Research And Application Of Firefly Algorithm With Neighborhood Multi-Topology

Since the 19th Party Congress was held,the goal of "further implementing the innovationdriven development strategy and accelerating the building of a strong science and technology nation" has been established.The Internet industry is gradually transitioning to an information technology era that combines software and hardware such as artificial intelligence,big data,5G communications and smart chips.From scientific theoretical breakthroughs to industrial applications,the optimization problems faced by human beings are becoming more and more complex,and traditional optimization techniques are often unsustainable due to their discrete,non-convex,non-trivial,multi-modal,largescale,high-dimensional and multi-objective characteristics.How to explore feasible solutions and analyse their feasibility in practical engineering problems is one of the major research issues that need to be addressed,and is also a "key trick" to realise the application scenarios.Firefly Algorithm(FA)is a relatively new meta-heuristic in swarm intelligence optimisation algorithms.It is constructed by simulating the flashing attraction behaviour of individual fireflies.Due to the simplicity of the firefly algorithm,the ease of programming,the small number of algorithm parameters and the good optimisation performance,it has been widely used in various optimisation problems.However,when solving multimodal sequential optimisation problems,the population diversity is significantly reduced by the complete attraction mechanism of the Firefly algorithm,resulting in the algorithm easily falling into local optima.At the same time,excessive attraction between individuals in the evolutionary process of the Firefly algorithm causes search oscillations,which in turn leads to problems such as premature convergence.In order to address the shortcomings of the firefly algorithm,we innovate the attraction mechanism between firefly individuals,improve the firefly individual position movement formula,study the neighbourhood multi-topology structure in depth,and verify its effectiveness through function test sets and practical application problems.The main work of this paper is as follows:1.A firefly algorithm based on the neighbourhood topology of a scale-free network is proposed.The characteristics of the scale-free network conform to the distribution pattern between the optimal and potential individuals of fireflies;the ring network effectively reduces the number of attraction between fireflies,which in turn reduces the computational complexity of the algorithm.This paper introduces a Gaussian chaotic mapping strategy to reduce the sensitivity of the algorithm parameters and uses a strategy to calculate the distance based on dimensional differences to reduce the algorithm running time.The diversity neighbourhood enhancement search strategy for firefly position update is improved to enhance the global search capability of the algorithm.In order to balance the exploration and exploitation capabilities of the algorithm,a phased algorithm is designed to balance the search mechanism,and the whole algorithm process is divided into a global search phase and a local search phase.The global search phase uses a scale-free network and a ring network neighbourhood topology population co-evolution mechanism,and the local search phase introduces a simplex strategy to fine-tune the firefly individuals locally in order to improve the problem solving accuracy.Finally,the proposed algorithm is verified to be superior through three experimental studies: parameter study,algorithm comparison and strategy evaluation.2.A firefly algorithm based on a small-world network neighbourhood topology is proposed.The "six-degree space theory" feature of the small-world network greatly enhances the connection between individuals,which in turn improves the individual’s ability to find the optimal value;the sex of individual fireflies is differentiated based on the male-female dyadic search mechanism,and the stability of the optimal search is balanced by the different degrees of bias of male and female individuals towards exploration and exploitation,and the position update of individual fireflies is improved by the In order to improve the solution accuracy of the male-female dyadic population,a dimensional exchange strategy and a differential variation strategy are introduced to update the feasible solutions of the optimal individuals through the variation of the dimensional and spatial relative positions of the firefly individuals;in the later stage of the algorithm,the chaotic search mechanism is added to the local optimal search stage,and a logical mapping function is used to generate a chaotic sequence.The chaos search mechanism is applied to individuals of subpopulations that have evolved through iterations of the small-world network topological population and the malefemale dyadic population.Finally,this paper analyses both theoretical and experimental results to verify that the proposed mechanism and strategy can effectively improve the performance of the firefly algorithm.3.A general framework based on the multi-topology neighbourhood firefly algorithm is proposed,and apply it to the agent-assisted optimization problem of breast cancer detection.Based on the two algorithms proposed in this paper,the general framework of the Neighbourhood MultiTopology Firefly algorithm is designed by condensing and summarising them;outlining the classification and optimisation problem of breast cancer detection,analysing the drawbacks of existing techniques and problem characteristics,constructing an agent-assisted optimisation model using Radial Basis Function(RBF),incorporating the general framework of the Neighbourhood Multi-Topology Firefly algorithm proposed in this paper,and combining with Support Vector Machine(SVM)The model is improved and optimized by using breast cancer related data sets as training data,and finally a breast cancer detection classifier with high solution accuracy and low computational cost is obtained.