MILLING ROBOT MULTI-MODAL FREQUENCY RESPONSE PREDICTION METHOD BASED ON SMALL-SAMPLE TRANSFER LEARNING

摘要

Disclosed is a milling robot multi-modal frequency response prediction method based on small-sample transfer learning. The method comprises the following steps: selecting several points and a tool nose point on a body of a milling robot at any two postures, so as to perform a hammering test, and also performing a hammering test on several points on the body at a target posture, so as to obtain transfer source data and transfer target data; constructing three-order complex tensors of a robot frequency response characteristic transfer source domain and target domain, and performing multi-order modal parameter identification on a multi-modal frequency response of the tool nose point on the basis of a least-squares complex exponential method, so as to construct a label of data in the transfer source domain; on the basis of an input tensor and an output vector of the transfer source domain, generating a virtual sample by means of an information expansion function based on triangular membership and a multi-objective grey wolf optimization algorithm; respectively extracting data features from a frequency domain, a time domain and a time-frequency domain, and on this basis, performing feature augmentation on the complex tensors in the source domain and the target domain; performing dimensionality reduction on the complex tensors in the source domain and the target domain by means of a naive tensor sub-space learning method; and constructing a complex kernel extreme learning machine based on a conjugate augmented input, so as to predict the multi-modal frequency response of the tool nose point at the target posture.