CN117269751A - GIS isolating switch switching position confirmation method - Google Patents

Abstract

The invention belongs to the technical field of power equipment monitoring, and relates to a GIS isolating switch switching position confirmation method, which comprises the steps of collecting real-time power of a motor of a GIS isolating switch, and performing data preprocessing to obtain an unknown action power curve; estimating probability density estimation curves corresponding to a standard opening power curve, a standard closing power curve and an unknown action power curve by using a nuclear density estimation method; adopting a cross-validation method to adaptively select the bandwidth of the kernel density estimation; calculating the similarity score of a nuclear density estimation curve corresponding to the unknown action power curve and the nuclear density estimation curves of the standard opening power curve and the standard closing power curve; and judging whether the unknown action power curve is opened or closed based on the similarity score. The invention estimates the probability density estimation curve by the kernel density estimation method and calculates the similarity to judge the opening and closing of the brake, and has the advantages of no dependence on the alignment process, wide applicability and capability of providing probability density information.

Description

A method for confirming the opening and closing position of GIS isolation switch

Technical field

The invention belongs to the technical field of power equipment monitoring, and specifically relates to a method for confirming the opening and closing position of a GIS isolation switch.

Background technique

Gas-insulated fully enclosed electrical appliances (GIS) are key components in the power system to achieve segmentation and isolation of electrical equipment. However, the mechanical structure of the GIS isolating switch is susceptible to assembly errors and structural damage caused by long-term use, leading to failure. Among them, the problem of insufficient opening and closing poses the greatest threat to the safety of the power system and personnel.

GIS isolation switch opening and closing position detection technology includes two types: direct detection method and status sensing method. Direct detection method, because the GIS isolation switch has a closed structure, small internal space, compact transmission mechanism, and operates in a high-voltage and high-current environment. When a sensor is introduced at a high potential, it is difficult to draw out the signal line. The signal is interfered by the strong electromagnetic environment, destroys the sealing of the original structure, and easily causes insulation hazards. It is not suitable for the detection of this object.

The state sensing method based on motor current and the state sensing method based on vibration signal theoretically have complete information about the mechanical state of the GIS isolating switch, and both can monitor the contact status of the GIS isolating switch. However, the state sensing method based on motor current cannot reflect the phase change relationship between voltage and current, and motor currents of different motor models have a certain degree of dispersion, resulting in inconspicuous current characteristics. Even under the premise that the current characteristics are relatively obvious, the existing diagnosis algorithm belongs to the physical analysis method. The defect analysis through the analysis of the current envelope belongs to the threshold judgment. This judgment will cause two problems. First, for different The threshold setting of the isolating switch is difficult to directly quantify. It requires expert intervention to complete the physical analysis of the movement process of the isolating switch and then perform threshold correspondence. At the same time, it is also difficult to adapt to the threshold value of the industrial site caused by the discreteness of the mechanical state of the switch. Differences arising from differences.

The state sensing method based on vibration signals only considers the vibration frequency of the isolation switch under the action of electromagnetic force and ignores the natural frequency of the GIS isolation switch body structure. The conclusions drawn are difficult to match the measured signals. The amount of data collected is large, the signal transmission time is long, and the storage space is consumed.

It is difficult for the above existing technologies to reliably confirm the opening and closing action instructions of the isolation switch. After the instruction is issued, the key focus is whether the equipment completes the corresponding action and can detect the abnormal status of the opening and closing position of the isolation switch in time to avoid possible accidents. Operational accident. This is crucial for the safe and stable operation of the power system.

Contents of the invention

In order to better judge the position of the isolating switch, the present invention provides a method for confirming the opening and closing position of the GIS isolating switch. The probability density corresponding to the standard opening power curve, the standard closing power curve and the unknown action power curve is estimated through the kernel density estimation method. Estimate the curve and calculate the similarity to judge opening and closing.

