Seth on SimulinkNovember 6th, 2009Generated Code for Variable Size SignalsAarti recently posted about Variable Size Signals in Simulink. Han responded with this comment: Aarti, Here is Aarti's response.
Hi Han, Thank you for your comment. In the case of a variable size signal, the generated code allocates the maximum possible size for the input/output variables. This allows signal sizes to vary during execution (as opposed to being hard-coded after model compilation). The generated code also propagates signal sizes through the model and accordingly assign the required dimension to signals. The algorithm code only needs to operate on the portion of memory required (thus saving execution time). The following model uses the Switch Block to change the size of its output signal.
The generated code looks like this: In the generated code, notice how the variable model_Xdim has the lower and upper bound pre-allocated as 1 and 5, 5 being the width of the 3rd Inport block. Using these values, the loop_ub variable for the Switch block is calculated. Thanks! Aarti Is this the code you expected? Leave a comment here and tell me about it. October 28th, 2009STEM Educational Initiative Signing, and Robotics CompetitionsIt is not every day that I attend a speech by political leaders of the Commonwealth of Massachusetts.
On October 14th, 2009 Massachusetts Governor Deval Patrick signed an executive order to establish the Science Technology Engineering and Math (STEM) Advisory Council. The MathWorks hosted the gathering of local political leaders. MathWorks CEO Jack Little introduced the Lieutenant Governor Timothy Murray and the Governor Deval Patrick to a room filled with MathWorkers. I could describe all the formalities and go into what this means for STEM education in Massachusetts, but other media sources covered those topics (like EDN, Mass High Tech and the Governors official website). Instead I would like to focus on some of the tangents to the main event.
Talking to the Governor My fellow blogger, Loren Shure, asked Governor Patrick what he thinks good corporate citizens can do to support STEM education.
The Governor gave a couple examples of how companies can get involved such as mentoring area students and engaging in student competitions. There already exist many opportunities to do this… Mentoring I have known many MathWorkers who volunteer at a local elementary school as math tutors in preparation for standardized tests. The same school has a Lego® Club where some of my colleagues mentor students on building robot using Lego® Mindstorms. Competitions In Governor Patrick’s response to Loren, he gave a specific example of a robotics competition, FIRST. Most people have heard of FIRST (For Inspiration and Recognition of Science and Technology). If you haven’t, browse around the FIRST website for a few minutes and get inspired about the great things available to the next generation of engineers and scientists. They have programs targeted at many different age groups, from the Junior FIRST Lego League for ages 6-9 years all the way up to “the varsity sport for the mind” FIRST Robotics Competition for ages 14 to 18 years. Targeted toward college students is the EcoCAR Challenge (which I wrote about in a recent post). MathWorks is a sponsor of the competition, and some of my colleagues act at mentors to teams. The goal is to modify GM donate vehicles in an attempt to reduce emissions, minimize fuel consumption AND maintain consumer appeal. Those college students involved are already on their way to careers in engineering. Another competition I am very excited about is ET Robocon, which MathWorks also sponsors. The site is in Japanese, but you can see videos like this one on YouTube (search for ET Robocon). I see ET Robocon as an embedded software design competition. Teams all start with a standard LEGO® NXTway robot (a two-wheel balancing robot built from LEGOs). They have to develop an autonomous control strategy and implement it to race robot around a pre-set track as fast as possible. Some of my colleagues from Japan developed the balancing algorithm using Simulink and Real-Time Workshop Embedded Coder (available on the File Exchange here). All the teams have to incorporate that balancing software into their larger design. I learned that there are many university teams as well as professional engineers from major companies involved in the competition. Now It’s Your Turn Are you a mentor? Are you involved in any of these or other competitions that encourage students to explore STEM fields? Leave a comment here and share your experience. October 16th, 2009Radar Tracking in Simulink: Variable Size SignalsThis week I welcome guest blogger Aarti Ramani to talk about a long requested feature, variable size signals in Simulink.
Depending on target ranges, radar systems operate in different modes (different lengths of data etc). Let’s assume we have two planes at two different altitudes. Now, what happens if there is a new plane in sight? Since the number of targets being detected by the radar is now changing, how would you use Simulink to add this plane to its tracking list?
