基于小波理论的文物多元时间序列分析研究

发布时间：2018-06-30 08:53

本文选题：文物多元时间序列 + 小波理论　；参考：《西安建筑科技大学》2017年硕士论文

【摘要】：不可移动文物作为我国古老文明的重要载体,具有极高的历史、文化和科学价值。由于自然环境变化等原因,不可移动文物出现了不同程度的病害,保存状态受到极大威胁。随着物联网技术在不可移动文物保护上的应用,避免了传统保护手段应对异常变化不及时、获取信息不全面、效率低下等问题。如何利用不可移动文物物联网数据实现对文物本体、环境等风险因素的分析、预测就显得十分必要,成为不可移动文物下一步信息化保护工作的关键。本文在分析不可移动文物物联网数据特点的基础上,研究了基于小波理论的相关分析方法和预测建模方法,设计了基于上述方法的文物病害预测分析系统。具体工作如下:(1)文物多元时间序列相关性分析。针对小波相关分析理论仅面向单元时间序列的局限性,以及定量分析时滞相关关系对研究文物多元时间序列的重要性,提出了一种基于小波理论的多元时间序列相关性分析方法。该方法利用定义的小波相关系数结合逐步回归向前思想实现了变量属性的约简,并通过构建的小波时滞相关系数公式定量描述了序列时滞相关关系。实验结果表明,该方法增强了小波分析方法在多元时间序列相关性分析的适用性,可以定量分析文物多元时间序列的相关性和时滞性。(2)文物多元时间序列预测建模。鉴于多元时间序列内部的繁杂相关性对预测建模的影响,及文物多元时间序列预测建模中考虑因素单一等问题,在小波相关分析的基础上进行变量选择,进而参考时滞性因素对预测建模方法进行了改进。首先根据小波相关分析理论结合最小相关准则的变量选择方法实现了特征变量的有效选择,在此之上利用得到的量化时滞相关关系构建了基于BP神经网络的预测模型,仿真结果表明该方法能够提高文物多元时间序列预测模型的精度。(3)文物病害预测分析系统。由于现有不可移动文物物联网在数据分析处理功能上存在的缺失,设计了基于本文算法的文物病害预测分析系统。该系统不仅可以为文物保护工作提供了科学可靠的依据,而且健全了不可移动文物物联网的应用体系,对文物预防性保护起到了重要的作用。
[Abstract]:As an important carrier of ancient civilization in our country, immovable cultural relics have high historical, cultural and scientific values. Because of the changes of natural environment, immovable cultural relics have appeared in different degrees of disease and the state of preservation has been greatly threatened. With the application of the Internet of things in the protection of immovable cultural relics, the traditional protection has been avoided. It is very necessary to use the data of unmovable cultural relic to realize the analysis of the risk factors such as the ontology and the environment of the unmovable cultural relic. It is the key to the information protection work of the unmovable cultural relics. This paper analyzes the unmovable text in this paper. On the basis of the data characteristics of the Internet of things, this paper studies the correlation analysis method based on the wavelet theory and the prediction modeling method, and designs a system for the prediction and analysis of cultural relics based on the above methods. The specific work is as follows: (1) the correlation analysis of the multiple time series of cultural relics. And quantitative analysis of the importance of time-delay correlation on the study of multiple time series of cultural relics, a method of multivariate time series correlation analysis based on wavelet theory is proposed. This method uses the defined wavelet correlation coefficient combined with stepwise regression forward to realize the reduction of variable dependent variables, and through the construction of wavelet time-delay correlation. The coefficient formula quantificationally describes the correlation of sequence delay. The experimental results show that this method enhances the applicability of wavelet analysis method in multivariate time series correlation analysis, and can quantitatively analyze the correlation and delay of multiple time series of cultural relic. (2) multi time sequence prediction modeling of cultural relics. The influence of miscellaneous correlation on prediction modeling, and the single factor in multi time series prediction modeling of cultural relic, variable selection on the basis of wavelet correlation analysis, and further improvement of the prediction modeling method based on time-delay factors. First, the selection of variables based on the wavelet correlation analysis theory combined with the minimum correlation criterion. The method realizes the effective selection of the characteristic variables. On this basis, the prediction model based on BP neural network is constructed by using the quantitative time-delay correlation. The simulation results show that this method can improve the precision of the multi time series prediction model of cultural relic. (3) the system of cultural relic disease prediction and analysis. In the absence of data analysis and processing function, a system based on this algorithm is designed. This system can not only provide scientific and reliable basis for the protection of cultural relics, but also improve the application system of the immovable cultural relic network, and play an important role in the preventive protection of cultural relics.
【学位授予单位】：西安建筑科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O211.61

【参考文献】