A method for confirming the opening and closing position of a GIS isolating switch of the present invention, the steps are as follows:

Step 1: Collect the real-time power of the motor of the GIS isolation switch, and perform data preprocessing to obtain the unknown action power curve;

Step 2: Use the kernel density estimation method to estimate the probability density estimation curves corresponding to the standard opening power curve, standard closing power curve and unknown action power curve;

Step 3: Use cross-validation method to adaptively select the bandwidth of kernel density estimation;

Step 4: Calculate the similarity score between the kernel density estimation curve corresponding to the unknown action power curve and the kernel density estimation curve of the standard opening power curve and the standard closing power curve; determine whether the unknown action power curve is opening or closing based on the similarity score brake.

Specifically, the kernel density estimation calculation process is:

;

;

in: is the probability density estimate at position x; _xi is the i-th data point, n is the total number of data points; K _h is the kernel function, and h is the bandwidth.

Specifically, the cross-validation method includes minimum mean square error cross-validation and minimum cross-validation estimation.

Specifically, the minimum mean square error cross-validation: divide the data set into a training set and a verification set, usually using k-fold cross-validation, and perform kernel density estimation on the training set for each bandwidth value; use the probability density estimation curve in Calculate the mean square error on the validation set; select the bandwidth that minimizes the mean square error as the optimal bandwidth.

Specifically, the minimum cross-validation estimation: divide the data set into a training set and a validation set; for each bandwidth value, perform kernel density estimation on the training set; use the probability density estimation curve to calculate the expected mean square error on the validation set; The bandwidth that minimizes the expected mean square error is selected as the optimal bandwidth.

Specifically, in step one, the voltage and current of the motor of the GIS isolation switch during operation are collected, and the real-time power of the motor is calculated based on the voltage and current to obtain an unknown action power curve.

Specifically, in step one, data preprocessing includes data cleaning, data formatting and data alignment.

Specifically, the data cleaning includes processing missing data, processing outliers and noise filtering.

Specifically, the data formatting includes: checking the time intervals between the data points of the standard opening power curve, the standard closing power curve and the unknown action power curve to ensure that they are consistent in time; checking the format of the data.

Specifically, the data alignment includes:

Time alignment: Based on the timestamp information, align the standard opening power curve, standard closing power curve and unknown action power curve to make them consistent in time;

Interpolation: If the data points are not evenly distributed in time, use interpolation methods to obtain data with uniform time steps.

The invention can perform status determination in real time during switch operation, ensuring stable operation of the power system. This invention uses the kernel density estimation method to perform power curve similarity analysis. Compared with traditional curve similarity analysis methods such as dynamic time warping (DTW), it has the following advantages:

Does not rely on the alignment process: Kernel density estimation methods do not require the process of curve alignment, while methods such as DTW usually require temporal or spatial alignment of curves before performing similarity analysis, which may involve complex calculations and parameter selection. Kernel density estimation methods allow free comparisons between curves without forced alignment by performing smooth estimates across the entire curve. This allows for greater flexibility in dealing with changes between curves and noise.

Wide applicability: The kernel density estimation method is a non-parametric method that does not depend on the specific shape or model of the curve, so it is suitable for various types of curve data, including continuous, discrete, unimodal, multimodal, etc. In contrast, methods such as DTW usually rely on the specific shape and characteristics of the curve, and different types of curves may need to adjust parameters or modify the algorithm.

Provide probability density information: The kernel density estimation method can not only be used to compare the similarity of curves, but also provide probability density information for each point on the curve. This means that the distribution properties of the curve can be understood in more detail, potential modes or peaks can be identified, and further statistical analysis can be performed. This probability density information is of great value in many application areas, such as anomaly detection, pattern recognition, and statistical analysis.

Non-contact monitoring: This method only needs to compare the existing power curve data and does not require additional physical contact or equipment intervention, reducing potential risks.

Description of the drawings

Figure 1 shows the standard opening power curve;

Figure 2 shows the standard closing power curve;

Figure 3 shows the unknown action power curve;

Figure 4 is a comparison diagram of the similarity of kernel density estimation curves.

Detailed ways

The present invention will be further explained in detail below with reference to the examples.

