520
P. Yang et al.
Begin
Set the parameters and initialize the population of Np nests xi(i=1,…, Np);
While (stop criterion is not met)
For each nest, perform Lévy flight to generate new solutions with Eq.(1);
Evaluate its fitness;
Abandon a fraction of solutions with a probability p0 and build new one with Eq.(7);
Evaluate its fitness;
Rank the solutions and keep the bestnest;
End
Output the best solution;
End
Fig. 45.1 The pseudocode of cuckoo search (CS)
of the individual cuckoo is determined by CS. In the process of optimizing the BP
neural network by the cuckoo search algorithm, each cuckoo individual represents a
combination of weights and thresholds, in other words, a neural network construction
method, and the merits of each individual are determined by the fitness function.
The cuckoo search algorithm finds the optimal fitness value corresponding to the
individual from performing the CS optimization process, and obtains the optimal
weight and threshold combination to construct the CSBP neural network. The flow
chart of CSBP is shown in Fig. 45.2.
45.3 Experiment
45.3.1 Experiment Platform
This experiment uses QPZZ-II type rotating machinery fault simulation platform.
The experimental platform is shown in Fig. 45.3. The experimental platform can
simulate various fault states of rotating machinery and provide the basic conditions
for comparative analysis and diagnosis of various states. As Fig. 45.3 shows, (a) is
the experimental platform, (b) is a normal bearing, (c) is the control platform.
45.3.2 Experiment Setup
In this experiment, the acoustic emission detection system adopts the US PCI-2
acoustic emission acquisition system, in which the gain of the pre-amplifier is 40 dB,
P. Yang et al.
Begin
Set the parameters and initialize the population of Np nests xi(i=1,…, Np);
While (stop criterion is not met)
For each nest, perform Lévy flight to generate new solutions with Eq.(1);
Evaluate its fitness;
Abandon a fraction of solutions with a probability p0 and build new one with Eq.(7);
Evaluate its fitness;
Rank the solutions and keep the bestnest;
End
Output the best solution;
End
Fig. 45.1 The pseudocode of cuckoo search (CS)
of the individual cuckoo is determined by CS. In the process of optimizing the BP
neural network by the cuckoo search algorithm, each cuckoo individual represents a
combination of weights and thresholds, in other words, a neural network construction
method, and the merits of each individual are determined by the fitness function.
The cuckoo search algorithm finds the optimal fitness value corresponding to the
individual from performing the CS optimization process, and obtains the optimal
weight and threshold combination to construct the CSBP neural network. The flow
chart of CSBP is shown in Fig. 45.2.
45.3 Experiment
45.3.1 Experiment Platform
This experiment uses QPZZ-II type rotating machinery fault simulation platform.
The experimental platform is shown in Fig. 45.3. The experimental platform can
simulate various fault states of rotating machinery and provide the basic conditions
for comparative analysis and diagnosis of various states. As Fig. 45.3 shows, (a) is
the experimental platform, (b) is a normal bearing, (c) is the control platform.
45.3.2 Experiment Setup
In this experiment, the acoustic emission detection system adopts the US PCI-2
acoustic emission acquisition system, in which the gain of the pre-amplifier is 40 dB,