基于改进神经网络的小型鱼塘水质检测仪设计与实现

Design and Implementation of a Small Fish Pond Water Quality Detection Instrument Based on Improved Neural Network

为提高小型鱼塘水质分析的准确率,提出一种基于BP神经网络的水质分析方法。该方法首先根据淡水养殖生物在不同水质的水体中的生长状况制定了水质等级评定标准,再利用LM算法改进的BP神经网络进行水质等级分析,并通过高斯隶属度衡量不同样本水质间的相似度,降低水质状况的模糊性和不确定性对水质等级评定的影响。仿真实验证明,改进BP神经网络的水质分类准确率可达99.3%,网络训练时长仅为4.967 s,具有较快的训练速度和等级分类性能;搭建水质硬件分析系统及应用表明,该方法对小型鱼塘水体水质的等级评定结果整体偏优,基本满足小型鱼塘水质状况的初步分析需求。

To improve the accuracy and speed of preliminary analysis of water quality in small fish ponds,a water quality analysis method for small fish ponds based on BP neural network is proposed.This method first establishes water quality rating standards based on the growth status of freshwater aquaculture organisms in different water quality bodies.Then,a BP neural network improved based on LM algorithm is used for water quality rating analysis,and the similarity between different sample water quality is measured through Gaussian membership degree to reduce the impact of fuzziness and uncertainty in water quality status on water quality rating.Through simulation experiments,it has been proven that the improved BP neural network has a water quality classification accuracy of 99.3%,and the network training time is only 4.967 seconds.It has a fast training speed and good water quality classification performance for small fish ponds;Through physical experiments,it has been proven that the overall evaluation results of the water quality level of small fish ponds using this method are relatively good,which basically meets the preliminary analysis requirements of the water quality status of small fish ponds.