The research on the modeling method of batch process based on OS-ELM-RMPLS

Nowadays, batch process is a typical mode of production and plays an important role in industries such as iron and steel, chemical engineering, biopharmaceuticals and semiconductors. Due to the process multiple variables, batch-to-batch variation, complexity and multiphase, the mechanism model is difficult to obtain. With the development and progress, soft-sensing is becoming a most widely used method in modeling and analyzing for batch processes. Extreme learn machine is a single hidden layer feed forward neural network with additive or RBF nodes, which tends to provide superior generalization performance and extremely fast learning speed, as well as less parameters to be identified by randomly choosing the input weights and analytically determining the output weights of the network. Combining ELM and MPLS, the new modeling method has the advantages of both MPLS and ELM, which reduces the dimension, handles the nonlinear and shows more rapid training speed. On this basis, the OS-ELM-RMPLS method is proposed for updating the model online to cope with batch-to-batch variation and unknown disturbance. It is proved an effective and applicable method by the simulation results.