基于改进遗传算法的BP神经网络的水体叶绿素a含量预测

Prediction of Chlorophyll a Content in Water Body Based on BP Neural Network with Improved Genetic Algorithm

采用BP神经网络是叶绿素a预测的主要手段之一。然而由于BP神经网络搜索的局限性,导致BP神经网络预测精度存在问题,因此,引入遗传算法对BP神经网络的权值和阈值进行优化,提升叶绿素a含量预测精度;由于遗传算法具有早熟和易陷入局部最优的缺点,因此,我们对遗传算法的选择/交叉以及变异算子进行改进。从结果可看出,改进的GA-BP神经网络训练精度比GA-BP神经网络提升1.23%,仿真精度提升1.66%,相对于BP神经网络训练精度提升2.71%,仿真精度提升5.03%,改进的GA-BP神经网络预测精度更优。基于Dimopoulos 因子敏感性分析,筛选出输入因子组合。

The prediction of chlorophyll content by BP neural network is one of the main methods for predicting chlorophyll a content.However,due to the limitations of BP neural network search,the prediction accuracy of BP neural network is problematic.Therefore,the genetic algorithm is introduced to optimize the weight and threshold of BP neural network to improve the prediction accuracy of chlorophyll a content.Because genetic algorithms have the disadvantages of premature and easy to fall into local optimum,we improve the selection/crossing of genetic algorithms and mutation operators.It can be seen from the results that the improved GA-BP neural network training accuracy is 1.23% higher than that of the GA-BP neural network,the simulation accuracy is improved by 1.66%,the training accuracy is improved by 2.71% compared with the BP neural network,and the simulation accuracy is improved by 5.03%.The GA-BP neural network has better prediction accuracy.Screening out input factor combinations based on Dimopoulos factor sensitivity analysis.