A method for confirming the opening and closing position of a GIS isolating switch. The steps are as follows:

Step 1: Collect the real-time power of the motor of the GIS isolation switch, and perform data preprocessing to obtain the unknown action power curve.

Collect the voltage and current of the motor of the GIS isolation switch during operation, calculate the real-time power of the motor based on the voltage and current, and obtain the unknown action power curve. Data preprocessing includes data cleaning, data formatting and data alignment.

Data cleaning is a critical step to ensure data quality. At this stage, the following tasks need to be performed:

Handle missing data: Check and fill in missing data points, or consider interpolation methods to fill in missing values;

Handle outliers: Identify and handle possible outliers or outliers to prevent them from affecting the accuracy of estimates;

Noise filtering: Consider applying smoothing or filtering techniques to reduce noise in your data.

Data formatting ensures data consistency and availability. At this stage, the following tasks need to be performed:

Data sampling frequency: Check the time intervals between the data points of the standard opening power curve, standard closing power curve and unknown action power curve to ensure that they are consistent in time;

Data format: Check the format of the data, such as whether the data is stored in the appropriate structure and type.

Data alignment is to ensure that the data of the standard opening power curve, standard closing power curve and unknown action power curve are aligned in time for subsequent comparison. Perform the following tasks to achieve data alignment:

Time alignment: Based on the timestamp information, align the standard opening power curve, standard closing power curve and unknown action power curve to make them consistent in time;

Interpolation: If the data points are not uniformly distributed in time, consider using interpolation methods to obtain data with uniform time steps;

Ensure that the aligned data sets of the standard opening power curve, the standard closing power curve and the unknown action power curve contain the same number of data points or time steps for subsequent kernel density estimation and comparison.

Step 2: Use the kernel density estimation method to estimate the probability density estimation curves corresponding to the standard opening power curve, standard closing power curve and unknown action power curve. The basic principle of kernel density estimation is to estimate the probability density by placing a kernel function near each data point and then superimposing them. The following is the basic formula for kernel density estimation:

;

in: is the probability density estimate at position x; _xi is the i-th data point, and n is the total number of data points. K _h is the kernel function, usually the density function of the standard normal distribution, the specific form is:

;

where h is the bandwidth, used to control the width of the kernel function.

Kernel density estimation is a non-parametric method for estimating the probability density estimation curve of data. It works by placing a kernel function around each data point and then stacking the kernel functions together to obtain an estimate of the entire data distribution. The kernel function is usually a symmetric function centered around zero that measures the contribution of a data point to the estimate. The density function of the standard normal distribution is usually used as the kernel function, but other kernel functions can also be selected, such as Epanechnikov kernel, trigonometric kernel, etc.

Kernel density estimation calculation: For the i-th data point x _i , calculate the value of the kernel function at position x . This value measures the contribution of data point _xi to the estimate at location x. Adding up the contributions of all data points and dividing by the sample size n yields an estimate of the probability density at location x .

Visualize the results of kernel density estimation and plot the estimated kernel density estimation curve. This will help to intuitively understand the distribution of the data and estimate the probability density of the unknown action power curve. Kernel density estimation is a useful tool for understanding data distributions and performing probability density estimates. Kernel density estimation is used to obtain the probability density estimation curves of the standard opening power curve, standard closing power curve and unknown action power curve to support subsequent similarity evaluation.

Step 3: Adaptive bandwidth selection.

In kernel density estimation, choosing an appropriate bandwidth is key because it directly affects the shape and smoothness of the probability density estimation curve. A cross-validation approach will be used to select the bandwidth to ensure the accuracy and robustness of the probability density estimates.

Cross-validation is a commonly used bandwidth selection method that selects the optimal bandwidth by performing multiple kernel density estimates on a data set and evaluating performance. Common cross-validation methods include minimum mean square error cross-validation and minimum cross-validation estimation.

Minimum mean square error cross-validation: The goal of this method is to choose the bandwidth to minimize the mean square error between the kernel density estimate and the true probability density curve. The specific steps are as follows: Divide the data set into a training set and a validation set, usually using k-fold cross-validation, where k is the fold number of cross-validation; for each possible bandwidth value, perform kernel density estimation on the training set; use probability density The estimated curve calculates the mean square error on the validation set; the bandwidth that minimizes the mean square error is selected as the optimal bandwidth.

