水产养殖环境因子数据动力学分析与智能预测

Dynamics Analysis and Intelligent Prediction of Aquaculture Data

进行水质环境因素分析对水产养殖的效益极为重要。研究水体因素的三个主要参数:水温、p H值以及溶解氧。针对检测仪取样的数据存在缺失、不准确等问题,通过高阶曲线插值较好地修复了数据,同时运用滤波方法划分了系统误差以及参数的自身节律;对不同水层、时间的参数分析,较好地吻合了实际水文情况,为工程养殖提供了可靠的依据;通过引入径向基函数神经网络方法跟踪主要参数的特征,弥补了非线性多项式插值的不足,实际数据证明了该方法全局跟踪有效以及局部节律刻画程度精细。

Data analysis on environmental factors data is crucial to aquaculture, in which three significant parameters were discussed, they are temperature, PH and dissolved oxygen. Fixing some missing data and inaccurate records in the sampling process by high-order curve fitting. Meanwhile, the use of filtering method was adopted to divide systematic errors and rhythms inside parameters. Analysis from different water layers and different time suited the true environment well, which provided effective references for engineering problems. Radial Basis Function Neural Networks was well applied in tracking the parameters trend both globally and locally.