基于BR算法和遗传算法优化生物柴油生产

Optimization of Biodiesel Production Based on BR Algorithm and Genetic Algorithm

以生物柴油成本为研究对象,综合考虑了原料、工厂产能、甘油置信度对生物柴油成本的影响。引入了人工神经网络BR(贝叶斯正规化)算法和遗传算法优化对收集到的数据进行预测。59组数据中的85%作为训练集,15%作为预测集,采用相关系数和均方误差作为预测结果的评价指标。研究结果表明:使用BR算法相关系数为0.9804,均方误差为0.0036,预测平均相对误差为4.47%。使用遗传算法的相关系数为0.8972,均方误差为0.0033,预测平均相对误差为4.89%。相较于L-M算法相关系数0.9741,均方误差0.0045,预测平均相对误差5.89%;两种算法的平均相对误差均有所下降,预测的效果更好。两种方法均有效验证了原料成本、工厂产能、甘油置信度对生物柴油最终成本的重要影响。

Taking the cost of biodiesel as the research object,the effects of raw materials,plant capacity and glycerol confidence on the cost of biodiesel were comprehensively considered.Artificial neural network(BR)algorithm and genetic algorithm are introduced to optimize the prediction of the collected data.85%of the 59 data sets were used as training set and 15%as prediction set.Correlation coefficient and mean square error were used as evaluation indexes of prediction results.The results show that the correlation coefficient of BR algorithm is 0.9804,the mean square error is 0.0036,and the average relative error of prediction is 4.47%.The correlation coefficient of genetic algorithm is 0.8972,the mean square error is 0.0033,and the average relative error of prediction is 4.89%.Compared with L-M algorithm,the correlation coefficient is 0.9741,the mean square error is 0.0045,and the average relative error of prediction is 5.89%;the average relative error of the two algorithms is decreased,and the prediction effect is better.Both methods effectively verified the important influence of raw material cost,plant capacity and glycerol confidence on the final cost of biodiesel.