A Tool for Determining Task-Specific Measurement Uncertainties In GDT Parameters Obtained from Coordinate Measuring Machines

摘要

Measurement uncertainty plays a vital role in establishing measurement traceability to national and international standards. Likewise, it is a critical factor in the arbitration of issues pertaining to product conformance. For dimensional metrology, it is necessary to associate with each GD&T parameter an estimate of its uncertainty at a specified level of confidence. As a result, for manufacturers to adhere to current standards, inspection reports must be supplemented with statements like, "The uncertainty of the diameter of this nominally 1/4-inch diameter hole is 0.0008 inches, at 95% confidence." Obtaining uncertainty statements for measurements derived from a coordinate measuring machine (CMM) is not trivial. This stringent requirement is not satisfied by merely reporting the CMM manufacturer's estimate of the single point uncertainty, or by B89 test suite results, or even by a full parametric characterization of the CMM. All potential sources of error must be considered and must be propagated in a statistically valid way to yield an uncertainty on each measurement result. In recent years, various sources of CMM uncertainty have been studied. In addition a broadly applicable statistical method for treating uncertainty sources individually or in tandem has been developed by researchers at NIST. In this paper we report on an implementation of this method for the concurrent treatment of uncertainty sources including the CMM itself, the probing system, environmental factors, sampling patterns, feature form errors, and geometric fitting algorithms. While the intrinsic versatility of the CMM has led to its dramatic success in dimensional metrology, this same versatility makes task-specific measurement uncertainty estimation problematic. The wide range of measurands and of measurement methods and conditions argue strongly against use of estimation methods commonly employed for simpler devices used in a narrow range of applications. Thus direct comparisons with calibrated artifacts or even attempts at error budgeting have been recognized as generally impractical to meet the needs of CMM users in typical industrial circumstances. The need for alternative approaches has been addressed. Among the legitimate approaches to CMM uncertainty estimation recognized by the International Standards Organization is that of modeling and simulation of the measurement process, including error sources. Using this approach, we have developed a convenient tool, PUNDIT/CMM, for the estimation of CMM-based task-specific measurement uncertainties. (PUNDIT is actually an acronym for Predicts UNcertainty in Dimensional Inspection Techniques.) PUNDIT/CMM addresses all the influential error sources cited above and does so with a modular architecture that facilitates enhancements of the error models as the state of the art advances. It incorporates a full solid model of the part to be measured as well as data structures to support tolerancing and datum reference frames and it has as a key principle of operation the statistical methodology called Simulation by Constraints (SBC) developed by researchers at NIST. SBC allows the treatment of data such as CMM or probe performance test results, which constitute a "bounding measurement set" while still not fully constraining the range of possible CMM (or probe) parametric errors. Thus users can estimate measurement uncertainties even when they have only quite limited data on their measurement systems. Only a brief review of SBC can be given here, as it applies to CMM errors. The reader is referred to for further details.