Optimization of the Signal Noise Ratio index using Simultaneous Perturbation Stochastic Approximation Algorithm.

This paper proposes a linear search Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm to maximize the signal-to-noise ratio (SNR) index in the least number of iterations and determine which succession measure converges and maximizes the signal-to-noise ratio (SNR) in the least number of iterations. The analysis and validation are performed with experimental simulation and validated with four case studies collected from the literature. The case studies were evaluated in ten experiments with different combinations of the succession measures a_k and c_k. The results show that the proposed SPSA is an iterative, efficient, and easy to use method to maximize the quality indexes of production processes, which is feasible to implement within the six sigma methodology. Also, the results show that experiments 3 and 5 converge to the best results in the four case studies analyzed. [ABSTRACT FROM AUTHOR]