Fault Prediction of Centrifugal Pump Based on Improved KNN

Table 4

Grid search cross-verifies the prediction accuracy in each case.

K value	400	800	1600	1800	2000	2200	3000	5000
Set 1	0.6183	0.6185	0.8183	0.8190	0.9187	0.8182	0.7186	0.6164
Set 2	0.7054	0.7049	0.9073	0.9072	0.9072	0.9071	0.8064	0.7058
Set 3	0.7289	0.7288	0.9289	0.9291	0.9283	0.8290	0.8287	0.7294
Set 4	0.7071	0.7097	0.9086	0.9088	0.9094	0.9095	0.8089	0.7091
Set 5	0.5534	0.5529	0.7540	0.7535	0.8537	0.7531	0.6536	0.5516
Set 6	0.6399	0.6385	0.8397	0.8398	0.8398	0.8401	0.7403	0.6399
Set 7	0.7436	0.7437	0.9442	0.9439	0.9440	0.9440	0.8439	0.7438
Set 8	0.5745	0.5755	0.7748	0.7759	0.8759	0.8754	0.6753	0.5749
Set 9	0.4991	0.5009	0.7017	0.7027	0.8025	0.7020	0.6026	0.5033
Set 10	0.5013	0.5025	0.7041	0.7038	0.9044	0.7040	0.6042	0.5066
Mean	0.6271	0.6276	0.8282	0.8283	0.8884	0.8283	0.7282	0.6278
Instability	0.0878	0.0875	0.0873	0.0872	0.0871	0.0874	0.0871	0.0868