A short and simple answer to this question is: Variable Size Signals. I’ve read many customer requests wanting to be able to change the dimensions of their Simulink variables while the simulation runs. Be it a radar system which needs to operate in different modes based on its target range to a simple combustion engine that needs to change its frame size based on the number of samples per cycle, each of these system dynamics work with signals whose sizes change during model execution. Although, earlier I had to tell our customers that Simulink did not have this capacity and you could not change a model’s signals dimensions during execution; with R2009b, Simulink now supports variable-size inputs and outputs in over 40 Simulink blocks. How does one create a variable size signal? There are several methods of creating a variable size signal: Switch blocks; Multi-Switch blocks with different input sizes; a Selector block indexing options or custom code blocks (S-functions and Embedded MATLAB blocks). A common way of generating variable-size signals is to use a Switch block for which each of the input signals differ in size. The Switch Block now allows variable size signals to be passed in as inputs:
By simply selecting the option “Allow different data input sizes”, you can obtain a variable size output signal whose dimensions change with time. As an experiment, let us look at a simple demo model “sldemo_varsize_basic.mdl”. This demonstration contains examples of how to use variable-size signals in a Simulink model, and to show what kind of operations you can apply to them.
Look at how the Switch block, allowing 2 inputs of different dimensions, can be used for operations like addition, vector concatenation etc. You can also use the “Selector” blocks to create a variable size signal. The Selector block generates as output selected or reordered elements of an input vector, matrix, or multidimensional signal. Using the “Starting and ending indices (port)” indexing option, you can allow inputs of variable dimensions to generate variable size signals.
You can use the Selector block to subreferece a matrix as shown below:
How can I get the current dimensions or width of a variable-size signal? You can use the Probe block to output the width of a signal.
The Probe allows you to visualize and save the signal dimensions and view the I/O and state data for blocks selected. This could also be used in a calculation that requires sample size information. How can I convert a variable-size signal into a fixed-size signal? To convert a variable size signal into a fixed size signal, you can use the “Assignment” block. The variable-size signal feeds into the second input port of the Assignment block and can be used to convert the incoming signal into a fixed size signal. By allowing signal sizes in Simulink models to vary during execution, one can easily model systems with varying environments, resources, and constraints. So, coming back to our radar system, using variable size signals, you can now create a conceptual air traffic control (ATC) radar simulation based on your radar range equation(s):
For more details on how this snazzy feature works, check out the Variable Sizing documentation. Now it’s your turn Do you plan to upgrade to R2009b and use variable size signals? Leave a comment here and tell me what you plan to model. October 6th, 2009MathWorks Hosts EcoCARRecently, when I arrived at work, I found this parked on the lawn:
This is the EcoCAR from the EcoCAR Challenge. MathWorks hosted about one hundred students from the EcoCAR teams for the EcoCAR Year 2 Fall Workshop. A Big Shiny Embedded Target When I first saw the EcoCAR sitting on the lawn, I wondered what system target file they might use:
I’m sure there are a lot of cool things these students will be able to do with a great piece of hardware like this, but in many ways, it’s just a big, shinny embedded target. What is EcoCAR? EcoCAR is an academic competition. A quote from ecocarchallenge.org sums it up:
The students had 5 days of training in different tracks to learn and apply MathWorks tools to the design problems facing them. If you read the Green Car Garage blog, you may have read about the event and the fun teams had during their visit to Boston. Nicole Lambiase from Argon National Lab wrote a post commenting that it really takes a village to make these competition events run smoothly. Some of the people in that village are members of the Model-Based Design community. A lot of my colleagues volunteer with EcoCAR Challenge. Many of them traveled from our Michigan office to lead trainings and mentor students that attended. The Trouble with Model-Based Design The trouble with Model-Based Design is that there are very few people and institutions available to help you learn it. I never had a class on Model-Based Design when I went to school. The EcoCAR Challenge is an opportunity to develop the next generation of engineers skilled in Model-Based Design. MathWorks is very committed to mentoring these future engineers. This aligns with our academic mission to accelerate the pace of learning, discovery, and research in engineering and science. How about you? How did you learn about modeling and Model-Based Design? Who was your mentor? Leave a comment here and share your story. September 25th, 2009Custom Components in Physical ModelsThis week I want to introduce guest blogger Steve Miller to talk about the Simscape language.
