摘要
为了提高BP神经网络模型对海洋藻类生长状态软测量的准确性,提出了一种基于遗传优化算法优化BP神经网络的软测量方法。利用遗传算法优化BP神经网络的权值和阈值,然后训练BP神经网络预测模型以求得最优解,再将该预测结果与传统BP网络预测模型的预测结果进行对比。对仿真结果进行有效性验证后,结果表明,通过这种软测量方法,经遗传算法优化后的BP神经网络可以在更短的时间里创造更高的预测准确性,大大提高了对海洋藻类生长状态预测的效率。
In order to improve the accuracy of the forecasts of marine algae growth state by BP neural network model,a method of soft sensing based on BP neural network optimized by genetic algorithm is presented.Use genetic algorithms to optimize the BP neural network weights and threshold,then train the BP neural forecasting model in order to achieve the optimal solution,and then the prediction results were compared with the predicted results with the traditional BP network prediction model.After validation of the simulation results,it shows that,with this soft measurement methods,BP neural network optimized by the genetic algorithm in a shorter amount of time can create a higher forecast accuracy,and improve the efficiency of marine algae growth state forecast greatly.
出处
《电子设计工程》
2013年第21期63-66,共4页
Electronic Design Engineering
基金
国家自然科学基金项目资助(61273068)
上海市自然科学基金项目资助(12ZR1412600)
上海市教委科研创新项目资助(13YZ084)
关键词
软测量
遗传算法
BP神经网络
状态预测
模型优化
soft sensing
genetic algorithm
BP neural network
state prediction
model optimization
