As a kind of clean energy, photovoltaic power generation can help reduce the dependence on fossil energy for agricultural production and promote the development of sustainable precision agriculture. In precision agriculture, improving the energy efficiency after photoelectric conversion is always a key research issue. By predicting the power generation of photovoltaic cells, the power load can be better arranged in agricultural production, and the agricultural equipment can be rationally scheduled, so as to improve the power utilization efficiency after photoelectric conversion. This paper not only presents a complete set of photovoltaic power prediction system, but also optimizes the prediction algorithm based on the traditional BP neural network. In view of the problems of traditional BP neural network, such as easy to fall into local extreme values and slow convergence speed, GA-IPSO hybrid algorithm is used to optimize the weight and threshold of BP neural network, and the prediction accuracy is compared with various algorithms. Through this method, higher prediction accuracy is obtained. Experiments show that the accuracy of GA-IPSO algorithm has higher advantages in power prediction. It can better assist the power supply scheduling work in precision agriculture.