In this paper, we implemented a network that predicts the vibration value of the driver’s seat, which is the vibration value of riding comfort. The input feature used to predict the ride comfort transfer function uses a total of 15 factors, x, y, an...
In this paper, we implemented a network that predicts the vibration value of the driver’s seat, which is the vibration value of riding comfort. The input feature used to predict the ride comfort transfer function uses a total of 15 factors, x, y, and z axes, respectively, for the vibration values of the four wheels and the engine vibration. The target value is the vibration value of the x and z axis of the driver’s seat. Through the following characteristic factors, we could experimentally find out that prediction through learning is possible even with the method using the deep learning technique, which is beyond the existing analysis method through mechanical modeling. In addition, network evaluation and verification were performed based on the degree of similarity of the comparison graph between the MSE and the actual result at the time of prediction, and the network with a deep design of the MLP showed the best performance through experiments on various network configurations.