I am glad that we have come to the SNEC exhibition in 2026. Today we met Liu Fuheng, general manager of Shandong Hengtong Vision Technology Co., Ltd. Hello.
Hello
Can you tell us what the company's main business is?
OK. Our company mainly focuses on the track of AI plus new energy. The company's main business is divided into two forms of products and services. Services, such as the intelligent inspection of photovoltaic power plants , as well as the shutdown and non-shutdown inspection of wind turbines, can be closed-loop in the whole process, from the early modeling, route planning, to the later data acquisition, AI analysis, to the final report. The product is mainly aimed at the intelligent inspection system of photovoltaic power plants. This system is a software and hardware collaboration, which can also support cloud and local deployment modes. It is also more flexible in deployment mode, in order to better adapt to different types of power plants. What do you think is the biggest pain point
for the current operation and maintenance of photovoltaic power plants?
At present, the pain points of the track of transportation and inspection in the photovoltaic industry can be roughly divided into several categories. First of all, the labor cost is relatively high. At present, the annual per capita operation and maintenance cost has probably exceeded 80,000 yuan per megawatt. Especially in some complex situations such as mountains and water, its operation and maintenance cost will be higher. Moreover, for photovoltaics such as mountains and water surfaces, there are safety risks such as electric shock and falling from high altitude, which are human costs. Moreover, the efficiency of manual inspection is also relatively low, about a 50 MW photovoltaic power station, manual inspection takes about 10 people a week, and it is difficult to find some hidden defects. Another aspect is that the quality of manual inspection is also difficult to guarantee. Manual inspection mainly relies on the naked eye, not through some intelligent tools, so it will lack some scientific basis and scientific maintenance programs. Moreover, a large number of operation and maintenance data of photovoltaic power plants have not been effectively utilized, and the value of data has been submerged. This problem is also due to the lack of professional data analysis capabilities, the overall level of intelligence in the photovoltaic industry needs to be improved, and people's acceptance of intelligent tools is also a process that needs to be improved. There are probably these pain points and problems.
As for technology and barriers, we start from these two aspects, one is data, the other is AI technology. In terms of data, we have accumulated a large number of hidden defect data, the number has reached about hundreds of thousands, through these high-quality hidden images, we can train high-precision recognition algorithms. First, the number of these data is relatively large, and then it covers more types of photovoltaic power plants, such as mountain, plane and water. In addition, we have carried out several rounds of data processing on these large amounts of image data, such as data annotation, data review and data pre-training. After training, there is a feedback, and we will carry out another round of modification and training. Therefore, in the case of a large amount of data training, our algorithm will also have a high accuracy. With high-precision algorithms, our self-developed intelligent inspection system for photovoltaic power plants also maximizes the use of lightweight design. With this design concept, we have integrated AI technology into all processes of photovoltaic operation and maintenance inspection. What
do you think the future of photovoltaic intelligent inspection will look like?
On this issue, we can first look at the top-level design of the national policy. From the AI plus energy released some time ago, to the two-way integration of AI and energy, and the 51 high-value application scenarios recently released, in fact, behind these policies are sending a core signal that the country is upgrading the integration of AI and energy from a technology application to a national-level strategic engine. This is from the policy point of view. From the perspective of industry development, the development of photovoltaic industry requires more and more refinement, so the original extensive mode of manual inspection may not adapt slowly. Another point can be seen from where, like the technology of UAV and AI big model is constantly breaking through and landing, which can really help our photovoltaic power plant operation and maintenance patrol to solve some pain points. So from these three perspectives, I think the future development trend is definitely to carry out a change and a development driven by data and AI under the mode of AI plus energy.
浙公网安备33010802003254号