Index screening method and device for studying low-correlation simulated driving indexes

Abstract

The invention discloses an index screening method and device for studying low-correlation simulated driving indexes. The method comprises the steps of obtaining a simulated driving index sequence; the method comprises the following steps: constructing # imgabs0 # matrixes; starting to traverse the current # imgabs 1 # matrix, and taking out a matrix to be screened from the current # imgabs 2 # matrix; determining whether the current to-be-screened matrix meets a correlation screening condition or not; if not, the next to-be-screened matrix is taken out from the current # imgabs3 # matrixes; if yes, adding 1 to the value of the current m, and then determining whether the value multiplied by the current m and j is smaller than a second threshold value or not; when the current # imgabs4 # matrixes are traversed, adding 1 to the value of the current j, and determining whether the value of the current m multiplied by the j is smaller than a second threshold value or not; if the value multiplied by current m and j is smaller than a second threshold value, returning to construct # imgabs 5 # matrixes; if the value multiplied by the current m and j is not less than a second threshold value, constructing # imgabs6 # matrixes according to the recently marked values of m and j; and determining a target matrix in the # imgabs7 # matrixes, and determining a target index in the target matrix.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular, to a method and device for screening indicators for studying low-correlation simulated driving indicators. Background Art

[0002] With the development of science and technology, the research in various disciplines has become increasingly in-depth, and the correlation problem between research indicators has gradually emerged. By deeply studying the correlation between indicators, the internal relationship between each indicator can be better understood, providing more accurate and comprehensive data support for scientific research.

[0003] In the simulated driving scenario, the number of simulated driving indicators available for research may be hundreds or thousands. However, in the face of indicators of this magnitude, the prior art is unable to determine which indicators need to be selected from these indicators for research, which has also led to a slow progress in the subsequent research on indicators.

[0004] Therefore, how to design an indicator screening method that can quickly select the most meaningful indicators from a large number of indicators as the subsequent research object has become a problem to be solved in this field. Summary of the Invention

[0005] In view of this, in the first aspect, this application proposes a method for screening indicators for studying low-correlation simulated driving indicators, the method comprising:

[0006] S100. Obtain a sequence of simulated driving indicators;

[0007] S200. In the sequence of simulated driving indicators, sequentially select simulated driving indicators in order to construct matrices; where m represents the matrix order, j represents a natural number starting from 1, represents a combination of taking m simulated driving indicators from the first m multiplied by j simulated driving indicators in the sequence of simulated driving indicators to construct a matrix;

[0008] S300. Start traversing the current matrices, and take out a matrix to be screened from the current matrices;

[0009] S400. Determine whether the current matrix to be screened meets the correlation screening condition, the correlation screening condition being: the absolute value of the correlation coefficient between each pair of indicators in the matrix is less than a first threshold;

[0010] If not, return to step S300, and take out the next matrix to be screened from the current matrices;

[0011] If satisfied, perform step S410. After marking the current values of m and j, increment the current value of m by 1, and then determine whether the product of the current m and j is less than the second threshold;

[0012] When finishing traversing the current matrix, and no matrix to be screened that meets the correlation screening condition is found in the current matrix, then

[0013] S420. After incrementing the current value of j by 1, determine whether the product of the current m and j is less than the second threshold;

[0014] If the product of the current m and j is less than the second threshold, return to step S200, and continue to construct matrices according to the current values of m and j;

[0015] If the product of the current m and j is not less than the second threshold, perform step S500. Construct matrices according to the most recently marked values of m and j;

[0016] S600. Determine the target matrix among the matrices, and determine the target index in the target matrix.

[0017] Preferably, step S100 includes:

[0018] S110. Generate a simulated driving index matrix according to the simulated driving index and the correlation coefficient between indexes;

[0019] S120. Count the number of absolute values of the correlation coefficients between indexes corresponding to each simulated driving index in the simulated driving index matrix that are less than the first threshold, and sort the simulated driving indexes in descending order according to the number to obtain a simulated driving index sequence.

[0020] Further preferably, before step S110, the method further includes:

[0021] S111. Preprocess the simulated driving index and the correlation coefficient between indexes.

[0022] Preferably, when the current matrix to be screened does not meet the correlation screening condition, the method further includes:

[0023] S421. Put the matrix that does not meet the correlation screening condition obtained by screening into the exclusion set.

[0024] Further preferably, step S600 includes:

[0025] S610. Screen out all first target matrices that meet the correlation screening condition among the matrices;

[0026] S620. Determine a second target matrix with the smallest sum of absolute values of the correlation coefficients between the metrics among all the first target matrices;

[0027] S630. Obtain the target metric in the second target matrix.

