Method of CSI algorithm data augmentation model based on big data of railway signal relays

Abstract

The present invention belongs to a CSI algorithm data augmentation model based on big data of railway signal relays. It supplements the data with zero actions by using the autocorrelation coefficient, performs data augmentation on the supplemented data set by using an improved CSI algorithm, and evaluates the data sets using the improved CSI algorithm and the original CSI algorithm with the Spearman correlation coefficient, obtaining the following conclusions: Completing the supplementation of the first data by using the autocorrelation coefficient can ensure the integrity of the data, and the CSI algorithm can only perform interpolation when there is data at both the beginning and the end; By performing data augmentation by using the improved CSI algorithm, the accuracy of data augmentation in the fluctuation section can be improved, and the error brought during data augmentation in the fluctuation section can be reduced; Evaluating the augmented data by using the Spearman correlation coefficient can make the data quality higher, and when performing correlation analysis by using the augmented data, the correctness and effectiveness of the method can be verified.

Description

Technical Field

[0001] The present invention relates to a method for measuring various parameters of a relay, in particular to the measurement of electrical parameters related to the electrical life of a relay. Specifically, it relates to a supplementary model for the first data based on the autocorrelation coefficient of big data technology of railway signal relays. After completing the supplementation of the first data, an improved CSI algorithm is used to perform a data expansion model at a certain data interval. Background Art

[0002] As an indispensable important component in the railway automatic control system, whether the railway signal relay operates reliably will directly affect the safety of the entire railway system operation. The research on the reliability of relays has become the focus of attention of people and the academic community, and predicting the life of relays is an important part of its reliability research. Since there are many parameters of relays and the measurement methods and measurement intervals of different parameters are different, when multiple parameters need to be selected for fusion and comprehensive parameter information is considered, there will be certain difficulties due to the differences in parameter dimensions and data volumes. Therefore, a certain method is needed to expand the data with fewer data dimensions so that its data dimension is the same as that of the data with the largest dimension.

[0003] In the electrical life experiment of relays, it can be seen that the measurement methods of various parameters of relays are not the same. For example, the parameter of time is recorded in real time through a low-level device; electrical parameters such as contact resistance are recorded through manual measurement. Each parameter plays an important role in the degradation analysis of relays. However, due to different measurement methods of different parameters, the recorded data dimensions will also be different. If multiple parameters are used for fusion analysis, the analysis results will have large errors due to the different data dimensions. Therefore, the problem of data expansion for the performance parameters with lower dimensions collected is the problem to be solved by the present invention.

[0004] When expanding relay data, most scholars use interpolation algorithms such as Newton interpolation, Lagrange interpolation, and CSI. However, due to the existence of problems such as noise in the collected data, the smoothness of the data is poor, that is, it has certain fluctuations. If the CSI algorithm is directly used to interpolate the data, the interpolation effect is poor because the CSI interpolation accuracy is low for data with poor smoothness.

[0005] Therefore, how to improve the CSI algorithm so that it can be applicable to both the fluctuating section and the non-fluctuating section data and improve the interpolation accuracy has become one of the urgent problems to be solved by those skilled in the art. Summary of the Invention

[0006] In view of the above problems, the purpose of the present invention is to provide a method for a CSI algorithm data expansion model based on big data of railway signal relays, which can ensure the integrity of data, improve the accuracy of data expansion in the fluctuation section, reduce the error caused by data expansion in the fluctuation section, and verify the correctness and effectiveness of the method.

[0007] The technical solution of the present invention is: a method related to a CSI algorithm data expansion model based on big data of railway signal relays, characterized by including the following steps:

[0008] (1) Since the parameters of the relay are measured at certain intervals of the number of operation times and the parameter values before the relay operates are not measured, that is, the parameter values corresponding to 0 operation times of the relay. First, it is necessary to supplement the 0 - time data values of the data set using the autocorrelation coefficient.

[0009] 1) Define the relay parameter data sequence, calculate the mean value of the data sequence, and the calculation formula for the mean value is:

[0010]

[0011] In the formula, represents the mean value of the data sequence; represents the total amount of data in the data sequence; x i represents the data value in the data sequence;

[0012] 2) Calculate the autocorrelation coefficient values between the data in the used data sequence. The autocorrelation coefficient calculation formula is:

[0013] In the formula, P k represents the autocorrelation coefficient value of the k - th data section; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; k represents the i - th autocorrelation coefficient, and n represents the total amount of data in the data sequence;

[0014] 3) Obtain the autocorrelation coefficient matrix of the data sequence, calculate the mean value of the autocorrelation coefficients, and the calculation formula for the mean value of the autocorrelation coefficients is:

[0015] In the formula, represents the mean value of the autocorrelation coefficients; K represents the total number of autocorrelation coefficients; P k represents the autocorrelation coefficient value of the k - th data section.

