Production capacity analysis and energy optimization of complex petrochemical industries using novel extreme learning machine integrating affinity propagation

With the rapid political and economic developments, energy saving and carbon dioxide emission reduction are now recognized as the most important goals worldwide. Therefore, this paper presents a production capacity analysis and energy optimization model that uses novel extreme learning machine integrating affinity propagation clustering. By using the affinity propagation method, clusters of raw data can be obtained automatically to reduce multi-dimensional data and fuse the high-similarity data. Then, the clustering results are taken as the training and testing sets of the extreme learning machine to improve prediction accuracy and analyze production capacity. Through comparisons with the extreme learning machine, back propagation, radical basis function and extreme learning machine integrated K-means clustering algorithm, the accuracy and validity of the proposed method are verified by using the University of California Irvine data sets. Finally, the proposed method is applied to build the production capacity analysis and energy optimization model of the ethylene and purified terephthalic acid production systems in complex petrochemical industries. The prediction accuracy of the ethylene and purified terephthalic acid production is approximately 99%, thus improving the energy efficiency of these complex petrochemical processes and achieving energy saving and carbon dioxide emission reduction.