Study On Real-time Estimation Of Road Grade Using OpenXC Data

The accurate and real-time knowledge of road slope is an important factor for supporting the vehicle dynamic control and driving decision while running on the road.In recent years,the vehicle information collection platform based on CAN bus can collect various driving data,which provides a new way to get the real-time road slope.However,it is difficult to establish the relationship of the road slope and driving data because of the uncertain dynamic noise of the driving circumstance and the braking state of driving operation.And the existing researches have not the corresponding solutions which consider about the above factors.Therefore,it is necessary to study on the real-time road slope estimation method more accurately and comprehensively under different driving operations and conditions,which has theoretical and practical significance for improving vehicle assistant driving control,stability control,safety and energy-saving control.The OpenXC platform which introduced by Ford can collect various real-time driving data.Based on the OpenXC data,a fuzzy adaptive extended Kalman filter is proposed to estimate the road slope aiming at the problem of estimation error and divergence due to the uncertain dynamic noise of the driving circumstance.On the other hand,it is difficult to realize road slope estimation through theory modeling during braking condition.To solve this problem,the auto-regressive prediction is adopted for slope estimation in short distance which extend the range of slope estimation application.Research work of this paper includes the following three aspects:Firstly,On the basis of the comprehensive driving state data from the OpenXC and considering some uncertain disturbance in complex driving circumstances,this paper propose real-time road slope estimation method with the fuzzy adaptive extended Kalman filter(FAEKF)algorithm from vehicle longitudinal dynamic model.And the simulations are conducted to verify the algorithm through CarSim platform.The results are in good agreement with the accurate estimation results and the anti-interfere performance for the unknown time-varying noises.Secondly,aiming at the problem that theory modeling with vehicle longitudinal dynamic is unsuited for braking conditions,we analyze the characteristic of OpenXC data and brake conditions,on this basis the auto-regressive prediction is proposed,which make use of the historical slope estimation to predict the slope information in short distance.By combining auto-regressive prediction algorithm with FAEKF algorithm to estimate road slope in comprehensive driving conditions,thus overcoming the big estimation errors when using single FAEKF algorithm in the braking condition,and extending the application of slope estimation method.Thirdly,considering that the small range of road slope variation,the accuracy of estimation result is of great importance.The analysis shows that there are two main factors which greatly affect estimation result,which are the outlier data and vehicle load.And the detection and correction methods based on the sliding time window are applied to process the outlier data.At the same time,the compensation of the slope deviation will be done for the change of the vehicle load,after the sensitivity analysis of their relation.In summary,by the consideration of OpenXC data characteristics and driving conditions,this paper presents a road slope estimation method which is suit for dynamic driving circumstance and comprehensive driving conditions.Both of the simulation and real driving experiment verify that the proposed method is effective,which can provide support and reference for the control and services of the vehicle driving.