The machine tool spindle fault diagnosis and inference method was proposed based on data-driven and ontology modeling to address the problems of single method and low level of intelligence in current CNC machine tool spindle system fault diagnosis.EMD was used to process and analyze the raw signals containing fault features collected by sensors,the original statistical features were extracted,and based on this,a DBN-RF diagnostic model was constructed to achieve deep feature adaptive mining and fault pattern recognition.The Protégé5.1 tool was used and combined with domain knowledge to construct a machine tool spindle fault ontology knowledge base,the fault identification results of the DBN-RF diagnostic model were semantically mapped with instances in the ontology knowledge base to achieve fault knowledge inference,the fault causes and fault resolution strategies were obtained.The effectiveness of the DBN-RF diagnostic model was validated based on actual collected bearing fault data under different working conditions,with the highest average fault diagnosis accuracy reaching 92.93%.The reusability and inference function of ontology knowledge base was verified through the construction of an instance.Finally,a CNC machine tool spindle health management service system was designed and developed to achieve real-time perception of spindle system status and fault diagnosis and inference.