Intelligent driving automobile dynamic characteristic similarity evaluation method and system

Abstract

The invention discloses an intelligent driving automobile dynamic characteristic similarity evaluation method and system, and the method comprises the steps: obtaining the performance index time series data of a real automobile and a simulation model in the same intelligent driving scene, carrying out the preprocessing of an original sequence and a simulation sequence of the real automobile, obtaining a sequence of the real automobile and the simulation model, and carrying out the calculation of the real automobile. The method comprises the following steps: calculating an overall difference distance based on a sequence with equal step lengths of a real vehicle and a simulation model, obtaining the similarity of vehicle dynamic characteristics based on the obtained overall difference distance, determining an effective comparison interval of the sequence to realize complete comparison of the vehicle dynamic characteristic sequence, and then constructing an equal-step-length sequence for a non-uniform sampling problem of the real vehicle; according to the method, the shape similarity and time sequence similarity evaluation methods are coupled, accurate description of the overall shape difference and the phase-amplitude microcosmic difference between the sequences is achieved, a global amplitude difference factor is introduced, a novel grey relational degree model considering and fusing multi-dimensional index generalization comparison is constructed, and fine quantitative description of the dynamic characteristics of the intelligent driving automobile is achieved.

Description

Technical Field

[0001] This invention belongs to the field of credibility evaluation of intelligent driving vehicle simulation, specifically involving a method and system for evaluating the similarity of dynamic characteristics of intelligent driving vehicles. Background Technology

[0002] A scientific and reasonable testing and evaluation system is crucial for the large-scale application and promotion of intelligent driving vehicles. High-reliability simulation testing, with its advantages of controllable environment, reproducible process, and scalable scenario generation, has become a key pillar for the R&D and access testing of intelligent driving vehicles. Intelligent driving simulation testing systems consist of multiple interdependent subsystems, including vehicle dynamics models, sensor models, and scenario models. High-precision model construction and accurate quantitative evaluation of reliability are currently the focus of research in simulation testing and evaluation. In particular, the construction and reliability evaluation of highly nonlinear, multi-dimensionally coupled, and time-varying vehicle dynamics models present difficulties in quantifying the similarity of temporal features. Therefore, a method for finely quantifying the similarity of vehicle dynamics features is urgently needed.

[0003] Refined quantification of temporal feature similarity is crucial for evaluating the credibility of vehicle dynamics models. Currently, domestic and international scholars primarily focus on research in areas such as battery anomaly detection, collision dynamics, and agricultural machinery dynamics, employing methods like Euclidean distance, Dynamic Time Warping (DTW), and grey relational analysis for multi-dimensional comprehensive evaluation. While relatively mature similarity evaluation methods have been developed in these fields, related research in vehicle dynamics remains scarce, with only a few exploratory works. For example, the geometric similarity of temporal features in vehicle dynamics can be characterized by the ratio similarity of sequence similarity elements, but this method only evaluates sequence differences from a geometric perspective and cannot accurately capture the differences between sampling times in the sequences. Another type of method uses statistical indicators such as root mean square error and mean absolute percentage error to quantify differences in dynamic features, but it is extremely sensitive to temporal misalignment or phase differences, and cannot characterize the similarity of vehicle dynamics sequences of unequal length or local differences at key peaks. The two methods mentioned above only focus on geometric similarity or statistical error, resulting in a single evaluation dimension. Furthermore, existing lockstep metrics such as Euclidean distance cannot characterize the similarity of sequences of unequal length and the local differences at key peaks. The DTW algorithm cannot accurately capture the differences between sampling times of sequences, and traditional grey relational analysis methods are only applicable to evaluating temporal alignment features. Due to the significant differences in the dynamic characteristics of intelligent driving vehicles under different operating conditions, their similarity evaluation suffers from problems such as temporal shift, unequal sequence length, and generalization comparison of multi-dimensional indicators. Existing single-metric methods are difficult to apply, and there is a lack of multi-dimensional, refined quantitative evaluation methods that combine generalization and stability. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for evaluating the similarity of dynamic characteristics of intelligent driving vehicles, so as to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for evaluating the similarity of dynamic features of intelligent driving vehicles includes the following steps: S1, acquire time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; S2, preprocess the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model respectively; S3, calculates the overall difference distance based on the sequence where the step size of the real vehicle and the simulation model are equal; S4, obtain the similarity of vehicle dynamic features based on the acquired overall difference distance.

