March 29, 2024

Introduction to the model method of the spring

The composite spring network model and method in this paper is different from the elastic network for solving optimization problems such as TSP. Claudia, Burr, and Stone, respectively, propose improved faster elastic network algorithms to solve TSP problems. The elastic network originates from an early model of establishing topological ordered mapping in the brain. The basic idea of ​​solving the optimization problem is that the elastic network is a single-loop tension ring formed by a rubber band with uniform thickness, and the target points passing through the ring. The location of the (city) is fixed and there is no interaction between the target points. The elastic network can be stretched into a certain geometry. An elastic network state that satisfies the optimization problem constraints and has the least elastic potential energy corresponds to a better solution to the problem. The MAS problem is much more complicated than the TSP, and is fundamentally different from the optimization problem solving such as TSP. There are concurrent random social interactions between MAS agents. The relative position (physical or logical) between agents is not fixed. Each agent has no global common goal and global consistency knowledge. Therefore, elastic network model and method Not suitable for MAS distributed problem solving.

The simulation experiment of multi-task self-organizing planning and resource self-organizing allocation of MAS and the experiment with Shehory-Kraus algorithm show that the composite spring network model method is far superior to Shehory-Kraus algorithm in many aspects such as computational complexity and applicability. . The composite spring network model solved without loss of generality, we take MAS distributed mission planning and resource allocation as an example to illustrate the mathematical physical model and algorithm of the composite spring network. Set the set of Agent A={A1,A2,...,An} and the set of tasks T={T1,T2,...,Tm}, the resource vector owned by AgentAi is ri=(ri<1>, ri<2> ,...,ri ), the resource vector required by task Tj is dj=(dj<1>, dj<2>,...,dj ). Multiple agents participating in the same task form an alliance. In order to utilize the composite spring network parallel search optimization solution and measure the merits of the solution results, we define a comprehensive evaluation function. When the total income of all agents is larger, the income of the agent with the lowest return is larger, and the sum of the remaining capabilities of each agent. The smaller the resource vector required by each task and the better the matching between the multiple resource vectors provided by multiple agents, the larger the value of the comprehensive evaluation function. We do not discuss how each agent decides what kind of social behavior to take against other agents based on its own environment and personality. The nodes of each agent are distributed around the center of the circle at equal intervals and in any order. The distance from the point to the center of the AgentAi is proportional to the gain of Ai in the current situation of the MAS. The radius of the circle is larger than the maximum possible return value of the single Agent. Each agent node is in the gravitational field formed by the circle. The gravitational field drives the various agent nodes to move along the radial direction as far as possible to the edge of the gravitational circle, thereby indicating that each agent tries to obtain its own maximum benefit. Each Agent wants to participate in the alliance of high-paying tasks, but at the same time it is subject to its own degree of concern for the overall interests of MAS, that is, in addition to the gravitational field, the force of the Agent node is also related to it. The personality of interest is related to autonomy.

Real Leather Care Solution

Guangdong Giant Fluorine Energy Saving Technology Co.,Ltd , https://www.tuwtech.com