模型误差的诊断及其补偿方法若干问题研究

发布时间：2018-11-19 08:00

【摘要】：测量是人类认识物质世界的重要手段,通过测量人们对客观事物在“质”的基础上获得了“量”的认识,从而总结出一般规律,建立起各种不同的计算公式和模型。但是,一般我们所建立起的模型,必然存在与客观实际不完全相符的情况,我们称之为模型误差。尤其是在现代数据采集的高精度化和自动化,改变了常规大地测量中的偶然性。本文系统的研究了模型误差理论及其诊断方法、模型误差的补偿效果方法、模型补偿在道路工程方面的应用。在软地基上修建高速路,造价高,标准高,对路基的沉降和稳定性要求高,而由于软地基自身透水性强、含水量大、强度低等特点,故沉降问题也成为一个修建高速路的核心问题,对于路面地基沉降的及时预测,对于整个施工团队的时间的控制具有重要的意义。由于目前对于模型误差的补偿技术研究不够深入,缺乏多种数学模型的对比研究及其精度的评定,对于其公路工程地基的沉降预测方面也有所不足。而目前,公认的比较合理的沉降预测方法是根据现场实测沉降值进行拟合的方法,由于算法的复杂性,要求实测数据的时间间隔要长,就不可避免的产生了误差,及时进行补偿,可以有效的提高模型精度。本文主要研究了模型误差理论和数据处理模型,以及用数学中统计的方法,来进行误差的诊断。在模型误差补偿方法上,重点研究了附加参数法、附加系统权、最小二乘配置、半参数法。通过一组实验拟合数据,用4种方法进行了补偿分析,得出相关结论,并提出了不同的回归模型的选择,对模型误差补偿效果影响不同的观点。最后,结合多个工程实例,对模型误差补偿技术在道路工程的沉降预测的应用,进行了详细的分析,经验公式法存在模型误差,本文将最小二乘配置法与解决公路工程实际问题相结合,对常用的利用实测数据建立模型的经验公式相结合,进行地基沉降的预测。经过试验分析表明,通过对比补偿效果图,可以发现基于最小二乘配置的模型补偿方法补偿效果良好。
[Abstract]:Measurement is an important means for human beings to understand the material world. By measuring people's understanding of objective things on the basis of "quality", we can conclude the general rules and establish various calculation formulas and models. However, in general, the model that we set up is not completely consistent with the objective reality, which we call model error. Especially, the high accuracy and automation of modern data acquisition have changed the contingency in conventional geodesy. In this paper, the theory of model error and its diagnosis method, the compensation effect of model error and the application of model compensation in road engineering are studied systematically. The construction of highway on soft foundation is of high cost and high standard, and requires high settlement and stability of subgrade. However, due to its strong permeability, large water content and low strength, the soft foundation is characterized by its strong permeability, high water content, low strength, and so on. So the settlement problem has become the core problem of the highway construction, which is of great significance for the timely prediction of the pavement foundation settlement and the control of the time of the whole construction team. Due to the lack of deep research on the compensation technology of model error and the lack of comparative study and accuracy evaluation of various mathematical models, the settlement prediction of its highway engineering foundation is also insufficient. At present, the generally accepted and reasonable settlement prediction method is based on the measured settlement value in the field. Because of the complexity of the algorithm, the time interval of the measured data is required to be long, and the error is inevitable. Timely compensation can effectively improve the accuracy of the model. This paper mainly studies the model error theory and data processing model, and uses the statistical method in mathematics to diagnose the error. In the model error compensation method, the additional parameter method, the additional system weight method, the least square collocation method and the semi-parameter method are studied. Through a group of experimental fitting data, the compensation analysis is carried out with four methods, and the relevant conclusions are drawn, and the selection of different regression models is put forward, which has different opinions on the effect of model error compensation. Finally, combined with many engineering examples, the application of model error compensation technology in settlement prediction of road engineering is analyzed in detail. The empirical formula method has model error. In this paper, the least square collocation method is combined with the practical problems of highway engineering, and the commonly used empirical formulas based on the measured data are combined to predict the foundation settlement. The experimental results show that the model compensation method based on least square collocation has good compensation effect by comparing the compensation effect diagram.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P207

【共引文献】