[0006] Preferably, the original sequence of the real vehicle and the simulation sequence are preprocessed, specifically including using the mean filtering method to smooth and reduce the noise of the time series data of the performance indicators collected from the real vehicle, so as to obtain the noise-reduced baseline sequence.

[0007] Preferably, the original sequence of the actual vehicle and the simulation sequence are preprocessed to obtain sequences with equal step sizes for the actual vehicle and the simulation model, specifically including: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model; Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes; The overall difference distance is calculated based on shape similarity and sequence temporal similarity.

[0008] Preferably, the DTW algorithm is used to measure the overall shape difference between the two sequences. The optimal matching path for the vehicle dynamics sequence is found through dynamic search. (3) In the formula, As a shape similarity index; and This represents the number of resampling points between the baseline and comparison sequences; Represents a sequence and The Middle and the The distance between points in time.

[0009] Preferably, the starting point of the two sequences is zeroed out, the comparison sequences are translated as a whole so that their left endpoints coincide, and then the right endpoints of the two sequences are connected to form a closed polyline. The enclosed region is discretized into a series of trapezoidal infinitesimal elements using the normalized step size. The area is calculated segment by segment and accumulated to obtain the integral value of the total area. Finally, the integral value is compared with... The ratio is used as a time series similarity index, as shown in equations (4) and (5): (4) (5) In the formula, It is a time series similarity index; This represents the maximum number of resampling points for the two sequences; This is the resampling time.

[0010] Preferably, the shape similarity index Similarity index with time series The weighted average yields the overall difference distance of the sequence. As shown in equation (6): (6) In the formula, This represents the number of path connections used in the DTW method. , For the similarity index weight, .

[0011] Preferably, the similarity of vehicle dynamic features is obtained based on the acquired overall difference distance: Introducing a global amplitude difference factor Combined with overall difference distance The new grey relational degree is calculated. , This is the similarity score between sequences, as shown in equations (7) and (8): (7) - (8) In the formula, For resolution coefficients, This is the global amplitude difference factor; As a reference sequence In the The value at time; For comparison sequences In the The value at any given moment.

[0012] A similarity evaluation system for the dynamic features of intelligent driving vehicles includes a data acquisition module, a data preprocessing module, a difference module, and a similarity module; The data acquisition module acquires time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; The data preprocessing module preprocesses the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model, respectively. The difference module calculates the overall difference distance based on a sequence where the step size of the real vehicle and the simulation model are equal. The similarity module obtains the similarity of vehicle dynamic features based on the acquired overall difference distance.

[0013] Preferably, the original sequence of the real vehicle and the simulation sequence are preprocessed, specifically including using the mean filtering method to smooth and reduce the noise of the time series data of the performance indicators collected from the real vehicle, so as to obtain the noise-reduced baseline sequence.

[0014] Preferably, the original sequence of the actual vehicle and the simulation sequence are preprocessed to obtain sequences with equal step sizes for the actual vehicle and the simulation model, specifically including: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model; Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes; The overall difference distance is calculated based on shape similarity and sequence temporal similarity.

[0015] Compared with the prior art, the present invention has the following beneficial technical effects: A method for evaluating the similarity of dynamic features of intelligent driving vehicles is proposed. This method acquires time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario. The original sequences of the real vehicles and simulation sequences are preprocessed to obtain sequences with equal step lengths for both. The overall difference distance is calculated based on these equal-step sequences, and the similarity of the vehicle's dynamic features is obtained based on this overall difference distance. By determining the effective comparison interval of the sequences, a complete comparison of the vehicle's dynamic feature sequences is achieved. Subsequently, to address the non-uniform sampling problem of the real vehicles, equal-step sequences are constructed. A method coupling shape similarity and temporal similarity evaluation is used to accurately characterize the overall shape difference and phase-amplitude micro-differences between sequences. A global amplitude difference factor is introduced, and a novel grey relational model considering the generalization comparison of multi-dimensional indicators is constructed to achieve a refined quantitative characterization of the dynamic features of intelligent driving vehicles. Attached Figure Description

[0016] This application can be better understood by referring to the description given below in conjunction with the accompanying drawings, which, together with the detailed description below, are incorporated in and form part of this specification. In the drawings: Figure 1 This is a flowchart of the method for evaluating the similarity of dynamic features of intelligent driving vehicles in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the similarity evaluation test of the dynamic characteristics of intelligent driving vehicles in an embodiment of the present invention. Figure 2 In this context, 'a' represents the result of a real-vehicle test. Figure 2 In this context, 'b' represents the simulation test result.

