Explicit MPC-Based RBF Neural Network Controller Design With Discrete-Time Actual Kalman Filter for Semiactive Suspension

Abstract

Many applications require fast control action and efficient constraint handling, such as in aircraft or vehicle control, where instead of the slow online computation of the model predictive control (MPC) the explicit MPC can be an alternative solution. Explicit MPC controllers consist of several affine feedback gains, each of them valid over a polyhedral region of the state space. The exponential blow-up of the number of regions with increasing the prediction horizon increases the searching time among the regions extremely which togetherwith the requirement of the full state measurement decreases its applicability for real systems. First, discrete-time actual Kalman filter is designed for the semiactive suspension and applied to explicit MPC controller that requires only measurement of the suspension deflection. Second, this paper presents a systematic way to design Gaussian radial basis functionbased neural network (NN) approximation of the explicit MPC controller and shows that a well-tuned NN with some neurons can replace the explicit MPC controller. This nonlinear statefeedback controller can ensure the fast control action but price of the approximation is some deterioration of the performance value. The complete novel nonlinear control system with Kalman filter is analyzed in detail. The derived controllers are evaluated through simulations, where shock tests and white noise velocity disturbances are applied to a real quarter car vertical model.

Recognition Method of Drinking-driving Behaviors Based on PCA and RBF Neural Network

“…Driving behavior factors influencing the recognition of drinking and driving are various, and there are complex relationships among the various factors, so the nonlinear fitting capacity of the drinking-driving recognition model should be high. The RBF neural network [9][10][11] is an adaptive dynamic system that is interconnected by many neurons. It is a type of a multi-level neural network that converges faster.…”

Globally robust explicit model predictive control of constrained systems exploiting SVM‐based approximation

“…Remark According to , the upper bound of the number of critical regions q is equal to where N , ι are the prediction horizon and the number of binary variables in the MPC problem, respectively.…”

“…Hence, it is necessary to choose appropriate training samples. Some promising methods are widely utilized such as uniform sampling, Chebyshev center sampling, and grid‐based sampling . In the overlapping regions, the control parameters in Eq.…”

“…But the training error of neural approximation is posterior, and amount of training samples are required to obtain precise approximation with high confidence level. Thus, aiming at reducing the complexity of the training process suffered in , we attempt to propose a novel method to obtain a unified approximate controller with small‐scaled training samples.…”

“…But the training error of neural approximation is posterior, and amount of training samples are required to obtain precise approximation with high confidence level. Thus, aiming at reducing the complexity of the training process suffered in [17], we attempt to propose a novel method to obtain a unified approximate controller with small-scaled training samples.For the second limitation, there are few studies in solving the infeasibility problem. More attention has focused on how to enlarge the feasible state space under state and control constraints.…”