Models Of Crystal-Cluster Behaviors:Theoretical Analysis And Its Application

Nature, it seems, will continue to be a muse for all those who dream, be it in art or in science. Optimization algorithms inspired by nature have been widely used to solve various scientific and engineering problems because of their intelligence and simplicity. However, the inherent problems, such as time-consuming, local optima, low-applicability, are easily caused by the complicated randomness of intelligent algorithms. Meanwhile, the gap between intelligent algorithms and their applications is getting larger with the increasing of scale and complexity of practicalities. Therefore, it is an emergence and of great significance to improve their performance.The essential properties of basic elements of intelligent algorithms, which are the intelligent agents and their cluster, are analyzed. Furthermore, parallel models and strategies related to their behaviors are established obeying the rules of simplicity and effectiveness.(1) A novel intelligent algorithm, Crystal Energy Optimizer (CEO) is proposed, which is based on the analysis and modeling of behaviors of intelligent crystals.Over interaction among intelligent agents usually affects the efficiency in traditional intelligent algorithm. To tackle with this problem, parallel models of intelligent crystals, which are inspired by the natural phenomenon of lake freezing, are established. Following the law of thermodynamics, CEO has strong parallelism and robustness. The efficiency of algorithm is improved significantly.(2) A novel intelligent algorithm, Wolves Optimization Algorithm is proposed, which is based on the analysis and modeling of behaviors of intelligent clusters.The applicability of Group Search Optimizer (GSO) is relatively low. When comes to dynamic and discrete problems, it is easier to get into local optima with a lower efficiency. To tackle with this problem, wolves-dynamic model, six separated space and discrete model are introduced into GSO. The improved GSO, Wolves Optimization Algorithm has not only increased the efficiency, but also expanded the applicability of GSO.(3) A novel intelligent algorithm, Crystal Cluster Optimizer is proposed, which is based on the analysis and modeling of behaviors of intelligent clusters and crystals.Complexity of intelligent computing is too high to handle the large-scale problems effectively. To tackle with this problem, the crystal-cluster models we built combine the advantages of both parallelism and intelligence from crystal models and cluster models respectively. It has a much higher performance in large-scale problems.(4) A novel behavioral strategy, Social Assimilative Network is established, which is based on the analysis and modeling of assimilative behaviors of intelligent crystals and clusters.Most intelligence for the existing heuristic models is inferior one from basic social behaviors of animials. To tackle with this problem, social behaviors of human beings are introduced into our crystal-cluster models. Social assimilative behavior models are built and their equilibrium is analyzed based on the dynamics. The intelligence as well as the performance of the network has been improved.(5) Network problems, especially including the Dynamic Network Programming and Recommendation System, are solved with the approaches above.Existing intelligent algorithm could barely solve the increasing of complex and scale of problems in reality. To tackle with this problem, on one hand, we improve the efficiency from basic elements from the perspective of intelligent algorithm. On the other hand, the strategies of network behaviors are built from the perspective of network. The outstanding results we obtained through simulation and theoretical analysis have proved that (a) the efficiency and applicability has been improved largely in Ad Hoc network and (b) the accuracy of recommendation system is increased with assimilative strategies.