Research On The Function Optimization Method Based On The Principle Of Artificial Emotion Memory

The optimization problem in engineering has the characteristics of complexity,multi-polarization, nonlinear, strong constraint etc, which conventional optimization is powerless, the emergence and development of swarm intelligence optimization algorithm make up the defects of the traditional optimization algorithm. However, face on the optimization problem of large population the swarm intelligence optimization algorithm are still shortcomings of the optimization time, in order to improve this shortcoming, a new function optimization method based on cognitive psychology,emotion theory and memory principle is proposed.First of all, by studying the characteristics of human emotion and the interaction of cognition and emotion, according to the basic idea of swarm intelligence optimization algorithm, a new algorithm for swarm intelligence optimization- emotion recognition optimization(ECOA) is proposed. The ECOA algorithm controls the optimization of each individual through the individual’s cognition and it effects on individual choice; The behavior choice operator and the choice probability operator are used to change the individual state; In the process of the interaction between emotion and cognition, the individual is optimized. The feature of ECOA is that the behavior selection operator on the exchange of information, only exchange a few variables of state information, the convergence rate of the algorithm can be improved greatly in the high-dimensional optimization problem. With the Griewank function as an example, a preliminary test on the performance of the algorithm, verified the good performance of the algorithm, but the optimization time is not drastically shorten.Then, improved the algorithm with combined the memory and emotion recognition, the artificial memory algorithm(AEMA) is designed. In the AEMA algorithm, each memory element corresponds to a tentative solution; The transfer of the memory element state is controlled by memory and forgetting memory, so as to realize the search for the optimization problem. The characteristic of AEMA is to associate the trial solution with the memory, the test solution will be forgotten in the course of evolution. With the evolution, the trial solution which do not need to deal will more and more, the AEMA algorithm will converge faster and faster. Taking the Ackley function as an example, testing the performance of algorithm, the results show that the AEMA algorithm has obvious advantages in the problem of large population size.Finally, fifteen benchmark functions are selected for testing function performance,comparing with four optimization algorithms, it shows that the convergence effect of AEMA and ECOA is better than the existing algorithms. Especially in the large population, the AEMA algorithm is fast and short for characteristics.