A Novel Semi-Supervised Method for Predicting Remanufacturing Costs of Used Electromechanical Devices Using Quality Characteristics

Remanufacturing cost is a key factor for making decisions on the remanufacturing of used electromechanical devices in the construction sector. Though, remanufacturing costs can vary significantly due to the diversity of quality characteristics, even for the same type of used electromechanical devices. To realize the prediction of the remanufacturing cost for used electromechanical devices relevant to construction, this paper proposes a semi-supervised remanufacturing cost prediction method based on quality characteristics. First, we establish a semi-supervised least squares support vector regression (SLSSVR) model. Then, a novel variable neighborhood search (VNS) algorithm is designed for SLSSVR parameter tuning and optimizing. To verify the performance of the VNS-SLSSVR, we provide three types of simulated examples and conduct a real case study on predicting the remanufacturing cost of used turbine worms. The experimental results show that the proposed methods are of high accuracy and reliability with a limited number of labeled samples and a substantial quantity of unlabeled ones.