基于灰狼算法的民航维修人为差错评价模型-《计算机技术与发展》

文章信息/Info

Title:: Human Error Evaluation Model of Civil Aviation MaintenanceBased on Gray Wolf Optimization

Author(s):: MA Ying; WANG Rui; School of Communication and Information Engineering,Shanghai University,Shanghai 200444,China

Keywords:: civil aviation security; human error; grey wolf optimization; radial basis function neural network; particle swarm optimization; evaluation model

摘要:: 随着民航运输业的迅猛发展,航空运输量和排班量大幅度增加。航空器在可靠性和安全性等诸多方面都有了大幅度提升。由机械故障导致的安全事故比例从 80% 下降到了 20% ,而维修过程中的人为差错占比却直线上升,成为影响民航安全、飞行安全及运行成本的重要因素。因此,民航业对于人为差错备受关注。? 为了降低民航维修中人为差错的发生几率,提高维修生产和适航质量,该文提出了 4 个层面、18 个影响民航维修人为差错的因子。以东航虹桥基地为例,采用了问卷调查收集数据;通过灰狼算法( grey wolf optimization,GWO) 结合粒子群算法( particle swarm optimization,PSO)以及增加三种改进策略,提出一种惯性自适应混合灰狼算法( inertial adaptive hybrid grey wolf optimization,IAHGWO) ;并构建了惯性自适应混合灰狼算法训练径向基函数神经网络( radial basis function neural network,RBFNN) 评价模型;结果表明该评价模型具有良好的实用性及准确性,弥补了现阶段民航企业适航质量监管体系对维修人员个体的人为差错管控中针对性、实时性、预见性上的不足。

Abstract:: With the rapid development of civil aviation transportation,the amount of air transportation and scheduling increased greatly.Aircraft have been greatly improved in many aspects, including reliability and safety. The proportion of safety accidents caused bymechanical failures decreased from 80% to 20% ,while the proportion of human errors during maintenance increased sharply,becomingan important factor affecting civil aviation safety,flight safety and operation cost. As a result,civil aviation industry is concerned abouthuman error. In order to reduce the probability of human error in civil aviation maintenance and improve maintenance production and air鄄worthiness quality,eighteen factors affecting human error in civil aviation maintenance from four levels are put forward. Taking HongqiaoBase of Eastern Airlines as an example, a questionnaire survey was used to collect data, and an inertial adaptive hybrid gray wolfoptimization ( IAHGWO) was proposed,which combined with particle swarm optimization ( PSO) and three improved strategies,and anevaluation model was constructed to use IAHGWO to train BP neural network. The results show that the proposed evaluation model hasgreat practicability and accuracy,which makes up for the shortcomings of the current airworthiness quality supervision system of civilaviation enterprises in the pertinence,real-time and predictability of human error control of maintenance personnel.