In order to realize the purpose of better inspection on the outer surface of high-rise buildings, comprehensively detect the damage of the building surface, so as to find various problems, such as cracks, bulge and falling off, a UAV inspection image acquisition and processing system based on convolutional neural network is designed. The UAV uses open MV to realize the efficient processing of visible light image information, and uses an embedded system for image fusion processing. Using convolutional neural network model and multi-object detection algorithm, extract the feature value of broken position and combine multiple sensors for 3 D position transformation. Data reads and 3 D coordinate map presentation were performed using the app software. For the safety problem of ceramic tiles, a marking device is installed on the UAV. Calculate the threat range and use the application to send location information to maintenance personnel and alert information to residents, and provide real-time location safety monitoring. The experimental results show that the designed uav can realize comprehensive, automatic, intelligent, efficient and preventive inspection of the building health status and problems of buildings through inspection, data collection and processing, intelligent analysis and early warning.

In this study, we analyzed tennis match data to answer various questions related to tennis. Firstly, we constructed a model to capture the process of the game and evaluate the performance of players. By preprocessing and analyzing the data, including eliminating outliers, filling in missing values, and standardizing the data, we obtained prepared analytical data. Subsequently, using principal component analysis, three representative principal components were identified. In addition, by using the entropy weight method, we calculated the weights, enabling us to derive a momentum formula for visualizing the temporal changes during the competition process. Secondly, we explored whether there is a correlation between match results and momentum. By using the momentum model we established, we used Pearson correlation analysis to calculate the correlation coefficient, thereby determining the relationship between momentum and game results.