Data Consistency Using the PreLook-Up Index Search Block

This demo shows one way to check a model to ensure that the prelook-up blocks have breakpoint data lengths that match the corresponding dimensions of table data for connected interpolation blocks. (Open the model: matlab:sldemo_bpcheck)

Contents

Use of the PreLookup Index Search Block

The PreLook-Up Index Search block allows you to minimize the number of index searches performed across a set of look-up tables and also to mix clipping, extrapolation, and index search algorithms within one table calculation. 

mdl = 'sldemo_bpcheck';
open_system(mdl)

Analysis of Breakpoint Data and Table Data for Consistency

This capability introduces a need to ensure that the breakpoint array length in the prelook-up blocks matches the corresponding table dimension in the connected interpolation block(s). The M-function matlab:edit('slcheckprelookups.m') can analyze a model and report on any mismatches it encounters between prelookup and interpolation blocks.

Errors have been deliberately introduced in the model to illustrate the use of a breakpoint checking tool that analyzes the model and ensures the breakpoint lengths match the table dimensions. 

chkblks = slcheckprelookups('sldemo_bpcheck')

Checked 5 interpolation blocks, found 4 issues to review.
Index list:  1, 2, 6, 10


chkblks = 

1x10 struct array with fields:
    interpBlkName
    interpBlkPort
    interpBlkParam
    interpBlkDimSize
    prelookupName
    prelookupParam
    prelookupSize
    mismatch
    errorMsg

This model has 3 blocks with data inconsistencies. Note that it is possible to have up to one inconsistency per dimension, so a 3-D table could have up to 3 inconsistencies. Here is the set of mismatched items from the full set of checked blocks, along with details on the first mismatch in the list: 

mismatches = find([chkblks.mismatch] == true)
chkblks(mismatches(1))

mismatches =

     1     2     6    10


ans = 

       interpBlkName: [1x86 char]
       interpBlkPort: 1
      interpBlkParam: 'reshape([1:300], [10,6,5])'
    interpBlkDimSize: 10
       prelookupName: [1x54 char]
      prelookupParam: '[0:10:100]'
       prelookupSize: 11
            mismatch: 1
            errorMsg: ''

[Close the model to reset the screen capture]

close_system(mdl)

The interpolation blocks with mismatch issues can be highlighted using Simulink's hilite_sytem() command:

open_system(mdl)
for k = 1:length(mismatches)
  hilite_system( chkblks(mismatches(k)).interpBlkName, 'error' );
end

NOTE: you can remove highlighting from a model using the View / Remove highlighting menuitem or this command:

set_param(mdl,'HiliteAncestors','none')

Copyright 1990-2004 The MathWorks, Inc.
Published with MATLAB® 7.0