Steering function safety controllability grading evaluation method based on simulation and clustering analysis

Abstract

The invention relates to the technical field of automobile chassis function safety, and discloses a steering function safety controllability grading evaluation method based on simulation and clustering analysis, and the method comprises the following steps: S1, defining vehicle parameters based on Carsim; s2, building a driver model; s3, a fault injection module is set up in the Simulink; s4, a vehicle operation scene is built, joint simulation is carried out based on Carsim and Simulink, and lateral deviation data is obtained; s5, controlling the driver model to control vehicle running states in different failure forms, traversing different vehicle initial states and different driver models, and running joint simulation to obtain multiple groups of test data; s6, extracting target data from the test data; s7, performing normalization processing on the target data; and S8, performing clustering analysis, and classifying data clusters formed by clustering to obtain different controllability levels. According to the method, the controllability can be comprehensively evaluated, the evaluation result is reliable, objective and accurate, and the method has actual reference value.

Description

Technical Field

[0001] The present invention relates to the technical field of automotive chassis functional safety, and particularly relates to a method for grading and evaluating the controllability of steering functional safety based on simulation and cluster analysis. Background Art

[0002] With the increasing integration of automotive electronics, failures of electronic and electrical systems occur frequently, and the status of functional safety has become increasingly prominent. Especially for systems such as automotive steering that are closely related to safety, their safety directly concerns the lives of passengers and road traffic safety. As an important part of functional safety, Hazard Analysis and Risk Assessment (HARA) identifies hazards and evaluates exposure, severity, and controllability. Among them, there are some quantifiable and highly operable specific bases for the evaluation of exposure and severity (such as standards like SAE J2980 and VDA 702). However, in the past, the evaluation of controllability mostly relied on subjective judgment methods such as expert experience. This method has obvious problems of lack of operability and large randomness, and it is difficult to meet the increasingly precise requirements for functional safety assessment.

[0003] From the perspective of the current development status, in terms of the evaluation of controllability, in addition to the conventional evaluation method relying on expert experience, existing technologies mostly use time to collision or FTTI for evaluation (such as a device and method for automatically rating the steering system S and C based on functional safety provided by the patent with the publication number CN115795686A, and a method for simulating vehicle dangerous behaviors based on quantitative analysis of functional safety HARA provided by the publication number CN115983001A). However, this method has many limitations. Firstly, the consideration dimension of the existing solutions is limited to a single collision impact factor, but collisions cannot fully cover out-of-control scenarios. Secondly, the existing solutions do not consider scenario impact factors. For example, in high-speed scenarios on highways and low-speed scenarios in parking lots, even if the time to collision is the same, the possible levels of harm are very different. In addition, there is a lack of basis for how time to collision or FTTI corresponds to a certain controllability level. Although the existing standard document ISO 26262 for functional safety can be used as the main basis for controllability evaluation, this standard is relatively general and only proposes a methodology, lacking specific guiding rules at the practical operation level. Summary of the Invention

[0004] The present invention aims to provide a method for grading and evaluating the controllability of steering functional safety based on simulation and cluster analysis, which can comprehensively evaluate the controllability, and the evaluation results are reliable, objective, and accurate, having practical reference value.

[0005] The basic solution provided by the present invention is: A method for grading and evaluating the controllability of steering functional safety based on simulation and cluster analysis, comprising the following steps:

[0006] S1. Define vehicle parameters based on Carsim;

[0007] S2. Build a driver model; the driver model includes a driver model based on torque control, and a post-failure operation module is provided in the driver model based on torque control; the post-failure operation module is used to input the operation modes of different drivers;

[0008] S3. Build a fault injection module in Simulink for the potential failure modes of the functional object;

[0009] S4. Build a vehicle operation scenario according to the potential failure modes of the functional object, and perform co-simulation based on Carsim and Simulink to obtain lateral offset data;

[0010] S5. Control the driver model to control the vehicle operation state under different failure modes, traverse different vehicle initial states and different driver models, and perform co-simulation to obtain multiple groups of test data;

[0011] S6. Extract the target data required for the controllability rating clustering analysis from the test data;

[0012] S7. Perform normalization processing on the target data;

[0013] S8. Use a clustering algorithm to perform clustering analysis on the normalized target data, and classify the formed data clusters to obtain different controllability levels.

[0014] The working principle and advantages of the present invention are as follows:

[0015] This solution first defines vehicle parameters based on Carsim, a vehicle dynamics simulation software, and builds a driver model based on torque control. Moreover, this model takes into account the operation modes corresponding to different drivers' physiological and psychological factors (corresponding to post-failure operation modes) and load limits (corresponding to torque control). By defining vehicle operation scenarios, vehicle operation states, and setting driver models under different hand torque upper limits, post-failure operation modes, etc. to replace different drivers, and through co-simulation of Carsim and Simulink, test data under different operation scenarios and different driver models can be obtained, and then clustering is performed using a clustering algorithm. Furthermore, in a machine learning manner, the controllability classification can be completed based on the data.