[0028] Further preferably, before the step S610, the method further includes:

[0029] S611. Exclude the matrices in the exclusion set from the matrices.

[0030] Further preferably, the initial value of m is 3.

[0031] In a second aspect, the present application further provides an index screening device for studying low-correlation simulated driving metrics, the device includes:

[0032] An acquisition module, configured to acquire a sequence of simulated driving metrics;

[0033] A processing module, configured to:

[0034] In the sequence of simulated driving metrics, sequentially select simulated driving metrics in order to construct matrices; where m represents the matrix order, j represents a natural number starting from 1, represents a combination of selecting m simulated driving metrics from the first m×j simulated driving metrics in the sequence of simulated driving metrics to construct a matrix;

[0035] Start traversing the current matrices, and take out a matrix to be screened from the current matrices;

[0036] Determine whether the current matrix to be screened meets the correlation screening condition, and the correlation screening condition is: the absolute value of the correlation coefficient between each pair of metrics in the matrix is less than a first threshold;

[0037] If not, take out the next matrix to be screened from the current matrices;

[0038] If so, after marking the current values of m and j, increment the current value of m by 1, and then determine whether the current value of m×j is less than a second threshold;

[0039] When finishing traversing the current matrices and there is no matrix to be screened that meets the correlation screening condition in the current matrices, increment the current value of j by 1, and then determine whether the current value of m×j is less than a second threshold;

[0040] If the product of the current m and j values is less than the second threshold, continue to construct matrices according to the current m and j values; if the product of the current m and j values is not less than the second threshold,

[0041] construct matrices according to the most recently marked m and j values;

[0042] A determination module is configured to determine a target matrix from among the matrices and determine a target index in the target matrix.

[0043] The index screening method for studying low-correlation simulated driving indices provided in this application constructs and traverses index matrices according to conditions based on a simulated driving index sequence, and finally screens out target indices from these matrices. This solution can quickly select the most meaningful simulated driving indices from a large number of simulated driving indices as the subsequent research object, providing a solid foundation for further studying the screened simulated driving indices.

[0044] Other features and advantages of this application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which form a part of this application, are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof are used to explain this application. In the drawings:

[0046] Figure 1 is a flowchart of the method for the index screening method for studying low-correlation simulated driving indices according to a preferred embodiment of this application;

[0047] Figure 2 is a schematic diagram of a partial content of a simulated driving index matrix according to a preferred embodiment of this application;

[0048] Figure 3 is a schematic structural diagram of an index screening device for studying low-correlation simulated driving indices according to a preferred embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The technical solutions of this application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0050] The correlation between indicators refers to a certain connection or dependence relationship existing between two or more indicators. This relationship can be direct or indirect, and may manifest as positive correlation, negative correlation, or no correlation. The correlation between indicators can be analyzed and quantified by various methods to calculate the correlation coefficient between indicators, such as the Pearson correlation coefficient, the Spearman rank correlation coefficient, etc. Studying low correlations between indicators helps to reveal which indicators are statistically independent, and the opposite is true for studying high correlations between indicators. These studies can be used to more accurately understand the respective roles and influencing factors of indicators.

[0051] In the field of simulated driving, the evaluation indicators involved in simulated driving are complex, diverse, and numerous. For example, hundreds of simulated driving indicators can be obtained using the Fronba driving simulator. Researchers often need to accurately screen out the indicators that make key contributions to the research problem from a large number of indicators.

[0052] Based on this, the present application first proposes an indicator screening method for studying low-correlation simulated driving indicators, as Figure 1 shown, including S100 - S600:

[0053] S100, obtain a sequence of simulated driving indicators;

[0054] Specifically, before screening the simulated driving indicators, a sequence of simulated driving indicators needs to be constructed. The indicators in this sequence need to be sorted in descending order according to the number of absolute values of the correlation coefficients between corresponding indicators that are less than a first threshold. The absolute value of the correlation coefficient between indicators being less than the first threshold indicates that the two simulated driving indicators corresponding to this correlation coefficient have low correlation. This first threshold can be set as needed.

[0055] It can be understood that the correlation coefficients between simulated driving indicators can be various types of coefficients, such as the Pearson correlation coefficient, the Spearman rank correlation coefficient, etc. Which method to specifically choose depends on the characteristics of the data and the analysis purpose, and the present application does not make a limitation.

