基于神经网络水质预测模型

发布时间：2019-07-09 09:51

【摘要】：近年来,水产养殖业正以惊人的速度增长,已经逐渐成为世界范围内的又一项主导产业。落后的水质管理方式和落后的水产养殖模式导致我国每年因水产病害损失高达140亿元,水质的恶化是引起水产品疾病暴发的主要因素。由于养殖池塘系统的非线性和复杂性,以及水质因子之间的相互影响,用传统的方法不能建立令人满意的水质预测模型。人工神经网络预测模型是以计算智能为核心的现代预测方法。虽然人工神经网络广泛的运用在各个领域,但是在水产养殖水质预测方面尚属初级阶段。本文以MATLAB2011为平台,采用了L-M算法改进的BP网络以及RBF网络对水质预测进行研究。本文以海口市某水产养殖场为例,通过实际检测获得其养殖池塘连续8d的6种水质因子的数据。分别建立了含有一个隐含层的3层BP神经网络以及RBF神经网络,两个网络的输出目标为溶解氧溶度,取另外的5种水质因子作为网络的输入向量,并用ATLAB软件建立了神经网络预测模型。通过对养殖池的溶解氧溶度的预测结果表明,BP神经网络以及RBF神经网络能够有效的运用在水质预测方面,具有较高的精度,具有很高的实用价值。
[Abstract]:In recent years, aquaculture is growing at an alarming rate and has become a leading industry in the world. The backward water quality management method and the backward aquiculture mode have resulted in the loss of the aquatic products by up to 1.4 billion yuan each year, and the deterioration of the water quality is the main factor causing the outbreak of the aquatic products disease. Due to the non-linearity and complexity of the culture pond system and the interaction between the water quality factors, a satisfactory water quality prediction model cannot be established by the conventional method. Artificial neural network prediction model is a modern predictor with intelligence as the core Method. Although the artificial neural network is widely used in various fields, it is the primary level in the water quality prediction of aquaculture. In this paper, using MATLAB2011 as the platform, the improved BP network and RBF network of the L-M algorithm are used to study the water quality prediction. In this paper, an example of an aquatic farm in Haikou City is taken as an example, and six water quality factors of the culture pond for 8 days are obtained by actual detection. The three-layer BP neural network with an implicit layer and the RBF neural network are respectively established. The output target of the two networks is the dissolved oxygen solubility, and the other five water quality factors are taken as the input vector of the network, and the neural network is pre-established by the ATLAB software. The prediction results of the dissolved oxygen solubility of the culture pond show that the BP neural network and the RBF neural network can be used effectively in water quality prediction and have high precision and high accuracy.
【学位授予单位】：海南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP183

【相似文献】