<p>Wheel shaping plays a vital role in grinding process. Nevertheless, accurately determining the shaping states of the grinding wheel during practical operations remains a significant challenge, hindering the achievement of an optimal balance between quality and efficiency in grinding processes. To address the difficulty of online assessment of the shaping states for diamond grinding wheels with super-hard abrasives, a novel monitoring method using acoustic emission (AE) is proposed. AE signals are decomposed into multiple intrinsic modal functions (IMFs) using the empirical modal decomposition (EMD) method. Representative signals are identified using correlation coefficients combined with time–frequency analysis. Common statistical features, such as variance, root mean square and cragginess of the IMFs, are extracted. Random Forest (RF) is employed to identify the features most relevant to the shaping states of the wheel. A classification model of the wheel shaping states is constructed by optimising the Support Vector Machine (SVM) parameters using Grid Search (GS). Shaping tests conducted on a diamond grinding wheel dressed with green silicon carbide oilstone achieved a combined accuracy of 98.4%, demonstrating the effectiveness of the proposed monitoring method in identifying wheel shaping states.</p>