优化的灰色马尔科夫模型在建筑物沉降预测中的应用

发布时间：2018-09-07 15:46

【摘要】：随着城市化进程的推进,高层建筑雨后春笋般拔地而起。存在建筑物的安全隐患,避免造成不必要的经济损失和人员伤亡,应定期对建筑物进行沉降观测,获得沉降变形数据,并对沉降变形数据进行处理分析。随着人们对变形监测和分析预测的重视与研究,目前国内外已出现多种预测方法,主要包括时间序列分析法、线性回归分析法、卡尔曼滤波分析法、小波理论分析法、支持向量机法、马尔科夫链、人工神经网络模型、灰色系统模型等。科学合理的沉降预测模型不仅有利于建筑物的安全,而且对制定相应决策也有重要意义。本文对建筑物监测的内容和意义、对建筑物沉降监测的技术手段以及建筑物沉降监测的要求以及目前数据分析和预测模型的研究现状进行讨论;基于马尔科夫预测基本理论对马尔科夫过程进行讨论,就灰色预测理论,讨论了灰色GM(1,1)模型精度和残差GM(1,1)模型。建立灰色马尔科夫预测模型及其应用该模型在建筑物沉降预测,验证了该模型在建筑物沉降量预测具可行性。重点讨论了对灰色马尔科夫预测模型的两种优化方法,分别是新陈代谢的优化方法和残差修正结合新陈代谢的优化方法,并用这两种优化后的模型对建筑物沉降量进行预测,且与其他方法进行比对试算分析,结果表明,优化后的模型不仅都能够相对有效地预测数据的范围,而且预测精度和准确度都有提高。
[Abstract]:With the development of urbanization, high-rise buildings sprang up like bamboo shoots. In order to avoid unnecessary economic losses and casualties, the buildings should be observed regularly to obtain the settlement deformation data, and the settlement deformation data should be processed and analyzed. With the attention and research of deformation monitoring and analysis and prediction, there are many prediction methods at home and abroad, including time series analysis, linear regression analysis, Kalman filter analysis, wavelet theory analysis, etc. Support vector machine method, Markov chain, artificial neural network model, grey system model and so on. A scientific and reasonable settlement prediction model is not only beneficial to the safety of buildings, but also important to making corresponding decisions. In this paper, the content and significance of building monitoring, the technical means of building settlement monitoring, the requirements of building settlement monitoring and the current research status of data analysis and prediction model are discussed. The Markov process is discussed based on the basic theory of Markov prediction. Based on the grey prediction theory, the precision of grey GM (1K1) model and the residual GM (1K1) model are discussed. The grey Markov prediction model and its application in building settlement prediction are established, and the feasibility of the model in building settlement prediction is verified. Two optimization methods of grey Markov prediction model are discussed, one is metabolic optimization method and the other is residual correction combined with metabolic optimization method, and the two optimized models are used to predict the settlement of buildings. Compared with other methods, the results show that the optimized models can not only predict the range of data effectively, but also improve the accuracy and accuracy of prediction.
【学位授予单位】：东华理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU433

【共引文献】