[0018] Figure 3 This is a schematic diagram illustrating the calculation of sequence difference distance in an embodiment of the present invention. Figure 3 In the diagram, 'a' represents a schematic representation of shape similarity calculation. Figure 3 In the diagram, b represents a schematic representation of temporal similarity calculation.

[0019] Figure 4 This is a schematic diagram comparing the results of sequence translation changes and grey relational degree changes in an embodiment of the present invention. Figure 4 In the figure, 'a' represents a comparison of sequence translation changes. Figure 4 In the figure, b is a comparison chart of the grey relational results. Detailed Implementation

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

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] like Figure 1 As shown, this invention provides a method for evaluating the similarity of dynamic features of intelligent driving vehicles, used to achieve a fine quantitative characterization of the dynamic features of intelligent driving vehicles, specifically including the following steps: S1, acquire time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; S2, preprocess the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model respectively; S3, calculates the overall difference distance based on the sequence where the step size of the real vehicle and the simulation model are equal; S4, obtain the similarity of vehicle dynamic features based on the acquired overall difference distance.

[0023] In a specific embodiment of the present invention, the original sequence of the real vehicle and the simulation sequence are preprocessed, specifically including denoising the time series data of the performance indicators of the real vehicle (i.e., the original sequence), specifically using the mean filtering method to smooth and denoise the time series data of the performance indicators collected from the real vehicle, to obtain the denoised baseline sequence.

[0024] To select effective intervals, a comparison was made between the noise-reduced baseline sequence and the simulation sequence. For the two original time series, to ensure the completeness of the comparison of the vehicle response process, an adaptive truncation method based on sliding window and cubic spline fitting was used to determine the boundaries of the effective intervals. The simulation sequence is sequence data obtained from the simulation model under the same intelligent driving scenario. Firstly, within the... A cubic spline fit is performed within a sliding window of sampling points, and the slope of each sampling point on the fitted curve is calculated. When determining the left boundary, the absolute value of the slope of the first point in the window and each sampling point to its right is compared with a preset threshold. If all values ​​are greater than the threshold, the first point of the window is taken as the left endpoint of the sequence; otherwise, the entire window is shifted one position to the right to continue the judgment. For the right boundary, the absolute value of the slope of the first point in the window and each sampling point to its left is compared. If all values ​​are greater than the threshold, the last point of the window is taken as the right endpoint of the sequence; otherwise, the window slides from right to left until the criterion is met. During local fitting, the relative error is used... To constrain and ensure that the accuracy of the fitted curve meets the requirements, the error calculation is shown in equations (1) and (2).

[0025] (1) (2) In the formula, This represents the number of sampling points in the sliding window. For the first in the sliding window Fitting error for each sampling point; This represents the number of sampling points in the original sequence. For the data sequence number True values ​​of each sampling point; These are the fitted values; These are the sampling points for the sliding window.

[0026] In a specific embodiment of the present invention, to address the issue of sequence timing inconsistency caused by non-uniform sampling of vehicle sensors, the reference sequence and the comparison sequence are resampled using linear interpolation within their start and end ranges, with a step size of [missing information]. The equally spaced sequences achieve the unification of time coordinates.

[0027] In a specific embodiment of the present invention, the original sequence of the actual vehicle and the simulation sequence are preprocessed to obtain sequences with equal step sizes for the actual vehicle and the simulation model, specifically including: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model. Specifically: The DTW algorithm is used to measure the overall shape difference between two sequences. By dynamically searching for the optimal matching path of the vehicle dynamics sequence, problems such as inconsistent length between sequences and time alignment deviation are solved, as shown in equation (3).

[0028] (3) In the formula, As a shape similarity index; and This represents the number of resampling points between the baseline and comparison sequences; Represents a sequence and The Middle and the The distance between points in time.

[0029] Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes, specifically including: Introducing time series similarity index To quantify the phase-amplitude micro-bias between two sequences, the starting points of both sequences are first zeroed out. The comparison sequence is then shifted as a whole so that its left endpoints coincide. A closed polygonal line is then formed by connecting the right endpoints of the two sequences. Using a normalized step size, the enclosed region is discretized into a series of trapezoidal infinitesimal elements. The area of ​​each element is calculated segment by segment and accumulated to obtain the integral value of the total area. Finally, the integral value is compared with... The ratio is used as a temporal similarity index, as shown in equations (4) and (5).

