Soft sensor modeling and optimization of Aluminum strip grain size based on PSO-BP

Aluminum strip grain size is one of the most important variables in the Aluminum electromagnetic casting-rolling process, and its on-line and real-time measure isn´t actualized at present. Analyzed the working principle introduction of aluminum electromagnetic roll-casting and the effect to product grain size of the cast-rolling factors and electromagnetic factors, the secondary variables and primary variable were determined on which. Then the soft senor modeling based on BP network was realized and PSO is used to optimize the modeling, which can detect the aluminum grain size on line. The results of simulation and research show that, the soft sensor modeling and optimization of aluminum strip grain size can be realized easily by BP network. After optimizing it by PSO, the stability and robustness are enhanced and the deterministic coefficient has significantly improved.