Research And Implementation Of Insulator Defect Detection Based On ZYNQ

As an important resource in our country,power resources play an extremely important role in ensuring rapid economic development and improving people’s lives.In recent years,my country’s power industry has always maintained a relatively high growth rate,so the operation and maintenance of the power system is an extremely important task.Among them,the normal function of the insulator of the transmission line plays a vital role in the safe and stable operation of the entire line.Therefore,accurately discovering and identifying faulty insulators during line inspection is of great significance for maintaining line safety.Traditional insulator defect detection is mostly based on embedded processors,CPUs and GPUs.Its embedded processors have poor computing power and cannot handle complex insulator defect detection algorithms.On the other hand,CPU and GPU platforms are used to exchange real-time data during circuit inspections through the network.Poor sex.Aiming at the shortcomings of traditional insulator defect detection application scenarios,this project is based on ZYNQ to implement insulator defect detection and detection applications,aiming to meet the current embedded application platform’s power consumption constraints to apply insulator defect detection algorithms.In order to reduce the calculation amount of the algorithm and meet the high efficiency of insulator defect detection,further research has verified the insulator defect detection convolutional neural network algorithm based on the improved YOLO v2-Tiny,which makes the improved insulator defect detection algorithm more suitable for the defect location of the insulator image Detection and positioning.Use the open source insulator data set to train the insulator defect detection algorithm model in this thesis.Finally,the insulator defect detection algorithm in this thesis is compared with the SSD and Faster R-CNN network models.The results show that the improved insulator defect detection algorithm in this thesis has better detection results.Based on the ZYNQ implementation of the insulator defect detection algorithm,this thesis analyzes the ZYNQ platform technology,the parallelism of the insulator defect detection algorithm,the convolutional data buffer implementation method,the activation function hardware implementation method,and the fixed-point design error,and then implements the algorithm for ZYNQ.Provide evidence.In order to realize the algorithm of this thesis based on the realization of ZYNQ,the hardware module design scheme of this algorithm is introduced in detail.First,the algorithm hardware modules are designed and optimized in the Vitis HLS advanced synthesis tool,so that each algorithm module can achieve the expected algorithm tasks and calculation accuracy.Then generate the corresponding IP cores for each algorithm module,and connect each IP core with the auxiliary IP core under the Vivado software to build a complete hardware system.At the same time,develop application layer software under the Vitis environment to realize the synthesis and algorithm of the algorithm module Task scheduling.Finally,the ZYNQ platform implementation of the algorithm in this thesis is compared with the implementation of this algorithm under the CPU and GPU platform.The final experimental results show that its computational power consumption has a very obvious advantage compared with other platforms,which improves the efficiency of target detection and provides power transmission for today The line inspection system provides a new solution.