A method for tire longitudinal slip data processing based on data density and trend residual analysis

Abstract

A method for processing tire longitudinal slip data based on data density and trend residual analysis belongs to the field of tire dynamics modeling and data processing. The method includes the following steps: Step 1, sorting the raw data according to slip ratio; Step 2, standardizing the sorted data and calculating the local density score and trend residual score respectively; Step 3, forming a comprehensive score based on a weighted combination of the local density score and trend residual score, and selecting high-quality data based on the comprehensive score; Step 4, performing a secondary smoothing filter on the selected high-quality data to obtain optimized longitudinal slip data curves. This invention significantly improves the usability and consistency of longitudinal slip data, reduces the interference of abnormal data on identification, and thus improves the fitting effect of semi-empirical models such as Pac and UniTire in longitudinal slip hysteresis and drive sections, thereby improving the tire model identification accuracy and the reliability of vehicle dynamics simulation and control applications.

Description

Technical Field

[0001] This invention relates to the field of tire dynamics modeling and data processing, and in particular to a tire longitudinal slip data processing method based on data density and trend residual analysis. Background Technology

[0002] With the rapid development of the automotive industry, the accuracy of tire dynamics models plays a crucial role in vehicle control strategy formulation, safety performance analysis, and vehicle dynamics simulation. While widely used semi-empirical tire models such as Pac or UniTire can effectively describe the steady-state characteristics of tires, these models cannot accurately describe the hysteresis characteristics generated during longitudinal slip, especially in the driving longitudinal slip region where significant vibrations and anomalies exist. This situation is particularly prominent in the longitudinal slip control and modeling of wider tires. Currently, there is no mature method in the industry to effectively handle this type of problem, severely limiting the accuracy and reliability of model identification. Summary of the Invention

[0003] The purpose of this invention is to solve the problems mentioned in the background art by providing a tire longitudinal slip data processing method based on data density and trend residual analysis, so as to effectively filter out abnormal data and improve the identification accuracy of the model.

[0004] A method for processing tire longitudinal slip data based on data density and trend residual analysis includes the following steps: Step 1: Sort the raw data according to the slip ratio; Step 2: Standardize the sorted data and calculate the local density score and trend residual score of the data respectively; Step 3: Generate a comprehensive score by weighting the local density score and the trend residual score, and then select high-quality data based on the comprehensive score; Step 4: Perform a second smoothing filter on the selected high-quality data to obtain the optimized longitudinal smoothing data curve.

[0005] In step two, the local density score is calculated by giving a higher weight to the longitudinal force direction than to the slip ratio direction.

[0006] In step two, the average distance within a set slip rate window is used in the trend residual score calculation process.

[0007] In step three, the local density score has a weight of 0.8, and the trend residual score has a weight of 0.2.

[0008] The secondary smoothing filtering method in step four is the moving average method (SMA).

[0009] The working process and working principle of this invention: 1. Data Input and Initialization This invention first receives input containing multidimensional data, with particular focus on data columns of slip ratio (Sx) and longitudinal force (Fx). The data is extracted into two variables: x represents the slip ratio, and y represents the longitudinal force. These data are then standardized for subsequent analysis. The purpose of standardization is to eliminate the influence between different units of measurement and ensure that all data are compared under the same standard.

[0010] 2. Smoothing Next, this invention processes the longitudinal force data using a smoothing algorithm. Specifically, a moving average method is used to smooth the data to remove noise and extract trends. The smoothing process extends the longitudinal force (Fx) data by sorting the slip ratio (Sx), thereby obtaining a smoothed longitudinal force curve.

[0011] 3. Local density scoring In this step, the invention calculates a local density score for each data point based on the relationship between the slip ratio (Sx) and the longitudinal force (Fx). The density of the data points is evaluated by calculating the distance between each data point and other data points, and based on a predetermined neighborhood radius. The main purpose of this process is to identify local clustering of data points, i.e., which data points are highly concentrated under similar slip ratios and longitudinal forces.

[0012] 4. Trend Residual Scoring To further evaluate the consistency between data points and their trends, this invention calculates the deviation of each data point from its corresponding smoothed trend curve. By setting a slip ratio window (x_window), trend data within the adjacent range of the current point are selected, and the distances between these points are calculated. A weighted distance evaluation is used to obtain a trend residual score for each point, reflecting the degree of deviation of that point from the trend curve.

[0013] 5. Calculation of total score This invention combines local density scores and trend residual scores to form a total score for each data point. The total score is synthesized using a weighted method, where the density score has a weight of w1 and the trend residual score has a weight of w2. This approach comprehensively considers both the local clustering of data points and their alignment with the overall trend.

[0014] 6. High-scoring region screening Finally, based on the overall score, this invention selects high-scoring data points. By setting a retention ratio, the selected data points with higher scores are chosen to ensure that the selected data points meet the quality requirements and provide high-quality samples for subsequent analysis.

[0015] The beneficial effects of this invention are: The method provided by this invention can significantly reduce data noise and outliers in the driving longitudinal slip region, especially significantly improving data processing accuracy within the ±10% slip ratio range. It significantly improves the fitting effect of Pac or UniTire models on tire longitudinal slip hysteresis characteristics, and helps to improve the accuracy of vehicle dynamics simulation and the effectiveness of control strategies. Attached Figure Description

[0016] Figure 1 This is a flowchart of an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the original data and preliminary filtering curve of tire composite longitudinal slip in an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram of data main trend extraction according to an embodiment of the present invention.

