A study on optimization of delayed production mode of iron and steel enterprises based on data mining

Abstract

Delayed production mode has been adopted by an increasing number of process production enterprises as a method to realize mass customization of multi-products. This paper used the convolutional neural network-long short-term memory artificial neural network algorithm (C-LSTM) in data mining technology to analyze and determine factors that have an impact on delayed production mode in the internal and external production and operation of enterprises. Combined with the actual production situation of iron and steel enterprises, a quantitative model of the delayed production was constructed. Lastly, data from a large iron and steel enterprise with good operation was used to verify the validity of the proposed model and analyze key influencing factors. According to the research, in scenarios of considering PDP alone, considering CODP alone, considering both PDP and CODP, considering PDP and CODP and using data mining technology to model, the matching degree of these methods with the actual situation of the enterprise is 31.8%, 61.4%, 71.6% and 86.6%, respectively. The numerical analysis results of the model based on data mining technology show that in delayed production, when customer service level improves or the delay penalty coefficient increases, the optimal locations of the product differentiation point (PDP) and customer order decoupling point (CODP) move toward the end of production, and the total cost increases gradually. When the difference in production cost or benefit of early delivery between the candidate locations of PDP and CODP is small, optimal locations of PDP and CODP are close to the beginning of the general and dedicated production processes. With an increase of cost difference or early delivery benefit, the optimal locations of PDP and CODP jumped to the end stage of the general and dedicated production processes, and the total cost begins to decrease.