[0016] 4) After calculating the mean value of the autocorrelation coefficients, set the mean value of the autocorrelation coefficients of the first parameter's data segment as the mean value to obtain the first parameter value, that is, the parameter value at 0 operation times. The calculation formula for the parameter value at 0 operation times is:

[0017]

[0018] In the formula, represents the mean value of the autocorrelation coefficient; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; x1 represents the first parameter value in the data sequence, that is, the parameter value when the action is 0 times, and is the value to be obtained.

[0019] (2) Calculate the value of the average smoothness of the overall data based on the supplemented data set. Take the average smoothness of the entire data section as the reference value and the average smoothness of the sliding window as the comparison value; if the comparison value is greater than the reference value, it is considered that the data segment has strong volatility, and if the comparison value is less than the reference value, it is considered that the data has strong smoothness.

[0020] 1) Calculate the value of the average smoothness of the overall data based on the supplemented data set. The formula for the average smoothness of the entire data section, that is, the reference value, is:

[0021]

[0022] In the formula, S m is the average smoothness of the entire data sequence; N is the length of the data sequence; y(i) is the corresponding data value.

[0023] 2) After calculating the average smoothness of the entire data section, calculate the value of the average smoothness of the data within each sliding window. The formula for the comparison value is:

[0024]

[0025] In the formula, S n is the average smoothness corresponding to the sliding window data; n is the selected data window value; y(i) is the corresponding data value.

[0026] (3) By comparing the value of the average smoothness of the sliding window data with that of the entire data section, determine the data using two piecewise functions. Use the improved CSI to expand the data. The formula for the improved CSI algorithm is:

[0027]

[0028] In the formula, S i (x) represents the interpolated data value; a i0 represents the constant term in the interpolation expression; a i1 represents the first-order term coefficient in the interpolation expression; a i2 represents the second-order term coefficient in the interpolation expression; a i3Represents the coefficient of the cubic term in the interpolation expression; S n Is the average smoothness corresponding to the sliding window data; S m Is the average smoothness of the entire data sequence; x and x t Both represent the data values of the data sequence;

[0029] (4) After using the improved CSI for data augmentation, in order to verify the feasibility of the application of the improved CSI, the concept of Spearman correlation coefficient is introduced to evaluate the datasets augmented with the improved CS algorithm and the CSI algorithm. The calculation formula of the Spearman correlation coefficient is:

[0030]

[0031] In the formula, ρ represents the correlation coefficient value calculated between the two data sequences; Represents the number of ranks X n , Y n The average value of.

[0032] The present invention supplements the data of the 0th order of the action by using the autocorrelation coefficient, augments the supplemented dataset by using the improved CSI algorithm, and evaluates the datasets using the improved CSI algorithm and the original CSI algorithm with the Spearman correlation coefficient, and obtains the following conclusions:

[0033] (1) Using the autocorrelation coefficient to complete the supplementation of the first data can ensure the integrity of the data. The CSI algorithm can only perform interpolation when there is data at both the beginning and the end;

[0034] (2) By using the improved CSI algorithm for data augmentation, the accuracy of data augmentation in the fluctuation section can be improved, and the error brought during data augmentation in the fluctuation section can be reduced.

[0035] (3) Using the Spearman correlation coefficient to evaluate the augmented data can make the data quality higher. When using the augmented data for correlation analysis, the correctness and effectiveness of the method can be verified. The present invention will be described in detail below in conjunction with the accompanying drawings of the embodiments and specific implementation cases. Brief Description of the Drawings

[0036] Figure 1 Is the flowchart of data augmentation based on the improved CSI algorithm of the present invention;

[0037] Figure 2 Is the original data curve of the pull-in voltage at 40°C;

[0038] Figure 3 Are the data curves of the pull-in voltage at 40°C after interpolation using the CSI interpolation algorithm and the improved CSI interpolation algorithm respectively;

[0039] Figure 4 The original data curve of the release voltage at 40°C;

[0040] Figure 5 The data curves after interpolation of the release voltage at 40°C using the CSI interpolation algorithm and the improved CSI interpolation algorithm respectively;

