بهینه سازی کنترل کننده فازی تطبیقی با استفاده از الگوریتم جدید ترکیب تکاملی و کاربرد آن در سیستم کنترل سیل لاگوس در جنوب نیومکزیکو

In fuzzy logic controllers (FLCs), optimal performance can be defined as performance that minimizes the deviation (error term) between the decisions of the fuzzy logic systems and the decisions of experts. A range of approaches – such as genetic algorithms (GA), particle swarm optimization (PSO), artificial neural networks (ANN), and adaptive network based fuzzy inference systems (ANFIS) – can be used to pursue optimal performance for FLCs by refining the membership function parameters (MFPs) that control performance. Multiple studies have been conducted to refine MFPs and improve the performance of fuzzy logic systems through the application of a single optimization approach, but since different optimization approaches yield different error terms under different scenarios, the use of a single optimization approach does not necessarily produce truly optimal results. Therefore, this study employed several optimization approaches – ANFIS, GA, and PSO – within a defined search engine unit that compared the error values from the different approaches under different scenarios and, in each scenario, selected the results that had the minimum error value. Additionally, appropriate initial variables for the optimization process were introduced through the Takagi–Sugeno method. This system was applied to a case study of the Diez Lagos (DL) flood controlling system in southern New Mexico, and we found that it had lower average error terms than a single optimization approach in monitoring a flood control gate and pump across a range of scenarios. Overall, using evolutionary algorithms in a novel search engine led to superior performance, using the Takagi–Sugeno method led to near-optimum initial values for the MFPs, and developing a feedback monitoring system consistently led to reliable operating rules. Therefore, we recommend the use of different methods in the search engine unit for finding the optimal MFPs, and selecting the MFPs from the method which has the lowest error value among them.