基于PCA-LSTM神经网络的凡纳滨对虾养殖水质预测

基于上海奉贤区2个水产养殖合作社2014—2018年和2021年的检测数据,选取水温(T)、溶解氧(DO)、高锰酸盐指数(I_Mn)、总磷(TP)、总氮(TN)、氨氮(TAN)、亚硝酸盐氮(NO₂^--N)、硝酸盐氮(NO₃^--N)共8个水质指标进行研究,提出了基于主成分分析(Principal component analysis, PCA)和长短时记忆神经网络(Long short-term memory neural network,LSTM)的预测模型。首先采用主成分分析法对数据进行特征提取和降维,选取高锰酸盐指数(I_Mn)和氨氮(TAN)作为水质预测指标,建立基于PCA法的LSTM模型;接着采用PCA-LSTM模型对不同养殖塘的水质进行预测;最后,将其与单一LSTM模型进行对比以验证模型的优劣。结果表明:PCA-LSTM模型可用于凡纳滨对虾养殖池塘水中I_Mn和TAN的预测, 预测结果优于单一LSTM模型。

Prediction of water quality in Litopenaeus vannamei aquaculture ponds based on the PCA-LSTM neural network model

Based on the detection data of the two farms in Fengxian District of Shanghai from 2014 to 2018 and in 2021,8 water quality indicators,including the water temperature (T), dissolved oxygen (DO), permanganate index(I_Mn), total phosphorus (TP), total nitrogen (TN),ammonia nitrogen (TAN), nitrite nitrogen (NO₂^--N) and nitrate nitrogen (NO₃^--N) were chosen to establish a prediction model based on principal component analysis (PCA) and long short-term memory (LSTM). Firstly, through principal component analysis which was used to reduce data feature extraction and dimension, I_Mn and TAN were determined to be the water quality prediction indexes to build a LSTM model based on the PCA analysis,then the PCA-LSTM model was used to predict the water quality of different ponds;Finally,comparison was carried out with a single LSTM model to verify the strengths and weaknesses of both models. The results show that the PCA-LSTM model can be used to predict I_Mn and TAN in Litopenaeus vannamei aquaculture ponds,and the prediction results are better than the single LSTM model.