[0041] Figure 6 The original data curve of the pull-in voltage at 65°C;

[0042] Figure 7 The data curves after interpolation of the pull-in voltage at 65°C using the CSI interpolation algorithm and the improved CSI interpolation algorithm respectively;

[0043] Figure 8 The original data curve of the release voltage at 65°C;

[0044] Figure 9 The data curves after interpolation of the release voltage at 65°C using the CSI interpolation algorithm and the improved CSI interpolation algorithm respectively. Specific embodiments

[0045] The following will describe in detail the embodiments of the present invention in conjunction with the accompanying drawings and embodiments, so as to fully understand how the present invention applies technical means to solve technical problems and achieve the implementation process of technical effects and implement accordingly. As Figure 1 shown, the present invention relates to a method for a CSI algorithm data augmentation model based on railway signal relay big data, which is characterized in that it includes the following steps:

[0046] (1) First, select the pull-in voltage and release voltage of two relays at two temperatures (40°C and 65°C) as the performance parameters for case analysis.

[0047] (2) Calculate the autocorrelation coefficient values of a total of four performance parameters under the two selected conditions respectively. Use the calculation formula for calculating the autocorrelation coefficient to calculate the autocorrelation coefficient values of each data sequence. The autocorrelation calculation formula is:

[0048]

[0049] In the formula, P k represents the autocorrelation coefficient value of the kth data section; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; k represents the ith autocorrelation coefficient, and n represents the total amount of data in the data sequence.

[0050] According to the above calculation formula, the autocorrelation coefficient values corresponding to multiple data sequences of the performance parameters are obtained. The detailed autocorrelation coefficient values corresponding to the four performance parameters are shown in Tables 1 - 4.

[0051] As Figure 2 、 Figure 3 、 Figure 4 and Figure 5 shown: 1) Select the lag number of the data sequence of the pull-in voltage at 40 °C to be 20, and calculate the autocorrelation coefficient between the data, as shown in Table 1.

[0052] Table 1 Autocorrelation Coefficient of Pull-in Time

[0053]

[0054] 2) Select the lag number of the data sequence of the release voltage at 40 °C to be 20, and calculate the autocorrelation coefficient between the data, as shown in Table 2.

[0055] Table 2 Autocorrelation Coefficient of Release Time

[0056]

[0057] As Figure 6 、 Figure 7 、 Figure 8 and Figure 9 shown: 3) Select the lag number of the data sequence of the pull-in voltage at 65 °C to be 10, and calculate the autocorrelation coefficient between the data, as shown in Table 3.

[0058] Table 3 Autocorrelation Coefficient of Pull-in Time

[0059]

[0060] 4) Select the lag number of the data sequence of the pull-in voltage at 65 °C to be 10. Calculate the autocorrelation coefficient between the data, as shown in Table 4.

[0061] Table 4 Autocorrelation Coefficient of Pull-in Time

[0062]

[0063]

[0064] (3) According to the obtained autocorrelation coefficient values of each data section, calculate the average value of the autocorrelation coefficient corresponding to each parameter. The calculation formula is:

[0065]

[0066] In the formula, represents the average value of the autocorrelation coefficient; represents the average value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; x1 represents the first parameter value in the data sequence, that is, the parameter value when the action is 0 times, and is the value to be obtained.

[0067] (4) Calculate the mean value of the autocorrelation coefficient under each performance parameter condition, and then the first parameter value corresponding to each parameter can be solved according to the following equation. The mean value of the autocorrelation coefficient of the four parameters and the calculated first action value are shown in Table 5.

[0068]

[0069] In the formula, represents the mean value of the autocorrelation coefficient; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; x1 represents the first parameter value in the data sequence, that is, the parameter value when the action is 0 times, and it is the value to be solved.

[0070] Table 5 Mean value of autocorrelation coefficient and first action value

[0071]

[0072] (5) The sliding time window value under two temperature conditions is selected as 3, that is, the average smoothness is calculated for three consecutive values; after calculating the data average smoothness of the sliding time window value, the average smooth value of the entire section of data is calculated, and the improved CSI algorithm is used for interpolation of different data values by comparing the magnitude relationship between the two.

[0073] (6) Use the improved CSI algorithm to supplement data at a certain interpolation interval. The interpolation interval of the pull-in voltage and release voltage at 40 °C is selected as 50 times, and the interpolation interval of the pull-in voltage and release voltage at 60 °C is also selected as 50 times, that is, data interpolation is performed every 50 times.

