Investigation of Data Augmentation Techniques for Disordered Speech Recognition