Force Update a Block Output at Minor Time Steps

cr
2013 10 11
1 个回答
更新时间：2014 8 21
13 次查看（30 天）

0 个投票

I just realised that the output of some blocks (e.g.source blocks like ramp and constant) are updated only at the beginning of a major time step. Is it possible to force update output signal of a block to update at the end of minor time steps too during a simulation?
Thanks.
P.s. I resubmitted this modifying the wordings of a Q I submitted earlier.

请先登录，再回答此问题。

回答（1 个）

Kaustubha Govind
Kaustubha Govind 2013-10-11

0 个投票

As far as I know, these blocks are internally configured to be Fixed in Minor Step, since it is a known fact that their output is not affected by the minor step. Perhaps you can explain more about why you'd like to force them to run in minor steps? It just seems inefficient.

3 个评论
显示 1更早的评论 隐藏 1更早的评论

cr
cr 2013-10-12
编辑：cr 2013-10-12
Well, for a constant block you are right. But for a ramp, the output does (or rather should) change even during minor time steps.
The reason I ask this is, I made two implementations of a problem (Lorenz attractor). One is direct math and integrator while the other is with a state-space (SS) block. Theoretically, the outputs from both implementations should match but they differ when simulated (in the same model, with same ICs). I can infer that the SS block implementation is the one thats off. The Lorenz attractor is a nonlinear model, but can be visualised as a time-varying linear system. To do that I'm using a simple s-function wherein I'm using update and output routines to change the A matrix of SS block as new outputs are computed. After quite a bit of investigation, I'm of conclusion that the outputs differ because the SS-block doesn't use its new A matrix during minor timesteps. Updation in A matrix is reflected only during major time step. Theres got to be some way to force update the output of a block.
I thought the experiment seems too complicated than it actually is and hence didn't mention it in the reposted question. I have now attached a simplified model that highlights the issue. In the mdl file, 'Difference' display blk should read zero, because the two integrators do the same job of integrating exp(x) from 0 to 1. It can be noticed that the smaller the time step, the smaller is the difference between the two outputs.
cr
cr 2013-10-14
Also, a choice of [0,0] sample time for a constant of ramp block seems to make no difference compared to [0 1]. Does the latter guarantee 'fixed in minor step'? Doesn't the former guarantee not 'fixed in minor step'?
Viren
Viren 2014-8-21
It's a pity you never got a proper answer for this question. The minor time steps in Simulink is a particularly dense concept with not much information available out there to help newcomers.

请先登录，再进行评论。

类别

在 帮助中心 和 File Exchange 中查找有关 Mathematics 的更多信息

产品

标签

提问:

cr
cr
2013-10-11

评论：

Viren
Viren
2014-8-21

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by