Minimum cross-validation estimate: The goal of this method is to choose a bandwidth that minimizes the expected mean square error of the cross-validation estimate. The specific steps are as follows: divide the data set into a training set and a validation set; for each possible bandwidth value, perform kernel density estimation on the training set; use the probability density estimation curve to calculate the expected mean square error on the validation set; choose to make the expected mean square error The bandwidth that minimizes the square error is taken as the optimal bandwidth.

The present invention comprehensively adopts the above two bandwidth verification methods to select the one with the best effect. The effect varies depending on the actual data.

The choice of bandwidth usually involves the use of cross-validation methods, with minimum mean square error cross-validation and minimum cross-validation estimation being common selection methods. The basic principle of these methods is to select the optimal bandwidth by estimating the kernel density function on the training set and then evaluating the quality of the estimate on the validation set.

The choice of bandwidth takes into account the trade-off between smoothness and adaptability. A smaller bandwidth will result in a probability density estimate curve that is closer to the original data, but may be too sharp. A larger bandwidth will result in a smoother probability density estimate curve, but may be too flat and lose the detail of the original data.

Through cross-validation, you can find a bandwidth that makes the probability density estimation curve have a smaller mean square error or expected mean square error on the validation set. This bandwidth can be used to estimate the probability density estimation curve of the unknown action power curve for subsequent similarity evaluation.

In summary, the goal of bandwidth selection is to find a value that enables the kernel density estimate to achieve the best balance between smoothness and adaptability to ensure that the kernel density estimate can accurately reflect the distribution characteristics of the data. Cross-validation methods are a common, data-based way to achieve this goal. The bandwidth adaptation method can automatically select an appropriate bandwidth based on the characteristics of the data without requiring manual adjustment or pre-set bandwidth values. This adaptability makes the method more general and can adapt to changes in different data sets. Bandwidth adaptation can better handle inhomogeneities and noise in data than traditional fixed-bandwidth methods.

Step 4: Calculate the similarity score between the kernel density estimation curve corresponding to the unknown action power curve and the kernel density estimation curve of the standard opening power curve and the standard closing power curve; based on the similarity score, explain whether the unknown action power curve is opening or closing. Possibility of closing, providing visualization or interpretability information of results to support final decision-making.

Application examples of the present invention are shown in Figures 1 to 4. Figure 1 is a standard opening power curve, in which the horizontal axis represents data points and the vertical axis represents numerical values, and the unit is watts (W). Figure 2 is the standard closing power curve, and Figure 3 is the unknown action power curve. Figure 4 is a comparison chart of the similarity of kernel density estimation curves. Curve 1 is the kernel density estimation curve corresponding to the standard opening power curve. Curve 2 is the kernel density estimation curve corresponding to the standard closing power curve. Curve 3 is the kernel density estimation curve corresponding to the unknown action power curve. The kernel density estimation curve of , the horizontal axis represents the data value, and the vertical axis represents the probability density, which is dimensionless. The similarity score between curve 3 and curve 1 is 0.0000000954, and the similarity score between curve 3 and curve 2 is 0.0000040416. Therefore, it is judged that Figure 3 is the closing power curve and the GIS isolating switch is closed.

This invention uses the kernel density estimation method to perform curve similarity analysis. Compared with traditional curve similarity analysis methods such as dynamic time warping (DTW), it has the following three main advantages:

Does not rely on the alignment process: Kernel density estimation methods do not require the process of curve alignment, while methods such as DTW usually require temporal or spatial alignment of curves before performing similarity analysis, which may involve complex calculations and parameter selection. Kernel density estimation methods allow free comparisons between curves without forced alignment by performing smooth estimates across the entire curve. This allows for greater flexibility in dealing with changes between curves and noise. This makes the kernel density estimation method easier to implement, especially for situations where there is noise or variation in the real data.

