针对现有的带式输送机关键部件监测系统存在的实时性差、盲点多等问题，以Thermovision A40型红外热像仪为基础，借助于Visual C++软件平台，融合红外热成像技术、计算机技术、图像处理及识别技术，研究开发基于红外热像的带式输送机关键部件监测与预警系统。本研究对于带式输送机关键部件温升故障实时诊断和预警具有重要意义。 研究适用于带式输送机关键部件红外图像的预处理算法，分别采用双边滤波和分段线性变换算法实现带式输送机关键部件红外图像的去噪和增强；研究适用于带式输送机关键部件红外图像的分割算法，通过改进的区域生长法对红外图像进行分割，并通过数学形态学运算对分割后的图像进行优化。实验结果表明，图像降噪增强以及红外图像分割效果良好。 针对带式输送机关键部件红外图像的特点，提出了基于形状特征和纹理特征相结合的BP神经网络关键部件识别算法，通过构造特征向量以及基于BP神经网络的分类器，进行实验验证，分类效果良好。 研究分析了红外热像诊断方法及带式输送机关键部件热故障，提出了带式输送机关键部件故障预警方案，利用SQL Server数据库，建立了用于故障预警系统的数据库，利用Visual C++，开发了带式输送机关键部件故障预警系统。 综合实验结果表明，本文设计的基于红外热像的带式输送机监测与预警系统能较好地解决带式输送机红外图像自动诊断和故障预警。

Aimed at some existing problems such as bad real-time performance and blind spots of the monitoring system for the key parts in conveyor belt, this thesis proposed the fault monitoring and predictive system based on infrared image, by Thermovision A40-M infrared camera. It’s combined the infrared thermal image technology, computer technology, image processing and recognition technology. This research has positive significance for the real-time temperature monitoring and fault early predictive. Infrared image preprocessing algorithm which is suitable for the key parts in conveyor belt has been studied, bilateral filter and piecewise linear transformation algorithm has been used for infrared image denoising and enhancement respectively; Infrared image segmentation algorithm which is suitable for the key parts in conveyor belt has been studied, the improved region growing method has been used for infrared image segmentation, and then through the mathematical morphology algorithm, the image after segmentation has been optimized. The experimental results show that infrared image denoising and enhancement algorithm and the infrared image segmentation algorithm have good effect. According to the characteristics of infrared image of the key parts in conveyor belt, feature vectors has been constructed through a combination of shape characteristic and texture features, and then the classifier which based on BP neural network has been builded. Experimental results show that feature vectors and the classifier which are used for infrared image of conveyor belt can reach an ideal classification effect. The fault early predictive scheme has been proposed by analysis of infrared image diagnosis method and thermal failure of key parts in conveyor belt, and database designed for fault early predictive system has been finished by using SQL Server database. Finally, fault early predictive system for key parts in conveyor belt has been designed and completed in the Visual C++ development platform. Experimental results show that conveyor belt fault monitoring and predictive system based on infrared image technology can well solve the problem of fault early predictive and automatic identification for key parts in conveyor belt.