基于主动测量模式的磨削加工几何精度融合预测控制理论与方法

Currently, there are some urgent problems need to be solved in the grinding system based on active measuring. The main problems are as follows: the reasonable grinding parameters selecting is quite arbitrary; dimension errors are compensated by means of manual method; the comprehensive control method for geometric errors is still lacking; and the uniformity of products accuracy is poor. To solve these problems, this project has proposed a novel prediction and control method for geometric accuracy based on active measuring. Firstly, on the study of active measuring grinding principle, optimization design method for grinding parameters based on weighted gray target decision is presented. Then, based on statistical learning theory, prediction model for geometric accuracy based on gray relational online support vector regression machine is built, and control system is also given. For real-time and accurate training of the forecasting models, active incremental learning mechanism, a rapid algorithm and parameters optimization method for online support vector regression machine are proposed. After that, verification operators of typical form errors base on geometrical product specification and implementation strategies are built, and a fusion control method for dimension and form errors is proposed, and then total control for geometric accuracy will be realized. Finally, typical grinding system with the new measuring and control method will be developed. This research result can provide the theoretical basis and technical support for improving geometric accuracy, increasing stability and intelligent of grinding system. In addition, this research result can also facilitate the integration of mechanical processing and the new generation of the geometrical product specification system.

针对现有主动测量磨削加工技术中急需解决的工艺参数调整随意性大、加工误差依赖人为补调、在线尺寸与形位误差综合控制手段欠缺、产品加工精度一致性较差等关键问题，提出一种新的基于主动测量模式的磨削加工几何精度预测与控制方法。以主动测量磨削加工机理为基础，提出基于加权灰靶决策的工艺参数智能优化设计方法；以统计学习理论为基础，构建基于灰色关联在线支持向量回归机的几何精度预测模型及智能化控制策略；针对模型训练学习的实时性及准确性要求，提出在线支持向量回归机的主动增量学习机制、快速算法及参数优化方法；建立符合新一代产品几何规范的典型形位误差检验操作算子及实施策略，进而提出尺寸与形位精度融合预测控制方法，实现两者的在线全面智能化控制；开发基于新测控模式的典型磨削加工系统。旨在为有效提高磨削加工几何精度、保证系统稳定性及改善智能化程度提供理论基础和技术支持，并促进机械制造与新一代产品几何规范体系的融合。

本项目提出了基于主动测量及统计学习理论的尺寸与形位精度融合预测控制方法，实现了磨削加工过程的智能化测量和控制。为解决磨削加工工艺参数的工艺参数调整随意性大的问题，本项目提出了基于加权灰靶决策的工艺参数智能优化设计方法，实现了工艺参数的最优设置。基于GPS规范及优化理论，分别提出了基于运动几何变换法、孪生支持向量机的圆度、圆柱度评定新方法。针对具有非光滑断续表面特征的零件磨削加工，提出了断续表面的在线尺寸精确预测测量方法。考虑到磨削加工几何误差的预测与补偿，本项目将灰色关联与在线支持向量机相结合，实现了不同磨削阶段的几何精度预测模型建立及几何误差融合控制策略研究，同时，完成了线支持向量回归机的参数优化方法及增量学习机制。开发完成了一套完整的集成新测控模式的典型磨削加工系统，由关联决策对评价指标分析确定试验工艺参数最优组合，并通过大量实验验证了在精密磨削加工中几何误差在线测量与融合控制的可行性。本项目的研究成果对积极推升磨削加工的精度、加工效率及智能化程度，具有重要的学术价值和工程应用意义。