Research On Knowledge Push Technology For Process Plan

Process plan needs to consider many factors,and the experience of process planner is highly demanded.Most process planners still need to refer to typical process cases and process knowledge retrieval to carry out process plan.This paper proposes a method of real-time push process based on the current state to improve the current situation.In this paper,by describing the relationship between processes to construct Markov model,and proposes two process pushing methods based on the Markov model.The first is process push method based on combined Markov models to solve the step selection problem of the Markov model.The combined Markov models use more historical information to make the accuracy higher.The second is process push method based on multi-model combined Markov models.Through analyzing the characteristics of the process,multiple Markov models are constructed according to its attribute characteristics,and then select the model according to the attribute characteristics of the plan target to carry out the process knowledge push.The dynamic parameter method is put forward in the combination parameters of the process push method based on combined Markov models.This method uses the information in the model to calculate the parameters,which is more suitable for the specific pushing task.The process push experiment was carried out on 2,348 aircraft sheet metal parts fabrication order,the Top1-Top5 accuracy rate of process name pushing is 0.9709,and the Top1-Top5 accuracy rate of process name and operation description pushing is 0.9431.Experimental results show that the process knowledge pushing method based on Markov model is effective.Focusing on the problems arising from the Markov model process push method and combining the advantage of Long Short-Term Memory(LSTM)network in learning the dependence characteristics of the long-distance process,a process push method based on Long Short-Term Memory(LSTM)network is proposed.The process push experiment was carried out on 2,348 aircraft sheet metal parts fabrication order,the Top1-Top5 accuracy rate of process name pushing is 0.9875,and the Top1-Top5 accuracy rate of process name and operation description pushing is 0.9884.Experimental results show that the process knowledge pushing method based on Long Short-Term Memory(LSTM)network is effective.Taking the aircraft sheet metal parts as an example,this paper realizes the knowledge pushing system,and provides the process knowledge push service for the process designer in the process planning.The system includes the functions of pushing similar process example,pushing process name and operation description.