【摘要】 为解决BOD传统测量耗时长、需要离线采样分析、实验操作复杂的问题,论文提出了一种基于特征重要性排序和LDA降维算法改进的随机森林模型用于BOD的快速软测量。改进随机森林模型将12维辅助特征向量降至3维特征向量,有效减少数据中存在的噪声与冗余信息,提升了随机森林模型的预测能力。仿真结果表明,改进后的随机森林模型相较于其他模型在快速BOD预测中有明显优势,其MSE为0.038 8,MAE为0.122 7,决定系数为0.967 6,预测时间0.485 0,说明该模型为快速BOD预测场景提供了一种可能的思路。更多还原

【Abstract】 In order to solve the problems of long time-consuming traditional BOD measurement, offline sampling and analysis, and complicated experimental operation, the paper proposes a random forest model based on feature importance ranking and LDA dimensionality reduction algorithm for fast soft measurement of BOD. The improved random forest model reduces the 12-dimensional auxiliary feature vector to a 3-dimensional feature vector, effectively reducing the noise and redundant information in the data, and improving the predictive ability of the random forest model. The simulation results show that the improved random forest model has obvious advantages in fast BOD prediction compared to other models. Its MSE is 0.038 8,MAE is 0.122 7,the coefficient of determination is 0.967 6,and the prediction time is 0.485 0,indicating that this model is a fast BOD prediction scenario Provides a possible idea.更多还原