Evaluation and prediction method of automotive electronic accelerator pedal based on support vector regression

In the process of automobile electronic accelerator pedal development, it is a critical and challenging issue to evaluate the rationality and comfort of the design of an automotive electronic accelerator pedal. Many factors influence the comfort of the accelerator pedal, such as the spatial layout, dynamic characteristics, and matching characteristics of the accelerator pedal and vehicle motion. Since comfort evaluation requires a lot of manpower and material resources, this paper proposes a prediction model based on support vector machine regression algorithm (SVR) for comprehensive evaluation of Chinese passenger car pedals. It uses the known evaluation results to predict the unknown evaluated accelerator pedal parameters to achieve a more efficient and accurate assessment of electronic accelerator pedal design. Firstly, the article performs pedal position scans, pedal static, and road tests to give criteria, limitations, and recommended design ranges for pedal operation. Then, the vehicle performance was predicted and evaluated using a support vector machine prediction model and back propagation (BP) neural network prediction model for comparison. The correlation coefficient for the prediction results of the SVR model was 0.9024 with a mean square error was 0.00195. The correlation coefficient for the BP neural network model prediction result was 0.8694 with a mean square error of 0.00582. Finally, the simulation results were analyzed, and the results showed that support vector regression outperformed the neural network in predicting the validity and reliability of pedal design and performance evaluation, and can facilitate automotive pedal design and development.