[0074] (7) Draw the curve graph after interpolation using the CSI interpolation and improved CSI interpolation algorithms.

[0075] (8) Use the Spearman correlation coefficient evaluation criterion to evaluate the data sequences interpolated by the two methods, and calculate the correlation coefficient value between the selected performance parameters and the action coefficient. The Spearman correlation coefficient values are shown in Table 5.

[0076] Table 5 Spearman correlation coefficient table

[0077]

[0078] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can be made, and all should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for a CSI algorithm data augmentation model based on big data of railway signal relays, characterized in that: It includes the following steps: (1) First, select the pull-in voltage and release voltage of two relays at two temperatures of 40°C and 65°C as the performance parameters for example analysis; (2) Calculate the autocorrelation coefficient values of the total four performance parameters under the two selected conditions respectively, and use the calculation formula for calculating the autocorrelation coefficient to calculate the autocorrelation coefficient values of each data sequence; Calculate the average autocorrelation coefficient corresponding to each parameter based on the obtained autocorrelation coefficient values of each data section; (3) Calculate the average autocorrelation coefficient under each performance parameter condition, then the first parameter value corresponding to each parameter can be solved according to the following equation. The average autocorrelation coefficients of the four parameters and the calculated first action values are shown in Table 5; In the formula, represents the mean value of the autocorrelation coefficient; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; x1 represents the first parameter value in the data sequence, that is, the parameter value when the action is 0 times, and is the value to be obtained; Table 5 Average autocorrelation coefficients and first action values (4) The sliding time window values under the two temperature conditions are selected as 3, that is, the average smoothness is calculated for three consecutive values; after calculating the data average smoothness of the sliding time window value, calculate the average smooth value of the entire section of data, and determine whether to use the improved CSI algorithm for interpolation for different data values by comparing the magnitude relationship between the two; (5) Use the improved CSI algorithm to supplement data at a certain interpolation interval. The interpolation interval for the pull-in voltage and release voltage at 40 degrees Celsius is selected as 50 times, and the interpolation interval for the pull-in voltage and release voltage at 60 degrees Celsius is also selected as 50 times, that is, data interpolation is performed every 50 times; (6) Draw the curve graphs after interpolation using the CSI interpolation and improved CSI interpolation algorithms; (7) Use the Spearman correlation coefficient evaluation criterion to evaluate the data sequences interpolated by the two methods, and calculate the correlation coefficient value between the selected performance parameters and the action coefficient.

2. The method of the CSI algorithm data augmentation model based on big data of railway signal relays according to claim 1, characterized in that: The above calculation formula obtains the autocorrelation coefficient values corresponding to multiple data sequences of performance parameters; The detailed autocorrelation coefficient values corresponding to the four performance parameters are shown in Tables 1 - 4; 1) Select the lag number of the data sequence of the pull-in voltage under the condition of 40°C as 20, and calculate the autocorrelation coefficient between the data, as shown in Table 1; Table 1 Autocorrelation coefficient of pull-in time 2) Select the lag number of the data sequence of the release voltage under the condition of 40°C as 20, and calculate the autocorrelation coefficient between the data, as shown in Table 2; Table 2 Autocorrelation coefficient of release time 3) Select the lag number of the data sequence of the pull-in voltage under the condition of 65°C as 10, and calculate the autocorrelation coefficient between the data, as shown in Table 3; Table 3 Autocorrelation coefficient of pull-in time 4) Select the lag number of the data sequence of the pull-in voltage under the condition of 65°C as 10, and calculate the autocorrelation coefficient between the data, as shown in Table 4; Table 4 Autocorrelation coefficient of pull-in time 3. The method of the CSI algorithm data augmentation model based on big data of railway signal relays according to claim 1, characterized in that: The pull-in voltage at 40°C is -0.2735V, the release voltage at 40°C is -0.2043V, the pull-in voltage at 65°C is -0.3020, and the release voltage at 65°C is 0.1640.

4. The method of the CSI algorithm data augmentation model based on big data of railway signal relays according to claim 1, characterized in that: The calculation formula for calculating the average autocorrelation coefficient corresponding to each parameter is: In the formula, represents the mean value of the autocorrelation coefficient; represents the mean value of the data sequence; h represents the selected data lag number; x i represents the data value in the data sequence; x1 represents the first parameter value in the data sequence, that is, the parameter value when the action is 0 times, and is the value to be obtained.

Images