基于BRISK的模板匹配在变电设备红外监测中的应用

为了提高电力系统变电设备红外监测的效率,同时对多个相似目标进行监测,提出了一种改进的基于BRISK（Binary Robust Invariant Scalable Keypoints,二进制鲁棒尺度不变特征）的模板匹配方法。首先在对目标模板图像与被监测图像预处理的基础上,利用BRISK算法提取模板与被监测红外图像的特征点,生成二进制描述子;利用滑动窗口在被监测图像上等步长滑动,检测出图像中存在的多个疑似目标区域,然后通过Hamming距离来判断正确匹配点对,从而正确识别变电设备。通过实验平台进行了相关实验,通过与模板匹配方法比较,该方法明显减少了运算量,充分满足系统实时性要求,提高了变电设备红外监测的效率。     

Phase transition and pitting corrosion behaviour for Mo—implanted aluminium

Molybdenum ions are implanted into aluminium with high ion flux and high dose at elevated temperatures of 200℃, 400℃ and 500℃. Due to the high temperature and high flux of vacancies and interstitial atoms, the atom diffusion and chemical effects are enhanced during the ion implantation. The effects increase with increasing ion flux and dose, so that new phase formation and phase transition emerge noticeably. X-ray diffraction analysis shows that when the aluminium is implanted with Mo ions at a low ion flux (25μA/cm²), the Al₅Mo alloy is formed. The atomic ratio of Mo/Al of the Al₅Mo phase is close to 20%. When the aluminium is implanted with Mo ions at a high ion flux (50μA/cm²), the phase transition from Al₅Mo to Al₁₂Mo appears, and the latter is dominant, which is determined to be the final phase. The ratio of Mo/Al in Al₁₂Mo is 7.7%. Rutherford backscattering spectroscopy indicates also that the Mo/Al atom ratio is ～7% to ～8% in Mo-implanted aluminium. The atomic ratios of the constituents in Al₅Mo and Al₁₂Mo are of stoichiometric composition for these alloys. The thicknesses of the Al₁₂Mo alloy layers for Mo-implanted Al with ion doses of 3×10¹⁷/cm² and 1×10¹⁸/cm² are 550nm and 2000nm, respectively. The pitting corrosion potential V_p increases obviously. It is clear that due to the formation of Al₁₂Mo alloy layer, the pitting corrosion resistance is enhanced.     

基于L0平滑和随机游走的变电设备红外图像分割

针对经典随机游走分割的不足,本文提出了基于L0平滑的随机游走分割方法。利用基于L0梯度最小化的图像平滑方法对红外图像进行预处理,在此基础上对图像进行随机游走分割。实验结果表明基于L0梯度最小化的图像平滑方法能够降低局部图像灰度值变化对经典随机游走分割路径的影响,在很大程度上提高了经典随机游走对变电设备红外图像分割的准确性。     

基于高斯尺度空间GHT的绝缘子红外图像识别方法

绝缘子作为输电线路中用量庞大的部件,其状态对输电线路的稳定运行至关重要。针对现场红外图像中绝缘子的识别问题,设计了一种高斯尺度空间 GHT 方法,基于此,提出了一种基于高斯尺度空间 GHT 的绝缘子红外图像的识别方法。首先利用形态学方法滤除图像中的小块和噪音,基于Canny检测算子提取图像边缘;然后对模板进行高斯尺度空间分解,将分解结果和输入图像进行基于GHT 的模板匹配;最后结合绝缘子的形状特征值识别出绝缘子。实验结果表明该方法可以准确地识别出绝缘子,并为后期绝缘子的故障诊断提供依据。     

电厂点检定修制研究

点检定修制是以点检为核心,协调全员对设备全过程进行动态管理的一种新型设备管理体制,本文介绍了点检定修制在某发电厂的实际应用情况和一些具体的措施,真正地把点检定修制的管理模式做好、做实、做顺,针对实行点检定修过程中出现的问题,通过落实责任制的考核来提高点检员的待遇,提出了加大点检定修培训力度,不断提高点检队伍的素质和水平,适应点检队伍不断扩大的需要,提高企业价值创造能力。