摘要: |
为预测团水虱的数量,减轻其对红树林植物的危害,采用多层感知器神经网络分析方法,对
2021 年 3 月至 9 月团水虱发生地的水温、风速、大气压、pH 值、溶解氧、高锰酸盐指数、氨氮、总磷、
总氮含量等因素进行分析。结果表明,总磷、水温是团水虱数量的主要因子,其拟合精度较好,平均绝
对百分误差 (MAPE) 为 0.120 2,均方误差 (MSE) 为 85.486 1。使用该模型对 2022 年 10 月团水虱数量进
行预测,其预测值为 34.42/dm3
,结果较为精确。 |
关键词: 团水虱 神经网络 多层感知器 预测模型 |
DOI: |
分类号: |
基金项目:海南省重点研发计划(ZDYF2021XDNY193) |
|
Construction of A Neural Network Prediction Model for Sphaeroma in Dongzhaigang National Nature Reserve, Hainan Province |
Cen Xuancai1, Shi Danni1, Zhong Mengying1, Li Shichuan2, Huang Danmin1, Guo Xia1
|
1.Hainan Academy of Foresty(Hainan Academy of Mangrove);2.Hai Nan Dong Zhai Gang National Nature Reserve Authority
|
Abstract: |
In order to predict the number of Sphaeroma and reduce its damage to mangrove plants, the
multi-layer perceptron-neural network analysis method was used to analyze the water temperature, wind speed,
atmospheric pressure, pH value, dissolved oxygen, permanganate index, ammonia nitrogen, total phosphorus, total
nitrogen content in the area of Sphaeroma occurrence from March 2021 to September 2022. The results showed
that total phosphorus and water temperature were the main factors of the number of Sphaeroma, and the fitting
accuracy was good, with the mean absolute percentage error (MAPE) of 0.120 2 and mean square error (MSE) of
85.486 1. This model was used to predict the number of Sphaeroma in October 2022, and the predicted value was
34.42/dm3
, which was accurate. The results provided a reference for predicting the number of Sphaeroma. |
Key words: Sphaeroma neural network multiple layer perceptron prediction model |