项目名称: 熟料质量稳健检测中的关键技术与并行实现方法研究
项目编号: No.61203016
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 陈华
作者单位: 湖南大学
项目金额: 26万元
中文摘要: 烧结工艺在诸如水泥、氧化铝等生产领域应用广泛,一直存在高能耗,粗放生产等问题,是节能减排工作的重点。其中熟料质量是各类烧结过程实现最优控制和节能降耗的关键工艺检测参数,但国内外均少有熟料质量的在线检测方法研究。 课题结合典型的工业回转窑烧结实例,在课题组前期研究基础上,针对现场熟料图像受光照、粉尘干扰较大,特征不稳定,现有分类检测稳健性不强的实际问题,研究熟料质量稳健检测中的关键鲁棒技术和快速并行实现方法。首先研究一种基于压缩感知的稳健视频纹理特征提取方法;其次研究利用稳健估计理论对ELM进行改进,结合视频纹理特征和窑前热工数据,对熟料质量进行鲁棒的融合检测;最后研究建立一套基于CUDA并行环境下的熟料质量机器视觉稳健检测系统。上述研究内容对于辨识烧结过程工况、优化控制参数、实现更加稳定高效的烧结过程控制具有重要的现实意义,同时对提升其他工业软测量系统的鲁棒性也有重要借鉴意义。
中文关键词: 回转窑;烧结过程;熟料检测;图像特征提取;并行计算
英文摘要: Sintering process is widely used in cement, alumina production areas, but its high energy consumption and extensive production is the main works of energy conservation and emission reduction. Especially the clinker quality is the key parameter to achieve optimal control and reduce consumption in sintering process, but the on-line measurement methods are rarely mentioned both at home and abroad. Based on the earlier research, aimed to the fuzzy image sequence interfered by lightness and dust, this project researches the steady measurement technology and fast parallel implement method for clinker quality. Firstly, CS (Compressive Sensing) is used to extract steady features of clinker sintering image sequence. Secondly, ELM (Extreme Learning Machine) combined with robust estimation theory is researched for the robust fusion measurement. At last, the steady machine vision measurement system of clinker sintering will be built in the CUDA parallel environment. The research is not only very important for optimization of sintering process, but also has significant reference for other robust soft computing measurement with industrial images.
英文关键词: rotary kiln;the process of sintering;clinker quality detection;image feature extraction;parallel computing