Particle Swarm Optimization and Neural Networks Application for Twin-Spirals Scroll Compressor Performance Prediction

Particle swarm optimization and neural networks (PSO- NN) was proposed for twin-spirals scroll compressor (TSSC) performance prediction. The method integrated evolutionary mechanism of PSO and self-learning, nonlinear approach ability of NN. In established NN the input variables were main structure parameters and the output variables were main performance parameters. PSO was used to train NN. The trained NN can predict the TSSC performance very well. The trained results showed that this kind of approach can converge to better solutions much faster compared with other reported approaches. It also overcomed the weakness of slow convergence and local minima. The PSO-NN offered a new method for TSSC performance optimization.