Introduction to the Simscape Language I’ve been working with MATLAB and Simulink since my college days, and one of the things I liked most about these tools is that I’m free to create just about anything with the commands and blocks. Whenever my boss asked me, “Do you think you could use MATLAB to...,” the enthusiastic “Yes!” was halfway out of my mouth before he even reached the end of the sentence. When I started working with the physical modeling products at The MathWorks, I was hoping that I’d have the same freedom there. Well, starting with Release 2008b, I do! Using the Simscape language (an enhancement to Simscape, which was first released in R2007a), I can create my own physical modeling blocks just like I created cool MATLAB scripts and Simulink subsystems to solve all those tricky problems my boss threw at me. What is the Simscape language? The Simscape language is a MATLAB-based, object-oriented language for modeling physical systems. The big difference here is that I can use this textual language to create components that have physical connections, such as hydraulic ports on valves and electrical terminals on resistors, instead of inputs and outputs like I’m used to doing in Simulink. Why is this important? Well the models built with physical connections are much easier to understand – models of electrical systems look like an electrical schematic. And, if I create something useful and I want to reuse it in another model, it is much easier, for I don’t have to worry about routing inputs and outputs -- I just plug my new electrical component into the circuit and try it. Here is an example of a component I have defined using the Simscape language. It’s a DC motor. You’ll see that it has four physical ports (not inputs and outputs) – two electrical and two mechanical. These physical ports let me plug the component into a physical network.
How does it work? You can use the Simscape language to define physical domains, components, and libraries of components. It’s usually easiest to look at an example, so let’s look at the definition of a DC motor created in the Simscape language. Here we see the first 11 lines of the file. This is where we define the component name, add a description, and define the physical ports:
You can see that we have two electrical (+ and – terminals) and two mechanical rotational ports (motor shaft and motor housing). These are reflected in the component shown above. In the next 6 lines of the file, we define the parameters (including units!). These are the values that I or another user may want to change:
You’ll notice that in the final component, the prompt, the default value, and the units all show up in the dialog box shown here:
Just like most Simscape blocks, there a link to the source code! In lines 18-23 of the file, we define the variables associated with the physical domains (electrical and mechanical) we need for the equations we will define in the equations section.
In the setup section, we set up the connections between the variables to the physical ports. Remember Kirchoff’s laws? Well, we’re essentially applying them to both the electrical and mechanical domains (yep, they work in other physical domains, too, if you define them properly).
You’ll also notice that we have used MATLAB to make sure that the value of the resistance in the dialog box is not less than zero. I can use MATLAB code in this section to do other things like calculate values or initialize variables. All that MATLAB I learned is not going to waste! In the last lines of the file, I define the equations for the motor. Take a look:
See the “==” symbol there? I’ve used that to define an implicit equation. This means I have set up a relationship between these quantities (including time derivatives – see the .der?). I’m not assigning values, nor does this represent a Boolean expression. I could have defined them as shown below and the results would be the same:
And there you have it! Once I have defined the file, I execute ssc_build and it produces a Simscape component I can connect to other Simscape components in my physical model. Videos of Simscape language in action I have made a video demo of modeling a nonlinear spring using the Simscape language. Take a look to see the blocks in use and the build process. For a more complete experience, watch this webinar on the Simscape language. Now it’s your turn Want a closer look at this example? Download it from the MATLAB Central File Exchange. Are you using the physical modeling tools? Have you tried the Simscape language? Post a comment here and tell us about it. September 18th, 2009How do I test for NaN in Simulink R2009b? (NEW!)Back in February 2009 I posted about how to test for NaN in Simulink. The approach I talked about was more of a logical experiment based on the special properties of NaN than an ideal software solution. In Simulink R2009b the Relational Operator block got an upgrade to include isNaN. Let’s see how it works. Relational Operator Block Upgrade The Relational Operator is often used to compare signals and test for greater than (>), equals (==), and more. In R2009b it has the following new single input modes: isNaN, isFinite, and isInf.
