Abstract:

To improve the prediction ability of the crab production in Jiangsu and make a scientific and reasonable plan for crab culture to reduce disordered competition and unbalanced supply and demand, meteorological data in seven observation stations near the crab producing areas from 2013 to 2017 and crab yield data of the region were collected. The interpolation method was used to make up for some missing values. The principal component analysis (PCA) was used to select the five principal components that best represented the characteristics of data. Back propagation (BP) neural network was used to explore its feasibility for predicting the changing trend of meteorological data in these areas. After dimensionality reduction processing by PCA, the results show that most of the characteristics of the data can be learned by the BP neural network. The correlation coefficient between the predicted value and the true value of yield is 0.82267, which means the BP neural network has a certain simulated relationship capacity between weather factor data and the yield.

Key words: meteorological factor, back propagation (BP) neural network, yield of crab, short-term prediction