Marine and port economy estimation is conducive to the understanding of the development law of marine economy degree. This study proposes a neural-learning network estimation model of the marine economy degree based on a priori architectural knowledge and adopts time-combined class array columns and multivariate modeling methods to estimate the indicators reflecting the development level of the marine economy degree in the ZJP region. The study adopts the PK_NN multivariate modeling method, taking cargo transportation volume, cargo turnover, cargo carrying value of ports near the sea, foreign trade throughput, port container throughput, and port container throughput as multivariate model inputs, and compared with the other modeling methods, the model of the time combination class array columns of type GM_11, which has a better comprehensive performance. Finally, the PK_NN time-combined class array column model is used to estimate the development level of the marine economy in the ZJP region near the seaport from 2011 to 2020, and the results show that the estimated value of the marine economy in the ZJP region is close to the actual planning value of the ZJP region. The algorithm was applied to estimate the economic degree curves of the five near-seaport areas of ABCDE under different harbor head construction art modes, and the results showed that the relative value error of the estimation was controlled between 4% and 10%, and the fluctuation ranges of each month's specific growth value area estimation were comparable. This proves the effectiveness and accuracy of the a priori marine neural-learning-based network algorithm in this paper.
Read full abstract