论文部分内容阅读
根据1998-2004年6-11月份我国鱿钓生产数据(月份、作业船数、经纬度和日产量)以及对应的海洋环境因子数据,即5 m水层的海水温度、46 m水层的海水温度、112 m水层的海水温度、317 m水层的海水温度、叶绿素a含量以及海平面高度距平值等,以经标准化后的单位捕捞努力量渔获量(CPUE)作为中心渔场指标,采用多种BP神经网络预报模型,对北太平洋柔鱼渔场进行了分析与比较。通过对13种神经网络预报模型的比较,以及实际CPUE的验证,以拟合残差最小的预报模型作为最优预报模型,认为结构为9-7-1的BP神经网络模型相对误差仅为0.008 570,可作为北太平洋柔鱼渔场的预报模型。
According to the production data of squid fishing in China from June to November in 1998-2004 (month, number of operating vessels, latitude and longitude and daily output) and corresponding marine environmental factor data, that is, seawater temperature of 5 m water layer, seawater temperature of 46 m water layer , Seawater temperature of 112 m water layer, seawater temperature of 317 m water layer, chlorophyll a content and sea level anomaly value, etc. The standardized catch unit effort (CPUE) A variety of BP neural network prediction models were used to analyze and compare the fishing grounds of Ommastrephes bartrami in the North Pacific Ocean. Through the comparison of the 13 kinds of neural network prediction models and the verification of the actual CPUE to fit the prediction model with the least residuals as the optimal prediction model, the relative error of BP neural network model with structure 9-7-1 is only 0.008 570, which can be used as the forecasting model for Ommastrephes fishes in the North Pacific Ocean.