Research On FMS Fault Diagnosis Based On Bayesian Network

Flexible manufacturing systems(FMS)as one of the cutting edge technology of automatic manufacturing technology in today's world,are widely used in machinery manufacturing,automobile,marine,aviation,etc.FMS fault diagnosis is a key technology to make FMS work safely and high efficiently.Currently,in complicated equipment fault diagnosis,particularly in FMS fault diagnosis there are some problems hard to tackle,such as fuzziness of fault information,uncertainty of Logic relationship between faults and polymorphism of fault state,however most of the fault diagnosis methods fail to deal with above problems.Bayesian networks based on probability theory and graph theory has a great deal of advantage in addressing uncertainty fault of complex equipment.So this paper proposes a Fuzzy Bayesian network approach.The main research as follows:1)This paper first establishes fault model through fault tree method,then transforms it into Bayesian network model.In addition this paper proposes observing nodes used to describe symptom information,it makes diagnosis results more reasonable and more accurate.2)Using fuzzy theory to obtain fault prior knowledge and set up conditional probability table is better to solve the problem that prior knowledge is hard to be acquired in practical engineering.3)Applying joint tree algorithm to achieve fault probability reasoning,in order to reduce complexity of reasoning and improve the efficiency of diagnosis system.4)The above methods are applied to FMS fault diagnosis,and achieve the reasoning of FMS fault diagnosis system by software HUGIN,the results show that this method is feasible.5)Developing a FMS diagnosis system based on Matlab and C#mix programming technology to apply to the practical fault diagnosis.