基于神经网络-遗传算法优化生物柴油制备工艺

根据生物柴油制备的实验数据,用人工神经网络(ANN)的反向传播(BP)算法建立了生物柴油转化率神经网络预测模型,提出了适宜的人工神经网络拓扑结构,讨论了BP算法中学习速率、动量系数及过拟合现象对网络的影响。实验数据检验表明,ANN方法能准确地关联生物柴油制备工艺条件与转化率的关系,转化率预测平均相对误差为1.917%,复相关系数R为0.9996;该神经网络预测模型用遗传算法优化,得到了最佳生物柴油制备条件。

Optimization for biodiesel production technology based on genetic algorithm-neural network

Based on the experiment data from biodiesel production in laboratory, an artificial neural network (ANN) model was developed for predicting the biodiesel conversion by using BP (Back-Propagation) algorithm. The appropriate topology of ANN was obtained. The learning rate, the momentum factor and overfitting phenomena in BP network were discussed. It was shown that the ANN model can correlate and predict the biodiesel conversion accurately after comparing the prediction results with the experiment data. The average prediction error of the biodiesel conversion was 1.917%,and the correlation coefficient R was equal to 0.9996. The ANN model was optimized by incorporating genetic algorithm. Optimal operation conditions of the biodiesel production were obtained by using the ANN model developed.