Font Size: a A A

Study And Application Of Emergency Decision On Case-based Reasoning Of Gas In Coal Mine

Posted on:2015-09-04Degree:MasterType:Thesis
Country:ChinaCandidate:L Y ZhangFull Text:PDF
GTID:2181330422487406Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
Gas accidents cause serious damage to coal industry in China, restricting thesustainable development of coal mining enterprises. To identify and analyze thefactors leading to gas dangerous and build an effective gas emergency decisionsystem based on case-based reasoning is one of the most important tasks in the coalmine safety research. It is quite significant for system construction to obtain a solutionof current problems by Case-based reasoning, which retrieves past similar problems incase base. The case base establishment and case retrieval are two most critical issuesof case-based reasoning that directly affect the accuracy and speed of case-basedreasoning. Therefore, building a good case base and carrying out case-based reasoningwith related attributes are of great significance.In order to solve these problems, specific studies are as follows:Firstly, the gas accident hazards and the process of emergency in accidents areanalyzed. Index and organization methods of case base in case-based reasoning arespecially researched. A multi-level-index structure is proposed, and case base of gasaccident case is indexed according to the structure which set accidental catastrophesin public emergencies as the first level of attribute. A two-level organizationalstructure of case base is presented, which takes advantage of cluster analysiscogitation, uses K-means clustering algorithm to cluster case base, finds typical caseand divides the case base into multiple sub-case base. An improved K-meansclustering algorithm, which constructs representative case base and divides case baseinto multiple sub-case base without cross case, is given to overcome the problem thatacross cases exist in case base after clustering which makes the search resultsinaccurate.Secondly, the contents of gas accident emergency cases are discussed in detailand expressed with framework methods. The fuzzy attributes with triangular fuzzynumber is quantized. The case retrieval process is studied particularly and a fuzzyhierarchy entropy engine method is proposed to calculate case feature attributeweights by using K-nearest neighbor algorithm to calculate the similarity between thecases and retrieve the most similar k cases with the target case.Finally, gas explosion instances are depicted, calculated and analyzed in detail byusing the attribute calculating method and case retrieval algorithm presented in thispaper. The results show that the algorithm is effective and easier to find a similar case and keep away with the dissimilar cases. A gas explosion accidents emergencycase-based reasoning system is built. The work process, modules and functions of thesystem are described, and the corresponding design interface is also given.
Keywords/Search Tags:gas, Case-based Reasoning, clustering, fuzzy hierarchy entropy, K-nearest neighbor algorithm
PDF Full Text Request
Related items