基于粒子群优化神经网络的煤炭成浆浓度的预测

Prediction of coal slurry concentration based on PSO-BP neural networks

利用一种可以避免搜索陷入局部极值的粒子群算法对BP神经网络进行了改进,并利用粒子群神经网络对不同制浆煤种的成浆浓度进行预测,并建立了考虑七因子影响因素后的煤炭成浆浓度的神经网络预测模型。结果表明,粒子群优化神经网络模型能够迅速而精确地预测不同制浆煤种的成浆浓度,误差只有0.18%。同时验证了选取七因子神经网络的合理性。

In this paper , a particle swarm optimization algorithm, which has strong global exploratory capability and efficiency, is employed to improve back propagation neural network. The PSO - BP neural network model with 7 input factors is set up for predicting the slurry concentration. The method in this paper is proved to feasible by the comparison with the result in the reference 2. The mean absolute error of the neural network model with 7 factors is 0.18 % . At the same time, the model with 7 input factors is feasible by researched in this paper.