基于多能态粒子数叠加求和煤灰分精确定量分析研究

For a period of time, due to the structure of China's energy resources, coal in China's primary energy consumption will still remain at about 60% and more than 4 billion tons of coal will be consumed each year. Therefore, a new fast coal quality analysis method that meets the national standard GB/T212-2008 is necessary to replace the traditional one which needs to take 2~4 hours. Laser-induced breakdown spectroscopy (LIBS) is a new quantitative compositional analysis method and a hot spot in spectral analysis field. However, the key problem that restricts the popularization and application of LIBS technology is the measurement repeatability of ash content in coal cannot satisfy the national standard by using the existing methods such as the local thermal equilibrium state theory, the optimized data processing, etc. This project intends to theoretically deduce a criterion of non-self-absorption effect (or very weak) presented in plasma based on the principle of physics. The criterion is independent of any unsure physical parameters or equipment acquisition settings, and is able to determine the best spectra acquisition time window as well as the reliable plasma temperature. By combining the temperature with the special recursive method, the number of multi-ionized particles of different elements in the plasma as well as the compositional concentrations can be obtained.

由于中国能源资源结构的原因，煤炭在我国一次能源消耗中未来一段时间内仍然会达到60%左右。我国每年煤炭消耗总量在40亿吨以上，传统的煤质分析方法需耗时2~4小时显然已不能适应目前大量的快速的煤质分析要求，急需一种满足国标GB/T212-2008新的快速煤质分析方法。利用激光诱导击穿光谱对物质成分进行定量分析是光谱分析领域一种崭新的分析手段和热点。但在该技术分析煤中灰分的实践中，通过利用局部热平衡态理论及优化分析算法等传统的方法和措施，其测量重复性指标仍达不到国标要求，这成为制约该技术推广应用的关键问题。本项目拟从物理学原理出发，理论推导出一个基于LIBS的无自吸收（或很弱）效应等离子态的判据，该判据不包含无法确定的物理参数和设备采集参数等因素，从而能确定出一个最佳光谱时间窗口，进而得到较高精度的等离子体温度，再用特殊的方法递归计算出各元素在等离子体中各能态粒子数，即可得到该物质各成分含量。

项目组通过理论分析获得等离子体无自吸收效应判据，并形成了可方便操作的实验判别流程和方法，这为后继数据分析提供有效荧光光谱；研究了多能态粒子数叠加求和反演元素含量的方法，基本消除了由于激光激发能量起伏和等离子体空间分布变化等因素所导致的单个能态粒子数波动的影响；设计了可输出反馈的最小均方自适应谱线增强算法，优化去相关参数，评判算法收敛速度、稳态误差及跟踪能力，在不损坏突变光谱前提下，通过滑动平均滤波来去除非高斯噪声，改善等离子体荧光光谱信噪比。这些工作使基于LIBS分析煤灰分的测量重复性指标能满足国标GB/T212-2008的要求。利用以上理论和技术成果，通过时间分辨率光谱采集和控制，采用基于等离子体光学薄判据的自吸收免疫激光诱导击穿光谱技术，利用多能态粒子数反演、主成分降维、小波变换及支持向量机，建立了激光诱导光谱至燃煤灰分、挥发分及发热量指标的转化模型，对燃煤工业指标中高位发热量、灰分、挥发分分析的均方根误差分别达到0.85MJ/kg、1.82%和1.22%，其中对灰分测量的重复性为0.14%。项目实施期间，项目组共发表SCI论文14篇，授权国家发明专利6项，毕业博士2人、硕士2人。总体完成了项目考核任务要求，达到了预期目标。