide real-time feedback on switchgear thermal failures is urgently needed in industry.

[0004] Existing methods for retrieving internal hot spot temperatures using surface temperature measurement technology in switchgear can achieve ground and low potential installations and uninterrupted power supply installations. However, due to the dual time lag of temperature, the surface temperature measurement-based retrieval technology may have limitations.

[0005] Chinese patent document CN110220602A, published on September 10, 2019, discloses a method for identifying overheating faults in switchgear. This method involves using temperature sensors arranged inside and on the surface of the switchgear to acquire temperature data during both fault and normal operation. An electro-magnetic-fluid coupling simulation model of the switchgear is established, and this model is modified based on the collected temperature data. Training samples are obtained by simulating various overheating fault types using the modified electro-magnetic-fluid coupling simulation model. A decision tree is constructed based on the training samples, and the decision tree is used to identify switchgear faults, achieving accurate indirect diagnosis of the internal temperature and potential fault types of the switchgear through the cabinet temperature. However, the decision tree series used does not consider the time lag of temperature rise transfer. Heat transfer from hot spots inside the switchgear to the cabinet surface via convection and radiation takes time, which may cause the thermal fault to escalate to a level that threatens equipment safety. Therefore, the real-time monitoring capability needs further improvement.

[0006] Chinese patent document CN111307306A, published on June 19, 2020, discloses a non-invasive temperature measurement health assessment method, device, and storage medium for use in power equipment. This method involves setting multiple temperature measurement points in each compartment of the equipment to monitor the temperature rise of each part and the load current. The theoretical temperature rise is calculated based on the measured current, and the theoretical and measured temperature rises are compared to determine the health status of each part. However, this patent does not provide the core technology of calculating the temperature rise of the measurement points using the load current, nor does it mention its applicability to transient or steady-state temperature monitoring and defect diagnosis. Therefore, it is impossible to directly commercialize this patent.

[0007] Chinese patent document CN112417686A, published on February 26, 2021, discloses a method for simulating thermal faults by correcting the temperature field distribution of switchgear. However, this method can only correct the temperature field distribution of switchgear in steady-state operation and does not consider the time lag of surface temperature rise relative to load change, which can be as long as hours. Therefore, the calculated results cannot meet the needs of real-time monitoring of hot spot temperature inside switchgear. Moreover, faults caused by switchgear defects often occur under sudden load changes, so this method has limited effectiveness in fault diagnosis. Summary of the Invention

[0008] To overcome the shortcomings of the prior art, this invention designs a transient inversion method for internal temperature mapping of switchgear. It adopts a temperature inversion method based on the weight definition of historical temperature contribution rate. The method eliminates the error and time delay of the inversion technique at the current moment by analyzing the temperature information or current information of the time potential before the current moment to the temperature at the current moment, thereby realizing real-time temperature inversion.

[0009] A method for transient inversion of internal temperature mapping in a switchgear includes the following steps:

[0010] Step 1: Obtain the internal and external temperature data streams of the switch cabinet to obtain the model training group and the model validation group;

[0011] Step 2: For each data stream in the model training group and model validation group formed in Step 1, set four sets of variables: temperature calculation traceability period, number of period segments, initial weights, and preset calculation accuracy to obtain training samples;

[0012] Step 3: Input all four sets of variables from the training samples obtained in Step 2 into the regression algorithm for training. During the training process, adjust the initial weights of these variables appropriately to ensure regression convergence. Simultaneously, change the preset precision to prevent overfitting even when the regression converges, thus avoiding the problem of accurately predicting future data. Once the regression algorithm converges, further adjust the length of the tracking time period or the number of segments to finally obtain the following convergence formula:

[0013] T n =α1T h1 +α2T h2 +α3T h3 +……

[0014] In the formula, T n T represents the current temperature. h1 T h2 ... represents the temperature of each segment set in step two, and α1, α2... represent the temperature contribution weights of each segment obtained through model training;

[0015] Step 4: When the algorithm fully converges and the accuracy meets the requirements, the obtained convergence formula is applied to the experimental samples as validation group data to check whether the same accuracy range is met for the non-training sample data. If it is met, the full temperature value formula at the current moment is given directly. If it is not met, feedback adjustment is performed to reset the length of the traceback time period and the number of segments.

[0016] Furthermore, the regression algorithm used in step three is either the support vector machine regression algorithm or a multiple linear regression model.

[0017] This invention designs a temperature mapping technology suitable for real-time detection of internal thermal faults in high-voltage switchgear using surface temperature measuring points. By collecting past temperature change trends inside and outside the switchgear as training samples, the technology trains to determine the contribution of internal temperature to external temperature at different time periods and measuring points. This results in a switchgear temperature inversion calculation model that considers the time lag in temperature transfer from internal hotspots to the switchgear surface. This model avoids the problem of surface temperature rise information lagging behind internal hotspot temperatures caused by thermal convection, fundamentally improving the timeliness of inverting internal hotspot temperatures through surface temperature rise monitoring. It is suitable for estimating internal temperatures and diagnosing defects in switchgear with large daily load fluctuations. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the transient inversion method for temperature mapping inside the switchgear of the present invention.

