Surrogate modeling and knowledge extraction in ga applied to a parameter's estimation case

The Genetic Algorithm is a well-known heuristic-based optimization technique, but by using it in expensive problems, the excessive use of the fitness function can produce heavy computational loads. The purpose of this paper is to avoid this drawback by employing surrogate algorithms, providing approximate but sufficiently accurate solutions. Particularly the surrogate is based on granular computing and fuzzy logic. The difference from our related previous works consists into treat the algorithm's search, not as a black box, but extracting certain knowledge of granules' behavior represented by a new form of fuzzy aptitude functions. The primary objective is to manage the process intelligently, not only avoiding unnecessary usage of fitness evaluations but also improving the convergence with the extracted knowledge and the parameters' update with a newly built neural network structure. The algorithm shows satisfactory results in saving unnecessary evaluations and in time; in this case, we proved the optimization process related to parameters' estimation of a permanent magnet synchronous machine (PMSM) and some common benchmark functions used in GA assessments.

[1]  Witold Pedrycz,et al.  Granular Computing: Perspectives and Challenges , 2013, IEEE Transactions on Cybernetics.

[2]  Hugo Jair Escalante,et al.  A note on "Adaptive fuzzy fitness granulation for evolutionary optimization" , 2015, Int. J. Approx. Reason..

[3]  Wilfrido Gómez-Flores,et al.  On the selection of surrogate models in evolutionary optimization algorithms , 2011, 2011 IEEE Congress of Evolutionary Computation (CEC).

[4]  Thomas Villmann,et al.  Can Learning Vector Quantization be an Alternative to SVM and Deep Learning? - Recent Trends and Advanced Variants of Learning Vector Quantization for Classification Learning , 2017, J. Artif. Intell. Soft Comput. Res..

[5]  Teuvo Kohonen,et al.  The self-organizing map , 1990 .

[6]  Anthony J. Jakeman,et al.  A review of surrogate models and their application to groundwater modeling , 2015 .

[7]  Jie Tian,et al.  A self-adaptive similarity-based fitness approximation for evolutionary optimization , 2016, 2016 IEEE Symposium Series on Computational Intelligence (SSCI).

[8]  Carlos A. Coello Coello,et al.  Evolutionary hidden information detection by granulation-based fitness approximation , 2010, Appl. Soft Comput..

[9]  Hayde Peregrina-Barreto,et al.  Parameter Identification of PMSMs Using Experimental Measurements and a PSO Algorithm , 2015, IEEE Transactions on Instrumentation and Measurement.

[10]  Pilar Gómez-Gil,et al.  Genetic algorithms based on a granular surrogate model and fuzzy aptitude functions , 2016, 2016 IEEE Congress on Evolutionary Computation (CEC).

[11]  Witold Pedrycz,et al.  Allocation of information granularity in optimization and decision-making models: Towards building the foundations of Granular Computing , 2014, Eur. J. Oper. Res..

[12]  Hugo Jair Escalante,et al.  Surrogate modeling based on an adaptive network and granular computing , 2016, Soft Comput..

[13]  Thomas Bäck,et al.  Online selection of surrogate models for constrained black-box optimization , 2016, 2016 IEEE Symposium Series on Computational Intelligence (SSCI).

[14]  Héctor Pomares,et al.  Evolutive Identification of Fuzzy Systems for Time-Series Prediction , 2002, PPSN.

[15]  Naser Pariz,et al.  Adaptive fuzzy fitness granulation for evolutionary optimization , 2008, Int. J. Approx. Reason..

[16]  Carlos A. Coello Coello,et al.  A Fitness Granulation Approach for Large-Scale Structural Design Optimization , 2012, Variants of Evolutionary Algorithms for Real-World Applications.

[17]  Hugo Jair Escalante,et al.  Improved Learning Rule for LVQ Based on Granular Computing , 2015, MCPR.

[18]  Yaochu Jin,et al.  Surrogate-assisted evolutionary computation: Recent advances and future challenges , 2011, Swarm Evol. Comput..

[19]  Jianchao Zeng,et al.  Surrogate-Assisted Cooperative Swarm Optimization of High-Dimensional Expensive Problems , 2017, IEEE Transactions on Evolutionary Computation.

[20]  Leifur Leifsson,et al.  Surrogate-Based Methods , 2011, Computational Optimization, Methods and Algorithms.