电力线除冰机器人基于粒子群优化的小波神经网络障碍物识别方法
发布时间：2019-06-24 22:43
【摘要】：由于除冰机器人工作在覆冰的电力线上,障碍物的识别存在着各类障碍物区分较难,准确率较低等不足。为提高机器人自主识别能力,设计一种自适应阀值的小波变换边缘提取算法来提取出障碍物的图像边缘,并针对电力线障碍物结构特点,在障碍物边缘提取过程中设计一种基于电力线位置约束的有效剔除部分干扰背景的方法;引入小波矩,通过提取边缘图像的小波矩作为障碍物的特征匹配数据;根据神经网络和粒子群算法的原理,设计一种粒子群优化的小波神经网络进行障碍物的识别分类,该方法用粒子群算法取代传统的梯度下降法,并改进惯性权重因子,优化小波网络的各个参数。试验结果表明所提出的方法对电力线上防震锤、悬垂线夹和耐张线夹等障碍物能够有效地识别,并具有比普通识别方法更高的识别精度。
[Abstract]:Because the deicing robot works on the power line covered with ice, there are some shortcomings in the recognition of obstacles, such as difficult to distinguish all kinds of obstacles, low accuracy and so on. In order to improve the autonomous recognition ability of the robot, an adaptive threshold wavelet transform edge extraction algorithm is designed to extract the image edge of the obstacle, and according to the structural characteristics of the obstacle, an effective method based on the power line position constraint is designed to eliminate the interference background, and the wavelet moment is introduced to extract the wavelet moment of the edge image as the feature matching data of the obstacle. According to the principle of neural network and particle swarm optimization algorithm, a wavelet neural network optimized by particle swarm optimization is designed to identify and classify obstacles. In this method, particle swarm optimization algorithm is used to replace the traditional gradient descent method, and the inertia weight factor is improved to optimize the parameters of wavelet network. The experimental results show that the proposed method can effectively identify obstacles such as shock hammer, drape clamp and tension clamp on the power line, and has higher recognition accuracy than the common recognition method.
【作者单位】： 湖南大学电气与信息工程学院;邵阳学院多电源地区电网运行与控制湖南省重点实验室;
【基金】：国家科技支撑计划(2015BAF11B01) 湖南省科技计划(2016TP1023) 湖南省教育厅科研(14C1015)资助项目
【分类号】：TP183;TP242
[Abstract]:Because the deicing robot works on the power line covered with ice, there are some shortcomings in the recognition of obstacles, such as difficult to distinguish all kinds of obstacles, low accuracy and so on. In order to improve the autonomous recognition ability of the robot, an adaptive threshold wavelet transform edge extraction algorithm is designed to extract the image edge of the obstacle, and according to the structural characteristics of the obstacle, an effective method based on the power line position constraint is designed to eliminate the interference background, and the wavelet moment is introduced to extract the wavelet moment of the edge image as the feature matching data of the obstacle. According to the principle of neural network and particle swarm optimization algorithm, a wavelet neural network optimized by particle swarm optimization is designed to identify and classify obstacles. In this method, particle swarm optimization algorithm is used to replace the traditional gradient descent method, and the inertia weight factor is improved to optimize the parameters of wavelet network. The experimental results show that the proposed method can effectively identify obstacles such as shock hammer, drape clamp and tension clamp on the power line, and has higher recognition accuracy than the common recognition method.
【作者单位】： 湖南大学电气与信息工程学院;邵阳学院多电源地区电网运行与控制湖南省重点实验室;
【基金】：国家科技支撑计划(2015BAF11B01) 湖南省科技计划(2016TP1023) 湖南省教育厅科研(14C1015)资助项目
【分类号】：TP183;TP242
【相似文献】
相关期刊论文 前5条
1 刘鸿雁;;一种新的车辆倒车突发障碍物规避方法仿真[J];计算机仿真;2014年08期
2 赵哲;马晓s，
本文编号：2505421
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2505421.html
