Estimation of Road Roughness Based on Both the Sprung and Unsprung Response of a Moving Vehicle Over Ordinary Roads: Modeling, Experiments and Discussions

The study proposes a Kalman Filter Family-schemed algorithm to identify the vehicle’s parameters and evaluate the road roughness simultaneously, based on the response of an adapted monitoring vehicle. The vehicle is simplified as a three-dimensional multi-body system consisting of the unsprung mass of wheels and the sprung mass of the car body, connected by springs and dampers that represent the suspension system. Accelerometers and a gyroscope measure the accelerations or angular velocity of different positions of the vehicle during vibrations. The Kalman filter (KF) with unknown inputs enables the prediction of the car body’s accelerations from those of wheels. On the other hand, the Augmented KF can compute the accelerations of wheels based on the car body’s accelerations. Thus, an inner-systemic objective function incorporating the vehicle’s parameter is built between the predicted and measured accelerations and solved in an optimized manner with the Genetic Algorithm, so that an individual complicated impact test or bump test is avoided. The numerical examples carry out field tests on a 900[Formula: see text]m long standardized test road under two scenarios of with and without a bump. The road roughness estimated by these two KF methods from different measurements is used to calculate the International Roughness Index (IRI) and further the Riding Quality Index, which are compared with those directly provided by a standardized laser IRI profilometer. Accordance in between confirms the accuracy of the proposed algorithm under different scenarios. The analysis underlines the measurements from unsprung wheels that contact directly with the road enable a more accurate estimation of road roughness than those from the sprung car body. In addition, a vehicle speed lower than 40[Formula: see text]km/h can provide a better estimation of the road profiles.