[0056] In a specific embodiment, the specific steps of S100 include S111 - S120:

[0057] S111, preprocess the simulated driving indicators and the correlation coefficients between indicators;

[0058] Specifically, the preprocessing includes steps such as data cleaning, missing value processing, outlier processing, dimensionless processing, and standardization processing.

[0059] S110, generate a simulated driving indicator matrix according to the simulated driving indicators and the correlation coefficients between indicators;

[0060] Specifically, if N simulated driving metrics are obtained, an N×N simulated driving metric matrix can be formed. The first row and the first column of this matrix are the simulated driving metrics, and the element values in the matrix are the correlation coefficients between the metrics.

[0061] In an embodiment of the simulated driving metric matrix as Figure 2 shown, the simulated driving metrics include "time, timeStamp, trafficTime, scenarioTime, type, model, ID, customID, description, positionX, position Y, position Z, yawAngle, pitchAngle, rollAngle, direction X, direction Y, direction Z, bodyPitchAngle, bodyRollAngle, RPM, transmissionState, gearNumber", and so on. The correlation coefficient between the metrics is a value between "-1" and "1". The smaller the value, the smaller the correlation. A positive value indicates a positive correlation between two simulated driving metrics, and a negative value indicates a negative correlation between two simulated driving metrics.

[0062] S120. In the simulated driving metric matrix, count the number of the absolute values of the correlation coefficients between the metrics corresponding to each simulated driving metric that are less than the first threshold, and sort the simulated driving metrics in descending order according to this number to obtain a simulated driving metric sequence;

[0063] Specifically, count the number of the absolute values of the correlation coefficients between the metrics corresponding to each simulated driving metric that are less than the first threshold, and sort all the simulated driving metrics in descending order according to this number.

[0064] In a specific embodiment of the simulated driving index sequence, the first threshold is 0.4, and some values in the simulated driving index sequence include: "('trailerWheelbase' (trailer wheelbase), 104), ('appliedThrottle' (actual throttle), 103), ('lightState' (light state), 102), ('rawThrottle' (raw throttle), 99), ('latestRoad' (current road), 96), ('clutch' (clutch), 91), ('trailer' (trailer), 91),

[0065] ('dragForce' (drag force), 90), ('trailerAngle' (trailer angle), 90),

[0066] ('distanceAlongRoad' (distance along road), 90), ('model' (model), 89), ('throttle' (throttle), 89), ('rawBrake' (raw brake), 88)”. Among them, the first string in each pair of parentheses is the name of the simulated driving index (the content in parentheses is the Chinese translation of the index), and the second is the number of all correlation coefficients involved in the simulated driving index within ±0.4. For example, the number of correlation coefficients between the index 'trailerWheelbase' and other indexes within ±0.4 is 104.

[0067] S200, in the simulated driving index sequence, sequentially select simulated driving indexes in order to construct matrices;

[0068] Specifically, in this application, according to the order of the indexes in the simulated driving index sequence, select indexes one by one from front to back to gradually construct a set of correlation coefficient matrices of the simulated driving indexes, and search for matrices that meet the requirements in the constructed set of correlation coefficient matrices of the simulated driving indexes. The number of matrices in this matrix set is where m represents the matrix order, j represents a natural number starting from 1, represents the combination of taking m simulated driving indexes from the first m times j simulated driving indexes in the simulated driving index sequence to construct a matrix.

[0069] Since the 2nd-order matrix is very simple and it is easy to find a matrix that meets the requirements. Therefore, the initial value of m in this application is "3", and the initial value of j is "1", that is, start the analysis from a 3rd-order matrix. When m = 3 and j = 1, select the first m times j simulated driving indexes with a higher order, select any m indexes from the m times j simulated driving indexes for combination, and each combination corresponds to a matrix. All combinations form a matrix set, and this matrix set includes a matrix.

[0070] S300, start traversing the current matrix, and take out a matrix to be screened from the current matrix;

[0071] Specifically, after taking out a matrix to be screened from the current matrix, execute step S400 to determine whether the current matrix to be screened meets the correlation screening condition. If all the current matrices have been traversed and no matrix to be screened that meets the correlation screening condition is found, then execute step S420.

[0072] It can be understood that starting to traverse the current matrix does not necessarily mean that the traversal of the current matrix is completed. If a matrix that meets the correlation screening condition is found in the current matrix, the traversal terminates, which can improve the screening efficiency in this way.

