An iterative learning control approach to sensor fault-tolerance in Takagi–Sugeno systems

This paper presents a solution to the problem of effective control of a system that is affected by both sensor faults and disturbances (and noises). It is assumed that the model of the system is given in the form of a fuzzy Takagi–Sugeno system. The main goals of the designed control scheme are: achieving a prescribed reference signal at the output, minimizing the impact of disturbances and the ability to respond to faults affecting the system sensor. To achieve the assumed control goals an Iterative Learning Control (ILC) scheme combined with the Fault Tolerant Control (FTC) approach is implemented. Such a combination allows detecting and using information of the faults affecting the system as soon as these are estimated. That in turns speeds up ILC with driving the system to the prescribed reference. Additionally, to determine the estimate of the fault signal and the faulty-free state vector, the observer providing these is designed and implemented. To minimize the impact of disturbances on the estimator, the H∞ methodology is used. The determined estimate of the fault-free state signal is then introduced into the ILC scheme in order to improve its operation in the presence of fault. To determine gains in the feedback loop of the ILC scheme, it is formulated in the form of a Discrete Linear Repetitive Process (DLRP), and then a methodology designed for that subclass of 2D systems is applied to ensure the so-called stability along the trial (which simultaneously means that the underlying ILC scheme tracking error converges to zero and, consequently, the system considered is driven to the requested reference signal). In order to minimize the impact of disturbances and noises on the designed ILC scheme, the H∞ methodology is used again. The obtained results are verified practically in the control process of the two-tank system with the assumed scenarios of emerging faults.