超声波测厚是维护管道安全性的重要无损检测技术，其测量原理是利用超声波脉冲反射法在线监测石油管道的厚度和腐蚀情况。超声波脉冲反射法测厚的精度主要依赖于声速，而声速受环境温度的影响较大，故对超声波在固体中的传播速度和温度关系进行了大量的试验，在这些研究的基础上，提出了温度与声速的神经网络误差补偿模型和线性回归误差补偿模型。把两种模型融入到超声波脉冲反射法中，对两种模型的误差补偿效果进行了对比分析，开发出了超声波声速自校正误差补偿在线监测系统，在一定程度上消除了温度对超声波测厚造成的影响。

Ultrasonic thickness measurement is an important nondestructive testing technology for maintaining pipeline safety. Its measurement principle is to use the ultrasonic pulse reflection method to monitor the thickness and corrosion of oil pipelines online.The accuracy of ultrasonic pulse reflection thickness measurement mainly depends on the speed of sound, whereas the latter is greatly affected by the ambient temperature. Therefore, the author has performed a lot of experimental studies on the relationship between the speed of propagation of ultrasonic waves in solids and different temperature. Based on these studies, a neural network error compensation model and a linear regression error compensation model for temperature and sound velocity are proposed. The two models are integrated into the ultrasonic pulse reflection method, and the error compensation effects of the two models are compared and analyzed. An ultrasonic sound velocity self-correction error compensation online monitoring system is developed, which eliminates the error caused by temperature influences on the ultrasonic thickness measurement to a certain extent.