基于神经网络的LF炉钢液温度的预测模型

针对现有炼钢厂在LF炉钢液温度的控制方面,大多采用人工测温的方式,存在测量精度差,调节时间长等缺点,不利于钢的质量稳定且不利于降低成本。以42CrMo4钢种为例,通过进行相关性分析,筛选出影响42CrMo4的LF炉终点温度的因素,以这些影响因素作为输入,LF炉终点温度作为输出建立了神经网络的LF炉温度预测模型,探讨给定这些输入量,是否能得到较为准确的终点温度,实验表明神经网络模型能够很好地预测LF炉钢液温度。

10kV高压开关柜电气过热预警在线监测系统的应用

研究了实现10kV开关柜内关键点温度实时在线监测的可行性和可靠性,以及过热预警在线监测系统代替传统的人工巡检测温的可行性。通过在某110kV变电站十几个10kV开关柜内安装电气过热预警在线监测系统,观察其上传数据的正确性、可靠性、稳定性以及运行期间设备的安全性并开展相关研究。该系统的安装可实现10kV开关柜内关键点温度的实时在线监测,数据上传稳定连续,经和实测温度比较,大部分数据正确可靠,目前未发现该系统的安装对设备运行造成安全隐患。但还是存在少部分数据不能正确反映实时温度的现象,同时在运行维护方面还存在诸多问题。因此,10kV开关柜电气过热预警在线监测系统还不能代替传统的人工巡检测温,还需要不断研究和完善。

Infrared radiation method for measuring ice segregation temperature of artificially frozen soils

In order to study the evolution of the freezing fringe and final lenses of frost susceptible soils and advance the understanding of frost heave and mechanism of frost heave control, we used an open one-dimensional frost heave test system of infrared radiation technology, instead of a traditional thermistor method. Temperatures of the freezing fringe and segregated ice were measured in a non-contact mode. The results show that accurate and precise temperatures of ice segregation can be obtained by infrared thermal imaging systems. A self-developed inversion program inverted the temperature field of frozen soils. Based on our analysis of temperature variation in segregated ice and our study of the relationship between temperature and rate of ice segregation in cooling and warming processes during intermittent freezing, the mechanism of decreasing frost heave of frozen soils by controlling the growth of final lenses with an intermittent freezing mode, can be explained properly.