关键词: 溶解氧, 溶解氧, 粒子群算法, BP神经网络, 优化, 罗非鱼池塘养殖

Abstract: The paper aims to provide support for the agricultural disaster prevention and mitigation, and for the fine forecast and early warning. Based on the data of diverse strong precipitation in summer (May to September) in Pukou during 1981-2017, we analyzed the characteristics of strong precipitation from time and amounts by adopting the methods of trend analysis, mutation test and the periodic analysis. The results showed that: (1) early and mid of July were peak period of continuous rainstorm and heavy rainstorm, the frequency of heavy rainstorm with 100- 150 mm was the highest and the heavy rainstorm had no obvious mutation in summer, but a remarkable 6-year cycle; (2) strong precipitation which reached the level of rainstorm earlywarning often started at the midnight and obviously became weakened in the afternoon; the strong precipitation duration was about 12 hours; the short-time precipitation always happened around the afternoon or in the dusk; (3) 40% of the warning signals and over half of the strong precipitation happened in 2008-2017, however, there was no obvious increase of the maximum precipitation per hour; (4) 39% of the strong precipitation reached the level of orange early- warning, but only one reached red early- warning. The main causes that the strong precipitation intensity in Pukou is stronger than that in urban Nanjing are the location and the natural environment.