重金属形态预测建模中GEP算法的应用研究

发布时间：2018-07-06 12:04

本文选题：GEP算法 + 滑动窗口预测法　；参考：《河北工程大学》2017年硕士论文

【摘要】：城市生活垃圾焚烧底灰中重金属在填埋场中经历长期复杂的物理和化学反应,不断发生形态变化和迁移,对周围环境形成持续的潜在风险。由于重金属的形态决定了重金属在环境中的迁移转化行为,因此一些研究人员利用化学提取法、模型计算法以及仪器检测等方法来分析重金属在环境中的化学形态,并以此来诠释重金属在环境中的迁移转化规律,从而减小城市生活垃圾焚烧技术所带来的二次污染危害。然而,由于重金属形态的变化过程复杂且影响因素较多,从而导致以上几种方法存在研究结果不准确以及研究内容不全面等缺陷,加上现场调研数据有限以及实验室模拟试验成本高,从而给研究重金属在环境中的化学形态带来了不可逾越的困难。因此,本文旨在建立一个重金属形态预测模型,获取表征重金属在填埋场中长期迁移转化行为的数据,为研究重金属的长期迁移转化机理提供基础。随着计算机科学技术的发展,越来越多的研究者开始将人工智能类算法应用到预测模型的建立中。基因表达式编程(Gene Expression Programming,GEP)是近几年逐渐发展起来的一种新型的用于数学建模的人工智能方法,它在解决函数发现、分类分析以及时间序列分析等问题方面表现出优良的性能,适用于解决时间序列的预测问题。因此,本文主要研究GEP算法在建立重金属形态预测模型中的应用。本文首先通过与其他进化算法进行比较,来突出GEP算法在解决复杂问题方面的主要优势和特点,然后又详细介绍了GEP算法中的基本组成要素及其进化过程。然而,尽管与其他进化算法相比而言,GEP算法具有较快的进化速度以及较高的计算精度,但它本身仍然面临着容易陷入局部收敛以及迭代计算时间长等缺陷,进而影响其解决复杂问题的能力。因此,为了改善这种情况,本文分别从适应度函数和遗传操作两个方面对GEP算法进行了针对性的改进,并利用改进后的GEP算法来分析已有的时间序列数据,建立重金属形态预测模型,用来预测重金属形态在未来时间点的变化数据。由于本文已有试验数据是一个简单的时间序列,直接利用该数据所建立的预测模型,难以准确地预测填埋场中重金属形态随未来时间变化的数据。因此,为了进一步提高预测精度,本文又将滑动窗口预测法(Sliding Window Prediction Method,SWPM)引入到预测模型的建立中,与改进GEP算法相结合构成GEP滑动窗口预测法(GEP-SWPM),对已有试验数据进行相空间重构,从而建立一个多维的重金属形态预测模型。最后,本文分别利用标准GEP算法、改进GEP算法以及GEP-SWPM算法建立预测模型,来预测重金属形态在填埋场中的未来变化数据。实验仿真结果表明,与前两者相比而言,GEP-SWPM算法在建模效率和预测精度上都有了很大的提高,更适合于重金属形态变化的预测,有利于发现重金属形态随时间变化的规律,从而为研究重金属的长期迁移转化机理提供了良好的数据基础。
[Abstract]:Heavy metals in the bottom ash of municipal solid waste incineration have experienced long-term complex physical and chemical reactions in the landfill, constantly changing and migrating, forming persistent potential risks to the surrounding environment. Because the form of heavy metals determines the migration and transformation of heavy metals in the environment, some researchers use chemical extraction, Model calculation method and instrument detection method are used to analyze the chemical forms of heavy metals in the environment, and to explain the migration and transformation of heavy metals in the environment, thus reducing the two pollution hazards caused by municipal solid waste incineration technology. However, because of the complex process of heavy metal form changes and many factors affecting it, Because of the inaccuracy of the above methods and the incomplete research content, the limited field investigation data and the high cost of laboratory simulation test have brought insuperable difficulties to the study of the chemical forms of heavy metals in the environment. Therefore, this paper aims to establish a prediction model of heavy metal form. With the development of the computer science and technology, more and more researchers began to apply artificial intelligence algorithm to the establishment of prediction model with the development of computer science and technology. Gene expression programming (Gene Expression Programming) GEP) is a new artificial intelligence method, which has been developed gradually in recent years. It shows excellent performance in solving problems such as function discovery, classification analysis and time series analysis. It is suitable for solving the prediction problem of time series. Therefore, this paper mainly studies the GEP algorithm in the formation of heavy metal form. In this paper, the main advantages and characteristics of GEP algorithm in solving complex problems are highlighted by comparing with other evolutionary algorithms, and then the basic components and evolution process of the GEP algorithm are introduced in detail. However, although compared with other evolutionary algorithms, the GEP algorithm has a faster evolution. The speed and high computing precision, but it still faces the defects of easy to fall into local convergence and the long time of iterative computation, and then affect its ability to solve complex problems. Therefore, in order to improve this situation, this paper improves the GEP algorithm from two aspects of fitness function and genetic operation. Using the improved GEP algorithm to analyze the existing time series data and establish the prediction model of heavy metal form, it is used to predict the change data of heavy metal form in the future time point. Since the existing experimental data is a simple time series, it is difficult to predict the landfill site accurately by using the prediction model built directly by the data. In order to further improve the prediction accuracy, the sliding window prediction method (Sliding Window Prediction Method, SWPM) is introduced into the prediction model, and the GEP sliding window prediction method (GEP-SWPM) is combined with the improved GEP algorithm, and the phase space of the existing experimental data is carried out. In the end, this paper uses standard GEP algorithm, improved GEP algorithm and GEP-SWPM algorithm to establish prediction model to predict the future change data of heavy metal form in the landfill. Experimental simulation results show that compared with the previous two, the GEP-SWPM algorithm is efficient in modeling efficiency. And the prediction accuracy has been greatly improved, which is more suitable for the prediction of heavy metal morphological changes, which is beneficial to the discovery of the regularity of heavy metal morphologic changes with time, thus providing a good data basis for the study of the long-term migration and transformation mechanism of heavy metals.
【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X132;TP18

【相似文献】