County-level distributed photovoltaic (PV) power stations serve as a crucial carrier for the integrated development of new energy in rural areas

and the quality of their operation and maintenance (OM) directly determines the stability and economy of clean energy supply. To address the problems such as low OM efficiency

slow fault response

and persistently high costs caused by the "small-scale

scattered

and remote" characteristics of current county-level PV power stations

this paper systematically constructs a "five-in-one" intelligent OM theoretical system of "inspection

perception

analysis

decision

and execution". It also designs an overall architecture for the intelligent OM of county-level PV power stations based on UAV inspection

artificial intelligence

and big data technologies

which covers the UAV mobile perception layer

fixed monitoring layer

data service layer

and application service layer. Taking a PV cluster in a mountainous county in East China as the research object

the analysis shows that the system improves the accuracy of component defect identification to over 97% and reduces OM costs by approximately one-third. This effectively solves the key pain points in county-level PV OM and provides a reference model for the efficient operation of similar county-level PV power stations.