Research on obtaining dynamic, robust fault diagnosis rules

This paper is about fault diagnosis rules. Firstly, a redundancy-based rule model is presented, compared with the reduced rule model without redundancy by Rough Set Theory (RST). This model has a robusticity to resist the data loss by sensor failure, as well as the inaccuracy data by sensor error. Furthermore, considering the constraint of fault diagnosis cost, a dynamic optimization method on the redundancy-based rule model is proposed. The principle of the dynamic optimization model is to maximize the robusticity of rule model under cost constraint. An approach using Genetic Algorithm (GA) is expressed to execute the optimization. Finally, case study on hydraulic pump of civil aeroplane is presented to demonstrate the utility of the proposed model.