Optimized Design Of Surface Topography Measurement System For Reflective Fiber Optic Sensor

With the rapid development of precision instrument manufacturing industry,the demand for measuring the surface morphology of products is increasing in mechanical manufacturing enterprises.Effectively evaluating the surface morphology has important practical significance for product processing quality and usability.Traditional contact measurement methods directly contact the surface of objects,which can easily damage the original texture of the surface.Reflective fiber optic sensors have the characteristics of simple structure,low cost,and easy design,and are widely used in surface morphology measurement systems.This thesis designs a surface morphology measurement system based on a reflective intensity-modulated fiber optic sensor,and conducts in-depth research on three aspects:analyzing the theoretical basis of surface morphology measurement of reflective fiber optic sensors,designing the hardware and software aspects of fiber optic sensors’ light intensity compensation methods,and constructing a reflective fiber optic sensor measurement system.Firstly,the feasibility of one-dimensional parameter measurement and three-dimensional surface reconstruction of reflective fiber optic sensors is analyzed.Based on the intensity modulation characteristic curve of fiber optic sensors,combined with the principle of light scattering,the basic process of reconstructing three-dimensional surface morphology is analyzed through displacement measurement.Secondly,a light intensity compensation method is proposed for the nonlinear impact of light source fluctuations on measurement results,in both hardware and software aspects.Based on the design of the probe structure as a dual-path receiving type,neural networks are used to process the received optical power of the inner and outer circles of the fiber optic sensor to improve the accuracy of the measurement system.Finally,the design of the hardware module and Lab VIEW software module is completed to build a reflective fiber optic sensor surface morphology measurement system.Through the measurement experiment on the surface of the sample block,the linear interval of the displacement measured by the reflective fiber optic sensor of this platform is0.3mm-0.7mm.By comparing the intensity modulation characteristic curves of standard blocks with different roughness under the same processing method,it is found that the reflective fiber optic sensor has the ability to distinguish changes in surface roughness.In order to solve the problem of traditional BP neural network easily falling into local extremes,the artificial bee colony algorithm and the sparrow search algorithm are used to optimize the BP neural network separately.By compensating for the received optical power of the inner and outer circles,the average absolute error of the sparrow search algorithm optimized BP neural network model is0.002,and the root mean square error is 0.003.All parameter indicators are smaller than those of support vector machines,traditional BP neural networks,artificial bee colony algorithm optimized BP neural networks,and particle swarm algorithm optimized BP neural networks.The results show that the accuracy of the measurement system is improved by optimizing the light intensity compensation part,and the surface morphology of the workpiece can be restored with micron accuracy.