[0030] (4) (5) In the formula, It is a time series similarity index; This represents the maximum number of resampling points for the two sequences; This is the resampling time.

[0031] The overall difference distance is calculated based on shape similarity and sequence temporal similarity, specifically including: To comprehensively consider the morphological differences and temporal shifts between the two sequences, the shape similarity index is used. Similarity index with time series The weighted average yields the overall difference distance of the sequence. As shown in equation (6).

[0032] (6) In the formula, This represents the number of path connections used in the DTW method. , For the similarity index weight, .

[0033] In a specific embodiment of the present invention, the similarity of vehicle dynamics features is obtained based on the acquired overall difference distance. To facilitate multi-dimensional index generalization comparison, a novel grey relational analysis method for evaluating vehicle dynamics features is proposed. A global amplitude difference factor is introduced. To ensure that the reference frame is consistent before and after the sequence translation, and to take into account the overall difference distance. The new grey relational degree is calculated. , This is the similarity score between sequences, as shown in equations (7) and (8).

[0034] (7) - (8) In the formula, For resolution coefficients, This is the global amplitude difference factor; As a reference sequence In the The value at time; For comparison sequences In the The value at any given moment.

[0035] Example Time-series data of performance indicators for real vehicles and simulation models under the same typical intelligent driving scenarios were obtained. Typical intelligent driving scenarios include seven scenarios: deceleration and stopping, acceleration and start-up, low-speed turning (30km / h), medium-speed turning (60km / h), high-speed turning (100km / h), U-turn, and lane change. Eight performance indicator sequences—longitudinal velocity, longitudinal acceleration, lateral acceleration, roll angle, pitch angle, roll rate, pitch rate, and yaw rate—were collected from both real vehicles and simulation models using sensors and simulation software. Based on the seven typical intelligent driving scenarios, eight performance indicators characterizing vehicle motion were selected, and corresponding performance indicators were chosen for the longitudinal or lateral control dynamic driving tasks involved in different scenarios, as shown in Table 1.

[0036] Table 1 Output Performance Indicators

[0037] Mean filtering was used to preprocess the real vehicle data, with a filter window length of 11 sampling points, to obtain the denoised baseline sequence. For each original sequence, a filter was then applied to the data containing... Perform cubic spline fitting on the sampled curve within a sliding window of 100 sampling points, and calculate the slope of the fitted curve at each sampling point. Taking the selection of the left boundary as an example, the specific implementation is as follows: Let the input be a sequence. Number of sampling points in the sliding window Fitting error threshold Slope threshold Output: .

[0038] Sliding window traversal initialization: Set the starting position of the sliding window The traversal range is from 1 to (in For sequence (total length), sequentially for each Perform the following steps; Cubic spline fitting: for the current sliding window W (the window contains sequences) From the position arrive of Each sampling point, i.e. Perform a cubic spline fitting operation to obtain the fitted curve. ( ); Fitting error calculation and adjustment: Calculate the fitting residual for each sampling point within the calculation window. ( This represents the index of the sampling point within the window, with a value ranging from 1 to... The relative fitting error is calculated based on the residuals. ; If relative error Greater than the preset fitting error threshold Then adjust the cubic spline fitting parameters and re-execute the above fitting operation, repeating this process until... ; Slope calculation and determination: Calculate the position of the point in the window ( (This is the floor function). Based on the adjusted fitted curve, the calculation window is used to determine the position. arrive The slope of the fitted curve corresponding to all sampling points ; Judgment: If all the above slopes The absolute values ​​of all slopes are greater than or equal to the preset slope threshold. And all If the signs are completely identical, then the starting position of the current sliding window will be changed. Determined as the left boundary of the sequence And terminate all traversal operations; Output: Returns the final determined left boundary of the sequence. . After determining the effective comparison intervals for the baseline and comparison sequences, linear interpolation resampling is performed within their start and end ranges to construct the step size. An equally spaced sequence with a time interval of 0.01s.

[0039] Sequence Difference Distance Calculation Module: This embodiment uses the DTW algorithm to calculate the shape similarity index. The time series similarity index was calculated using the normalized area method. and shape similarity index Similarity index with time series The weighted average yields the overall difference distance of the sequence. .

[0040]

[0041] In the formula, This represents the number of path connections used in the DTW method. , For the similarity index weight, In this embodiment, the value is 0.5.

