Missing Data Imputation With OLS-Based Autoencoder for Intelligent Manufacturing

Motivated by the global economy that is greatly shaped by the landscape changes in energy and manufacturing where more and more devices and systems are interconnected, intelligent manufacturing in which data mining is of great importance is studied. In this article, an energy monitoring platform for small- and medium-sized enterprises developed by the point energy team (www.pointenergy.org) is first introduced, which monitors and records the energy consumption of manufacturing processes at various levels of granularity. In processing the collected data, the incompleteness in the data due to various factors needs to be addressed first otherwise it may lead to the inaccurate portrayal of the system and poor generalization of the resultant model trained by the data. Hence, a novel orthogonal-least-square-based autoencoder is proposed to generate new samples for the imputation of missing values. This approach is to learn the representative code from the original samples by constructing an improved encoder network in which the hidden neurons are orthogonal with each other. The new samples are then generated through the decoder network. The proposed approach selects the hidden neurons one by one based on the OLS estimation until an adequate network is built. The classical techniques and other generative models are compared to verify the effectiveness of the proposed algorithm. For these methods, the optimal parameters are estimated based on the performance metric of the cross-validation mean square error. In the experiment, two real industrial datasets from a baking process and a polymer extrusion process are adopted and the percentage of missing values varies from 0.02 to 0.25. The experimental results confirm that the proposed method offers stable performance in the presence of different missing ratios, and it outperforms significantly alternative approaches while the missing ratio is greater than 0.05.