基于麻雀算法优化支持向量机的NOx浓度预测

NOx Concentration Prediction Based on Sparro w Algorithm Optimized Support Vector Machine

煤炭作为火电厂发电的主要能源,其在锅炉内焚烧过程中会产生大量的氮氧化物。各电厂一般利用烟气自动监控系统对其浓度进行实时测量,但由于测量时存在较大迟延,不能准确地反映SCR系统NOx浓度的实时变化。因此提出了一种基于改进麻雀算法优化最小二乘支持向量机的NOX浓度预测方法。首先,引入余弦因子改进麻雀算法中的比例算子,将迭代次数信息引入到迭代过程中,平衡算法前后期的全局与局部搜索能力。其次,使用新的变异算子代替原算子,将混沌理论融合到麻雀算法,解决了算法全局搜索能力较差、初始化麻雀分布不稳定及发现者位置更新方式不足的问题。最后,采用改进麻雀算法(CDE-SSA)对最小二乘支持向量机(LSSVM)进行参数寻优。实验结果证明,方法在NOX浓度预测的精度和稳定性上均表现出了良好的性能。

As the main energy source for power generation in thermal power plants,coal gene rates large amounts of NOx during combustion in the boiler.Vario us power plants generally use automatic flue gas monitoring systems to measure their concentrations in real-time,b ut due to the large delay in measurement,they cannot accurately reflect the real-time changes in NOx concentration a t the SCR system.This paper proposes a NOx co ncentration prediction method based on an improved sparrow algorith m optimized by least squares support vector m achines.Firstly,the cosine factor is introduced to improve the pr oportional operator in the sparrow algorithm,and the information on the number of iterations is introduced into the i terative process to balance the global and lo cal search ability in the first and second stages of the algorithm.Secondly,a new variational operator is used instead o f the original operator,and chaos theory is integrated into the sparrow a lgorithm to solve the problems of poor global search ability,unstable distribution of initialized sparrows and insuffici ent ways to update the location of discoverer s.Finally,the improved sparrow algorithm(CDE-SSA)is used to perform parame ter search for the least squares support vect or machine(LSSVM).Experimental results demonstrate that the me thod shows good performance in terms of both accuracy and stability of NOx concentration prediction.