Research And Realization Of Liquid Level Detection And Control System In Float Glass Furnace

Float glass technology is the mainstream technology of flat glass production in building,photovoltaic,automobile and other industries.Melting kiln is the most important thermal equipment in float glass process line.It mainly undertakes the work of melting solid raw materials into qualified glass molten liquid.The height of the glass liquid level in the melting kiln will affect the thermal efficiency of the fuel,the life of the melting kiln and the rate of good glass.Therefore,continuous accurate detection and stability control of glass liquid level in melting kiln is of great significance to improve the level of production automation,reduce energy consumption,prolong the life of melting kiln and reduce the rate of defective products.However,there are some problems in the current liquid level detection methods,such as low detection efficiency,large detection error and high maintenance cost,which cannot accurately reflect the change of liquid level height in real time.At the same time,the current liquid level control of the furnace is linear control,the control error is large,the need for manual correction,the automation level is low.Aiming at the problems of low precision and great difficulty in liquid level height detection of melting kiln,an image-type glass liquid level height measurement system was established.The glass liquid level height was calculated by detecting the distance between the marker of melting kiln wall and the edge of glass liquid level,and combining with the amplification factor of pixel distance.Then,aiming at the problems that existing edge detection algorithms have large detection errors and decreased edge positioning accuracy in high temperature scenarios,a CHED model based on multi-scale feature fusion is proposed.By improving the feature extraction network and multi-scale side output layer and integrating adaptive Canny algorithm,the accuracy and detection speed of edge detection are improved.Finally,Zernike moment method was used to fit the sub-pixel edge to further improve the accuracy of liquid level detection and reduce the detection error.Aiming at the problems of large error and low automation level in liquid level control of melting furnace,a liquid level prediction control model combining BP neural network and fuzzy PID control was proposed by analyzing the influencing factors of glass level in melting furnace.Firstly,the BP neural network was improved by adding momentum factor and correcting learning rate to improve its training convergence speed and alleviate training shock phenomenon.Then the improved BP neural network is used to predict the height deviation of the liquid level,and the PID controller parameters are set online according to the set fuzzy rules to complete the optimization control of the liquid level in the melting kiln.The simulation results show that the overshoot of the proposed control model is significantly reduced and the time to steady state is shorter,which can reduce the influence of the disturbance of the glass level on the glass forming quality and fire consumption.Finally,based on the CHED model integrated with multi-scale features and the improved BP& fuzzy PID control model,combined with industrial cameras,alarm devices and industrial computers,the liquid level detection and control system of melting kiln glass is developed.The results show that the system improves the efficiency and accuracy of liquid level detection and the effectiveness of liquid level control in melting furnace.