ABSTRACT
As the manufacturing industry moves towards rapid response design, it is essential for manufacturers to upgrade their existing products. Traditionally, design improvement is performed with product failure data. However, there are limitations when dealing with little failure data. Many kinds of research considering product time-varying performance data are emerging for design improvement. To expand the application of this data in the design improvement, there are two critical problems to solve. One is to identify the abnormal time instances (i.e. outliers) of field data. The other problem is to analyse function modules which are related to these abnormal time instances. To solve the above questions, a novel method to transform performance data into useful design improvement information is proposed. Based on performance data, an anomaly detection model is developed according to the outlier analysis theory. Next, outlier instances detected by the model are used to analyse the degradation degree of each performance feature. Finally, the redesign module can be identified using the correlation matrix between performance features and function modules. The effectiveness of this proposed method is demonstrated through a case study with design improvement of a wind turbine.
Disclosure statement
No potential conflict of interest was reported by the authors.