PhD Thesis Presentation
Ride comfort and handling qualities of a vehicle have attained so much attention in the field of automobile. Vehicle manufacturers are trying to introduce as much ride comfort as possible without compromising handling qualities of vehicle. The two characteristics of vehicle are linked with suspension system of vehicle. Good ride comfort means the suspension system does not let road disturbance to reach the body of vehicle, while good handling qualities means vehicle has good road grip and maneuverability. Main objective to improve ride comfort is to elude physical fatigue and tiredness of driver while improvement in handling qualities will minimize chance of road accidents. Different suspension systems like passive, semi-active and active have been studied. Semi-active suspension system can provide change in damping coefficient to fulfill requirements of good suspension system in a simple, easy, without requiring change in vehicle and cost effective way. Therefore, semi-active damper systems have gain more popularity in automobiles field.
Designing control algorithm for semi-active suspension system is a complicated and challenging task. There have been many control algorithms to control damping coefficient, but no algorithm has utilized semi-active dampers to their full potential. This study investigates shortcomings of existing approaches, root cause of the problems and provides solution to the problem in an optimum way. A suspension system involves different regions of operations which can be separated on the bases of operational constraints and stability of system. Existing methods do not use stability for designing their controllers or separating different regions. Furthermore, existing control strategies are based on linearized models and requires gain adjustment and tuning to get better performance. Through analysis, it was found that there are three distinct regions. Damping force requirements for each region are different and critical. If wrong damping force is generated in any region, it may induce more acceleration/disturbance on vehicle, which is the main problem that existing methods suffer. Proposed method generate right amount of force required at right time to suppress vibration by designing better control algorithm based on energy equations and dynamic equations. The proposed methods are validated by using simulations and experiments.
For experimental validation of suspension system and attaining higher performance, accuracy of measurement system and parameters involved cannot be overlooked. Therefore, detailed investigations on accurate measurement system and parameter estimation methodology for suspension system were conducted. Thorough analysis on existing measurement system revealed critical errors in existing measurement system. To find out root cause of problems an analytical formula for evaluation of measurement system was derived. Through detailed analysis root cause of problems and its effects on measurements were highlighted. To overcome measurements issues it is required to use time as scalar base, instead of encoders. This is not possible in existing methods, as sampling time of existing methods is dependent on other associated systems. Therefore, new measurement system is proposed, which uses high frequency clock and high sampling frequency to measure accurate time as soon as change in displacement occurs. This is same as using a stopwatch with a resolution as low as Nano-seconds. Simulations and experiments proved the proposed strategy and its effectiveness. The achieved accuracy from proposed method is comparable to the speed measurements system based on laser interferometer system.
After having accurate measurement system for suspension system, accuracy of systems’ parameters are evaluated from accuracy point of view. The estimation problem is a mathematical problem in which number of independent equations are less than parameters to be estimated, which makes this problem unsolvable and get an analytical solution. Existing approaches use excitation method to generate more relation and convert system to an overdetermined system. By which problem is converted to an optimization problem. Getting overdetermined system is not a solution to the problem as independent equations are still the same. Furthermore, existing methods do not take care of coupling effects, eliminating or isolating noise and parameters from each other. Therefore, parameters estimation is not accurate. This motivated to develop an algorithm to estimate parameters with more accuracy and reliability. The proposed method converts estimation problem into a control problem and the parameters are updated by generating step-by-step update based on state feedback and projecting input on error of states or their functions. Properly designed desired trajectories for different parameters are used to decouple the parameter from other parameters. A systematic way of updating one parameter at a time in different cycles ensured good convergence of parameters to true value even if absolute decoupling between parameters is not possible for complicate and nonlinear systems. The proposed method is validated, for linear and nonlinear problems by simulations and experiments.
(Supervisor: Prof. Lilong Cai)