(A) study for the control of passenger car suspension for ride comfort

Abstract

The objective of this study is to develop a control logic of suspension for the ride comfort in a passenger car. The current trend of the functional requirement of the road vehicles is not confined to transportation but includes the ride comfort and stability. Advancement in the design and development of the automotive sensor systems and controllers makes it possible for the vehicles to be equipped with active or semi-active suspensions. Especially, the controllers using preview control logic, neural network and/or artificial intelligence control begin to be implemented in commercial vehicles. The incorporation of a frequency-dependent weighting matrix allows us to emphasize or de-emphasize the variables related to the vibration at the specific band of frequencies. So a control logic is derived using a frequency and time mixed shape performance index with the control forces divided into feedforward and feedback ones. The performance index is constructed to improve the variables related with human comfort in the frequency domain and reduce the variables such as tire deflections and rattle spaces in the time domain. Here, a speed and road-adaptive preview logic is developed based on the general road shapes. This preview logic is an approximate to Riccati equation and gives better solution than non-preview logic for those road shapes tested. Because the amount of calculation for the feedforward control force is massive, it is difficult to implement in real time. The road data is stored in a shift register and multiplied by the feedforward gain matrix. In this procedure, two methods can be applied : one is the sequential calculation and the other is the paralell calculation. The latter is adequate for realization and the neuro-controller can perform this. The input patterns for teaching the neural network are the road data set stored in the shift register and the output patterns are the feedforward control force calculated from the optimal controller. In the design...