for real time implementation, your chose depends on the amount of calculus in your design, if it is a simple controller you need only a micro-controller and so as. the most important thing it is that if your system needs a control action in each time of T0, your hardware (Mc, dsp, ...) must give its response within this time T0,

The CortexM4 based STM32F4 is to be mentioned here, as there is a blockset for Simulink available and the RTOS implementation is easily to be done:

Another possibility is the usage of the Infineon XMC4500 (a Cortex M4 as well - interesting because of the FPU, so calculus should not be a bottleneck)

I would suggest using LabVIEW, specifically, if you have built your algorithm in simulink or an environment like that. However, Matlab environment has its own I/O for establishing data connection and thus, implementing real-time control of the system.

Develop this application using c language in MATLAB so there is a real time control by using user friendly control

you can use Embedded Coder/ Embedded Targets/ Processors/ Texas Instruments C2000/ C2802x in matlab

If you want to use the MCU with your computer, then you can try Arduino. MATLAB supports I/O from Arduino. Arduino is easy to use and can be programmed using MATLAB.

You can also integrate Arduino with SIMULINK.

The question is not very clear. If you can write details about the plant and the hardware you are using then I may able to answer your question.

I mean the following

1. Plant

2. Power electronics setup

3. Controller equations, with constants and variable definitions

4. The computing system (microcontroller board, DSP, PC and the operating system if applicable.

Thanks to all for your responses…I need lot of discussion on this topic …I will post my further question soon...

Hey all,

I am currently implementing a kalman-based filter for attitude estimation, particularly UKF, on an embedded computer. I wrote my code in MATLAB. Then , I used the MATLAB coder to convert my code to C/C++ code in order to be uploaded to my embedded computer. I also used some time-computing function that helps me to asses the computational time expenses that are needed on the used embedded computer.

I learnt the following,

1- from the code side: you need to optimize the code, as much as possible, in order to meet you processor computational power.

2- from the hardware side: you need to make sure that processor frequency provides you with a sufficient computational time, even higher for any possible overhead processing.

3- make sure to minimize any overhead process that are not really needed. e.g. close all un-needed applications that eats your processor power .

4- make sure that the operating system that you are using is giving you an optimized handling of your application run (e.g. application priority in processing)

5- keep an eye on the memory allocation in your code in order to avoid any possible overflow which, in result, might cause undesired false results.

Thanks,

Mohamed

sorry for any possible typos, I just woke up!! :D

If you distribute your various simulation models on separate computers and need to share the data between the models in real time, GE Intelligent Platforms Reflective Memory offers a hardware solution.

http://www.ge-ip.com/reflectivememorynetworks

You can also use the real time OS (XPCTARGET) of mathworks. It is very good with simulink and not very expensive. I used it for my real time experiments for robust control and adaptive control and regulation.

Thanks for the contributors for this questions...There is lot of suggestion about the real time implementation of the simulation based controller..... I want to create the standard procedure for real time implementation.....that means like a package ....

Hai..one of my student currently, do the same thing as you are. My advice, you need to understand between realtime (t) and simulation time (T) which is T is proportional to the computer sampling time. Another thing is looping & recursive factor, you need to reduce this factor unless you have high speed processor with high memory. You need to reprogram the code suited for uC implementation. Directly converting mfile to .c file is not a solution. 10 cens from me

Irawan & Robert Unguran: How do we design the higher order H-infinity controller through C-program?... i.e.,Theoretically, we made some mathematical calculation to find the robust controller (for example H-infinity loop shaping robust controller)…My question is can we implement that same mathematical calculation in machine language (i.e C , C++,etc)...

Robert Unguran: Thanks for your contribution…Can u give me the technical paper which describe that program?...It will be very helpful for my research....

You may attend difficulties that did not exist in simulation, because models are approximate and reality is much more complex. But do not quit. If you have a controller in Laplace Transform, you must discretize it (there are several methods of discretization (see for example the Tustin Transform) and then obtain a difference equation that you program. If you have a continuous controller in state space, you must also discretize it . See in Matlab the function c2d. Good luck.

Robert Unguran: Thanks robert.. I saw your blog..it is very nice..but due to time factor am not able to read throughly...But surely within two days i will read and try to implement the H-infinity filter....If i have find problems ..i need your help..can you?...

Antonio Dourado: Thats the idea am trying to implement the controller as you said..I think Robert also follow the same thing...thanks for your encouragement....

Adam sobczyk: Please suggest me a processor that can able to process complicated equations and has faster execution time....

You may try Cerebot MX7cK...

You can try to use Real Time Devices DM6430 , Support for the RTD DM6430 I/O board. Real Time Devices DM6430 the General Description pf it "The DM6430 is an ISA PC/104 I/O board with 16 single or 8 differential analog input (A/D) channels (16-bit) with a maximum sample rate of 100 kHz, 1 analog output (D/A) channel (16-bit), 16 digital I/O lines, and 2 counter/timers (16-bit)"

Depending on the expression of the control action that you have, you can program any types of microcontroller accordingly through using assembly, C/C++, ...etc and according to the microcontroller that you have. However, be cautious about the processing time of the microcontroller.

The best approach for you is to use a real-time simulator, such the one from OPAL-RT Technologies. It allows you to compile your code (Matlab or Simulink code) for real-time execution and allows you to use fixed time step solver with continuous equation. So you can directly use your Laplace transfer function without any modification by using a explicit or implicit method and higher order

All types of I/O are supported (analog, digital, pwm, resolver, encoder, ...) and you can use various protocols to interface your controller with your plant (ethercat, can, lin, arync, rs-232, ...)

From my point of view, this is the best way to do rapid control prototyping...

See the following link for more information:

http://www.opal-rt.com/rapid-control-prototyping

http://www.opal-rt.com/rapid-control-prototyping-general-overview

http://www.opal-rt.com/rapid-control-prototyping-benefits

If you model your controler in simulink you can also use TargetLink from dSpace or the embedded Coder from Mathworks. The benefit of using such tools is you can run your simulink simulation and evalute them in real time by PIL or HIL.

Two significant issues which should be considered in the implementation such controller are :

1- Electromagnetic noises caused to problems in communication between Micro-controller and the the other connected diveces.

2- Delay in response (for measurement devices and actuators) which should be considered in the algorithm.

a well known way is to use MATLAB’s xPC Target two PC-type desktop computers in a host-target configuration, and a NI BNC-2110 data acquisition card. as in "Robust H ∞ Controller for High Precision Positioning System, Design, Analysis, and Implementation" .Intelligent Control and Automation 01/2012; 333030:262-273. DOI:10.4236/ica.2012.33030

Hi! Could you please let me know how to design a robust H infinity controller for the PLL in simulink/matlab toolbar?Thanks