[0073] S400, determine whether the current matrix to be screened meets the correlation screening condition;

[0074] Specifically, in this application, the correlation screening condition is: the absolute value of the correlation coefficient between each index in the matrix is less than the first threshold, which corresponds to the sorting requirement of the indexes in the simulated driving index sequence. Determining whether the current matrix to be screened meets the correlation screening condition is equivalent to determining whether the absolute value of each element in the matrix is less than the first threshold.

[0075] If the matrix to be screened meets the correlation screening condition, then execute S410. If the matrix to be screened does not meet the correlation screening condition, put the current matrix to be screened into the exclusion set, and then take out the next matrix to be screened from the current matrix, and continue to confirm whether the absolute value of each element in the next matrix to be screened is less than the first threshold, that is, then return to step S300. If after traversing the current matrix, still no matrix to be screened that meets the correlation screening condition is found, then execute step S420.

[0076] S410. If the current matrix to be screened meets the correlation screening condition, then after increasing the current value of m by 1, determine whether the product of m and the value of j is less than the second threshold;

[0077] Specifically, if the absolute value of each element in the current matrix is less than the first threshold, then mark the current values of m and j, then increase the current value of m by 1, and determine whether the product of the incremented m and j is less than the second threshold. If it is less, return to step S200 and continue to construct according to the current values of m and j a matrix. If it is not less than, then step S500 is executed.

[0078] Here, the purpose of determining whether the value of m multiplied by j is less than the second threshold is to set a calculation upper limit to save computing power. Therefore, the second threshold can be set according to factors such as computing power, data volume, and target requirements.

[0079] S420. When traversing the current matrix, and there is no matrix to be screened in the current matrix that meets the correlation screening condition, then after adding 1 to the current value of j, determine whether the current value of m multiplied by j is less than the second threshold;

[0080] Specifically, after traversing the current matrix, add 1 to the current value of j, and determine whether the current value of m multiplied by j is less than the second threshold. If it is less, it means that the amount of calculation is within the allowable range of computing power, then return to execute S200, and continue to construct the matrix according to the current values of m and j. If it is not less than, then execute step S500.

[0081] That is to say, in the above steps S410 and S420, as long as the value of m or j is incremented by 1, a computing power threshold judgment is required to avoid the situation where the computing power cannot be borne as more and more screening matrices are involved in the loop of index screening.

[0082] It can be understood that at this time, the value of m and / or the value of j is different from the initial value, and the matrix constructed is not the matrix set in the initial state. Through repeated loop searches and markings, the present application can find the simulation driving index matrix with the maximum order that meets the correlation screening conditions among a large number of simulation driving indexes within the allowable range of computing power, that is, find as many simulation driving indexes that best meet the requirements as possible.

[0083] S500. If the current value of m multiplied by j is not less than the second threshold, then construct the matrix according to the most recently marked values of m and j;

[0084] Specifically, if the current value of m multiplied by j is not less than the second threshold, it means that the current amount of calculation is at the upper limit allowed by the computing power, and the number of matrices in the matrix set should not be increased anymore, that is, there is no need to increase the values of m or j to construct a new matrix set. Then, determine the values of M and J according to the most recently marked values of m and j, and construct the matrix, and use this matrix set as the set to determine the final simulation driving index based on this matrix set in subsequent steps.

[0085] In a specific embodiment, the specific steps of S500 include S510 - S520:

[0086] S510, If the current product of m and j is not less than the second threshold, then confirm the values of m and j of the most recently marked ones.

[0087] Specifically, if the current product of m and j is not less than the second threshold, then the node backtracks to determine the values of m and j among the most recently marked nodes.

[0088] S520, Construct a matrix based on the values of m and j of the most recently marked ones;

[0089] Specifically, here the value of M is equal to the value of m of the most recently marked one, and the value of J is equal to the value of j of the most recently marked one, indicating a combination of selecting M simulation driving metrics from the first M×J simulation driving metrics in the simulation driving metric sequence to construct a matrix.

[0090] S600, Determine the target matrix among the matrices, and determine the target metric in the target matrix;

[0091] In a specific embodiment, the specific steps of S600 include S611 - S630:

[0092] S611, Exclude the matrices in the exclusion set among the matrices;

[0093] Specifically, before screening among the constructed matrices, the matrices in the already marked exclusion set can be excluded first. The data in these matrices do not meet the relevance screening conditions, so excluding these matrices from the matrix set of the matrices is beneficial for faster subsequent screening.

[0094] S610, Screen out all first target matrices that meet the relevance screening conditions among the matrices;

[0095] Specifically, traverse the matrices, and screen out the matrices whose absolute value of the correlation coefficient between the metrics within all matrices is less than the first threshold as the first target matrices.

