A Fault Prediction and Cause Identification Approach in Complex Industrial Processes Based on Deep Learning

Table 3

F1-score from the testing dataset using the approach proposed in this paper.

States	Type	Maximum of F1-score	Minimum of F1-score	Average
Fault1	Step	0.91	0.78	0.84
Fault2	Step	0.85	0.71	0.78
Fault3	Step	0.67	0.53	0.6
Fault4	Step	0.98	0.95	0.97
Fault5	Step	0.77	0.62	0.71
Fault6	Step	0.99	0.95	0.97
Fault7	Step	0.99	0.96	0.98
Fault8	Random	0.89	0.78	0.83
Fault9	Random	0.7	0.51	0.6
Fault10	Random	0.98	0.92	0.96
Fault11	Random	0.991	0.975	0.982
Fault12	Random	0.86	0.78	0.8
Fault13	Slow drift	0.97	0.86	0.89
Fault14	Sticking	0.98	0.89	0.92
Fault15	Sticking	0.21	0.18	0.19
Fault16	Unknown	0.23	0.15	0.17
Fault17	Unknown	0.99	0.94	0.97
Fault18	Unknown	0.88	0.81	0.85
Fault19	Unknown	0.98	0.95	0.97
Fault20	Unknown	0.85	0.76	0.81