These functions are familiar to those who program in MATLAB and they work the same way. When the input signal has the properties we are testing for, the block outputs true. How does it work? I talked to my colleague Omar who upgraded the block and he shared with me some key points. He mentioned that the block handles:
These new operators are only meaningful with data types that can represent Inf and NaN. So how does it work with non-floating point types? If you pass in an integer, or a fixed-point type that cannot represent NaN of Inf, the block returns false. This is a key point for workflows where you might find yourself switching the data types in your design. This often happens when selecting proper fixed-point data type properties. If you are running simulations to gather ideal double precision results, you may need to handle Inf and NaN. When converting the model to fixed-point for implementation, the algorithm does not require any modifications. Generated Code The code generated for the isNaN function looks like this:
34 /* Outport: ‘<Root>/Out1’
incorporates: The function rtIsNaN is part of rt_nonfinite.c: 61 /* Test if value is not a number */ 62 boolean_T rtIsNaN(real_T value) 63 { 64 65 #if defined(_MSC_VER) && (_MSC_VER <= 1200) 66 67 /* For MSVC 6.0, use a compiler specific comparison function */ 68 return _isnan(value)? 1U:0U; 69 70 #else 71 72 return((value!=value) ? 1U : 0U); 73 74 #endif 75 76 }
Further Optimization using Target Function Library If your embedded target environment has a special function for handling tests for NaN, Inf or Finiteness you can use the Target Function Library to replace the call to rtIsNaN with that call. What do you think? Will you use the isInf, isNaN and isFinite mode of this block? Leave a comment here to tell me how. September 9th, 2009New Stuff – Simulink R2009bLast week the MathWorks released the R2009b family of products. There are many new capabilities in the latest release and with this blog post, I want to highlight a some of the features in Simulink I’m really excited about. While you read this, start installing the latest products from the MathWorks.com downloads area (login and license required). Reading the release notes is so R2007b! Did you know that there is a presentation containing highlights and screen shots from R2009b Simulink? If you have missed this in the past, you can go back and check out R2008a, R2008b and R2009a highlights. Try it yourself, and browse through all the cool new features in R2009b Simulink.
Model Reference Protected Models! Share your model functionality without sharing your model intellectual property! Anyone with R2009b Simulink can use a protected model. If you have a license to Real-Time Workshop, you can create a protected model.
Model Reference Variants If you have multiple implementations of your component model, variant objects enable you to control the implementation used. This allows you to globally control and coordinate switching between variant implementations of your model.
Tabs in the Mask Editor! Now you can create tabs in your custom block masks.
Variably Sized Signals! Special blocks and Embedded MATLAB now support dynamically sizing signals during simulation!
The SIM command can return a single output! The SIM command has a new single output syntax so all your results are part of a single SimulationOutput object. This enables SIM to be called within a PARFOR loop, thus enabling easy parallel Simulation using the Parallel Computing Toolbox. paramNameValStruct.SimulationMode = 'rapid';
New Library Links Tool for fixing broken links! After making changes to a model and disabling library links, the Library Links tool will enable you to reestablish those links to your library. Each change in the model can be pushed back into the library, or the block can be restored from the original library source.
Now it’s your turn Have you downloaded R2009b? Leave a comment here and tell me about your favorite new features. August 28th, 2009BEEP, Simulink versus MATLABDid you ever notice that a BEEP in Simulink means something different from a BEEP in MATLAB? BEEP, What BEEP? What I’m talking about is that bell character you sometimes hear. In older version of MATLAB (Pre R14), you could actually display a beep on the screen. >> fprintf('\a')
In more recent version, the command window doesn’t display the BEEP character. Instead, you can call the function BEEP. >> beep
You hear it under different circumstances for Simulink versus MATLAB. MATLAB BEEP == Error In MATLAB, the BEEP alerts you to an error. Sometimes you get it because your calling syntax is wrong, sometimes when the tab completion can’t find anything. In MATLAB, BEEP is bad. >> plot(x,y
Simulink BEEP == Done In Simulink, the BEEP happens when your simulation is finished. It is a beautiful sound. The BEEP is an alert to tell you the work is done. The fruit of your simulation is ready to harvest. It is time to review your results. In Simulink, BEEP is a good sound. >> vdp
Leave your comments, after the BEEP. (THANK YOU Jeannette for alerting me to the difference.) August 20th, 2009A Better If-Else ConstructIn a previous post, I answered a question about how to model an If-Else behavior. Here I will restate the algorithm I want to create: if(sel==0) The first answer I gave relied upon a Switch block and the Conditional Input Branch Execution optimization to get an efficient If-Else construct in the model. While this works, I don’t like reliance on an optimization to provide good behavior. Sometimes, small changes to the model prevent Simulink from applying an optimization. I think it is best to implement requirements explicitly. In this post, I want to show you a way to model explicitly an If-Else conditional execution behavior. Conditionally Executed Subsystems and Merge The If-Else construct requires decision logic to control the execution of algorithm contained within the expression. One way to do this is using the If block (from the Ports & Subsystems library), combined with the If Action Subsystem. The If Action Subsystem executes based on the conditional expression in the If block. If you have used Function Call subsystems, this is very similar.