[0019] Figure 4 This is a complete schematic diagram of time-domain low-pass filtering after data screening according to an embodiment of the present invention. Detailed implementation method: Please see Figures 1 to 4 The image shown is an embodiment of the present invention.

[0020] A method for processing tire longitudinal slip data based on data density and trend residual analysis includes the following steps: Step 1: Sort the raw data according to the slip ratio; Step 2: Standardize the sorted data and calculate the local density score and trend residual score respectively. The local density score is calculated by giving a higher weight to the longitudinal force direction than to the slip ratio direction. The trend residual score is calculated by using the average distance within a set slip ratio window. Step 3: A comprehensive score is formed by weighting the local density score and the trend residual score, and high-quality data is selected based on the comprehensive score; the local density score has a weight of 0.8 and the trend residual score has a weight of 0.2 in the comprehensive score. Step 4: Perform a secondary smoothing filter on the selected high-quality data to obtain the optimized longitudinal smoothing curve; the secondary smoothing filter method is the moving average (SMA) method. Data standardization and density score calculation % Parameter settings r = 0.03; % Neighborhood radius in the normalized domain w1 = 0.8; % Density score weights w2 = 0.2; % Smooth residual score weights keep_ratio = 0.97; % Keep the top 70% of high-scoring data points % Disassembly Data x = data(:,12); % Sx slip ratio y = data(:,18); % Fx Longitudinal force n = length(x); % === Standardization === x_mean = mean(x); x_std = std(x); y_mean = mean(y); y_std = std(y); x_norm = (x - x_mean) / x_std; y_norm = (y - y_mean) / y_std; Trend residual score calculation % === Trend Residual Score === % Trend curve coordinates x_fit = x_sort; y_fit = y_smooth; % parameter: Search window in the slip ratio direction x_window = 1.5; % Unit: slip ratio, e.g., ±1.5% % Weighting parameter (optional: the y-direction can be multiplied by the weight wy) wx_Distance = 1.0; % Horizontal dimension is only used for point selection; scores are not multiplied by weight. wy_Distance = 1.0; % Weight for distance evaluation in the y-direction % Initialization d_all = zeros(n, 1); for i = 1:n Step 1: Select points on the smooth curve whose x-axis falls within ±x_window. x_left = x(i) - x_window; x_right = x(i) + x_window; in_window = (x_fit>= x_left)&(x_fit<= x_right); if sum(in_window) == 0 d_all(i) = inf; % No comparable points, set to maximum continue? end Step 2: Select a point on the curve within the window and calculate its y-distance from the current point. y_ref = y_fit(in_window); x_ref = x_fit(in_window); dy = y(i) - y_ref; dx = x(i) - x_ref; Step 3: Evaluate the average distance of the "window segment" (evaluate only the magnitude of the distance, without considering direction). d = sqrt(wx_Distance dx.^2 + wy_Distance dy.^2); d_all(i) = mean(d); % You can also use median(d) or rms(d) end % Trend residual score (reversed after normalization) score_trend = 1 - d_all / max(d_all); Comprehensive scoring and data filtering % ===Total Score & High-Scoring Areas=== total_score = w1 score_density + w2 score_trend; threshold = quantile(total_score, 1 - keep_ratio); index_data = total_score>= threshold; Data smoothing filtering % ===Smoothing Filter Curve=== method = 'moving'; [x_sort, sort_idx] = sort(x); y_sort = y(sort_idx); y_ext = [y_sort(1:span); y_sort; y_sort(end-span:end)]; y_smooth_ext = smooth(y_ext, span, method); y_smooth = y_smooth_ext(span+1:end-span-1); In summary, this invention proposes a tire composite longitudinal slip data processing method based on data density and trend residual analysis. This method characterizes the clustering characteristics of data points through local density scoring and measures the deviation of data points from the main trend curve using trend residuals. Furthermore, a weighted fusion is employed to form a comprehensive score, thereby achieving automatic identification and high-quality data filtering of jitter points and anomalies in the driving longitudinal slip region. Finally, the filtered results undergo secondary shift smoothing to output a longitudinal slip data curve with lower noise and a more stable trend. Through the above process, this invention significantly improves the usability and consistency of longitudinal slip data, reduces the interference of anomalies in identification, and thus improves the fitting effect of semi-empirical models such as Pac and UniTire in longitudinal slip hysteresis and driving sections, enhancing the tire model identification accuracy and the reliability of vehicle dynamics simulation and control applications.

Claims

1. A method for processing tire longitudinal slip data based on data density and trend residual analysis, characterized in that, Includes the following steps: Step 1: Sort the raw data according to the slip ratio; Step 2: Standardize the sorted data and calculate the local density score and trend residual score of the data respectively; Step 3: Generate a comprehensive score by weighting the local density score and the trend residual score, and then select high-quality data based on the comprehensive score; Step 4: Perform a second smoothing filter on the selected high-quality data to obtain the optimized longitudinal smoothing data curve.

2. The tire longitudinal slip data processing method based on data density and trend residual analysis according to claim 1, characterized in that: In step two, the local density score is calculated by giving a higher weight to the longitudinal force direction than to the slip ratio direction.

3. The tire longitudinal slip data processing method based on data density and trend residual analysis according to claim 1, characterized in that: In step two, the average distance within a set slip rate window is used in the trend residual score calculation process.

4. The tire longitudinal slip data processing method based on data density and trend residual analysis according to claim 1, characterized in that: In step three, the local density score has a weight of 0.8, and the trend residual score has a weight of 0.

2.

5. The tire longitudinal slip data processing method based on data density and trend residual analysis according to claim 1, characterized in that: The secondary smoothing filtering method in step four is the moving average method.

Images