Methods and mechanisms to perform automated classifications of anomalous trace shapes

A system, comprising:

• a memory device;
• and
• a processing device, operatively coupled to the memory device, to perform operations comprising:
• obtaining current trace data associated with a substrate processing system;
• providing the trace data as input to a first predictive subsystem trained to detect anomalies using a first technique;
• responsive to detecting an anomaly in the trace data, providing the trace data as input to a second predictive subsystem trained to detect anomalies using a second technique;
• providing output data obtained from the first predictive subsystem and the second predictive subsystem to a third predictive subsystem;
• obtaining, from the third predictive subsystem, output data reflective of a trace shape associated with the anomaly;
• and
• identifying, based on the trace shape, a type of issue that caused the anomaly in the trace data.

A non-transitory computer-readable storage medium comprising instructions that, when executed by a processing device operatively coupled to a memory, performs operations comprising:

• obtaining current trace data associated with a substrate processing system;
• providing the trace data as input to a first predictive subsystem trained to detect anomalies using a first technique;
• responsive to detecting an anomaly in the trace data, providing the trace data as input to a second predictive subsystem trained to detect anomalies using a second technique;
• providing output data obtained from the first predictive subsystem and the second predictive subsystem to a third predictive subsystem;
• obtaining, from the third predictive subsystem, output data reflective of a trace shape associated with the anomaly;
• and
• identifying, based on the trace shape, a type of issue that caused the anomaly in the trace data.