The If block provides control over If and ElseIf conditions, and there is even an option to provide an Else signal.
This model will give you the conditional execution of the two subsystem, however, the subsystems write their outputs to separate signals. How do you get two subsystem to write to the same signal? Merge the Signals The Merge block provides a way for both subsystems to write to the same signal.
In many ways, the Merge block doesn’t behave like a block with the traditional input/output relationship. I think of it more like a jumper across multiple wires. The merge block specifies that all signals connected to it have the same value and actually share the same memory. This is the programming practice of specifying multiple writers to the same variable. Generated Code with Merge The code generated from the above model looks a lot like our original algorithm. 34 /* If: '<Root>/If' incorporates:
Correct use of Merge Because Merge blocks are a memory specification rather than an algorithmic construct, there are some guidelines for using merge blocks. The documentation shows some correct and incorrect usage patterns. If you use merge block in your model, I suggest you run the Model Advisor Check for proper Merge block usage. Now it’s your turn Do you use the Merge block in your models? Leave a comment here and share your experience. August 11th, 2009How Many Blocks are in that Model?My colleague Sam and I wanted to find an example model with about 10,000 blocks. I suggested the HL-20 model from the Aerospace Blockset, asbhl20.mdl. Sam decided to measure the total number of blocks by running sldiagnostics: >> [t,s] = sldiagnostics('asbhl20','CountBlocks');
Sam quickly sent me a message saying that the HL-20 model only had 5700 blocks. What gives? What is SLDIAGNOSTICS? First, let us discuss a little history of sldiagnostics. Over the years, Simulink developers have worked with many customers to learn about their modeling styles and assist in debugging problems. Unfortunately some customers are unable to share their models because they might contain proprietary information. The solution these developers created were a series of scripts that would quickly gather information about a model without any model algorithm information. These collected scripts then became the sldiagnostics function. Because the output of these scripts was generally devoid of any proprietary data, even customers who could not share their models were able to provide the diagnostic information instead. Most of the checks done by sldiagnostics help you gather an understanding of the model as a whole. This is a great help in learning about models rather than having to browse through every system one-by-one. CountBlocks The diagnostic information Sam gathered was the output of CountBlocks. Here is a part of the text report for the HL-20 model. Finished counting blocks in 'asbhl20'.
HL-20 Includes Model Reference The HL-20 model includes 9 ModelReference blocks. (I introduced some of the basics of model reference a previous post.) You can see the hierarchy of the model in the Dependency Viewer instance view.
When I think about a model, I imagine the net collected functionality represented by it. I think the number of blocks used to create that model is a simple measure for how much functionality/algorithm there is. Model reference blocks contain the functionality of the entire model that they reference, however, sldiagnostics counts ever Model Reference block as only one block. The measure I am really looking for is the subsystem equivalent implementation of all the blocks in the model. To get this for the HL-20 model, you actually need to count the blocks inside each model referenced and multiply by the number of instances. I wrote a MATLAB function to do this: function totalBlocks =
mdlrefCountBlocks(mdl)
Total Blocks Here are the results of running this function: >> total = mdlrefCountBlocks('asbhl20');
Now it’s your turn In the end, we can see the HL-20 model has almost 11K blocks. Do you use model reference? How many total blocks are in the hierarchy of your largest model? Use the script above to gather this data and post it in the comments here. |
|
|||
|
These postings are the author's and don't necessarily represent the opinions of The MathWorks. |
Recent Comments