Actions
Forum Thread Details
  • State Not Answered
  • Replies 2 replies
  • Subscribers 473 subscribers
  • Views 569 views
  • Users 0 members are here
Related

Implementation of Neural network (trained in MATLAB) in Microcontroller

Former Member

Hi all,

I want to dump my neural network on any controller like PIC AVR or ARM..

I made the network in MATLAB using NFTOOL now,I have simulink file and I want to generate equivalent  "C" code.

and that code compile I want to dump in micro-controller..

I am having Matlab 7.12 R2011a

(Lpcxpresso kit with LPC1114/302 microcontroller

Arduino kit with ATmega 32)

 

Guys.-- Please Help me out....

Parents
  • 0

    Hey

     

    I am doing a very similar project. Built a simulink model also using Neural Network Control but am having problems linking my simulink to the Arduino Atmega 2560. Mathworks offers a plugin to allow conversion from the simulink model to a code but i am having problems with it as it only offers 2 board options and the 2560 is not an option. What you can try is the real time workshop function. It takes your simulink model and converts it into a C code.

     

    let me know if this helps or if you find help cause having the same problems and this for my fourth year design project.

     

    Cool

  • 0 in reply to Former Member

    you can build c code from the model using real time workshop, then simply incorporate the generated c files into a project file and compile via your standard ide / compiler usage.  The advantage seen with boards that are part of the Matlab target support packages is that you can invoke the compiler / create the entire project from the model.  When you use a non-supported target, you have to do a bit more leg work, to bring the auto generated code to the compiled / ready to flash step but it is still a useful tool.

Reply
  • 0 in reply to Former Member

    you can build c code from the model using real time workshop, then simply incorporate the generated c files into a project file and compile via your standard ide / compiler usage.  The advantage seen with boards that are part of the Matlab target support packages is that you can invoke the compiler / create the entire project from the model.  When you use a non-supported target, you have to do a bit more leg work, to bring the auto generated code to the compiled / ready to flash step but it is still a useful tool.

Children
No Data