[0019] Figure 2 This is a detailed technical roadmap of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The basic idea of ​​this invention is that, since the heat generated by the load current inside the switch cabinet has a large time lag on the surface temperature measurement point of the switch cabinet, the internal temperature of the switch cabinet at different times before the current moment has different contributions to the surface measurement point temperature at the current moment.

[0022] Please see Figure 1 This invention provides a method for transient inversion of internal temperature mapping in a switchgear, the method comprising:

[0023] Step 1: Obtain the internal and external temperature data streams of the switch cabinet to obtain the model training group and the model validation group.

[0024] Specifically, temperature measurement data streams from multiple sensors are acquired during the load change process of the switchgear operation through switchgear temperature rise experiments or on-site measurements in substations where temperature sensors are already installed inside the switchgear. Each data stream includes external temperature data from multiple temperature sensors on the top of the switchgear and internal temperature measurement data from contacts, busbar joints, and cable joints inside the switchgear. The data stream must start from one steady-state temperature and end at another, encompassing the complete temperature change process during that period. These data streams are randomly divided into a model training group and a model validation group; alternatively, all the data can be used as the model training group, followed by on-site measurement data as the model validation group.

[0025] Step 2: Set the temperature calculation traceability period, the number of time period segments, the initial weights, and the preset calculation accuracy.

[0026] For each data stream in the model training group and model validation group formed above, the weights of the forward tracing time period length, the number of time period segments, and the contribution of the internal average or highest temperature in each time period to the temperature of the switch cabinet surface measuring points are set as the initial conditions for calculation, thus obtaining the training samples.

[0027] Step 3: Regression Algorithm Calculation

[0028] The four sets of variables (tracing time period, number of segments, initial weights, and preset precision) from the training samples obtained in step two are all input into the regression algorithm for training. The regression algorithm used can be SVR (Support Vector Machine Regression), multiple linear regression, etc. During the training process, the initial weights are adjusted appropriately to ensure regression convergence, while the preset precision is changed to prevent overfitting even when regression converges, thus avoiding the phenomenon that would hinder accurate prediction of future data.

[0029] After the regression algorithm converges, the final convergence result may differ depending on the specific algorithm used. This could result in a calculation that does not meet accuracy requirements. Based on the algorithm's characteristics, further adjustments can be made to the length of the tracking time period or the number of segments, ultimately leading to the following convergence formula:

[0030] T n =α1T h1 +α2T h2 +α3T h3 +……

[0031] In the formula, T n T represents the current temperature. h1 T h2... represents the temperature of each segment set in step two (which can be the average value, the maximum value, or a combination of other temperature information), and α1, α2... are the temperature contribution weights of each segment obtained through model training.

[0032] Step 4: Parameter Setting and Adjustment

[0033] When the algorithm fully converges and the accuracy meets the requirements, the obtained convergence formula is applied to the experimental samples as validation group data to check whether the same accuracy range is met for the non-training sample data. If it is met, the full temperature value formula at the current moment is given directly. If it is not met, feedback adjustment is performed to reset the length and number of segments of the traceback time period, and closed-loop operation is performed.

[0034] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for transient inversion of internal temperature mapping in a switchgear, characterized in that: Includes the following steps: Step 1: Obtain the internal and external temperature data streams of the switch cabinet to obtain the model training group and the model validation group; Step 2: For each data stream in the model training group and model validation group formed in Step 1, set four sets of variables: temperature calculation traceability period, number of period segments, initial weights, and preset calculation accuracy to obtain training samples; Step 3: Input all four sets of variables from the training samples obtained in Step 2 into the regression algorithm for training. During the training process, adjust the initial weights of these variables appropriately to ensure regression convergence. Simultaneously, change the preset precision to prevent overfitting even when the regression converges, thus avoiding the problem of accurately predicting future data. Once the regression algorithm converges, further adjust the length of the tracking time period or the number of segments to finally obtain the following convergence formula: T n =α1T h1 +α2T h2 +α3T h3 +…… In the formula, T n T represents the current temperature. h1 T h2 ... represents the temperature of each segment set in step two, and α1, α2... represent the temperature contribution weights of each segment obtained through model training; Step 4: When the algorithm fully converges and the accuracy meets the requirements, the obtained convergence formula is applied to the experimental samples as validation group data to check whether the same accuracy range is met for the non-training sample data. If it is met, the full temperature value formula at the current moment is given directly. If it is not met, feedback adjustment is performed to reset the length of the traceback time period and the number of segments.

2. The transient inversion method for temperature mapping inside the switchgear as described in claim 1, characterized in that: The regression algorithm used in step three is either the support vector machine regression algorithm or the multiple linear regression model.

Images