Applying artificial intelligence to predict the fishing performance of stow net constructed with noctilucent sticks

Abstract

The application of artificial light to attract marine organisms was demonstrated to enhance fishing gear efficiency. This study presented a novel stow net design incorporating noctilucent sticks to optimize catch performance. Specifically, the research examined the impact of different colores of noctilucent sticks on the efficiency of catch in stow nets. The results revealed the use of noctilucent sticks could significantly increase the catch weight of stow nets (p < 0.01). Notably, the color of the noctilucent sticks influenced their effectiveness, with olive green sticks increasing catch weight by 40.65 %, followed by azure sticks (12.57 %) and bluish green sticks (8.88 %). The predominant species caught was the small yellow croaker, whose catch proportion rose from 19.18 % to 22.80 % due to the noctilucent sticks. In addition, a comprehensive analysis was conducted using the generalized additive model (GAM) to assess the impact of noctilucent sticks, lunar phases, and tidal cycles on stow net catch weights, complemented by a backpropagation (BP) neural network for predictive modeling of catch weights. It was confirmed that lunar phase, tidal cycle, and noctilucent stick presence significantly affected stow net catches. While the BP neural network predictions closely matched the measured data, with the accuracy exceeding 89.63 % and 91.72 %. This study provided theoretical guidance for stow net fishing practices in the Yellow Sea.