Integrated Model-Based Control Allocation Strategies Oriented to Predictive Maintenance of Saturated Actuators

Predictive Maintenance approaches are gaining popularity in the new Industry 4.0 paradigm as they offer superior benefits in terms of time and money savings when it is required to assess the current working capabilities of operating equipment to carefully schedule maintenance operations. This work deals with a control allocation strategy inspired by Model Predictive Control ideas and able to address the loss of effectiveness of actuating equipment arising from their continuous usage. The scheme here presented comprises two modules: a prognostic unit for monitoring the reliability conditions of the actuators and a re-configurable control allocation block that operates according to the deterioration degree of the present actuators. The benefits of the proposed approach are testified by the numerical simulations carried out on both an unstable system and a tanks network. In particular, it can be observed that the proposed method is capable of suggesting a time-window for maintenance interventions that prevents either stability or feasibility issues. Note to Practitioners—Nowadays, in the viewpoint of financial and technical issues in the industries, the predictive maintenance (PM) plays the main and important subject. In the variety of industries, such as oil, gas, petrochemical, power plant, or transmission and distribution infrastructures, which are using the actuators and pipelines, predictive maintenance (PM) is a critical topic to prevent the unwanted shutdown and unrequired maintenance costs. However, most PM research and solutions have paid attention mainly to the estimation of the Remaining Useful Life (RUL) of critical devices (i.e., pipelines) only. On the contrary, in this work we focus on the possibility of influencing the RUL by properly acting on the actuators’ effort. The proposed method consists in a dual-mode fault-tolerant control allocation technique developed for discrete-time systems subject to input saturation. The proposed predictive model-based control solution can be considered as a simple, structural and practical approach to minimize the maintenance cost and reliability by decreasing the number of maintenance interventions and optimizing the time for their repairing, by preventing or decreasing the probability of occurrence of unwanted failures events. The main limitation in this approach is that it assumes the RUL of the critical assets under investigation known. As a future research, this work will be extended to include a related RUL estimator hinging on online data.