An improved AEA algorithm with Harmony Search(HSAEA) and its application in reaction kinetic parameter estimation

• An improved AEA algorithm with Harmony Search (HSAEA) is proposed.
• Estimation of distribution algorithm is used to generate the compared population in AEA.
• The moving step length in AEA is improved to vary with different phases of the iteration.
• Harmony search is introduced to improve the quality of population.

Alopex-based evolutionary algorithm (AEA) is one kind of evolutionary algorithm. It possesses the basic characteristics of evolutionary algorithms as well as the advantages of gradient descent methods and simulation anneal algorithm, but it is also easy to trap into a local optimum. For the AEA algorithm, the unreasonably settings of compared population and step length are two typical drawbacks, which lead to the lack of communication between individuals in each generation. In this paper, estimation of distribution algorithm (EDA) is employed to generate the compared population. Then the moving step length in AEA is improved to vary with different phases of the iteration during the actual operation process. And more importantly, harmony search algorithm (HS) is introduced to improve the quality of population of every generation. By compared with original AEA, the performance of the improved algorithm (HSAEA) was tested on 22 unconstrained benchmark functions. The testing results show that HSAEA clearly outperforms original AEA for almost all the benchmark functions. Furthermore, HSAEA is used to estimate reaction kinetic parameters for a heavy oil thermal cracking three lumps model and Homogeneous mercury (Hg) oxidation, and satisfactory results are obtained.

Alopex-based evolutionary algorithm (AEA) possesses the basic characteristics of evolutionary algorithms as well as the advantages of gradient descent methods and simulation anneal algorithm. According to the communication between individuals in each generation, the compared population is generated by estimation of distribution algorithm (EDA) and the moving step length in AEA is improved to vary with different phases of the iteration during the actual operation process. More importantly, harmony search algorithm (HS) is introduced to improve the quality of population of every generation. By compared with original AEA, the performance of the improved algorithm (HSAEA) was tested on 22 unconstrained benchmark functions. The testing results show that HSAEA clearly outperforms original AEA for almost all the benchmark functions. Furthermore, HSAEA is used to estimate reaction kinetic parameters for a heavy oil thermal cracking three lumps model and Homogeneous mercury (Hg) oxidation, and satisfactory results are obtained.Figure optionsDownload as PowerPoint slide