Wide applicability: The kernel density estimation method is a non-parametric method that does not depend on the specific shape or model of the curve, so it is suitable for various types of curve data, including continuous, discrete, unimodal, multimodal, etc. In contrast, methods such as DTW usually rely on the specific shape and characteristics of the curve, and different types of curves may need to adjust parameters or modify the algorithm. Methods such as DTW are often based on specific curvilinear models or assumptions, which may limit their applicability. If the shape of the curve does not meet these assumptions, performance may degrade.

Provide probability density information: The kernel density estimation method can not only be used to compare the similarity of curves, but also provide probability density information for each point on the curve. This means that the distribution properties of the curve can be understood in more detail, potential modes or peaks can be identified, and further statistical analysis can be performed. This probability density information is of great value in many application areas, such as anomaly detection, pattern recognition, and statistical analysis.

Although embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that these embodiments may be modified without departing from the principles and spirit of the invention. Various changes, modifications, substitutions and variations are possible, the scope of the invention being defined by the appended claims and their equivalents.

Claims (10)

1. The GIS disconnecting switch switching position confirmation method is characterized by comprising the following steps:

step one: collecting real-time power of a motor of the GIS isolating switch, and performing data preprocessing to obtain an unknown action power curve;

step two: estimating probability density estimation curves corresponding to a standard opening power curve, a standard closing power curve and an unknown action power curve by using a nuclear density estimation method;

step three: adopting a cross-validation method to adaptively select the bandwidth of the kernel density estimation;

step four: calculating the similarity score of a nuclear density estimation curve corresponding to the unknown action power curve and the nuclear density estimation curves of the standard opening power curve and the standard closing power curve; and judging whether the unknown action power curve is opened or closed based on the similarity score.

2. The method for confirming the on-off position of a GIS isolating switch according to claim 1, wherein the core density estimation calculation process is as follows:

；

；

wherein:is a probability density estimate at location x; x is x _i Is the ith data point, n is the total number of data points; k (K) _h Is a kernel function and h is bandwidth.

3. The method of claim 1, wherein the cross-validation method comprises minimum mean square error cross-validation and minimum cross-validation estimation.

4. A GIS isolation switch on-off position confirmation method according to claim 3, wherein the minimum mean square error cross-validation: dividing the data set into a training set and a verification set, generally adopting k-fold cross verification, and carrying out kernel density estimation on the training set for each bandwidth value; calculating a mean square error over the validation set using the probability density estimation curve; the bandwidth that minimizes the mean square error is selected as the optimal bandwidth.

5. A GIS disconnector switching position confirmation method according to claim 3, characterized in that the minimum cross-validation estimate: dividing the data set into a training set and a verification set; for each bandwidth value, performing kernel density estimation on the training set; calculating an expected mean square error over the validation set using the probability density estimation curve; the bandwidth that minimizes the desired mean square error is selected as the optimal bandwidth.

6. The method for confirming the switching position of the GIS isolating switch according to claim 1, wherein in the first step, the voltage and the current of the motor of the GIS isolating switch in the running process are collected, and the real-time power of the motor is calculated according to the voltage and the current to obtain an unknown action power curve.

7. The method for confirming the opening and closing positions of the GIS isolating switch according to claim 1, wherein in the first step, the data preprocessing comprises data cleaning, data formatting and data alignment.

8. The method of claim 7, wherein the data cleaning includes processing missing data, processing outliers and noise filtering.

9. The method for confirming the opening and closing positions of a GIS isolation switch according to claim 7, wherein the data formatting includes: checking the time intervals among the data points of the standard opening power curve, the standard closing power curve and the unknown action power curve to ensure that the data points are consistent in time; the format of the data is checked.

10. The method for confirming the opening and closing positions of a GIS isolation switch according to claim 7, wherein the data alignment comprises:

time alignment: according to the timestamp information, aligning a standard opening power curve, a standard closing power curve and an unknown action power curve to make the standard opening power curve, the standard closing power curve and the unknown action power curve consistent in time;

interpolation: if the data points are not evenly distributed in time, interpolation methods are used to obtain data for even time steps.