Abstract:  The surface roughness and surface dimensional accuracy of blades had important influence on the overall performance and service life of the aero-engines. In order to solve the influences of the changes of normal contact forces on the surface roughness and the uniformity of material removal depth on the cut-in or cut-out stages of robotic abrasive belt grinding and the changes of normal contact forces on the parts with large curvature changes， orthogonal center combination design （CCD） test was designed and width learning algorithm was used to establish the prediction model of grinding process parameters， surface roughness and material removal depth. The prediction model of normal contact force and surface roughness and material removal depth measured by sensors was used to solve the adaptive processing parameters by combining multiple learning backtracking search algorithm. Finally， grinding tests were carried out with the obtained processing parameters. The maximum errors between the test values of surface roughness and material removal depth and that of the model predictions are as 14.2% and 13.4% respectively. The proposed method may ensure good consistency between surface roughness and material removal depth. 

Key words:  , robot, abrasive belt grinding, normal contact force, processing parameter, adaptive technology

摘要： 叶片表面粗糙度与型面尺寸精度对航空发动机整体性能和使用寿命有重要影响，为了解决机器人砂带磨削的切入切出阶段及曲率变化较大部位的法向接触力变化对表面粗糙度和材料去除深度一致性的影响，设计正交中心组合试验并用宽度学习算法建立了磨削工艺参数与表面粗糙度和材料去除深度的预测模型，利用传感器测量的法向接触力与建立的预测模型，结合多重学习回溯搜索算法求解了自适应工艺参数，最后用得到的工艺参数进行了磨削试验，表面粗糙度和材料去除深度的试验值与模型预测值最大误差为14.2%和13.4%，表明所提方法可以保证表面粗糙度和材料去除深度的一致性良好。

关键词: 机器人, 砂带磨削, 法向接触力, 工艺参数, 自适应技术