摘要：AbstractModern steel rolling process is commonly designed to produce products with a variety of grade specifications. There exist enormous historical data for the typical products in large batch production; while for small-batch customized products, the lack of sufficient historical data may prevent successful application of traditional data-based process modeling and quality prediction methods. In this paper, a practical and effective strategy is developed for fast modeling and flatness defect prediction of a steel rolling process. The key idea is based on model migration, assuming that high-performance quality prediction models (defined as Base Models) have been available for the typical products. A Principal Component Analysis (PCA) similarity indicator is adopted to measure the difference between new operating modes and the typical operating modes, based on which, new operating modes are divided into non-significantly-changed modes and significantly-changed modes. For the former, the new flatness defect prediction model is developed by screening the similar data in the typical modes and augmenting them into the modeling data set for the new mode. For the latter, the new model is obtained by reconstructing the base model via parameters’ shifting and scaling. Case study results can show the validity of the proposed method.
关键词：KeywordsFast modelingQuality predictionModel migrationPartial Least Squares (PLS)Steel rolling process
