基于BP神经网络的飞机装配车间异常事件预警

摘要

In the aircraft assembly workshop, because of a variety of factors, such as the movement of persons, equipment failures, shortage of materials, manu-facturing process and so on, it often brings about some work delays according to plan, even some abnormal events like job stoppages. After comprehensive analysis of the ab-normal events in aircraft assembly workshop, we analyzed the typical inlfuential factors of abnormal events based on the method of 4M1E, which represents Manpower, Ma-chine, Material, Method and Environment. Then we estab-lished the prediction model for abnormal events in aircraft assembly workshop based on BP neural network to provide strong supports for information monitoring and response. Finally, we veriifed the application of the prediction model using the real data of Shanghai Aircraft Manufacturing Company, and the experimental results showed that the model has a high performance.%在飞机装配现场工作中，因为车间人员流动、设备故障、物料短缺或者工艺等因素影响，常常造成工作未能按计划进行，甚至造成工作等待等异常事件。在综合分析飞机装配车间异常事件的原因后，基于4M1E方法（人、机、料、法、环）分析异常事件影响特征因素，并利用BP神经网络建立飞机装配车间异常事件预警模型，实现对车间异常事件的分类预警，支持企业及时、快速地对异常事件做出响应。最后利用上海飞机制造有限公司装配车间异常事件数据对本模型的应用进行验证，结果显示该模型有很高的检测正确性，可用于飞机装配车间实际应用。