基于图像识别与语音交互的变压器发热缺陷检测系统设计

Design of transformer heating defect detection system based on image recognition and voice interaction

变压器发热缺陷识别缺少中值滤波预处理,导致系统检测信号解调率过低。因此,研究基于图像识别与语音交互的变压器发热缺陷检测系统设计。硬件部分:构建变压热敏传感器,分析热敏线的结构,对缺陷检测接口电路进行设计。软件部分:利用至导电杆采集变压器发热缺陷特征,对变压器发热缺陷信号进行处理,将处理后的特征结果转换为图像格式,通过中值滤波预处理,并进行图像识别,划分缺陷特征。利用语音交互,搭建缺陷自检辅助机制,基于此完成变压器发热缺陷检测。至此,完成对基于图像识别与语音交互的变压器发热缺陷检测系统的设计。实验结果分析:传统方法1系统检测信号解调率最高为0.39%,传统方法2系统检测信号解调率最高为0.38%,传统方法3系统检测信号解调率最高为0.44%,本方法系统检测信号解调率最高为0.59%,因此,本方法更佳。

Due to the lack of media filter pre-treatment, the demodulation rate of the system detection system is too low. Therefore, the design of a transformer heating defect detection system based on image recognition and speech interaction is being examined. Analyze the structure of the thermal line and design the interface circuit for fault detection. Software party: by using the control bar to collect the characteristics of transformer heat effects, the signal is processed from transformer heat effects, the processed functional results are converted to image format, through preparation of the media filter, and image recognition, the defect properties are shared. Using speech interaction, the support mechanism for defective self-testing is built, The design of a system for detecting defects in transformer heating based on image recognition and speech interaction has been completed. Experimental result analysis: the demodulation speed of the traditional method 1-detection signal is 0.39%, the traditional method 2-detection signal demodulation rate is 0.38%, the traditional method 3-detection signal demodulation rate is 0.44%, this method system detection signal demodulation rate is 0.59%, therefore this method is better.