[0096] S620, Determine the second target matrix with the minimum sum of the absolute values of the correlation coefficients between the metrics among all the first target matrices;

[0097] Specifically, determine the matrix with the minimum sum of the absolute values of the correlation coefficients between the metrics among all the first target matrices as the second target matrix. That is, the second target matrix is the matrix with the minimum average correlation coefficient among all the first target matrices.

[0098] S630, obtain the target metrics in the second target matrix;

[0099] Specifically, use the simulated driving metrics in the second target matrix as the target metrics, so as to obtain all the simulated driving metrics that finally meet the requirements of the correlation coefficient.

[0100] It can be understood that if the initially obtained simulated driving metrics are N, forming an N×N simulated driving metric matrix, then finally the largest M×M sub-matrix that meets the requirements can be found in this N×N matrix. The M simulated driving metrics corresponding to this sub-matrix are the set of simulated driving metrics in the largest range we want to study.

[0101] In a specific embodiment, the second threshold is "30", currently m = 8, j = 4, and at this time m×j = 32 > 30. Assume that the recently marked value of m is 7 and the value of j is 3. Then, M = 7, J = 3, and select the top-ranked M×J = 21 simulated driving metrics to construct a combined matrix. Assume there are 7100 matrices in the elimination set. Eliminate the 7100 matrices in the elimination set, and there are 17000 matrices remaining. Assume that among these 17000 matrices, a total of 5 first target matrices are screened out where the absolute value of the correlation coefficient between the metrics in the matrix is less than the first threshold. Finally, among these 5 first target matrices, the simulated driving metrics in the second target matrix with the smallest sum of the absolute values of the correlation coefficients between the metrics are: "lightState (lighting state), rawThrottle (original throttle), latestRoad (current road), trailer (trailer), dragForce (drag), model (model), centerOfGravityHeight (center of gravity height)". These 7 simulated driving metrics are the finally screened target metrics.

[0102] The metric screening method for studying low-correlation simulated driving metrics provided by this application can quickly select the most meaningful simulated driving metrics from a large number of simulated driving metrics as the subsequent research object by constructing and traversing the metric matrix according to the simulated driving metric sequence and finally screening out the target metrics in these matrices, providing a solid foundation for further studying the screened simulated driving metrics.

[0103] In a second aspect, this application also provides a metric screening device for studying low-correlation simulated driving metrics. The device is used to implement the metric screening method for studying low-correlation simulated driving metrics described in the first aspect above, as Figure 3 shown. The device includes:

[0104] An acquisition module 310, configured to acquire a sequence of simulated driving metrics;

[0105] A processing module 320, configured to:

[0106] In the sequence of simulated driving metrics, sequentially select simulated driving metrics in order to construct matrices; where m represents the matrix order, j represents a natural number starting from 1, represents a combination of selecting m simulated driving metrics from the first m times j simulated driving metrics in the sequence of simulated driving metrics to construct a matrix;

[0107] Start traversing the current matrices, and take out a matrix to be screened from the current matrices;

[0108] Determine whether the current matrix to be screened meets the correlation screening condition, where the correlation screening condition is: the absolute value of the correlation coefficient between each index in the matrix is less than a first threshold;

[0109] If not, take out the next matrix to be screened from the current matrices;

[0110] If it meets, after marking the current values of m and j, increase the current value of m by 1, and then determine whether the current m times j value is less than a second threshold;

[0111] When finishing traversing the current matrices and there is no matrix to be screened that meets the correlation screening condition in the current matrices, increase the current value of j by 1, and then determine whether the current m times j value is less than a second threshold;

[0112] If the current m times j value is less than the second threshold, continue to construct matrices according to the current values of m and j; if the current m times j value is not less than the second threshold, then

[0113] Construct matrices according to the most recently marked values of m and j;

[0114] A determination module 330, configured to determine a target matrix among the matrices and determine a target index in the target matrix.

[0115] Other preferred embodiments of the index screening device for studying low-correlation simulated driving metrics provided by this application, the technical problems that can be solved, and the technical effects that can be achieved are the same as those of the above-mentioned index screening method for studying low-correlation simulated driving metrics, and will not be elaborated here.

[0116] The preferred embodiments of the present application have been described in detail above. However, the present application is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present application, various simple modifications can be made to the technical solutions of the present application, and these simple modifications all fall within the protection scope of the present application.

