Early Detection of Product Reliability Based on the Parameters of the Production Line and Warranty Data

Reliability is one of the key dimensions of the quality of services and products that should be always evaluated. Growth and development of industries can be achieved by appropriate reliability engineering of products. Companies should evaluate and predict the reliability of products and accordingly find and fix the potential problems. In this regard, early detection of reliability problems based on the parameters of the production line or quality test results can prevent future warranty costs. Early detection of reliability problems based on production process and test data has not gained much attention in the literature. Therefore, an early detection model for predicting the reliability of products according to their quality test results is proposed in this research. For this purpose, hot test and warranty data of car engines manufactured by an automotive company are utilized. This data are prepared to predict engine reliability after preprocessing and removing inefficient data. Then, engines are divided into two homogeneous clusters using particle swarm optimization (PSO) clustering algorithm. Afterwards, the data in these clusters are used to feed the Artificial Neural Network (ANN) to predict the reliability of the engines. The obtained results show that the proposed ANN-based method is able to predict the reliability of the engines based on engine kilometers operated and hot test milks. Also, it is shown that the proposed method outperforms the Cox proportional hazards model which has previously been used for early detection of product reliability.