可靠度计算方法的改进及其在核电厂防波堤稳定性中的应用

发布时间：2019-04-04 09:13

【摘要】：核电工程中的防波堤主要为核电厂的厂区及循环冷却水和核安全用水等的取排水设施提供保护,使其免受波浪的袭击,保护泵房取水不受波浪的影响。其地基通常是工程性质差的海岸海床地基土,在地震等外界作用下,易造成地质灾害,其稳定性对厂区及取排水设施的正常运行有很大影响。防波堤地基土以及筑堤材料强度变异性较大,仅仅用安全系数评价其稳定性是不够的,而目前对核电工程中防波堤的可靠度研究比较少。本文研究了一种新的样本取样方法,在传统的响应面法和改进的Kriging模型基础上构造功能函数,再利用Hasofer-Lind可靠度指标计算方法进行可靠度分析。首先利用了传统的二次多项式响应面进行了防波堤的可靠度研究。由于采用响应面法计算结构可靠性时都需要迭代,直至可靠度指标收敛。迭代过程中每次都是在当前验算点附近抽取新的样本来构造响应面。对于复杂问题,每次抽样都需要大量的计算时间。本文提出了一个响应面构造的改进方法——样本累积法,即用新样本和本次迭代以前所有的样本一起来构造响应面,从而最大程度地利用已有样本点的信息,避免了可靠度迭代陷入局部极值或往复震荡难以收敛的问题,提高了计算效率。如果仅选择离极限状态曲面很近的样本点参与累积和响应面的构造,可以进一步提高计算效率。通过两个算例验证了样本累积法的准确性和计算效率。最后采用改进方法分析了某核电工程防波堤的稳定可靠性。传统的响应面法常采用二次多项式构造功能函数,难以拟合非线性程度高的隐式的极限状态曲面,且对抽样中心的选择以及抽样方法敏感。本文利用粒子群算法的全局优化能力改进了Kriging模型,解决了相关参数的优化求解问题,能够更好地模拟非线性程度高的隐式的极限状态曲面。在样本选取上仍采用了样本累积的方法。算例表明基于样本累积的改进Kriging模型能够很好地拟合极限状态曲面,效率和精度较传统响应面法有较大提高。进一步研究发现选择离极限状态曲面近的样本点进行样本累积和功能函数构造能够提高计算效率。根据边坡稳定可靠度分析的实例,发现选取安全系数在[0.9,1.1]范围内的样本点进行累积是合适的。基于样本累积的改进Kriging模型对抽样中心和抽样方法不敏感,具有更好的适应性。最后分析了某核电工程防波堤稳定的可靠性。
[Abstract]:The breakwater in the nuclear power project mainly provides protection for the plant area of the nuclear power plant and water collection and drainage facilities such as circulating cooling water and nuclear safety water to protect it from the wave attack and to protect the pumping station from the influence of the wave. The foundation is usually shoreline subsoil with poor engineering properties, which is easy to cause geological disasters under earthquake and other external effects, and its stability has a great influence on the normal operation of the plant area and drainage facilities. It is not enough to evaluate the stability of breakwater foundation soil and embankment material by safety factor, but the reliability of breakwater in nuclear power engineering is less studied at present. In this paper, a new sample sampling method is studied. On the basis of the traditional response surface method and the improved Kriging model, the function is constructed, and then the reliability analysis is carried out by using the Hasofer-Lind reliability index calculation method. Firstly, the reliability of breakwater is studied by using the traditional quadratic polynomial response surface. Since the response surface method is used to calculate the reliability of structures, it is necessary to iterate until the reliability index converges. In each iteration, new samples are taken near the current check point to construct the response surface. For complex problems, it takes a lot of calculation time for each sampling. In this paper, an improved method of response surface construction-sample accumulation method is proposed, that is, the new sample is used to construct the response surface together with all the samples prior to this iteration, so that the information of the existing sample points can be used to the maximum extent. The reliability iteration can avoid the problem that it is difficult to converge in local extremum or reciprocating oscillation, and the computational efficiency is improved. If only the sample points close to the limit state surface are selected to participate in the construction of cumulative and response surfaces, the computational efficiency can be further improved. Two examples are given to verify the accuracy and efficiency of the sample accumulation method. Finally, the stability and reliability of the breakwater of a nuclear power project are analyzed by the improved method. The traditional response surface method often uses quadratic polynomial to construct the function, which makes it difficult to fit the implicit limit state surface with high nonlinearity, and is sensitive to the selection of sampling center and sampling method. In this paper, the global optimization ability of particle swarm optimization (PSO) is used to improve the Kriging model, solve the optimization problem of related parameters, and can better simulate the implicit limit state surface with high nonlinearity. The method of sample accumulation is still used in the selection of samples. An example shows that the improved Kriging model based on sample accumulation can fit the limit state surface very well, and the efficiency and accuracy are greatly improved compared with the traditional response surface method. Furthermore, it is found that selecting the sample point close to the limit state surface to construct the sample accumulation and function can improve the computational efficiency. According to the example of slope stability reliability analysis, it is found that it is appropriate to select the sample point with safety factor in the range of [0.9, 1.1] to accumulate. The improved Kriging model based on sample accumulation is not sensitive to sampling center and sampling method, and has better adaptability. Finally, the reliability of breakwater stability in a nuclear power project is analyzed.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM623;U656.2

【参考文献】