Error handling and controller design for controller area network-based networked control system

Abstract

Networked Control System (NCS) is a feedback control system which dynamic process is running via the communication channel. Surrounded by many choices of network types that can be used to establish an NCS, Controller Area Network (CAN) is a popular choice widely used in most real-time applications. Under harsh environment, fault at transmission line for CAN-based NCS is more prominent compared to fault in network nodes. Fault in bus line of CAN will induce data error which will result in data dropout or/and time delay which consequently lead to performance degradation or system instability. In this thesis, strategies to handle fault occurrence in CAN bus are proposed in order to properly analyse the effect of fault to CAN-based NCS performance. To implement the strategies, first, fault occurrences are modelled based on fault inter-arrival time, fault bursts duration and Poisson law. By using fault and message attributes, Response Time Analysis (RTA) is performed and the probability of NCS message that misses its deadline is calculated based on Homogeneous Poisson Process (HPP). A new error handling algorithm per-sample-error-counter (PSeC) is introduced to replace native error handling of CAN. PSeC mechanism is designed based on online monitoring and counting of erroneous sensor and control signal data at every sampling instance and it gives a bound parameters known as Maximum Allowable Number of Data Retransmission (MADR). If the number of retransmission for NCS message violates the value of MADR, the data will be discarded. With the utilization of PSeC mechanism to replace the Native Error Handling (NEH) of CAN, the probability of NCS message that misses its deadline can be translated to the probability of data dropout of NCS message. Despite the PSeC has prevented network from congestion which can lead to prolonged loop delay, it also introduces one-step loop delay and data dropout. Therefore, the controller that is able to compensate the effect of delay and data dropout should be introduced. Thus, a control algorithm is designed based on Lyapunov stability theory formulated in Linear Matrix Inequality (LMI) form by taking into account network delay and data dropout probability. In order to proof the efficacy of the strategies, Steer-by-Wire (SbW) system is used and simulated in TrueTime MATLAB R /Simulink environment. Simulation results show that the strategies of introducing PSeC mechanism and the designed controller in this work have superior performance than NEH mechanism for CAN-based NCS environment in terms of integral of the absolute error (IAE) and energy consumption.