The TK6932 super-heavy-duty CNC floor boring and milling machine is important basic equipment for super large parts machining such as million kilowatts level of pressurized water reactor nuclear power station,hydrogenation reactor and super large cold and hot rolling mill.As the spindle carrier,the ram is extended and retracted in the spindle box.Its mechanical characteristics and reliability directly affect the machining quality of parts.Lightweight design is necessary while guaranteeing its reliability.Only one machine is assembled until now,the problem is non-probabilistic reliability with minimum sample.An efficient optimization technique for approximate model and accelerated micropopulation with double-layer nested genetic algorithm was proposed to solve the problem of high cost of approximate model fitting of large complex structure and inefficient reliability optimization model with two-layer nested.