Road adhesion coefficient estimation by multi-sensors with LM-MMSOFNN algorithm

Accurate and efficient road adhesion coefficient estimation is the premise for proper functioning of vehicle active safety control system. With increased application distributed drive vehicles on-board sensors, a multi-module self-organizing feedforward neural network (LM-MMSOFNN) based on improved Levenberg-Marquardt (LM) learning algorithm proposed online estimation. In this method, dynamics model Dugoff tire were well established, input output variables obtained by Principal Component Analysis (PCA) method. To improve accuracy, Extended Kalman Filter (EKF) Moving Average (MA) used to denoise measured signal. On basis, was established. Both sides coefficients are calculated simultaneously. Through increase decrease neurons LM algorithm, computational complexity system hardware storage reduced, exhibits good adaptability different roads. Simulation experiments show that method can fully extract multi-sensor information adapt characteristics changes under driving condition. As compared with Kmeans it has higher accuracy stronger varying speed.