Based on the personalized customization model in bearing steel enterprises, it is studied in this paper how historical data can be fully utilized to predict orders in the face of multi-variety and small-batch order models.. In order to overcome the bottleneck of information exchange that cannot be obtained from pure historical data based on diverse small batch customer needs. A demand fusion mechanism based on local rules is proposed, taking fuzzy customer demand preferences as “individuals” and introducing historical data such as order area and external environmental factors. First, based on the Hegselmann-Krause (HK) model, a personalized customized customer demand preference prediction model is proposed to simulate and predict the evolution process of large-scale customer demand preferences. Secondly, according to the fuzzy rules, the customer's demand preference in the historical data is used as the center point to introduce the fuzzy individual, and the standard deviation is determined based on the order area. Finally, the effectiveness of the proposed method was verified through simulation results.