Precision Control Technology Of Micro-Prism Array Based On Online Optical Sensing During Rapid Hot-Embossing

The rapid hot-embossing process can be used to fabricate micro-prism array on the surface of PMMA light guide plate,which can improve the luminous efficiency and uniformity from spot light to surface light.However,in the rapid hot-embossing micro-forming process,there is a non-linear relationship between the micro-forming accuracy and the micro-forming process parameters.Pressure sensor accuracy drift,equipment wears,and environmental changes will cause fluctuations in micro-forming control.Moreover,the forming accuracy is determined offmachine detection,which leads to a lag for the control of the fabricating process.The problem of controlling the micro-forming accuracy during the process is: How to sense the forming accuracy of the micro-prism array on the macro surface during the hot-embossing process within 1?3 seconds?Therefore,the reflection and refraction of white light generated in the micro array structure is utilized,resulting in directional changes of luminous flux,identifying the micro-prism array height,and correlating the hot-embossing process parameters.The technical method is to decide the process parameters through the Naive Bayes algorithm based on the hot-embossing process experience database,and then correct the parameters by online optical sensing of the microprism array height.Wherein,the key is to integrate the empirical data in onsite process and the characterization data of the optical sensing illuminance,and finally realize the online control of rapid hot-embossing process.First,an online detection platform for the macro surface of the micro-prism array based on optical sensing was designed.An automatic communication system apparatus was built to connect the hot-embossing machine,online optical sensing data and the computer control center.Then,experiments were performed to analyze the effect,and set up the hot-embossing process experience database.the optical sensing illuminance curve in the hot-embossing process was characterized,and the height of the micro-prism array was matched with the characteristic value of optical-sensing illuminance.After verification,the fitting error is 1.07 ?m on average.Subsequently,a process parameter selection model was proposed with BP?RBF neural network algorithms.The experimental results show that the RBF neural network can better reflect the change of the original empirical data curve.In order to make error correction feedback on database updates timely,a parameter selection model of the Naive Bayes algorithm was constructed.The results show that the average prediction error is 1.52 ?m,while it is 2.20 ?m for the RBF neural network algorithm.Finally,a precision control model of the micro-prism array during rapid hot-embossing was established,combined with the intelligent parameter selection by the Naive Bayes algorithm and the parameter adjustment by online optical sensing and pressure correction algorithm.The experimental results show that: For light guide plate with expected micro-prism array height of 70 ?m,in 20 experiments,the micro-prism array forming height can be controlled within an error range of ±3 ?m,which is realized through the intelligent temperature/pressure/holding-time selection by the empirical database and the pressure adjustment by the repeated optical sensing online test and pressure correction algorithm.As a conclusion,the precision forming process of hot-embossing for the micro-prism array on the 85 mm×85 mm light guide plate surface has been achieved.