基于物联网的浮标水质监测系统与溶解氧浓度预测模型

为促进近海养殖业信息化发展，更好地实现对近海养殖环境的监控，设计了基于浮标平台的环境监测系统。利用STM32L475微控制器定时采集光照、温度、pH值、溶解氧浓度等信息，通过物联网技术将数据传输至云监测平台，实现了多区域环境信息远程监测和多终端访问。提出了改进遗传算法BP神经网络的溶解氧浓度预测模型，实现对近海养殖环境的预测；根据所采集的数据，利用改进遗传算法对初始权重和阈值进行优化得到最优参数，在此基础上构建BP神经网络溶解氧浓度预测模型。通过试验验证了该系统海洋环境信息采集的准确性与可靠性，以及溶解氧浓度预测模型的有效性；与传统遗传算法BP神经网络预测模型相比，平均误差由0.0778mg/L降至0.0178mg/L，能够满足近海养殖的实际需求。

Buoy Water Quality Monitoring System and Prediction Model Based on Internet of Things

In order to promote the informatization development of offshore aquaculture, realize the monitoring of offshore aquaculture environment more accurately and conveniently, and solve the problems of poor prediction accuracy and robustness of traditional offshore aquaculture water quality prediction methods, an environmental monitoring system was designed based on buoy platform, which realized the remote collection and data storage functions of multi-regional environmental information monitoring data. On this basis, an improved genetic algorithm was proposed to optimize the offshore dissolved oxygen prediction model of BP neural network to realize the prediction of offshore aquaculture environment. The STM32L475 microcontroller was used to collect information such as illumination, temperature, pH value, dissolved oxygen and so on with the help of sensor network, and transmitted the data to the cloud monitoring platform through the Internet of things technology, thus realizing remote monitoring of multi-regional environmental information and multi terminal access. Through the analysis and research of classical prediction algorithms, a dissolved oxygen prediction model based on traditional algorithm optimization was proposed to realize the accurate prediction of offshore aquaculture water quality environment. According to the collected data of aquaculture environment, the initial weights and thresholds were optimized by improved genetic algorithm to obtain the optimal parameters, and then the BP neural network dissolved oxygen prediction model was constructed. Through experiments, the accuracy and reliability of marine environmental information collection and the effectiveness of dissolved oxygen prediction model were verified. Compared with the traditional neural network prediction model, the average error was reduced from 0.0778mg/L to 0.0178mg/L, which can meet the actual needs of offshore aquaculture.