CAJ | 학술논문

针对目前营养盐检测主要是通过化学方法实现，无法获得在线检测的问题，利用营养盐与其影响因子之间的关系，提出结合自适应遗传算法与弹性BP神经网络的预测模型。利用改进的自适应遗传算法，通过交叉、变异获取弹性BP神经网络的初始权值与阈值，加速预测过程。该模型通过营养盐影响因子数据，预测亚硝酸盐浓度。仿真结果表明：基于弹性BP神经网络的预测模型预测营养盐浓度是可行的，其预测得到的亚硝酸盐浓度值的相对误差主要集中于0~30%；结合自适应遗传算法与弹性BP神经网络的预测模型的预测效果好于基于弹性BP神经网络的预测模型。
침대목전영양염검측주요시통과화학방법실현，무법획득재선검측적문제，이용영양염여기영향인자지간적관계，제출결합자괄응유전산법여탄성BP신경망락적예측모형。이용개진적자괄응유전산법，통과교차、변이획취탄성BP신경망락적초시권치여역치，가속예측과정。해모형통과영양염영향인자수거，예측아초산염농도。방진결과표명：기우탄성BP신경망락적예측모형예측영양염농도시가행적，기예측득도적아초산염농도치적상대오차주요집중우0~30%；결합자괄응유전산법여탄성BP신경망락적예측모형적예측효과호우기우탄성BP신경망락적예측모형。
Currently nutrients are detected by the chemical method. A chemical method cannot get online detection. To solve the problem, based on the relationship between nutrients and their impact factors, a prediction model which combined Adaptive Genetic Algorithm and Elastic BP Neural Network is put forward in this paper. Using the improved Adaptive Genetic Algorithm, the initial weights and thresholds of Elastic BP Neural Network are obtained by the crossover and mutation to accelerate the prediction process. The imporoved model predicts the nitrite by using the data of its impact factors. Simulation results show that it is feasible to predict the nutrient concentration by using the prediction model based on the Elastic BP Neural Network. The relative error of nitrite concentration value mainly focuses on 0-30%. The prediction model based on Adaptive Genetic Algorithm and Elastic BP neural network is better than that based on Elastic BP Neural Network.