Integrating an artificial neural network with genetic, particle swarm optimization algorithms, and fuzzy rule to optimize the surface roughness in dry turning AISI 1045 steel

Abstract A strategic plan for manufacturing systems in the fourth industrial revolution requires smart manufacturing to enhance the performance of manufacturing technologies. Consequently, the machine learning technique is one of the major techniques to meet these requirements. This research aims to propose optimization methods including a hybrid artificial neural network (ANN) with genetic algorithms (GA) particle swarm optimization (PSO), and Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict and optimize surface roughness during dry machining of AISI 1045 steel. To achieve this experimentally investigates and statistical analyses on how different cutting parameters affect the performance of machining AISI 1045 steel were performed. Experimental trials were conducted using a full factorial design, and an analysis of variance (ANOVA) was used to determine the impact of each variable on the outcomes of the process. ANN-GA, ANFIS, and ANN-PSO techniques were used to develop a prediction model of surface roughness. In addition, The ANN-GA, ANFIS, and ANN-PSO algorithms are used for determining the optimal process parameters for the turning of AISI 1045 steel. The results show that ANFIS, ANN-PSO algorithm, and ANN-GA presented a high-accuracy model for predicting surface roughness. In addition, the ANFIS model demonstrates superior performance compared to the ANN-PSO algorithm, and ANN- GA. Ra of 0.202 μm was achieved by using tool-type wiper carbide inserts, a speed of 80 m min −1 , a depth of cut of 0.5 mm, and a feed rate of 0.045 mm rev −1 as the input process parameters with a relative error of less than 3%.