Research And Application On Optimization And Simulation Algoirthms Of Swarm Intelligence

The Swarm Intelligence gains the attention of more and more scholars, this method usesthe individual ability and their corporation to achieve the system object. The algorithm andmodel of this field have the characteristics of self-government and non-centralized control, itreflects the power of collaboration between the groups, and provides a new idea and a newmethod for modeling of complex systems, complex problem solving. Classic applicationssuch as: the genetic algorithm, Multi-Agent Systems and Particle Swarm algorithm.Autonomy-oriented computing (AOC) is a new approach for the practicalproblem-solving and modeling complex systems, and is particularly effective for modelingcomplex systems. an AOC system mainly consists of a group of autonomous entities and theirliving environment or system. Based on this technology, the algorithm can solve complexproblems, such as the wireless sensor network routing or clustering. This paper researches thecorresponding application, designs and implemented the algorithm to achieve the purpose ofreducing network energy consumption, extend network life cycle.The research of the CDA market environment model and the bidding algorithm is thecore content of the field. The static CDA market, also known as the stability CDA market, itssupply and demand curves no significant changes in the market, the market presents abalanced state, in which the relationship between market supply and demand and price is arelatively static state, which is different with the market model in the real world, and bringscertain limitations to the research work. Focus on this matter, this paper proposes dynamicCDA market model theory and designs the model parameters, and evaluates the classicbidding algorithm, then proposes an improved model by analyzing and comparing the results.This model can adapt to a variety of markets environmental, also has the better performance.This article studies the model and algorithm of swarm intelligence, proposes the theCDA market bidding model based on Agent, and verifies the result of the dynamic and staticmarket environment; proposes a genetic algorithm-based wireless sensor network routingalgorithm for data transmission and collection; proposes AOC-based wireless sensor networkclustering algorithm for reducing energy consumption and extending network life cycle. Themain works are as follows:1) Propose dynamic CDA market environment model. The relationship between supplyand demand of the static CDA market is small, which does not meet the actual marketsituation, and bring some limitations to such research. The key feature of a dynamic marketenvironment is that the market supply and demand curve and the market price are changing. The main types include: price rise, price fall and price fluctuation. For fully simulating thedynamic market price fluctuations, we research and simulate these three types. We simulateequilibrium price rise, fall and shock by controling the trading time and price, assess theclassical GD and GDX algorithm performance in the dynamic market, analyze the reasonsand improve the innovation idea at last.2) Propose the Agent-based CDA bidding model. This work is the continuation of part1),because the classical bidding algorithms have some limitations in the dynamic market, basedon GDX model, we propose an improved algorithm: GDXP, in which agent dynamicallyadjusts its bidding strategy according to the environmental changes, this model also supportsthe individual to make a more rational decision. The discount rate parameters in the modeldecide the agent’s performance, it reflects the forecast of the traders on market trends, traderswill face greater loss of profit if this value is set incorrectly. Generally, the value should be adynamic variable, the traders should be appropriately adjusted according to marketconditions, based on this idea, this paper proposes the bidding models to adapt to the dynamicCDA market environment: GDXP. The algorithm uses the least-squares fitting to predictmarket trends by analyzing the transaction sequence, while proves the correctness of thetheory, and further evaluates the effectiveness of the method.3) Propose a genetic algorithm-based wireless sensor network routing algorithm. Thedata collection and information fusion in wireless sensor networks are the focus in the field.Using the Mobile Agent, sending the data processing code to the owner of the original datasensor node, instead of the data to the processing nodes, the method can save networkbandwidth and reduce the amount of data transferred, and reduce transmission delay. Thispaper uses the evolutionary algorithm theory, proposes the mobile agent routing algorithmbased on genetic algorithm: MAGA algorithm, the algorithm uses one layer natural numbercoding, can fully access the target area, do not cause the information missing, and overcomesthe shortcomings of unable to complete collection of information; transmitting radius Rdetermines the connectivity between nodes, which can adapt to the random distribution of thewireless sensor network nodes, has higher self-adaptive and more common applications; thenode position is randomly generated by the system, while supporting the artificial setting,closer to the real-world sensor nodes dispenser.4) Propose a AOC-based wireless sensor network clustering algorithm. The energyconsumption of sensor network protocols and algorithms is the main indicator of theassessment. The sensor node clustering method follows this standard, which has been provedto be an effective way. Based on the Autonomy-oriented computing, this paper propose asolution to the wireless sensor network clustering problem. In this model, every sensor nodehas the corresponging agent, the agent can sense the environment within a certain range, use acommunication mechanism to interact with other agent, and complete the clustering processthrough the shrink and enlarge. The experimental sensor network consists of100nodes,transmission distance is20meters. According to the comparison with classic LEACHalgorithm, the results show that the energy consumption of the AOC-w is smaller and the proposed algorithm is more to extend the network life cycle.Now, the computer scientist pay more and more attention to the research of the SI computingalgorithm and application, especially the wireless sensor network, the social network and theautonomy computing model, etc. the research works of the thesis will greatly enrich thestudies of this area. We hope that the work of this paper can play a certain role in promotingthe development in this field.