Compressive sensing–based optimal sensor placement and fault diagnosis for multi-station assembly processes

Developments in sensing technologies have created the opportunity to diagnose the process faults in multi-station assembly processes by analyzing measurement data. Sufficient diagnosability for process faults is a challenging issue, as the sensors cannot be excessively used. Therefore, there have been a number of methods reported in the literature for the optimization of the diagnosability of a diagnostic method for a given sensor cost, thus allowing the identification of process faults incurred in multi-station assembly processes. However, most of these methods assume that the number of sensors is more than that of the process errors. Unfortunately, this assumption may not hold in many real industrial applications. Thus, the diagnostic methods have to solve underdetermined linear equations. In order to address this issue, we propose an optimal sensor placement method by devising a new diagnosability criterion based on compressive sensing theory, which is able to handle underdetermined linear equations. Our method seeks the optimal sensor placement by minimizing the average mutual coherence to maximize the diagnosability. The proposed method is demonstrated and validated through case studies from actual industrial applications.