Comparison of classical and optimal active suspension control systems
2017-05-25T13:39:29Z (GMT) by
British Rail has been designing active suspensions for some 16 years, starting with tilt systems for the Advanced Passenger train. These have been designed using classical control techniques requiring a combination of experience, intuition and frequency response stability techniques, such as Nichols' plots. In order to see if a more systematic approach to control system design could produce improvements in performance and implementation the current investigation was instigated in which controllers designed using classical techniques are compared with controllers designed using linear optimal control theory. The active suspension used for the investigation was an Electro Magnetic active vertical suspension fitted to a service MkIII coach. Design of the actuators is described in the thesis along with the design of analogue and digital control systems. Two classical control systems were designed. a simple "Sky Hook" damper control system and a more sophisticated position control system developed from British Rail’s experience with Maglev suspensions. A regulator designed using linear optimal control theory was found to give very good results in theory. However to implement the regulator it was necessary to design a system observer. In order to achieve a practically realisable observer considerable rationalisation of the vehicle model was required, which drew heavily on experience gained designing classical control systems. The classical control systems proved to be much easier to commission than the optimal controllers as they were designed with implementation in mind. During track testing problems of interaction between vehicles were encountered, as a result the biggest improvements in ride were obtained with the simple Sky Hook damper, as it was less specific to the vehicle than the other configurations. With further development one of the optimal control systems considered will probably turn out to be the most effective as it draws on the attributes of both classical and optimal design techniques.