5.3 Physical Anti-Collision Based on Wavelet
193
Repeat the above steps in Section 5.3.4. Each detected tag is matched
with the template, respectively, and the relative distance between the tag
and the template in the vertical direction is calculated. Then, we obtain
the vertical coordinate of each tag.
5.3.3 RFID Multi-tag Wavelet Neural Network
(1) Construction of wavelet neural network model
A large number of test data show that there is a complex nonlinear relationship
between the 3D coordinate distribution of RFID multi-tag network and the corresponding reading distance. Some scholars have used GA-BP to model the multi-tag
network [33]. In this paper, a new RFID multi-tag wavelet neural network model is
proposed. The wavelet neural network model does not need to construct a mathematical function-based model, and it can approximate any nonlinear function. Therefore, the wavelet neural network is used to model the complex nonlinear relationship between the 3D coordinate distribution of RFID multi-tag network and the
corresponding reading distance. The wavelet neural network is an improved neural
network. The wavelet basis function is used as the transfer function of the hidden
node. The error of the signal is transmitted back at the same time as the forward
propagation.
Assuming X 1 , X 2 , . . . , X k are the input parameters of wavelet neural networks.
Y 1 , Y 2 , . . . , Y m are predictive outputs of wavelet neural networks. Ω ij and ω jk
are wavelet neural network weights. When the input signal sequence is x i (i =
1, 2, . . . , k), the output of hidden layer is shown in Eq. (5.47).
h( j) = h j
⎡
⎢
⎢
⎢
⎣
k
i=1
ω i j − b j
a j
⎤
⎥
⎥
⎥
⎦
j = 1, 2, 3, · · · , l
(5.47)
In Eq. (5.47), h( j) is output value for the j th node of the hidden layer. ω i j is the
connection weight between the input layer and the hidden layer. H j is the wavelet
basis function. B j is translation factor of wavelet basis function h j . a j is stretching
factor of wavelet basis function h j .
In this paper, we use the Morlet mother wavelet basis function as the activation
function. The equation of wavelet basis function is shown in Eq. (5.48).
y = cos(1.75x)e
−x
2 /2
(5.48)
The output layer of wavelet neural network is calculated in Eq. (5.49).
193
Repeat the above steps in Section 5.3.4. Each detected tag is matched
with the template, respectively, and the relative distance between the tag
and the template in the vertical direction is calculated. Then, we obtain
the vertical coordinate of each tag.
5.3.3 RFID Multi-tag Wavelet Neural Network
(1) Construction of wavelet neural network model
A large number of test data show that there is a complex nonlinear relationship
between the 3D coordinate distribution of RFID multi-tag network and the corresponding reading distance. Some scholars have used GA-BP to model the multi-tag
network [33]. In this paper, a new RFID multi-tag wavelet neural network model is
proposed. The wavelet neural network model does not need to construct a mathematical function-based model, and it can approximate any nonlinear function. Therefore, the wavelet neural network is used to model the complex nonlinear relationship between the 3D coordinate distribution of RFID multi-tag network and the
corresponding reading distance. The wavelet neural network is an improved neural
network. The wavelet basis function is used as the transfer function of the hidden
node. The error of the signal is transmitted back at the same time as the forward
propagation.
Assuming X 1 , X 2 , . . . , X k are the input parameters of wavelet neural networks.
Y 1 , Y 2 , . . . , Y m are predictive outputs of wavelet neural networks. Ω ij and ω jk
are wavelet neural network weights. When the input signal sequence is x i (i =
1, 2, . . . , k), the output of hidden layer is shown in Eq. (5.47).
h( j) = h j
⎡
⎢
⎢
⎢
⎣
k
i=1
ω i j − b j
a j
⎤
⎥
⎥
⎥
⎦
j = 1, 2, 3, · · · , l
(5.47)
In Eq. (5.47), h( j) is output value for the j th node of the hidden layer. ω i j is the
connection weight between the input layer and the hidden layer. H j is the wavelet
basis function. B j is translation factor of wavelet basis function h j . a j is stretching
factor of wavelet basis function h j .
In this paper, we use the Morlet mother wavelet basis function as the activation
function. The equation of wavelet basis function is shown in Eq. (5.48).
y = cos(1.75x)e
−x
2 /2
(5.48)
The output layer of wavelet neural network is calculated in Eq. (5.49).