基于机器学习的浮标实时自动化赤潮预警研究

实现赤潮预警对于减轻海洋环境灾害、避免海洋产业特别是海洋渔业重大经济损失具有重要意义。针对当前水文监测数据海量却难以实现实时自动化监测与预警,特别是难以利用传统监测手段实现对危害更大的赤潮的精准实时预测这一显著问题,提出利用浮标数据作为依据,借助机器学习在大数据分析和智能决策方面的优势,建立一种新颖的双重递进式赤潮预警机制的方法。首先,通过相关算法分析历史数据,以确认赤潮初步预警阈值;其次,对叶绿素a、pH、溶解氧等重要监测指标的当前和阶段性变化进行初步分析,判断是否达到预警触发条件;然后,进一步联合分类、回归、聚类、神经网络等机器学习相关方法,对数据进行深度挖掘;最后,通过这种递进式的机制对短期内是否会发生赤潮作出判断,以实现赤潮自动化预警预报。在此基础上,利用宁波梅山湾实际监测数据,证实了该方法在赤潮实时自动化预警中的有效性。

Research on Real-Time Automatic Red Tide Warning of Bathing Buoy Based on Machine Learning

It is of great significance to realize red tide early warning for reducing marine environmental disasters and avoiding major economic loss of marine industry,especially marine fishery.Aimed to the problem that there are a large amount of hydrological monitoring data,which is difficult to achieve real-time automatic monitoring and early warning,especially is difficult to achieve accurate and is the challenge for real-time prediction of red tide to outbreak with greater harm by traditional monitoring means.It is proposed to establish a novel method of double progressive red tide early warning mechanism based on the basis of buoy data and the advantages of machine learning in big data analysis and intelligent decision-making.Firstly,the essay points out that the historical data needs to be analyzed by relevant algorithms to confirm the preliminary warning threshold of red tide.Secondly,the current and periodic changes of important monitoring indicators such as chlorophyll-a,pH and dissolved oxygen should be preliminarily analyzed to determine whether the trigger conditions for early warning were met.Then,the data are further mined by combining the relevant methods of machine learning such as classification,regression,clustering,and neural network.Finally,the automatic warning and forecast of red tide could be realized through this progressive mechanism to judge whether red tide will occur in a short time.On this basis,the effectiveness of this method in real-time and automatic early warning of red tide is confirmed by the actual data of Meishan Bay in Ningbo.