| Popis: |
In the case of industrial solutions, in the context of Industry 4.0, Artificial Neural Networks (ANNs) currently dominate almost all tasks related to the optimisation of manufacturing processes. For manufacturing enterprises, a quick and precise response to customers needs is crucial for gaining a competitive advantage, therefore, managers should invest in new technologies, such as Additive Manufacturing (AM) technologies. However, these activities are very costly, so it seems that a good solution would be to encourage enterprises to cooperate with specialized research laboratories offering material research in the field of the AM technologies used. This paper proposes a new framework to determine the types and quantities of devices in the field of AM technologies, with which such a research laboratory should be equipped. The proposed AI-based Decision Support System (AI-DSS) integrates the results of a survey of 250 manufacturing companies in western Poland, the use of the Ensemble Neural Network (ENN) Model and the use of a Genetic Algorithm (GA) to determine the profitability of investing in research laboratories, especially in AM technologies. Firstly, in order to reduce variance and thus improve generalization of the proposed model, the combination of outputs of several networks forming ENN was proposed. For this purpose, the bootstrap technique was used, resulting in a model with 75% accuracy being obtained. Next, using the fitness function in a GA algorithm, the research laboratory optimal configuration was determined. Finally, its practicality is presented by applying AI-DSS in the design of the current and future workload of devices, in order to find the optimal configuration in the research laboratory. |
