The application of 0-1 mixed integer nonlinear programming optimization model based on a surrogate model to identify...

Area of Science:

• Environmental Science
• Operations Research
• Computational Science

Background:

• Traditional non-linear programming models struggle with accurate identification of pollution source location due to continuous variables.
• Increasing numbers of pollution sources lead to exponential increases in computational load and decreased accuracy in existing models.

Purpose of the Study:

• To develop an improved optimization model for simultaneous identification of pollution source location and release intensity.
• To reduce the significant computational load associated with complex environmental simulations in pollution source identification.

Main Methods:

• Implemented a 0-1 mixed integer nonlinear programming (MINLP) optimization model.
• Integrated a Kriging surrogate model to approximate computationally intensive simulation components.
• Utilized a genetic algorithm (GA) to solve the optimization problem.

Main Results:

• The 0-1 MINLP model successfully identified both the integer location and continuous release intensity of pollution sources.
• The Kriging surrogate model significantly reduced computational load while maintaining simulation accuracy.
• The genetic algorithm effectively solved the problem, yielding accurate pollution source locations and release intensities with small relative errors.

Conclusions:

• The proposed Kriging-assisted 0-1 MINLP model offers a highly accurate and computationally efficient solution for pollution source identification.
• This approach overcomes the limitations of traditional models in handling integer variables and high computational demands.
• The study demonstrates a powerful methodology for environmental pollution source analysis.