Operation Parameters Optimisation of a Machine Swarm Using Artificial Intelligence

Due to improper setting of operating parameters, cigarette machines are subject to a high unqualified production rate. For this reason, in this study, a multiobjective optimisation (MOP) method based on the metaheuristic intelligence optimisation is proposed in this study. First, to eliminate interference parameters, the random forest (RF) is used to analyse the parameter importance of the cigarette machine and select the most important operation parameters for the multiobjective optimisation. Second, an artificial neural network (ANN) optimised by the grey wolf optimiser is designed to establish a mirror model of the cigarette machine to fast calculate the machine output quality factors, including the rod break rate, single cigarette weight, and circumference index. Lastly, an improved multiobjective grey wolf optimisation algorithm is used to optimise these three quality factors simultaneously to obtain the optimal operating parameters of the cigarette machine. A machine swarm (including four cigarette machines) in the real world is used to evaluate the developed optimisation method, and the testing results demonstrate that the proposed multiobjective optimisation method is able to improve the three quality factors by at least 50 %, which greatly reduces the unqualified rate of cigarettes.