摘要
为了有效解决电缆、发电设备人工巡检作业难度大、危险系数高、效率低等问题,结合电力巡检无人机与视频处理关键技术,提出基于改进卷积神经网络与图像特征融合处理技术相结合的电力设备异常状态识别算法,并将其应用于发电设备及电缆巡检中.所提技术采用图像融合技术,通过图像匹配、特征处理后,提出冗余信息,保留分类特征项.采用视频压缩技术,对经过关键数据提炼后的视频数据,作为卷积神经网络模型的训练样本构建无人机巡检视频分析模型,识别设备异常状态.同时,使用交叉熵函数优化卷积神经网络模型,改善分类效果.将提出的算法与其他图像识别算法进行比较,结果表明所提出的算法准确率和处理性能均为最佳,具有一定的应用推广价值.
Abstract
In order to effectively solve the problems of difficult,high risk factor and low efficiency of manual inspection opera-tions of cable and power generation equipment,this paper combines the key technologies of power inspection drones and video processing,and proposes an abnormal state recognition algorithm for power equipment based on the combination of improved convolutional neural network and image feature fusion processing technology,and applies it to the inspection of power genera-tion equipment and cables.The technique uses image fusion technology to propose redundant information and retain classifica-tion feature terms after image matching and feature processing.Further,video compression technology is used to construct the UAV inspection video analysis model as the training sample of convolutional neural network model to identify the abnormal state of equipment for the video data after key data refinement.Meanwhile,the cross-entropy function is used to optimize the convolutional neural network model and improve the classification effect.The proposed algorithm is compared with other image recognition algorithms,and the results show that the accuracy and processing performance of the proposed algorithm are the best and have certain application promotion value.
维普
万方数据