Research On The Key Technique Of The CMM Spatial Dynamic Error Correcting Method

Higher speed and accuracy are required to CMM with the development of modern manufacturing of CMM. The measurement acceleration and structure inertia force of CMM are increased due to the increasing measurement velocity, which cause aggravated vibration and the declined measurement accuracy of the CMM. There exists a contradiction between high measuring efficiency and high precision of CMM. To resolve the contradiction, dynamic error sources of CMM were analyzed in depth, the dynamic error separation and experimental devices of CMM was developed. The dynamic errors were separated, which were used to build the correcting model by using the effective mathematical modeling.The various components structures of CMM were studied in detail and the main dynamic errors sources and their relations were analyzed in depth. The influence to the dynamic error of CMM air guide was studied chiefly. The work principle of air guide and errors transmitting paths were studied. The dynamic error separating experiment systems and measuring acceleration testing system were developed. The spectrum analysis was applied to analyzing the experimental data, which made the conclusion that the air guide has an important impact on dynamic errors of the CMM and it was one of the main dynamic error sources of the CMM.In this paper, PLS was applied to analyze the separated dynamic errors data of CMM, which determined the main and subordinate influencing factors of dynamic error. The X, Y and Z coordinate were the main influencing factor and the measurement velocity was the subordinate influencing factor. So the needed modeling dimension of CMM dynamic errors model was reduced due to the conclusion, which can help to build the high precision error correcting model that fit for the actual environment by using the effective modern mathematical methods.Based on the determined main effective dynamic error influencing factors, the error separation experiment system was built, which was used to separate the dynamic error on the appropriate selected spatial measuring point. The principles of the BP neural networks, support vector machines and wavelet neural network were introduced. The software of BP neural networks, support vector machine and wavelet neural network were programmed. The spatial dynamic error correction models were built by using the separated dynamic error data on the different measuring point and these modeling methods. These models were verified by the dynamic errors which were not used to build model and the modeling accuracy of these modeling methods were obtained. Support vector machine has the highest accuracy which reaches 0.02um . The modeling accuracy of Wavelet neural network reaches 0.05um while the modeling accuracy of BP neural network reaches 1um .