Abstract:Due to the complex composition of inspection data of new energy unattended field station, the reliability of inspection results is difficult to be guaranteed. Therefore, an automatic inspection method of new energy unattended field station based on AI machine vision technology is proposed. When a primary learner composed of multiple basic classifiers is used for the initial judgment of visual image categories, random forest is taken as a specific integrated learning method, and multiple decision tree classifiers are constructed to determine the category of new energy unattended station images based on the labels that appear most frequently in the output results. After selecting the partition features of the visual images of the optimal new energy unattended station based on information gain, ID3 decision tree algorithm is used to output the concrete inspection results. In the test results, mAP is always stable above 0.95. From the overall perspective, the mean value of mAP is higher than that of the control group by 0.0733 and 0.0903, respectively, with high reliability.