为了提高三维片上网络(three dimensional network-on-chip,3D NoC)测试效率,结合3D NoC测试的特点建立了一种变权时间Petri网的测试模型,设置算法权值与变迁时延相关联,对Petri网变迁进行动态变权处理。在此基础上,将变迁激发序列作为IP核并行测试任务规划方案,采用简化蝙蝠位置更新方程的改进蝙蝠算法对其进行优化求解。将蝙蝠的位置更新规则融入到Petri网进化规则中,简化了推理过程,避免算法陷入早熟,提高了收敛速度。采用ITC’02测试基准作为实验对象,仿真结果表明,与其他算法相比,模型可以有效的描述3D NoC测试规划问题,最大时间优化率达到13.9%,提高了测试效率。
To improve the testing efficiency of 3D NoC,a variable weight time Petri net model was established in combination with the characteristics of 3D NoC test.The weight of the algorithm was related to the transition delay.Meanwhile,the Petri net given the dynamic transition.On this basis,we planned the scheme by taking transition firing sequences as parallel test tasks of IP cores,and used the improved bat algorithm based on the simplified bat location update equation to solve this model.The rules of bat location update were integrated into the Petri net evolution rules to simplify the reasoning process,avoid the algorithm premature and improve the convergence rate.The ITC'02 test benchmarks were used as the experimental objects.The simulation results show that the 3D NoC test planning problem can be effectively described in this proposed model,and 13.9% of the test time optimal rate and improved test efficiency can be obtained.