全光交换网的高吞吐量和信道利用率要求全光缓存器具有大的动态延迟范围。建立了基于平行排列3×3光纤耦合器的双环耦合全光缓存器(DLOB)的分析模型。分析了限制DLOB延迟范围的各个因素。分析发现,通过级联DLOB,可以实现大动态延迟范围。实验验证级联型DLOB的可变延迟范围达到1～9999T,延迟粒度为25ns,输出信号的误码率低于10-9。得到理论模型与分析方法对其他基于半导体光放大器(SOA)光开关的反馈式光学处理结构同样具有借鉴作用。 All-optical switching network high throughput and channel utilization require all-optical buffer has a large dynamic range of delay. The analytical model of a dual-loop coupled all-optical buffer (DLOB) based on parallel arranged 3 × 3 fiber couplers is established. The various factors that limit the range of DLOB delays are analyzed. The analysis found that with cascaded DLOBs, a large dynamic range of latency can be achieved. Experimental results show that the variable delay range of cascaded DLOB reaches 1 ~ 9999T, the delay granularity is 25ns and the bit error rate of the output signal is less than 10 ~ 9. Obtaining the theoretical models and analysis methods can also be used as reference for other optical processing structures based on semiconductor optical amplifier (SOA) optical switches.