Research On Noise Problem Of Limited-aware Swarming Cluster Model

Collective behavior is a highly abstract of biological groups and labor groups of real-life. A single intelligent individualcan only complete simple tasks, but the multi-agent system composed by a lot of smart individuals able to complete more complex tasks and save more difficult problems in reality through the interaction between individuals. In the current society, multi-agent system has been widely applied to all aspects of people’s lives, such as environmental monitoring, disaster search and rescue, military operations and industrial operations, and many other fields, so it has important theoretical and practical significance to study multi-agent system.In order to analyze the nature of intelligence community, people construct the limited perception swarming cluster model. The model describesstate of motion of the i-th individual of N constantly interacting agents randomly distributed in space. It is built on attracting / repulsive force function, the basic principle to follow is "attractive work as far, exclusion as near.” All the intelligent individuals described by the model can converge in a finite period of time to a bounded region, with good local stability. When we performed theoretical calculations, Gravity / repulsion function requires take both of mode of gravity item g_a（|y|） and mode of repulsion term g_r（|y|） into account,therefore, the process is complexrelatively, it is difficult to get the dataquickly in actual operation, so the efficiency is low relatively. Moreover, there exist situation that noise interferenceoccurredin practice, in this paper, the uncontrollable of noise is proved through the study of the noise problem, at this time, noise interference term also affect the stability of the model. Consider that state equation of model and Consistency is essentially the same, whether can we introduce the noise control strategy of Consistency into equation of state of the model, to study the specific model?To solve these problems, this paper does the work as follows:Limited perception swarming cluster model was carefully studied and analyzed, and we discover that the gravity / repulsion function in the model need to take both of mode of gravity item g_a（|y|） and mode of repulsion term g_r（|y|） into account, so it is difficult to calculate relatively, thus slow the efficiency of calculation. Because of this, we simplify the function by merge the gravity / repulsion functions into a function, which make the operation process simpler. Then introduce the weighting factor G that is gravity into the model, in this case, the farther the distance between the agents, the greater the attractive force between them, whicheffectively improve the efficiency of calculation of the model. Through the scientifictheoretical analysis, prove that the model keep local stability during the simplify.This paper do research on noise problem in Consistency algorithm, prove that noise is uncontrollable, then propose a Noise Control Strategy based on noise suppression operator ε（t）,prove that when the high-order infinitesimal of t^-0.5 when ε（t） taken, the noise is controllable afternoise suppression. In the absence of noise, whent in ε（t） take low-order infinitesimal of t^-1, the algorithm after noise suppression can make the final state of agents converges to X^*; In noisy situations, when t values between t^-0.5 and t^-1, the final status of multi-agent will be in X^* centered normal distribution. And the feasibility of noise control strategy based on noise suppression operator ε（t） is verified through By DHA algorithm as an example to be discussed and verified.Consider that state equation of model and Consistency is essentially the same, we introduce the noise suppression operator ε（t） into equation x_m（t+1） of state of the model, prove that in the absence of noise, when t take low-order infinitesimalof t^-1 or high-order infinitesimal of t^-0.5, algorithm after noise suppression can make the final state of agents converges to X^* centered normal distribution, that agents in model eventually converge to the bounded region.In summary, this paper does deep research on the limitedperceptionswarmingclustermodelandnoiseproblem in Consistency algorithm. And summary the advantages and disadvantages of the model, study and prove the uncontrollable of noise in Consistency algorithm, put forward effective solutions to deal with these problems, demonstrate the effectiveness of the program through theoretical analysis and experimental results.