[0042] Similarity Calculation Module: To achieve generalized comparison of multi-dimensional indicators, this embodiment proposes a novel grey relational analysis method for evaluating vehicle dynamics features, introducing a global amplitude difference factor. To ensure that the reference frame is consistent before and after the sequence translation, and to take into account the overall difference distance. The new grey relational degree is calculated. As shown in the following formula.

[0043]

[0044] In the formula, For resolution coefficients, in this embodiment ; This is the global amplitude difference factor.

[0045] Figure 2 A schematic diagram of an evaluation test for the similarity of dynamic characteristics of intelligent driving vehicles is shown. Figure 2 In this context, 'a' represents a real vehicle test. Figure 2 In this example, 'b' represents the simulation test. The Hongqi E-HS3 was selected as the test vehicle. Real-vehicle performance time-series data was collected using a Vbox device at a frequency of 100Hz. A high-fidelity dynamic model with over 90% prediction accuracy of key vehicle characteristic indicators was established using Carsim software. Vehicle performance time-series data was collected with a simulation step size of 100Hz.

[0046] Before the test, the Vbox device was fixed to the center of the rear floor of the vehicle, turned on, and calibrated. After the test began, the driver activated the vehicle's intelligent driving function, and the vehicle autonomously completed lane-changing maneuvers, while the Vbox device simultaneously recorded the vehicle's performance time-series data. Subsequently, the same driving trajectory and accelerator pedal opening as in the real-vehicle test were input into the Carsim software, and the vehicle's performance time-series data was recorded.

[0047] Figure 3 A schematic diagram of sequence difference distance calculation is shown. This embodiment uses the time series data of real vehicle and simulated yaw rate under the same lane-changing scenario as the benchmark sequence and comparison sequence, respectively, to demonstrate the technical solution of this application. Figure 3 In the diagram, 'a' represents a schematic diagram of shape similarity calculation. The DTW algorithm is used to dynamically search for the optimal matching path of the sequence, and the matching path connection is shown. Figure 3 b in the figure is a schematic diagram of time series similarity calculation, showing the enclosed area composed of the left endpoint starting point zeroing, the right endpoint closed polyline and a series of trapezoidal infinitesimal elements.

[0048] Figure 4 A schematic diagram comparing the results of sequence translation changes and grey relational degree changes is shown. Compared with the traditional Dönbrunn correlation coefficient, the method in this embodiment ensures that the correlation coefficient changes with the overall difference distance of the sequence. The increasing and monotonically decreasing characteristic makes the model more stable in the face of sequence shifts and delays. Since the true similarity value cannot be obtained when performing similarity evaluation, it is difficult to quantitatively compare the performance of different methods based on the true error. To demonstrate the superiority of this embodiment in the case of sequence shifts, two sequences are randomly selected, the benchmark sequence is fixed, and the comparison sequence is shifted point by point from left to right. The novel correlation model considering only shape similarity is calculated for each sequence. A novel association model that only considers temporal similarity ( A novel association model considering shape-temporal similarity ( The correlation results between the traditional Dun correlation model and the traditional Dun correlation model are shown. The results indicate that the traditional Dun correlation method shows increased similarity when the sequence undergoes a small shift, which is illogical. The improved method considering only temporal similarity shows a significant decrease in similarity after a shift, then tends to plateau. The improved method considering only shape similarity remains unchanged after a shift, indicating that shape similarity results are not affected by temporal shifts. The shape-temporal similarity evaluation method falls between the first two methods, maintaining sensitivity to sequence differences while also exhibiting good stability.

[0049] In a specific embodiment of the present invention, an intelligent driving vehicle dynamics feature similarity evaluation system is provided, including a data acquisition module, a data preprocessing module, a difference module and a similarity module; The data acquisition module acquires time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; The data preprocessing module preprocesses the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model, respectively. The difference module calculates the overall difference distance based on a sequence where the step size of the real vehicle and the simulation model are equal. The similarity module obtains the similarity of vehicle dynamic features based on the acquired overall difference distance.

[0050] Preferably, the original sequence of the real vehicle and the simulation sequence are preprocessed, specifically including using the mean filtering method to smooth and reduce the noise of the time series data of the performance indicators collected from the real vehicle, so as to obtain the noise-reduced baseline sequence.

[0051] Preferably, the original sequence of the actual vehicle and the simulation sequence are preprocessed to obtain sequences with equal step sizes for the actual vehicle and the simulation model, specifically including: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model; Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes; The overall difference distance is calculated based on shape similarity and sequence temporal similarity.

