生产线故障诊断知识引擎系统的集成推理方法

针对大型复杂连续性生产流水线设备所具有的多部件结构、多层次故障症状、故障的不确定性等特点及对其状态监测和故障诊断的实时性要求,在设计了基于综合动、静隶属度运算的模糊识别法和基于神经网络的并行运算方法的基础上,研制了生产线故障诊断知识引擎系统的集成推理机制,突破了只有单一推理方法的诊断系统在领域知识和推理能力方面的局限性,提高了智能诊断系统对偶然性和不确定性问题的诊断能力,同时在一定程度上解决了诊断系统对生产流水线所具有的大量的状态参数和控制信号等知识获取的"瓶颈"问题以及由此而来的快速推理难题,使该智能诊断系统具有在线实时诊断能力.

Integrated inference technique of knowledge engine system for fault diagnosis of product line

In particular consideration of such characteristics of the large complex continuously-operating mechanical device of product line as multi-component structures,multi-level fault symptom and uncertainty of faults,and the needs of the device for real-time monitoring and diagnosing,this paper develops an integrated inference technique of the knowledge engine system for fault diagnosis of product lines by the design of a fuzzy recognition method based on the comprehensive static-dynamic membership computing and a parallel computing method based on neural networks.Breaking through the limits in domain knowledge and reasoning capability of the intelligent diagnosis system with a single method of reasoning,the present technique improves the capability of diagnosing faults of this system,which are characterized by contingency and uncertainty,and partly solves the problem of "bottleneck" in knowledge acquisition of the intelligent system caused by a large number of state parameters and control signals of product lines,as well as the puzzler of quick inference caused by a lot of data,so that the intelligent diagnosis system can be endowed with the capability of real-time on-line diagnosis.