The data breadth and data depth are enriched substantially at the age of the mobile internet and social network. Efficient mining and analysis of the mass big data will bring about huge commercial value and profound social benefit for the social development. The fast-growing economic society gives a new definition for the data analysis value, i.e. diversity and temporality and spatiality of data value. As a smart pipe for the mobile internet and social network, the information communication network has rich data resources. On the basis of the operation and maintenance management for the information communication network, the research involves application of the swarm intelligence theoretical method for mining and analysis of the association rules for network alarms, puts forward the association rule on mining algorithm for the network alarms based on the swarm intelligence and the behavior characteristics of the typical colonial organisms (insects-ant colony, birds-particle swarm, mammalia-wolf pack). The research achievements of the paper are of great significance for the theoretical and practice research on the analysis of the association rules for alarms in operation and maintenance management for the information communication network. The main research contents and work of this paper include:(1) In combination with the evolvement of the operation and maintenance management for the information communication network, the thesis puts forward the six patterns of the network alarm management and the alarm management system and four stages of the network system development specific to the network management supporting system. The paper clarifies the network alarms, alarm standardization, alarm standardization fields, network system alarm standardization etc. in the information communication network, puts forward and discusses the data characteristics and preprocessing flow of the alarms in the information communication network, and allows the logic correlation of the network alarms to be present both within and among the network devices. In addition, the thesis applies the Apriori algorithm-based concept and model and the FP-Growth algorithm idea to the analysis of the association rules for the alarms in the communication network, and evaluates and analyzes the algorithm performance from the three aspects of number of alarm association rules, algorithm operation duration, and algorithm memory consumption, and puts forward the engineering test analysis method with the correlation rate and association strength being the analytical indexes.(2) The thesis finds out the strategy for selecting the shortest foraging routine based on pheromone by observing and analyzing the biological behaviors of the ant colonies, such as nesting, asymmetric double bridge experiment etc.. It raises and achieves the SN-APLAC algorithm (Sharing mechanism Niche-Apriori Leaping Ant Colony) by combining the sharing mechanism niche technology in biology population characteristics of the Indian leaping ants. The SN-APLAC algorithm eliminates the invalid "path points" in the TSP undirected graph. All subsets of all items in the "paths" in the undirected graph are considered as the potential frequent item sets. The "stimulation points" are obtained by judging whether the frequent item sets meet the minimum support. A leaping ant path diagram containing "stimulation points" is formed. The new potential association relationships are mined in the raw data by means of "crossover and mutation operations" of the sharing niche. The overall quality of the SN-APLAC algorithm is evaluated in a comprehensive manner from the two aspects of performance tests (the relationship between number of rules, number of ants, number of iterations and memory resources, performance differences between the SN-APLAC algorithm and Apriori algorithm/FP-Growth algorithm) and engineering tests (association rate and association strength).(3) Based on the fundamental principle of achieving the particle swarm algorithm by the flock foraging behavior (Apriori Particle Swarm Optimization), the APPSO algorithm is achieved by establishing sample particle swarm, candidate particle swarm, and rule particle swarm. Finally, the paper carries out the APPSO algorithmic logic by sequencing and numbering of the network alarm ID, computing the sparse chain table-based support, and optimizing the nature of the Apriori algorithm. Besides, The thesis conducts a performance test analysis (relationships between support/confidence coefficient/particle swarm size and number of association rules and relationship between number of iterations and operation duration) and an engineering test analysis (association rate and association strength).(4) The paper presents the basic logic of the AWPS algorithm based on alpha wolf, detective wolf, and fierce wolf (Apriori Wolf Pack Search) by analyzing the group behaviors of the wolf pack, such as social rank, foraging, attack etc. In combination with the distribution characteristics of the alarms in the communication network, the thesis proposes the "equal width" boundary compression division method with the compression function being the benchmark, applies the method to the preying and hunting processes of the detective and fierce wolves, and finally develops the AWPS algorithm. Combining with the network alarms, the paper conducts a performance test for the AWPS algorithm (relationship between wolf pack size, number of iterations and number of association rules, relationship between distribution of types of wolf packs and number of the association rules, performance comparison between AWPS and Apriori algorithm/FP-Growth algorithm) and an engineering test (association rate and association strength). |