[0117] In addition, it should be noted that, in the case of no conflict, the various specific technical features described in the above specific embodiments can be combined in any appropriate manner. To avoid unnecessary repetition, the present application will not separately describe various possible combination manners.

[0118] Furthermore, any combination can be made between various different embodiments of the present application, as long as it does not violate the idea of the present application, and it should also be regarded as the content disclosed by the present invention.

Claims

1. An index screening method for studying low-correlation simulated driving indexes, characterized in that The method includes the following steps: S100. Obtain a sequence of simulated driving metrics; S200. Sequentially select the simulated driving metrics in the simulated driving metric sequence in order to construct matrices; where m represents the matrix order, j represents a natural number starting from 1, represents a combination of taking m simulated driving metrics from the first m×j simulated driving metrics in the simulated driving metric sequence to construct a matrix; S300. Start traversing the current matrix, and take out a matrix to be screened from the current matrix; S400. Determine whether the current matrix to be screened meets the correlation screening condition, where the correlation screening condition is that the absolute value of the correlation coefficient between each pair of metrics in the matrix is less than a first threshold; If not satisfied, return to step S300 and take out the next matrix to be screened from the current matrix; If it meets the condition, then perform step S410. After marking the current values of m and j, increment the current value of m by 1, and then determine whether the product of the current value of m and the value of j is less than a second threshold; When finishing traversing the current matrix and there is no matrix to be screened in the current matrix that meets the correlation screening condition, then S420. After incrementing the current value of j by 1, determine whether the product of the current value of m and the value of j is less than a second threshold; If the current value of m multiplied by j is less than the second threshold, return to step S200 and continue to construct matrices according to the current values of m and j. If the current value of m multiplied by the value of j is not less than the second threshold, then perform step S500, constructing matrices according to the most recently marked values of m and j; S600. Determine a target matrix among the matrices, and determine a target index in the target matrix.

2. The method according to claim 1, characterized in that The step S100 includes: S110. Generate a simulated driving metric matrix based on the simulated driving metrics and the correlation coefficients between the metrics; S120. In the simulated driving metric matrix, count the number of absolute values of the correlation coefficients between the metrics corresponding to each simulated driving metric that are less than the first threshold, and sort the simulated driving metrics in descending order according to this number to obtain a sequence of simulated driving metrics.

3. The method according to claim 2, wherein Before step S110, the method further includes: S111. Preprocess the simulated driving metrics and the correlation coefficients between the metrics.

4. The method according to claim 1, wherein If the current matrix to be screened does not meet the correlation screening condition, the method further includes: S421. Put the matrix that does not meet the correlation screening condition obtained by screening into an exclusion set.

5. The method according to claim 4, wherein The step S600 includes: S610. Among the matrices, all first target matrices that meet the correlation screening criteria are screened out; S620. Determine a second target matrix with the smallest sum of the absolute values of the correlation coefficients between the metrics among all the first target matrices; S630. Obtain the target metric in the second target matrix.

6. The method according to claim 5, wherein Before the step S610, the method further includes: S611. Exclude the matrices in the exclusion set from the matrices.

7. The method according to any one of claims 1-6, characterized in that The initial value of m is 3.

8. An index screening device for studying low-correlation simulated driving indexes, characterized in that, The device includes: An acquisition module configured to obtain a sequence of simulated driving metrics; A processing module configured to: In the sequence of simulated driving indicators, select the simulated driving indicators in sequence to construct matrices; where m represents the matrix order, j represents a natural number starting from 1, represents the combination of taking m simulated driving indicators from the first m×j simulated driving indicators in the sequence of simulated driving indicators to construct a matrix; Start traversing the current matrix, and take a matrix to be screened out from the current matrix; Determine whether the current matrix to be screened meets the correlation screening condition, where the correlation screening condition is that the absolute value of the correlation coefficient between each pair of metrics in the matrix is less than a first threshold; If not satisfied, take out the next matrix to be screened from the current matrix; If it meets the condition, then mark the current values of m and j, increment the current value of m by 1, and then determine whether the product of the current value of m and the value of j is less than a second threshold; When finishing traversing the current matrix, and there is no matrix to be screened in the current matrix that meets the correlation screening condition, then after adding 1 to the current value of j, determine whether the product of the current m and the value of j is less than the second threshold; If the current product of m and j is less than the second threshold, continue to construct matrices according to the current values of m and j; if the current product of m and j is not less than the second threshold, then Construct according to the most recently marked values of m and j matrices; Determination module, configured to determine a target matrix among the matrices and determine a target index in the target matrix.

Images