Feature Constrained Encoding Generative Adversarial Prediction Network for Quality Measurement Modeling Under Multi-time Scales Data Imbalance In Process Industry

Data imbalance problem hinders the quality indicators data driven soft sensing modeling in process industry. This problem is caused by production complexity, which makes the pivotal variables can only be manually sampled and detected rather than directly detected by sensors. Aiming at this problem, we propose a feature constrained encoding generative adversarial prediction network model (FCEGAPN) for quality soft sensing in cement clinker calcination process, which is constructed by an Encoder, a Generator, a Discriminator and a Predictor. The Encoder, Generator and Discriminator mine the features of multi-variable time series data and produce fake samples under the constructed three-player adversarial learning mechanism. The produced multi-variable time series samples are mixed with actual samples for data augmentation, so that the amount of quality indicator is increased and its feature space is enlarged and the data imbalance problem is eliminated, whose effectiveness is validated by statistics and clustering methods. Using augmented data to train Predictor and ultimately improves the soft sensing performance of quality indicator. The advantages in accuracy, validity and robustness of the proposed method are demonstrated by the experiments implemented by cement production data gained from a cement manufacturing enterprise.