This project belongs to the field of electrical engineering. With the rapid development of new energy sources, my country's cumulative installed capacity of photovoltaics has reached 170 million kilowatts. At this stage, the front-runner projects have basically achieved affordable Internet access, and the photovoltaic industry is developing towards high-quality and high-efficiency. However, the current grid-connected efficiency of some photovoltaic power stations is still low, and refined intelligent operation and maintenance technology is relatively lacking, which has become a key technical bottleneck restricting photovoltaic access to the Internet at affordable prices. How to improve efficiency and reduce costs from the two technical levels of control and operation and maintenance has become a core content that determines the future development of the photovoltaic industry. In order to solve technical problems in grid-connected control and intelligent operation and maintenance, and enhance the competitiveness of the photovoltaic industry, this project is funded by the National Natural Science Foundation of China, the National Key Basic Research and Development Plan, and the Shanghai City Science and Technology Commission. After ten years of research, a series of results have been achieved. The main innovations are as follows: (1) Aiming at the problem of low grid-connected efficiency of inverters under light load, comprehensive efficiency improvement technologies for photovoltaic grid-connected inverters of various power levels are proposed. A low-frequency model predictive control technology for grid-connected photovoltaic inverters is proposed to achieve optimal control of multiple technical indicators such as maximum power tracking, switching loss reduction and narrow pulse suppression. A current-source micro inverter based on navigation converters suitable for rooftop photovoltaics has been developed. It has high power density, small energy conversion loss, and a maximum efficiency of 95.02%. Invented and developed a multi-functional photovoltaic simulation system with a mixed current-source and voltage-source power. The conversion efficiency exceeds 85% in the power range of more than 70%, meeting the efficiency testing needs of various photovoltaic inverters. (2)Aiming at the voltage fluctuations caused by intermittent photovoltaic power generation, a multi-modal photovoltaic inverter cluster reactive power regulation technology is proposed. Build a theoretical model of reactive power dispatchable, tap the reactive power voltage regulation potential of inverter clusters, and suppress voltage fluctuations at parallel points. A multi-modal intelligent switching control is proposed to enhance the flexibility of reactive power regulation. According to different time scales of reactive power regulation, combined with voltage zoning and optimal selection of dominant nodes, the power grid voltage stability level has been improved, and the number of automatic actions of capacitors and transformer taps has dropped by 58% year-on-year. (3)Aiming at the reliability requirement of photovoltaic power station condition detection, a multi-parameter state evaluation and real-time detection technology for photovoltaic system is proposed. Build a high-speed switching characteristic model based on compressed sensing, complete 25-fold downsampling, and accurately and efficiently carry out device state assessment and life prediction. A technology for detecting the electrical characteristics and state of the outer end of the power module has been developed, which breaks through the problem of low accuracy in indirect measurement of the junction temperature of power devices. An empirical mode decomposition method based on clustering is proposed to improve the accuracy of DC arc fault feature extraction and recognition. (4)Aiming at the defect of low efficiency of manual operation and maintenance inspections, an intelligent operation and maintenance monitoring system has been developed. The cloud-based platform system can effectively remotely monitor each photovoltaic module. A short-term photovoltaic power generation combined prediction method based on historical data is proposed, which improves the prediction accuracy of photovoltaic power generation and enhances the dispatchability of photovoltaic power generation. An operation and maintenance system based on failure probability analysis was created, reducing downtime by approximately 10%. The project applied for 33 invention patents, including 15 authorized invention patents and 3 authorized utility model patents, published 39 papers (27 included in SCI/EI), participated in 1 local standard, and published 1 monograph. In the past three years, the new output value has been 806 million yuan and the new profit has been 146 million yuan. Some of the results have outstanding substantive characteristics and significant progress through novelty search. The project has been accepted by the expert group and has reached the international advanced level as a whole. The results have been promoted and applied to more than 100 photovoltaic projects in Jiangsu, Shanghai, Zhejiang, Anhui, Henan, Liaoning and other places, and have been applied to countries along the Belt and Road Initiative such as Bulgaria and Thailand, effectively improving photovoltaic power generation efficiency and achieving significant social and economic benefits, making positive contributions to the implementation of energy conservation and emission reduction.