Research on Mechanical Fault Prediction Method Based on Multifeature Fusion of Vibration Sensing Data

. 2019 Dec 18;20(1):6. doi: 10.3390/s20010006

Algorithm 1: PSO-ANN algorithm.

Input: All the eigenvalues of the optimal feature combination.

Output: The best position of the particle swarm Gbest, and the best prediction accuracy.

01: Set the parameters {n,

i t e r a t i o n_{m a x}

v_{m a x}

v_{m i n}

x_{m a x}

x_{m i n}

}
02: for i = 1 to n do /* n is the number of particles */
03: Initialize

v_{i}

= (

v_{i 1}, v_{i 2}, v_{i 3}, v_{i 4}

x_{i}

= (

x_{i 1}, x_{i 2}, x_{i 3}, x_{i 4}

P b e s t_{i} = x_{i}

04: end for
05: Acquire training set

X_{t r a i n}

Y_{t r a i n}

and test set

X_{t e s t}

Y_{t e s t}

06: Set the particle with best

fitness (P b e s t_{i})

to be

Gbest

07: for k = 1

to i t e r a t i o n_{m a x}

do
08: Update

w

with Equation (3)
09: Update

c_{1}

c_{2}

with Equation (4)
10: for i = 1 to n do
11:

a n n_{i} =

ann_model(learning_rate =

x_{i 1}

, hidden_layer_ neurons =

x_{i 2}

,
12: momentum_parameter =

x_{i 3}

, rmsprop_parameter =

x_{i 4}

)
13:

a n n_{i}

.fit(

X_{t r a i n}

Y_{t r a i n}

) /* Training ANN model */
14:

l o s s_v a l u e_{i}

a n n_{i}

.loss_value
15:

p r e d i c t i o n_a c c u r a c y_{i}

a n n_{i}

.score(

X_{t e s t}

Y_{t e s t}

)
16: if (

l o s s_v a l u e_{i}

> fitness(

P b e s t_{i}

).loss_value and
17:

p r e d i c t i o n_a c c u r a c y_{i}

< fitness(

P b e s t_{i}

).prediction_accuracy) then
18:

P b e s t_{i} = x_{i}

19: end if
20: if (

l o s s_v a l u e_{i}

> fitness(

Gbest

).loss_value and
21:

p r e d i c t i o n_a c c u r a c y_{i}

< fitness(

Gbest

).prediction_accuracy) then
22:

Gbest = x_{i}

23:                end if
24:                for j = 1 to 4 do
25:

v_{i j}^{k + 1} = w v_{i j}^{k} + c_{1} r_{1} (P b e s t_{i j}^{k} - x_{i j}^{k}) + c_{2} r_{2} (G b e s t_{j}^{k} - x_{i j}^{k})

26:

x_{i j}^{k + 1} = x_{i j}^{k} + v_{i j}^{k + 1}

27: end for
28: end for
29: end for