With the increasing penetration of renewable energy, the power quality disturbance (PQD) problem faced by the power grid has become more complicated. The traditional classification method based on one-dimensional PQD signals makes it difficult to extract and identify periodic and trend disturbances at the same time. To address this problem, this paper proposes a PQD classification method based on time-series two-dimensional transformation and multi-scale Transformer. Firstly, the time-series two-dimensional transformation is used to convert the one-dimensional PQD time series into a set of two-dimensional tensors based on multiple periods, to deeply mine the characteristic information contained in the PQD signals in the two-dimensional space. Then, the multi-scale feature map of the PQD signal is extracted through the multi-scale Transformer encoder module. And the multi-scale Transformer decoder module is used to splice and fuse the multi-scale feature maps, for effectively merging the feature maps extracted at different scales. Finally, the PQD classification task is accomplished through a fully connected layer and a Softmax classifier. To verify the effectiveness of the proposed method, a dataset containing 24 kinds of PQD is established to test the model. The results indicate that the proposed method has a high classification accuracy and noise robustness for PQD signals.