PHM Technology Of Roll To Roll Equipment Based On Fuzzy Neural Network

In recent years,China’s manufacturing is going through significant progress with the announcement of the plan entitled “Made in China 2025”.Consequently,the standard requirement for equipment performance raises.The conflict,however,exists between growing production and performance degradation of equipment.To resolve the conflict,more attention is paid to reduce safety hazards and extend service life of equipment by adopting scientific maintenance strategies.Flexible material roll-to-roll(R2R)production equipment has been widely used in industries like PET,PVC,floor heating film,transfer paper,PCB/FPC,membrane switch,solar cell,glass,etc.As a multiaxial continuous production system,the performance of the key component-roll shaft,will have a direct impact on the equipment performance.But R2 R equipment has hundreds of roll shafts working simultaneously to complete different processes.This makes it difficult to apply suitable performance index to represent and analyse the equipment status.This paper,titled “PHM Technology of R2 R Equipment Based on Fuzzy Neural Network”,is designed to investigate the connection between the vibration feature parameter of roll shaft and the performance degradation of equipment,so as to provide valid theoretic and technical support to the intelligent maintenance and intelligent manufacturing of R2 R equipment.Taking the analysis of different vibration features of roll shafts,the collecting of PCA performance degradation index and the modeling of R2 R equipment degradation prediction as focal points,this paper involves following contents:I.To analyse the technological process of flexible material R2 R manufacturing system and identify different roll shafts,therefore illustrate that the performance of roll shafts can have great impact on equipment performance.Then to demonstrate the connection between the equipment degradation and the vibration signals which contains performance features and are produced by roll shaft vibration.II.The amount of interrelated vibration feature parameters such as time domain and frequency domain surpass other parameters.If all these parameters are put into the follow-up prediction model,the system computation will definitely bear more burden.Thus the PCA technique will be used here to tackle this problem.Results show that this technique can efficiently identify different vibration characteristics,indicate precisely the performance status of roll shafts,and reduce input parameters of the prediction model.III.In this paper,a neural-network prediction model is proposed which combines PCM and TS-FNN,to predict the performance degradation of R2 R equipment.This model takes performance degradation PCA index of roll shafts from different positions as inputs,by going through multiple iteration processes to achieve fuzzy classification of the sample and self-organizing adjustment of structure parameters.Experiment shows that PTS-FNN model has outstanding prediction performance no matter on prediction accuracy or convergence speed.