| The pantograph is the key equipment for trains to obtain electric energy from the power grid,and its quality determines the safety of the train’s electric energy transmission and operation.The current pantograph detection technology is difficult to meet the real-time and accurate requirements.Therefore,this thesis designs and improves a target tracking algorithm to quickly detect the pantograph dropout fault while monitoring the operational state of the pantograph in real time.Based on the KCF algorithm,this thesis focuses on the optimization and improvement of the KCF algorithm to solve the problem of its incompetence to meet the current requirements of the train operation scene change,and then designs an algorithm that can be practically applied to the safe state detection of the train pantograph.Specific contents include:(1)Aiming at the scale change,complex background,light intensity,occlusion and target loss in the KCF algorithm,this paper proposes to utilize the maximum response average and the redefined Response value to judge the KCF algorithm model and propose a solution.Aiming at the problem of the target loss and the irretrievability of the KCF algorithm,this thesis introduces the YOLOv3 detection mechanism to correct the KCF algorithm.At the same time,an early warning mechanism for target loss is introduced which will be activated when the target detection fails.The experimental results show that the KCF target tracking algorithm based on the YOLOv3 detection algorithm can effectively solve the problems of target scale change,complex background,target occlusion mechanism and target loss.(2) By applying the improved KCF algorithm to the rapid detection of the safe state of the pantograph,the algorithm can accurately detect the pantograph under the conditions of scale conversion,occlusion,and complex background.And the algorithm can give early warnings in time when the pantograph falls off abnormally or fall-off faults occur.In the same batch of pantograph test data sets,with mAP as the evaluation index,the KCF-YOLOv3 algorithm proposed in this thesis is better than the other two algorithms.Among them,the mAP of the KCF-SSD algorithm is 0.9953,the mAP of the KCF-YOLOv2 algorithm is 0.9968,and the mAP of KCF-YOLOv3 can reach 1.The experimental results show that the improved KCF algorithm can track the pantograph more accurately and provide timely warnings of its falling state.In conclusion,this thesis uses the YOLOv3 detection algorithm to optimize and improve the KCF target tracking algorithm and has verified the accuracy and effectiveness of the improved algorithm by experiments.Based on ensuring the speed of the algorithm,the accuracy of the algorithm is improved,the existing problems of KCF are effectively solved and improved the robustness of the algorithm.Finally,experiments are designed to verify the effectiveness of the improved algorithm,and the improved KCF algorithm is applied to the safe state detection of the pantograph. |
