Bidirectional Branch and Bound Solvers for Worst Case Loss Minimization

Two branch and bound solvers using worst case loss criterion to select controlled variables.

Description

It is desired for modern process systems to achieve optimal operation. However, operation at a pre-designed nominally optimal point may not necessarily be actually optimal due to realtime disturbances, measurement and control errors and uncertainties. Self-optimizing control aims to tackle this problem through feedback control. By carefully selecting or synthesizing controlled variables, a self-optimizing control system can
achieve optimal or near optimal operation in spite of the presence of disturbances, measurement and control errors and uncertainties.

Selection of controlled variables is a NP-hard problem due to its combinatorial nature althrough theoretical criteria have been developed for self-optimizing control. This package includes two solvers using the local worst case loss criterion. Both solvers are based on the bidirectional branch and bound approach developed recently by the author and Dr. Kariwala. The principles, efficiency and applications of these solvers are available in the recent paper to be published in Computers and Chemical Engineering, and available online at http://dx.doi.org/10.1016/j.compchemeng.2009.01.014
 For a preprint copy of the paper, please contact the author.

Acknowledgements

The author wishes to acknowledge the following in the creation of this submission:
Bidirectional Branch and Bound Minimum Singular Value Solver (V2)
This submission has inspired the following:
Bidirectional Branch and Bound for Average Loss Minimization

• MATLAB release: MATLAB 7.7 (R2008b)

[Xu Wings]

It doesnt work.... rating 1

[Yi Cao]

Dear Xu Wings,

Thank you for giving your feedback. However, the sentence "it doesnt work" doesnt help anyone although I dont mind you gave the rating 1. You need to explain what you have done and what error happened so that you got this conclusion.

Have you tested the examples provided in b3wc.m and pb3wc.m files? Make sure you test these examples by copying and pasting these examples to the command window. If it still does not work, please let me know what matlab version you are using and what error message you have obtained.

Regards,
Yi

[John D'Errico]

It seems that Xu Wings is virtually in love with Shahab Anbarjafari. So almost anything that author does is given glowing ratings, while anyone who says bad things about those works are given retributively bad nonsense ratings by Xu Wings.

As for this author, I have a tremendous amount of respect for his submissions on the FEX. I've also tested this one, and it did work on my tests. The help is well done, etc.

[Kenneth Eaton]

 

[Michael Jordan]

 

[V. Poor]

 

[Yi Cao]

Mohammad,

If you are a real user who is concerned about this issue, please drop me an email. I will clarify this to you.

Yi

[Kazim Hanbay]

I want to selecet optimal features from 20 statistics.GLCM(gry level co-occurence matrix) contains the second -order statistical information. of neighboring pixels of an image.Features are generally selected by search procedures.Especially,branch and bound techniques have been fully exploited to extract the feature from mamogram images.I have attempt to optimize the feature og GLSM by fast branch and bound technique to extrcat most relevant features fot my classification work.How do I use your program(Bidirectional Branch and Bound (B3) for Average Loss Criterion).How do I select the 6 optimal feature from 20 features using your program.