The World's Source For Fuzzy Logic Solutions
Dynamic debugging lets you test your fuzzy system as it interacts with a simulated or physical plant. You can set up a simulation or a physical system which provides inputs to your fuzzy system and have the outputs from the fuzzy system sent back to the simulated or external system. You can then make changes to the rules or membership functions of your fuzzy system and have those changes immediately updated and observe their effects. The dynamic debugging tools are:
TILShell comes with a set of monitoring tools during static or dynamic debugging. These tools provide information about the internal states of the fuzzy system. The monitoring tools are:
The Quick Test Window is a static debugging tool used to test the output of a fuzzy system for a set of specific inputs. The values for each of the inputs to the system may be specified, and the values for the outputs of the system will be displayed.
This is especially useful if you know exactly what the outputs of your system should be for a specific set of inputs, and you wish to tune the system until the fuzzy system's outputs match the ones you want.
An example Quick Test window is shown below.

The Control Surface Window provides a 3-dimensional view of the relationship between two inputs and one output of your fuzzy system. This lets you check the behavior of the outputs across the entire range of possible input combinations.
The Control Surface Window also allows the plotting of more complex systems by setting all but two of the inputs to constant values and selecting a single output variable to plot.
An example Control Surface Window is shown below.

The Operator Visualization window is a static debugging tool that provides a 3-dimensional view of the relationship between the inputs and output of an operator (e.g. AND, OR). This lets you "see" the behavior of an operator.
The Operator Visualization window plots any of the operators which have been included in the FPL.INI file and properly implemented in the DLL userops.dll. See the TILShell Tutorial for additional information on user defined operators.
An example Operator Visualization window is shown below.

TILShell provides the capability of simulating the physical plant or information system in which your fuzzy system will eventually be incorporated. The purpose of this capability is to allow you to interactively debug, tune, and validate your system as it is running, immediately observing the effects of any changes made to the system.
The model of the external system is described in the FPL Script Language, which is stored in a Model object. When a simulation is started, this model, along with the fuzzy system definition, is interpreted by TILShell to provide the running simulation.
An example for a simulation setup is shown below. Click on one of the windows to obtain moreinformation about it.

Once the fuzzy system is implemented on an external processor
TILShell cantune it through the online facility. The online facility
is similar to the simulation facility, with the difference being
that an online system is an actual external system rather than
a mathematical model of the system.
The Online Debugging Interface is intended to allow the user to
connect a host computer running TILShell to a target computer
executing a debugging version of a fuzzy system via a physical
link. Once this physical link has been connected, the user can
enter "Online Debugging Mode" in TILShell. In this mode,
selected data is transferred from the target to the host, and
anychanges the user makes to the system in TILShell are immediately
transferred from the host to the target while the system is running.
This allows you to tune a system while it is running.
The diagram below shows the data flow for a typical online tuning session.

The Rulebase Tuning Window displays information about the rules in a rulebase as the rules are executed, lets you enable and disable rules, and lets you change their weights. You can use the Rulebase Tuning Window to observe the activation level of each rule in the rulebase to determine which rules affect the output the most for a specific set of input conditions.
You can also use the Rulebase Tuning Window to determine which rules fire the most often and to what degree. This lets you determine which rules are the "most important" for your system, and lets you experiment to see if any rules can be omitted from your system or simplified.
An example Rulebase Tuning window is shown below.

The Rule Window presents the interaction between the inputs and outputs of a single rule in a graphical format. These windows allow you to individually determine the validity of the rules in a rulebase. Using a Rule Window together with the Quick Test Window provides a means of testing and observing the behavior of a single rule for each possible combination of input conditions.
An example Rule Tuning window is shown below.

The Var/Member Editor provides the debugging capabilities for observing variables and their membership functions. During a debugging session, the Var/Member editor displays a vertical line to represent the current value of the variable, and displays the degree of belief for each membership function in the legend.
An example Var/Member window is shown below.

Chart Windows allow you to graph the values of variables in your system either as a function of time, or as a function of another variable. This gives you considerable flexibility in observing the performance of your system during debugging and tuning.
An example Chart window is shown below.

Triggers allow you to set watchpoints to control the storage of information from a simulation or online debugging session to the Backtrace buffer. When the specified conditions are met, the iteration is marked in the Backtrace buffer, and a user-specified number of samples are kept before and after the trigger event in the Backtrace buffer. For example, you could set a trigger to watch for an input variable from an online system going out of range.
Triggers are disabled until the Start Capture command is given, and once a trigger has fired, are disabled again until the Start Capture command is given again. For a complete description of triggers and their syntax, see the FPL Guide.
The Watch Window allows you to observe the numeric value of any variable or other object in the fuzzy system, simulation, on online system. This provides the most basic level of information on the behavior of your system.
An exampleWatch window is shown below.

The Logging feature of TILShell allows you to keep a record of all the values of the objects in the Watch list in either a window or a file on disk. While logging is activated, the values of the objects in the Watch Window are written either to the Log window and/or the log file.
An example Logging window is shown below.

The Backtrace facility of TILShell acts as a tape recorder or storage oscilloscope. The values of the items in the Watch window are stored in the TILShell host's memory while a simulation is running and the values of the items in the Autosend/Capture list while an online debugging session is running. After the simulation or online session is halted, the values stored in the Backtrace buffer may then be replayed (forwards or backwards) to observe what happened during the simulation or online session.
As the samples taken during the simulation or online session are replayed, the open debugging windows of TILShell are updated to match the values of the current sample of the Backtrace buffer. However, the simulation graphic is not updated as the buffer contents are replayed.
The Simulation Control Window controls all TILShell dynamic debug modes. It can reset, start and stop the simulation and it controls the back tracing facility.

The Simulation Graphics Windows shows a figure of the plant being simulated by TILShell in the closed loop simulation. The Graphics Windows must be supplied by the user in a Windows dynamic link library.

This site is http://www.ortech-engr.com/fuzzy/debug.html.
Send inquiries, comments, or suggestions to webmaster@ortech-engr.com.
| Page Stats: Visitors -- Last Modified on: -- Thanks for visiting from: . |