Cumulative diagnosis strategy for predictive maintenance decision support

We propose a new diagnosis strategy, here called “cumulative diagnosis”, for advanced decision support to predictive maintenance. It is based on the cumulative damage principle and the use the degradation laws of the considered components. The main objectives of this strategy include the reduction of the cost of diagnoses per time unit and the improvement of the systems' availability. The strategy requires establishing and composing three models: resources allocation to the diagnosis tasks under exclusiveness constraint; diagnosis tasks scheduling under precedence constraints; and a dynamic model of tasks' planning in real-time over periodic, a-periodic and stochastic time windows. The obtained models are integrated to support the predictive maintenance decisions. The new diagnosis strategy has several advantages and its performances may be appreciated through the experimental results of evaluation.