Online Operation Optimization System Design And Implementation In Ethylene Cracking Furnace

Ethylene cracking furnace (ECF), the key device to ethylene production, not only directly affects the quantity and quality of ethylene production, but also has a significant impact on the production of downstream chemical products. Therefore, research on operation optimization in ECF is essential. With the development of computer technology, data-based intelligent model has become an important method of modeling and optimization in ECF, to make the running of ECF safer and operation more stable and to improve the yields of pyrolysis products.Based on OPC technology, the two-direction communication between local computer and DCS system of ECF is achived to read data from DCS and write data to it. Then use the data preprocess technology, including data integration, data filling, data anomaly detection, data filtering and data normalization, to process the collection data for the following modeling and optimization.Based on RBF neural network, PSO algorithm and K-Means clustering algorithm, DMPSO-RBF algorithm is proposed as a new training method of neural network, to establish the online soft-measurement model of the yields of pyrolysis products in ECF. At the same time, the soft-measurement model is optimized by PSO algorithm. The applications show that the yields of ethylene and propylene are increased after operation optimization, so DMPSO-RBF algorithm is proved to be effective.On depth study of concept lattice, FCA construction algorithm is improved to greatly reduce the complexity of concept lattice, at the same time, target property based operation rule mining algorithm is given to make the mining process more targeted, greatly improving the mining efficiency of operation rule.Based on .Net Framework, COM and design pattern, online operation optimization system in ECF is developed. The system has many features such as friendly user interface, rich system functionality, portability, maintainability, extensibility and so on. The applications show that the system can meet the requirements of actual operation optimization, and has achived good results in the actual production process.