Research on Dynamic Planning for Production Decision-Making Based on Support Vector Machines and Neural Networks

Under the competitive environment of globalisation and digitalisation, enterprises need to respond quickly to changes in market demand, in which production decision optimisation not only empowers the modern enterprise manufacturing industry, but is also an important means of improving enterprise production efficiency. In the actual production process, enterprises need to consider whether to test spare parts, the follow-up treatment of unqualified products, and the rationalisation of the production process, efficiency issues. Therefore, the design of excellent production decisionmaking programme becomes extremely important in the production process of enterprises. In this paper, support vector machine (SVM) models are applied on the basis of traditional statistical models. By training these algorithms to identify the fluctuation pattern of the defective rate, the sampling strategy can be dynamically optimised during the sampling and inspection process to reduce the number of unnecessary inspections, and at the same time improve the accuracy of the defective rate prediction. In this paper, a multi-stage decision optimisation model is constructed to perform real-time optimisation based on the actual data of spare parts inspection, product assembly and defective products processing, and to make the optimal balancing decision between the cost of expenses and sales revenue. In this paper, we use neural networks to predict the defective rate and the final finished product benefit of multi-process and multi-parts using deep learning. It helps enterprises to identify key risk factors in the production process in advance, so as to customise a more reasonable profit-optimal strategy.