[0016] The method for evaluating the controllability rating of steering functional safety based on simulation and clustering analysis of the present invention can comprehensively evaluate the controllability, and the evaluation results are reliable, objective, and accurate, having practical reference value. The key points are as follows:

[0017] This solution proposes an evaluation method that can be accurately quantified, with objective and intuitive evaluation. Compared with the existing subjective evaluation methods for functional safety controllability, which rely on human subjective judgment and experience, usually cannot provide specific numerical indicators, are difficult to conduct reliable comparison and analysis, and limit the accurate assessment of system safety. There are also very few related patents that mainly provide the collision time as an evaluation index, without considering influencing factors such as drivers and different scenarios, and the mapping relationship between time and the controllability level lacks a source basis. Although it can quantify the controllability level, this evaluation method is actually not accurate and has a low reliability. The controllability quantification method designed in this solution does not have the above problems and expands the consideration dimension of influencing factors, achieving controllability quantification with objective basis.

[0018] Among them, this solution specifically designs a driver model based on torque control, integrating the operation methods and load limits corresponding to different drivers' physiological and psychological factors, and breakthroughly introduces the driver as an influencing factor into the controllability assessment. In the existing solutions, the effect of the driver in vehicle functional safety is often ignored. In actual driving scenarios, due to the different hand force ranges of drivers of different genders, different preferences of drivers for choosing to brake or steer first when encountering danger, etc., the existence of these differential characteristics will directly affect the controllability of the vehicle. This solution discovers this problem and conducts targeted design, which also makes the evaluation process of this solution closer to the actual driving situation. Based on this driver model, different driver influencing factors (configured with different upper limits of driver hand torque and different operation methods) can be set, and then the operation of the vehicle in various situations can be more accurately simulated, which helps to improve the reliability and authenticity of the assessment.

[0019] Secondly, when obtaining test data, this solution traverses different vehicle initial states and different driver models, and adopts the combined simulation of Carsim and Simulink, which can effectively combine the advantages of the two software, more accurately simulate the dynamic behavior of the vehicle and the control logic of the steering system, and can also introduce scene influencing factors, covering diverse scenarios during the traversal process, and then obtaining more reliable and richer test data, providing a solid data basis for the subsequent controllability assessment.

[0020] In addition, this solution introduces a clustering analysis method to map the controllability evaluation. By using a clustering algorithm to perform clustering analysis on the normalized target data, where the normalization process can eliminate the interference of different scenarios and drivers on the controllability evaluation; the clustering algorithm can classify the target data and map the controllability level based on evidence. Moreover, this data-driven method can extract valuable information from a large amount of target data, automatically classify the controllability according to the characteristics of the data, reduce the interference of human factors, improve the objectivity and accuracy of the evaluation, and effectively reduce the dependence on expert experience. At the same time, the machine learning algorithm has certain self-adaptability and learning ability, and can optimize the classification result of the controllability as the data accumulates and updates continuously, ensuring that the controllability evaluation is more objective and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 FIG. is a schematic flowchart of an embodiment of the method for grading and evaluating the controllability of the steering functional safety based on simulation and clustering analysis of the present invention;

[0022] Figure 2 FIG. is a schematic diagram of the application process of the method of an embodiment of the method for grading and evaluating the controllability of the steering functional safety based on simulation and clustering analysis of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The following is a more detailed description through specific embodiments:

[0024] The embodiment is basically as shown in the attached Figure 1 and Figure 2 The method for grading and evaluating the controllability of the steering functional safety based on simulation and clustering analysis includes the following steps:

[0025] S1. Define vehicle parameters based on Carsim. Specifically, the vehicle parameters include vehicle model, drive form, sensors, etc.

[0026] S2. Build a driver model. The driver model includes a driver model based on torque control.

[0027] Specifically, in the driver model interface provided by Carsim, select the driver model of open-loop torque control, and here the torque is selected as Zero torque, that is, no torque control is performed.

[0028] In Simulink, develop a driver model based on torque control.

[0029] The driver model based on torque control is provided with a post-fault operation module, a longitudinal control module, and a front-wheel steering angle conversion torque module. The post-fault operation module is used to input the operation modes of different drivers. The longitudinal control module is used to control the throttle opening and pedal of the vehicle for longitudinal speed control; the front-wheel steering angle conversion torque module is used to calculate the real-time front-wheel steering angle of the vehicle and convert it into real-time torque.

[0030] Further, in Simulink, different upper limits of hand torque and different post-fault operation modes are set for the driver model based on torque control when the driver steers. In this embodiment, the operation modes of the driver include continuing driving operation and deceleration operation. The setting range of the upper limit of the driver's hand torque is 8 - 15 N·m.