[0052] In another embodiment of the present invention, a terminal device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used for the operation of an intelligent driving vehicle dynamics feature similarity evaluation method.

[0053] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the intelligent driving vehicle dynamics feature similarity evaluation method in the above embodiments.

[0054] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0055] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0056] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0057] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for evaluating the similarity of dynamic characteristics of intelligent driving vehicles, characterized in that, Includes the following steps: S1, acquire time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; S2, preprocess the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model respectively; S3, calculates the overall difference distance based on the sequence where the step size of the real vehicle and the simulation model are equal; S4, obtain the similarity of vehicle dynamic features based on the acquired overall difference distance.

2. The method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 1, characterized in that, Preprocessing is performed on the original sequence of the real vehicle and the simulation sequence. Specifically, the mean filtering method is used to smooth and reduce the noise of the time series data of the performance indicators collected from the real vehicle to obtain the noise-reduced baseline sequence.

3. The method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 2, characterized in that, Preprocessing is performed on the original sequence of the actual vehicle and the simulation sequence to obtain sequences with equal step sizes for both the actual vehicle and the simulation model. Specifically, this includes: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model; Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes; The overall difference distance is calculated based on shape similarity and sequence temporal similarity.

4. The method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 3, characterized in that, The DTW algorithm is used to measure the overall shape difference between two sequences. The optimal matching path for the vehicle dynamics sequence is found through dynamic search. （3） In the formula, As a shape similarity index; and This represents the number of resampling points between the baseline and comparison sequences; Represents a sequence and The Middle and the The distance between points in time.

5. A method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 4, characterized in that, The two sequences are initialized to zero at their starting points. The comparison sequences are then shifted as a whole so that their left endpoints coincide. A closed polygonal line is formed by connecting the right endpoints of the two sequences. The enclosed region is discretized into a series of trapezoidal infinitesimal elements using a normalized step size. The area of ​​each element is calculated segment by segment and accumulated to obtain the integral value of the total area. Finally, the integral value is compared with... The ratio is used as a time series similarity index, as shown in equations (4) and (5): （4） （5） In the formula, It is a time series similarity index; This represents the maximum number of resampling points for the two sequences; This is the resampling time.

6. A method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 5, characterized in that, Shape similarity index Similarity index with time series The weighted average yields the overall difference distance of the sequence. As shown in equation (6): （6） In the formula, This represents the number of path connections used in the DTW method. , For the similarity index weight, .

7. The method for evaluating the similarity of dynamic features of intelligent driving vehicles according to claim 1, characterized in that, The similarity of vehicle dynamic features is obtained based on the acquired overall difference distance: Introducing a global amplitude difference factor Combined with overall difference distance The new grey relational degree is calculated. , This is the similarity score between sequences, as shown in equations (7) and (8): （7） - （8） In the formula, For resolution coefficients, This is the global amplitude difference factor; As a reference sequence In the The value at time; For comparison sequences In the The value at any given moment.

8. A similarity evaluation system for the dynamic characteristics of intelligent driving vehicles, characterized in that, It includes a data acquisition module, a data preprocessing module, a difference module, and a similarity module; The data acquisition module acquires time-series data of performance indicators of real vehicles and simulation models under the same intelligent driving scenario; The data preprocessing module preprocesses the original sequence of the real vehicle and the simulation sequence to obtain sequences with equal step sizes for the real vehicle and the simulation model, respectively. The difference module calculates the overall difference distance based on a sequence where the step size of the real vehicle and the simulation model are equal. The similarity module obtains the similarity of vehicle dynamic features based on the acquired overall difference distance.

9. The intelligent driving vehicle dynamics feature similarity evaluation system according to claim 8, characterized in that, Preprocessing is performed on the original sequence of the real vehicle and the simulation sequence. Specifically, the mean filtering method is used to smooth and reduce the noise of the time series data of the performance indicators collected from the real vehicle to obtain the noise-reduced baseline sequence.

10. A similarity evaluation system for the dynamic characteristics of intelligent driving vehicles according to claim 9, characterized in that, Preprocessing is performed on the original sequence of the actual vehicle and the simulation sequence to obtain sequences with equal step sizes for both the actual vehicle and the simulation model. Specifically, this includes: Shape similarity is calculated based on sequences with equal step lengths between the real vehicle and the simulation model; Based on the temporal similarity of sequences between the real vehicle and the simulation model with equal step sizes; The overall difference distance is calculated based on shape similarity and sequence temporal similarity.

Images