Earticle

In order to improve the slow convergence speed and avoid falling into the local optimum in ant colony optimization algorithm, an improved quantum ant colony optimization (IMAQACO) algorithm based on combing quantum evolutionary algorithm with ant colony optimization algorithm is proposed for solving complex function problems in this paper. In the IMAQACO algorithm, the quantum state vectors are used to represent the pheromone, the adaptively dynamical updating strategy is used to control pheromone evaporation factor, the quantum rotation gate is used to realize the ant movement and change the convergence tend of quantum probability amplitude, quantum non-gate is used to realize ant location variation, so the IMAQACO algorithm has better global search ability and population diversity than ACO algorithm. In order to test the optimization performance of IMAQACO algorithm, several benchmark functions are selected in here. The tested results indicate that the IMAQACO can effectively improve the convergence speed and avoid falling into the local optimum, and has a stronger global optimization ability and higher convergence speed in solving complex function problems.

최근 양자컴퓨터의 개발과 더불어 양자역학의 특성을 응용한 양자기반 탐색기법의 개발과 공학 문제에의 적용은 매우 흥미로운 연구주제 중 하나로 부각되고 있다. 이 알고리즘은 기본적으로 0과 1이 중첩되어진 양자비트를 이용하여 정보가 저장되고, 양자게이트 연산을 통해 해에 접근하게 된다. 이 과정에서 알고리즘은 탐사와 개척 두 가지 탐색 특성간의 균형이 자연스럽게 유지되며, 진화정보가 계속 누적된다는 장점으로 기존의 탐색법과 차별되어 새로운 알고리즘으로 평가되었다. 본 연구에서는 이와 같은 양자기반 진화알고리즘을 평면 트러스의 구조최적화에 적용하여 최소중량설계 기법을 제안하였다. 최적화 수리모형에서 비용함수는 최소중량이며, 제약함수는 변위와 응력에 관한 함수로 구성하였다. 진화정보의 누적과 수렴 과정을 알아보기 위해서 10부재 평면 트러스와 17부재 평면트러스 예제를 수치예제로 채택하여 결과를 분석하였다. 수치예제의 구조최적설계 결과에서 볼 때, 기존의 고전적 탐색기법의 연구결과와 비교해서 더 나은 최소중량 설계의 결과를 얻을 수 있었으며, 진화정보의 누적된 결과로 해의 정밀도를 관찰할 수 있었다. 또한 누적된 진화정보인 양자비트의 확률적 표현은 종료시점을 쉽게 판단할 수 있다.

With the development of quantum computer, the development of the quantum-inspired search method applying the features of quantum mechanics and its application to engineering problems have emerged as one of the most interesting research topics. This algorithm stores information by using quantum-bit superposed basically by zero and one and approaches optional values through the quantum-gate operation. In this process, it can easily keep the balance between the two features of exploration and exploitation, and continually accumulates evolutionary information. This makes it differentiated from the existing search methods and estimated as a new algorithm as well. Thus, this study is to suggest a new minimum weight design technique by applying quantum-inspired search method into structural optimization of planar truss. In its mathematical model for optimum design, cost function is minimum weight and constraint function consists of the displacement and stress. To trace the accumulative process and gathering process of evolutionary information, the examples of 10-bar planar truss and 17-bar planar truss are chosen as the numerical examples, and their results are analyzed. The result of the structural optimized design in the numerical examples shows it has better result in minimum weight design, compared to those of the other existing search methods. It is also observed that more accurate optional values can be acquired as the result by accumulating evolutionary information. Besides, terminal condition is easily caught by representing Quantum-bit in probability.