[0031] S3. For the potential failure modes of the functional object, a fault injection module is built in Simulink.

[0032] The potential failure modes of the functional object include excessive assistance, insufficient assistance, loss of assistance, and unexpected assistance for the electric power steering system; when building the fault injection module, the following sub-steps are included:

[0033] Close the EPS assistance characteristic curve built in Carsim, externalize it to Simulink through a look-up table module, and then generate a step signal through Matlab.

[0034] The step signal is adjusted according to the failure mode and the failure occurrence time; for the failure mode of loss of assistance, the step signal jumps from 1 to 0; for the failure mode of insufficient assistance, the step signal jumps from 1 to 0.5; for the failure mode of excessive assistance, the step signal jumps from 1 to 2; for the failure mode of unexpected assistance, the step signal jumps from 1 to -1. The moment of generating the step signal is consistent with the failure occurrence time. For example, if the failure (fault) occurs at the 5th s of the vehicle operation, the step signal is generated at the 5th s.

[0035] S4. According to the potential failure modes of the functional object, a vehicle operation scenario is built, and based on the joint simulation of Carsim and Simulink, lateral offset data is obtained.

[0036] When building the vehicle operation scenario, the vehicle operation scenario is set as double lane change. And the initial vehicle speed is set, and a fault is injected at the 5th s of the vehicle operation during the simulation process.

[0037] S5. Control the driver model to control the vehicle operation state under different failure modes, and through the script in Matlab, traverse different vehicle initial states and different driver models, and run the joint simulation to obtain multiple groups of test data.

[0038] The test data includes driver failure modes, driver hand torque, lateral offset, initial vehicle driving speed, and operation mode after failure.

[0039] S6. Extract the target data required for the controllability rating clustering analysis from the test data.

[0040] In this embodiment, the target data includes lateral offset, driver hand force, and initial vehicle driving speed under the double lane change test condition.

[0041] S7. Perform normalization processing on the target data, including:

[0042] err lat_norm (i) = abs(err lat (i) / 3.5); where err lat (i) represents the lateral offset of the i-th data;

[0043] T driver_norm (i) = abs(T driver (i)) / max(abs(T driver (i))); where T drivrr (i) represents the driver hand force of the i-th data;

[0044] V init_norm (i) = (V init (i)-40) / 6; where V init (i) represents the initial vehicle driving speed of the i-th data.

[0045] S8. Use a clustering algorithm to perform clustering analysis on the normalized target data, and classify the data clusters formed by clustering to obtain different controllability levels.

[0046] When performing clustering analysis, use the K-Means clustering algorithm model for clustering - use the L2 distance to calculate the distance between the feature vectors corresponding to two types of data points, and divide the data points with closer distances into one cluster. Finally, the best effect is achieved when the distance between the clusters is the largest. Among them, the k value of the K-Means clustering algorithm model is set to 4, and the model is trained until it converges.

[0047] The L2 distance calculation formula is as follows: L2 = sqrt(a 2 +b 2 ). Where a and b are two different feature vectors.

[0048] Specifically, the steps for obtaining the controllability level include:

[0049] Step1, set the number of clustering categories to 4, corresponding to the controllability levels C0, C1, C2, and C3, and randomly initialize 4 center points;

[0050] Step2, calculate the distances from each data point to the 4 center points respectively, and assign each data point to the center class that is closer to this data point;

[0051] Step3, recalculate the center points of the 4 categories;

[0052] Step4, repeat Step2 and Step3 until the change of the center point of each category after each iteration is less than the threshold;

[0053] Step5, refer to the definitions of C0, C1, C2, and C3, and classify the controllability levels of the four clusters formed by clustering.

[0054] Among them, the controllability gradually decreases from C0 to C3, that is, it becomes more and more uncontrollable.

[0055] This embodiment also provides a steering functional safety controllability rating evaluation system, including:

[0056] A memory for storing computer-executable instructions and data used or produced when executing the computer-executable instructions;

[0057] A processor communicatively connected to the memory and configured to execute the computer-executable instructions stored in the memory;

[0058] When the computer-executable instructions are executed, they implement a steering functional safety controllability rating evaluation method based on simulation and clustering analysis as described above.

[0059] A steering functional safety controllability rating evaluation method based on simulation and clustering analysis provided by this embodiment can comprehensively evaluate the controllability, and the evaluation results are reliable, objective, and accurate, having practical reference value.

[0060] The above are only embodiments of the present invention. Specific structures and characteristics and other common knowledge in the art are not described in detail herein. Those of ordinary skill in the art know all the common general technical knowledge in the technical field to which the invention pertains before the filing date or the priority date, are able to know all the prior art in this field, and have the ability to apply conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, combine their own abilities to improve and implement this solution. Some typical well-known structures or well-known methods should not become an obstacle for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent.

