Optimization-based online estimation of vehicle mass and road grade: Theoretical analysis and experimental validation

The gross vehicle mass (GVM) and the road grade are two factors that both have a substantial influence on performance of vehicle’s powertrain. In this paper, we propose novel model-based estimation method for GVM exploits entire sequences powertrain measurements at once is formulated as nonlinear program (NLP). estimator based simple model longitudinal dynamics with only few intuitive parameters. By assuming to remain constant during certain sections trip by describing profile in distance domain, achieve separation scales, which enhances disturbance rejection significantly lowers number optimization variables. resulting thoroughly analyzed analytically numerically. We show closed-form solution exists function GVM. Furthermore, if can be assumed journey considered, problem translated scalar NLP finding Although rigorous proof missing, our experiments practice, objective quasi-convex reasonable interval values thus unique exists. robustness sensitivity studies conducted, where various perturbations considered controlled environment, including uncorrelated correlated noise, sensor offset, mismatches. Compared well-known recursive filters described literature, shows higher respect all perturbations. Finally, validate real data from an electric city bus. proposed outperforms algorithms achieves average relative error 3.4%. On standard personal computer, driving phase around one hour solved roughly 7.5 s, while representing 75 s 12 ms. Both results indicate real-time applicability algorithm.