中文摘要：

以边坡稳定性计算的Janbu条分法为基础，提出一种改进的遗传算法（MIGA），在无需对边坡滑动面形状进行假设的前提下能自由搜索最危险滑动面并计算其安全系数。MIGA法模拟了生物进化的过程，克服了传统方法容易陷入局部极小值的缺点，是一种全局优化算法，且具有比简单遗传算法（SGA）更高的搜索效率。在工程实例中，通过与其他方法进行对比分析，用MIGA法计算得到的安全系数最小，且搜索到的最危险滑动面最接近实际情况，证明了该方法的优越性。

英文摘要：

Based on the Janbu slice method for the calculation of integral stability of slope, an improved genetic algorithm, named MIGA was proposed, which can freely search the most dangerous failure slope surface and calculate the corresponding minimum safety factor without supposing the geometric shape of the most dangerous non-circular failure surface. This improved genetic algorithm can simulate the genetic evolutionary process of organisms and avoids local minimum value compared with the classical methods. It is a global optimal algorithm, so it has higher searching efficiency than simple genetic algorithm （SGA）. The safety factor calculated by MIGA is the least compared with that by other methods in engineering, and the most dangerous failure surface searched by MIGA agrees well with the reality, so its advantages are verified.