关键词: 最小二乘支持向量机, 变精度粗糙集, 顾客需求, 权重预测, 产品设计

Abstract: To understand the customers' requirements more accurate and provide reliable decision-making in product design process, the product designers needed to determine not only the current weight of customers' requirements but also predicted the future weight so as to master the dynamic trend of customers' requirements. At present, there were few studies on predicting future weight of customers' requirements and dynamic trend analysis. Aiming at this problem, an integrated method based on the preprocessing of Variable Precision Rough Set (VPRS)and Least Squares Support Vector Machines (LS-SVM) regression theory was presented to predict the weight of customer requirements. In this method, the weight of customers' requirements was calculated firstly based on β-classification accuracy method of VPRS, and then the time sequence weight was constructed by calculating each customer equariment weight periodically. The LS-SVM regression theory was introduced to process time sequence of customers' requirements weight so as to determine concrete representation of prediction model. According to the historical, current and future weight of customers' requirements, the dynamic trend column diagram of customers' requirements was constructed for analysis. A case study of high-angle belt conveyor was provided to prove the feasibility and effectiveness of the proposed method.

Key words: least squares support vector machines, variable precision rough set, customer requirement, weight prediction, product design