A Multichannel LSTM-CNN Method for Fault Diagnosis of Chemical Process

Figure 11

Visualization of training process using t-SNE: (a) the input data, (b) the output of the LSTM, (c) the output of the concatenate, (d) the output of the pooling, (e) the output of Conv, (f) the output of FC₁, and (g) the output of FC₂.

(a)

(b)

(c)

(d)

(e)

(f)

(g)