Automobile Suspension Prediction Model Based on Neural Network and Grey Neural Network

Abstract

Vehicle Vibration is a major issue in automobile industry which hinders the stability of the vehicle and also adds to the discomfort levels of the passengers. The proposed work models a suspension prediction system using a combination of both Neural Network and Grey Neural Network for more efficient predictions. The research was divided into three stages. First stage is to develop a hardware to acquire the vibrational acceleration data of the test vehicle at different suspension types and road types. The data acquired was then used to train the designed neural model which was the stage two of the project. The final stage consisted of testing the prediction model by acquiring dynamic data with unknown parameters including the road type and the suspension tune settings. The research is effectively efficient to predict suspension tune settings for a vehicle traveling on dynamic road conditions with very little mean absolute percentage error. The training, learning, and the testing of the prediction model was done on a real-time system. The real-time vibration prediction model allows easy suspension tunings and improved drive experience for both the driver and the passenger. The incorporation of machine learning allows the reduction of MAPE thus, producing highly efficient results.