A Study on Production Decision Making Based on Simulated Annealing Algorithm and Bayesian Optimisation Algorithm

This paper investigates a production decision optimisation model based on heuristic algorithms, using genetic algorithms, simulated annealing algorithms, Monte Carlo algorithms, etc., with the aim of comparing the optimal algorithms to help enterprises make reasonable production decisions. This study is aimed at an electronic product manufacturing enterprise, which involves a number of links such as spare parts procurement, finished product inspection and non-conforming products processing. Firstly, the Bayesian optimisation algorithm is introduced to reduce the number of unnecessary inspections by dynamically adjusting the sample size and combining the historical data, which finally achieves the purpose of inspection with the smallest possible sample size. Secondly, on the basis of cost and benefit analysis, the "exhaustive search" algorithm in the Boolean decision problem is applied to optimise the decision-making in each production stage, encode different decision combinations, and search for the optimal strategy under the complex production environment quickly. Then, the simulated annealing algorithm is introduced to compare the benefits and costs of dismantling semi-finished and finished products, and it is concluded that in some cases dismantling can bring additional benefits to the enterprise, while in other cases, the enterprise should prioritise detection to reduce the subsequent processing costs. Finally, the Monte Carlo simulation method was redesigned using a Monte Carlo simulation approach to enable dynamic adjustment of the detection strategy and a visualisation tool to visualise the optimal detection and treatment options under different conditions. The optimisation algorithms used, such as Bayesian optimisation, genetic algorithm and Monte Carlo simulation, are robust and efficient enough to cope with the uncertainty problems in real production environments.