Claims

1. A steering function safety controllability grading assessment method based on simulation and cluster analysis, characterized in that: The following steps are involved: S1. Define vehicle parameters based on Carsim; S2. Building a driver model; the driver model includes a driver model based on torque control, wherein the driver model based on torque control is provided with a post-fault operation module; The post-fault operation module is used to input the operation modes of different drivers; S3. Build a fault injection module in Simulink for the potential failure forms of the functional object; S4. According to the potential failure forms of the functional objects, the vehicle operation scenario is built, and the lateral offset data is obtained by joint simulation based on Carsim and Simulink; S5. Control the driver model to control the vehicle operating state under different failure modes, traverse different vehicle initial states and different driver models, and run joint simulation to obtain multiple sets of test data; S6. Extracting target data required for controllability rating cluster analysis from the test data; S7, normalizing the target data; S8. Use a clustering algorithm to perform cluster analysis on the normalized target data, and classify the data clusters formed by clustering to obtain different controllability levels.

2. The steering function safety controllability grading evaluation method based on simulation and cluster analysis according to claim 1 is characterized in that: The driver's operation modes include continuing driving operation and decelerating operation.

3. The steering function safety controllability grading evaluation method based on simulation and cluster analysis according to claim 1 is characterized in that: In the driver model based on torque control, the setting range of the upper limit of the driver's hand torque is 8 to 15 N·m.

4. The steering function safety controllability grading evaluation method based on simulation and cluster analysis according to claim 1 is characterized in that: The torque control-based driver model also includes a longitudinal control module and a front wheel angle conversion torque module; the longitudinal control module is used to control the throttle opening and pedal of the vehicle to perform longitudinal speed control; the front wheel angle conversion torque module is used to calculate the real-time vehicle front wheel angle and convert it into real-time torque.

5. The steering function safety controllability grading evaluation method based on simulation and cluster analysis according to claim 1 is characterized in that: The potential failure forms of the functional object include excessive power assistance, insufficient power assistance, power assistance loss, and unexpected power assistance. When building a fault injection module, the following sub-steps are included: Close the built-in EPS power assist characteristic curve in Carsim, externalize it to Simulink through the table lookup module, and then generate a step signal through Matlab; The step signal is adjusted according to the failure mode and the time when the failure occurs; for the failure mode of loss of power assistance, the step signal jumps from 1 to 0; for the failure mode of too little power assistance, the step signal jumps from 1 to 0.5; for the failure mode of too much power assistance, the step signal jumps from 1 to 2; for the failure mode of unexpected power assistance, the step signal jumps from 1 to -1; the time when the step signal is generated coincides with the time when the failure occurs.

6. The method for evaluating the safety controllability of steering functions based on simulation and cluster analysis according to claim 1, characterized in that: When building a vehicle operation scenario, set the vehicle operation scenario to double lane change.

7. The method for evaluating the safety controllability of steering functions based on simulation and cluster analysis according to claim 6 is characterized in that: The test data includes driver failure mode, driver hand torque, lateral deviation, vehicle initial speed, and post-fault operation mode; the target data includes lateral deviation, driver hand force, and vehicle initial speed under double lane change test conditions.

8. The method for evaluating the safety controllability of steering functions based on simulation and cluster analysis according to claim 7 is characterized in that: The normalization processing of the target data comprises: err lat_norm (i) = abs(err lat (i) / 3.5); where err lat (i) represents the lateral offset of the i-th data; T driver_norm (i) = abs(T driver (i)) / max(abs(T driver (i))); where T driver (i) represents the driver's hand force of the i-th data; V init_norm (i)=(V init (i)-40) / 6; where V init (i) represents the initial speed of the vehicle at the i-th data point.

9. The method for evaluating the safety controllability of steering functions based on simulation and cluster analysis according to claim 1, characterized in that: When performing cluster analysis, the K-Means clustering algorithm model is used for clustering. The L2 distance is used to calculate the distance between the feature vectors corresponding to two types of data points. Those with closer distances are divided into one cluster. The best effect is achieved when the distance between clusters is the largest. The k value of the K-Means clustering algorithm model is set to 4, and the model is trained until it converges.

10. The steering function safety controllability grading and evaluation method based on simulation and cluster analysis according to claim 9, characterized in that: In S8, the step of obtaining the controllability level includes: Step 1, set the cluster category to 4, corresponding to the controllability levels C0, C1, C2, and C3, and randomly initialize 4 center points; Step 2: Calculate the distance between each data point and the four center points respectively, and assign each data point to the center class that is closest to the data point; Step 3, recalculate the center points of the four categories; Step 4, repeat Step 2 and Step 3 until the center point of each class changes less than the threshold after each iteration; Step 5, refer to the definitions of C0, C1, C2, and C3, and classify the controllability levels of the four clusters formed by clustering.

Images