• Establish a three-layer mapping model from manufacturing tolerance to spatial error. • Propose a novel accuracy allocation method linked to manufacturing tolerances. • Save 20.4% cost by tolerance reduction in design balancing precision and cost. Accuracy allocation is a pivotal stage in the forward design of high-end machine tools, where the manufacturing tolerances of core components serve as the foundation determining the overall geometric accuracy. Conventional accuracy allocation methods typically only allocate to the level of geometric errors, such as the 41 geometric error components in five-axis machine tools, making it challenging to directly guide the manufacturing of machine tool components. This paper proposes a novel accuracy allocation method by tracing back to the source: the manufacturing tolerances of critical components. Firstly, utilizing the Small-Deformation Assumption and Superposition Principle (SDASP), a three-layer mapping model is established among component manufacturing tolerances (exemplified by the flatness tolerance of the base mounting surface of the guide rail), geometric errors, and machine tool spatial errors. Secondly, the concept of manufacturing tolerance sensitivity is proposed to quantify the influence of the flatness tolerances of guide rails on the spatial error of the machine tool. Finally, aiming to minimize manufacturing costs, a Genetic Algorithm (GA) is introduced to implement accuracy allocation. Simulation verification is conducted using a swivel-head and rotary-table type machine tool. The results indicate that the optimized flatness tolerance allocation scheme reduces the estimated cost by approximately 20.4% while satisfying the spatial error accuracy requirement of 0.004 mm under a 95% confidence level, validating the method's effectiveness. This study traces accuracy allocation back from the abstract geometric error level to the component manufacturing tolerance level, enabling further synergistic optimization of machine tool accuracy design and manufacturing costs.
Dai et al. (Sun,) studied this question.