基于PCA-GA-BP网络B、Cu胁迫下油菜生理响应机制的研究

Study on Physiological Response for Rape under Stress from B and Cu Based on PCA-GA-BP Network

【目的】本文研究了油菜在B、Cu胁迫下的生理响应机制,提出一种基于PCA-GA-BP网络的智能化评价方法。【方法】采用5种不同浓度的Cu溶液和3种不同浓度的B溶液交叉处理盆栽油菜幼苗,测定油菜的叶绿素、蛋白质、硝态氮、淀粉酶、丙二醛含量以及过氧化物酶活性。首先采用主成分分析法分析不同处理下各指标的综合权重,优化油菜的生理指标。然后根据优化后的油菜指标和综合权重数据建立BP神经网络智能评价模型,并采用遗传算法优化BP网络的权阈值。【结果】T_(200)B_1组合时综合权重最高,即B 0.5 mg/L、Cu 200 mg/L配施时对各项指标最好。用主成分法优化油菜生长指标后,对评价结果影响不大。GA-BP网络测试误差不超过2.85%,测试效果很好。【结论】该方法可以避免人工干预,能更加客观地对油菜生理响应进行评价,同时具有自适应能力,能自动识别B、Cu配置情况,为油菜生理响应机制研究提供一种更加科学高效的智能化评价方法。

【Objective】This paper aimed to study the physiological response mechanism for rape under the stress of B and Cu,an intelligent evaluation method was proposed based on PCA-GA-BP(Principal Component Analysis-Genetic Algorithm-BP network).【Method】The potted seedlings of rape were dealed with 5 different concentrations of Cu solution and 3 different concentrations of B solution,and the indexes of chlorophyll,protein,Nitrate nitrogen,amylase,MDA content and peroxidase activity were measured.First,the comprehensive weights of the indexes under different treatments were analyzed,and the physiological indexes of rape by the principal component analysis method were optimized.Then the intelligent evaluation model based on BP neural network was established according to the optimized indexes and the comprehensive weights,and at the same time,the weights and thresholds of BP network were optimized by genetic algorithm.【Result】The comprehensive weight was the highest with the combination of T200B1,and the indicators with the configuration of B(0.5 mg/L)and Cu(200 mg/L)were the best.There were no effects on the results of the evaluation as the growth indexes were optimized by the principal component method.The errors of GA-BP network test were not more than 2.85%,thus the test results were very good.【Conclusion】The proposed method can avoid subjective intervention from experts,thus it can evaluate the physiological response of rapeseed more objectively.At the same time,it has the ability of self-adaptive and can automatically identify the configuration of B and Cu,which provides a more scientific and efficient way for intelligent evaluation on physiological response mechanism of rapeseed.