Research On Measurement Technology Of Blade Surface And Error Analysis

Blade as an important part of turbines, gas turbines, aero-engines and otherequipments, the influence of its surface quality and dimensional accuracy to themachine performance cannot be ignored. Efficient and accurate measurement of theblade surface data is the key technology to ensure the performance and service life ofthe machine. Research on measurement technology of blade surface and error analysis,which services to the manufacture and detection of complex and high performancesurfaces, has an important significance for the development of Chinese aerospaceand mechanical manufacturing industry.There are two main methods to obtain data on blade surfaces, contact way andnon-contact way. From the development trend of free-form surface on the detectiontechnology, the non-contact measurement system combined with multi freedomdegree takes both efficiency and accuracy into account, which will be the key researchdirection of blade parts detection. Based on the non-contact measurement system asthe carrier, the blade surface as the study object, this paper proposed three slice dataacquisition methods: local and equal length step, equal time step and feature points.The intelligent measurement system is developed and the program interface isdesigned to realize the intelligent and efficient measurement of complex surfaces.This paper presents a new method for quick measurement of blade surface dataaccording to the extracted feature points, which considered the comprehensiveadvantages of local and equal length step method with high precision and good edgedetection effect and equal time step method with high efficiency. On the premise of data compression theory and theoretical model point cloud, extraction of featurepoints is realized by the improved linear angle method of polygon approximationalgorithm. The extracted feature points distribute more in the inlet and outlet partswith large curvature and less in the blade basin and back with small curvature.Extraction error is calculated by the Hausdorff distance between the B-spline fittingcurve of discrete points and the theoretical curve.The cumulative chord length parameterization and the closest point projection isintroduced into the similarity theory to simplify the calculation of Hausdorff distancebetween discrete points and free-form curve, which provides theoretical basis forexperimental error. Characteristics of blade configuration are analyzed, blademeasurement experimental is designed and could points are transformed to obtain themeasurement model. Based on the equal step experiment, the influence of differentsampling units is studied compared with laser scan method. Results show that themeasurement errors of the three methods are higher than general non-contact scanningmethod. The feature method makes good balance between accuracy and efficiencyand the actual measurement error is in the ten micron level.The error factors is analyzed to search for the main system error, such likemachine motion error, workpiece surface color and tilt angle error as the study objects.Topological relationship of hardware system is studied according to the multi bodydynamics theory. The machine error model and the optical error model of laser probeare constructed based on Lambert rule. The tilt angle effect on energy centroidposition and receiving optical power is researched and the corresponding experimentwith standard ball is design to analyze the error distribution rule and establish theerror compensation function, which is finally verified by standard ceramic ballexperiment.