基于计算智能和GIS的暴雨型泥石流分析预测研究

发布时间：2018-05-04 07:01

本文选题：暴雨型泥石流 + GIS空间分析　；参考：《中国地质大学（北京）》2013年博士论文

【摘要】：泥石流是山区常见的自然灾害，具有很强的破坏性，直接威胁人民生命和财产安全，严重影响经济的可持续发展。本文基于GIS空间分析技术和计算智能理论建立模型，分析评价泥石流灾害风险，为泥石流预测预报提供科学依据和技术支持。泥石流灾害系统属于复杂非线性系统，存在模糊性和不确定性。本文依据泥石流孕育、发展过程不同阶段的特点，以计算智能理论和GIS技术集成泥石流灾害的多元影响因子，建立泥石流灾害的风险评价预测模型。运用研究区观测数据进行仿真，验证了模型的有效性。本文的研究工作与主要结论如下： (1)根据泥石流发展过程中的一系列动态过程事件，以事件树理论为基础构建了泥石流灾害发展过程的事件树模型，用模糊语言表达事件树节点事件的发生概率，，并对泥石流灾害的发生概率进行模糊评价，最后通过解模糊方法得到泥石流灾害风险概率。事件树模型体现了泥石流孕灾过程的阶段性，计算出的泥石流灾害事件的发生概率与实际吻合。 (2)将坡度、相对高差、植被覆盖度、沿沟松散物储量、前5天累计降雨量、最大小时雨强和当日雨量作为泥石流灾害预警模型的评价指标，根据评价指标特点制定相应的关联函数来计算关联度，运用可拓学理论建立泥石流预警模型，为泥石流灾害评价提出了形式化的理论方法。 (3)利用模拟退火遗传算法改进GMDH网络模型，并用改进的GMDH模型预测泥石流灾害。用KLDA判别分析方法选取关联度大的致灾因子，将其作为输入参数，泥石流一次最大冲出量作为输出参数，运用改进的GMDH网络模型进行泥石流灾害预测，模型精度比其他模型（如BP和ANFIS）高。 (4)提出了基于本地搜索策略的混合蚁群优化方法来优化贝叶斯网络结构学习，以改进贝叶斯网络模型，并将改进的模型用于泥石流灾害风险评估，计算出的泥石流灾害危险度与实际情况吻合。该方法为地学分析中不确定性问题、数据不完整问题的研究提供了一种新技术。
[Abstract]:Debris flow is a common natural disaster in mountainous area, which is very destructive and directly threatens the safety of people's life and property, and seriously affects the sustainable development of economy. Based on GIS spatial analysis technology and computational intelligence theory, this paper establishes a model to analyze and evaluate the risk of debris flow disaster, and provides scientific basis and technical support for debris flow prediction and prediction. Debris flow disaster system is a complex nonlinear system with fuzzy and uncertainty. According to the characteristics of different stages of debris flow gestation and development, the risk assessment and prediction model of debris flow disaster is established by integrating the multiple influencing factors of debris flow disaster with intelligent theory and GIS technology. The validity of the model is verified by using the observed data in the research area. The research work and main conclusions are as follows: 1) according to a series of dynamic events in the development of debris flow, the event tree model of debris flow disaster development process is constructed based on event tree theory, and the occurrence probability of event tree node event is expressed by fuzzy language. The probability of debris flow hazard is evaluated by fuzzy evaluation. Finally, the probability of debris flow disaster risk is obtained by solving fuzzy method. The event tree model reflects the stages of debris flow disaster occurrence process, and the calculated probability of debris flow disaster event is consistent with the actual situation. (2) the slope, relative height difference, vegetation coverage, reserves of loose materials along gullies, accumulated rainfall in the first 5 days, maximum hourly rain intensity and daily rainfall are taken as the evaluation indexes of debris flow disaster warning model. According to the characteristics of the evaluation index, the correlation function is formulated to calculate the correlation degree, and the early warning model of debris flow is established by using extension theory, which provides a formal theoretical method for debris flow disaster evaluation. The simulated annealing genetic algorithm is used to improve the GMDH network model and the improved GMDH model is used to predict the debris flow. The KLDA discriminant analysis method is used to select the disaster factors with high correlation degree, which is regarded as the input parameter, and the maximum flow amount of debris flow is taken as the output parameter. The improved GMDH network model is used to predict the debris flow disaster. The accuracy of the model is higher than that of other models, such as BP and ANFIS. (4) A hybrid ant colony optimization method based on local search strategy is proposed to optimize Bayesian network structure learning to improve the Bayesian network model, and the improved model is applied to debris flow risk assessment. The calculated hazard degree of debris flow is in agreement with the actual situation. This method provides a new technique for the study of uncertainty and incomplete data in geoscientific analysis.
【学位授予单位】：中国地质大学（北京）
【学位级别】：博士
【学位授予年份】：2013
【分类号】：P208;P642.23

【参考文献】