6σ uncertainty optimisation strategy for USV navigation performance based on reliability analysis

Abstract

This study presents a Design for Six Sigma (DFSS) optimisation strategy to improve the robustness and reliability of uncertainty parameters, focusing on the problem of navigation performance failure for unmanned surface vehicles (USVs) when uncertainty exists. The three aspects of USV performance are speedability, manoeuvrability, and seakeeping, and its multi-objective mathematical model is accurately constructed through validation. A design of experiment (DOE) technique is conducted to assign the weights using normalised processing. Based on this, a sensitivity analysis is implemented via the optimal Latin hypercube (OLH) method to reveal the correlation relationships between the variables and determine the main factors. A deterministic optimisation design (DOD) is performed using different exploratory optimisation techniques to obtain optimal certainty results. However, the design variables tend to leap across the boundary constraints once they experience a slight interference. Therefore, uncertainty optimisation is conducted to restrict the responses within the constraints. A six sigma (6σ) optimisation based on DFSS is adopted here to improve the reliability of the USV performance parameters. The DFSS results show a high credibility compared to the DOD results, indicating that the proposed 6σ optimisation strategy has a great degree of immunity to uncertainty in actual situations.