Predictive Of Gas Emission Based On CIPSO-ENN Coupled Algorithm

According to the characteristics of the project complexity, time varying and nonlinear,The paper proposed CIPSO-ENN coupling algorithm for the nonlinear dynamic model parameter identification and predictive control of gas emission. Thesis first proposed by Elman neural network prediction of gas emission, and analysis of the Elman neural network, for the slow convergence,easy-to-precocious of the neural network paper proposed the particle swarm algorithm to improve it. But when start particle swarm algorithm(SPSO) solution for multimodal, high dimensional, nonlinear optimization problems will fall into local optimization of defects in early .So introduced chaos theory and immune clone selection algorithm theory into SPSO proposed to improve it, proposed CIPSO algorithm. In the process of particle swarm evolution, the particles clonal selection algorithm to improve the speed of convergence, the particles on chaotic mutation cloned to enhance the ability of local search population. Numerical experiments show that, compared to the standard pso the algorithm accuracy and stability has improved. Significantly. After CIPSO algorithm and Elman feedback network integration, proposed CIPSO-ENN coupled algorithm. The coupling algorithm uses CIPSO train Elman network weights and threshold for optimization to improve network generalization, convergence speed and the nonlinear mapping capability, and then based on the coupling algorithm, with predictive intelligence, control theory, the establishment of the prediction of gas emission control system model based on CIPSO-ENN coupling algorithm. Using the historical data of mineactual monitoring to the simulation ,results show that the model compared to the Elman and BP neural network model the identification convergence faster 8 times, identification accuracy is improved by nearly 2 orders of magnitude, prediction accuracy is increased by 3 times. So algorithm has fast convergence, high forecast precision and robust characteristics. Starting from the data itself, using CIPSO-ENN establish model predictive control is a very effective method and can be extended to other areas.