Estimation of vehicle inertial parameters

University of Technology, Sydney. Faculty of Engineering.NO FULL TEXT AVAILABLE. Access is restricted indefinitely. The hardcopy may be available for consultation at the UTS Library.NO FULL TEXT AVAILABLE. Access is restricted indefinitely. ----- The inertial properties of all vehicles vary over time due to changes in load and load distribution. These changes in inertial properties affect the vehicles' dynamic performance, including resistance to rollover, yaw and directional stability, natural frequencies, damping ratios, vertical acceleration and passenger comfort levels and maximum available rates of longitudinal acceleration and deceleration. These changes can also undermine the effectiveness of a vehicle's active control systems, such as traction control and electronic stability programs. These vehicle controllers are optimised around a nominal set of vehicle inertial parameters. The inevitable changes to each vehicle's inertial parameters is likely to result in a control model that differs to the physical system. Ensuring that the vehicle's control systems operate using the actual, rather than nominal inertial characteristics, will improve performance levels with less control effort and faster response times. Treating the vehicle as a multi-body system and combining this with modal analysis theory, a methodology has been developed that can identify vehicle mass, centre of gravity locations and mass moments of inertia. Common, and in some cases existing, on-board sensors such as accelerometers and gyroscopes are used, which minimise cost and installation complexity. The method that will be presented in this thesis does not require the input forces to be known; this differs from the state of the art alternatives which require the uncertain and varying input forces such as aerodynamic drag, road friction, road grade or suspension displacement to be measured. This requires additional sensors such as GPS, tyre force transducers, displacement or pressure sensors. The reduction in the need for multiple sensors of different kinds yields both cost and performance benefits. The sensors used are also non-contact, so long-term reliability is improved. The response of the sprung mass when the vehicle encounters a series of random road disturbances is measured using accelerometers or gyroscopes, and provides all of the data required. From this data the vehicles' free decay response at each sensor location is extracted. The state variable method is then used to find the state transition matrix. From the state transition matrix an eigen analysis is performed to extract the natural frequencies, damping ratios and mode shapes. Some of these detected modes will not be those of the sprung mass, so a filtering algorithm is used to detect and reject false noise modes. From this data the system characteristic matrix can be formulated. A simplified model of the vehicle is then used to find the vehicle inertial properties. It is assumed that the wheel base, track width and equivalent spring stiffness values are known. The values of the vehicle dampers are not required. Using a least squares analysis the values of mass, mass moment of inertia and centre of gravity are varied and the characteristic matrix found. When the error has been minimised between the measured and estimated characteristic matrices the best estimate of the mass, mass moment of inertia and centre of gravity have been found. Simulation and experimental results indicate that the inertial parameters can be estimated within acceptable error levels