Abstract: A half flush-mounted S-duct inlet with a large amount of boundary layer ingestion (δ=0.2H，H is the height of the air intake entrance section) was designed. And then，to facilitate the geometry parameterization of the diffuser，the optimization software ISIGHT and the Non-dominated Sorting Genetic Algorithm were adopted as platform and optimization algorithm，respectively. At last，the aerodynamic performance of the prototype and the optimization are compared. It shows that the total pressure recovery improves a little and the swirl distortion reduces by about 49.31%. In addition，geometry of the optimization also changes a lot. The centerline tends to be more modest at the entrance while more rapid at the exit. The cross-section area ascends slowly at the starting?segment，but increases dramatically and results in rapid static pressure increase near the exit. Total pressure recovery coefficient，circumferential total pressure distortion and swirl distortion at exit all improve in different Mach number of the incoming stream，though change?trend of these parameters with Mach number are not the same before and after optimization.

Key words: Inlet；Design and optimization；Boundary layer；Circumferential total pressure distortion；Swirl distortion

摘要： 为了减小大量附面层(δ=20%入口截面高度)吸入的半埋入式S弯进气道出口畸变，在完成其设计的基础上，以ISIGHT优化软件为平台，应用非支配排序遗传算法对其扩压器部分进行优化，并将优化前后结果进行对比。结果表明:优化后，进气道总压恢复系数略有提高，旋流畸变改善最为明显，降低约49.31%；几何结构相对原型改变较大，中心线趋于前后缓急相当，截面面积开始缓慢递增，靠近出口时急剧增加，呈现出“后发力”的特点；不同马赫数下，优化后进气道出口截面总压恢复、周向总压畸变、旋流畸变都有所改善，但优化前后各参数随马赫数的变化趋势不尽相同。

关键词: 进气道；设计及优化；附面层；周向总压畸变；旋流畸变