† Corresponding author. E-mail:
Extracting characteristic signal from a continuous signal can effectively reduce the difficulty of analyzing the running states of a single-variable nonlinear system. Whether the extracted characteristic signal can accurately reflect the running states of the system is very important. In this paper, a method called automatic sampling method (ASM) for extracting characteristic signals is investigated. The complete definition is described, the effectiveness is proved theoretically, and the general formulas of the extracted characteristic signals are derived for the first time. Furthermore, typical Chuaʼs circuit is used to accomplish a lot of experimental research on the aspect of frequency domain. The experimental results show that ASM is feasible and practical, and can automatically generate a characteristic signal with the change of the original signal.
By monitoring the output signal of a nonlinear system, the change of the system’s running state can be directly identified. In order to analyze or apply the dynamic characteristics of a system, the traditional analytical methods have been applied to many research fields. For instance, the Lyapunov exponent is used to detect weak signals[1] and prove the intra-layer synchronization of the duplex network;[2] the numerical simulation method is used to study the magnetoelastic dynamic behavior of a rotating annular thin[3] and the Hopf bifurcation control of a modified Pan-like chaotic system;[4] the bifurcation is used to study the nonlinear dynamic behavior of the PMSG[5] and external-cavity multi-quantum-well laser;[6] the Poincaré section is used to analyze the limit cycle oscillation flutter and chaotic motion of a two degree-of-freedom aeroelastic airfoil[7] and the chaos properties of the time-dependent driven Dicke model.[8] All these methods may help people predict the behavior of these systems or avoid their negative effects. However, for a single-variable nonlinear system, the traditional methods are difficult to monitor its running states. In recent years, many methods for analyzing nonlinear time series have been proposed, which can be used to analyze a single-variable nonlinear system. For instance, reference [9] used the recurrence plot (RP) and recurrence quantification analysis (RQA) method of the volume data of the lane closest to the median to analyze nonlinear time series on traffic data for incident detection; reference [10] proposed a new algorithm which can automatically select an optimum Poincaré plane for transferring the maximum information from EEG time series to a set of Poincaré samples. Although these methods can effectively detect the dynamic characteristics of the single-variable nonlinear systems, they require a lot of numerical calculations to obtain results, which makes them unable to monitor the system in real time.
In 2013 and 2015, the 45-degree line and SP methods were respectively proposed in Refs. [11] and [12], which can monitor the running states of a nonlinear system in the screen of an oscilloscope by a special way. Extracting the useful characteristic signals from a continuous signal is the key to realize these methods. However, these methods neither give a complete definition of the extraction method nor demonstrate and validate it. These shortcomings affect their further application.
Therefore, a method which can automatically obtain characteristic signal according to the complexity of the tested signal is called automatic sampling method (ASM) in this paper. It is a fast and efficient sampling method.
The structure of this paper is as followed. In Section 2, the complete definition of ASM is given, and the extracted characteristic signals still retaining the frequency characteristics of the tested signal are analyzed and proved. Then the general formulas of characteristic signals which are extracted from periodic signals and non-periodic signals are derived. In Section 3, based on the ASM, the key issue on the choice of marking time points for extracting characteristic signal is studied, and its selection method is given. In Section 4, according to the above method, a hardware circuit system for extracting characteristic signal is designed, which is used to do the verification experiments of the extracted characteristic signals from the frequency domain perspective. Finally, the conclusions are given in Section 4.
Characteristic signal is a kind of signal which can reflect the characteristic information of the tested signal. In order to ensure the applicability of the characteristic signal, it is necessary to study how to obtain different characteristic signals according to the different tested signals. Therefore, a definite method of extracting characteristic signal is given below.
Automatic sampling method (ASM): for a bounded continuous and differentiable signal f(t), a horizontal straight line is selected within its boundary to intersect f(t) for
It should be noted that for signal f(t), if M is the set of its maximum; N is the set of its minimum. Then any
Combining with Fig.
(a) For a continuous period 1 signal
In order to determine t
i
Since the length of section two is too long, the method for determining C and t i will be given in the next section.
According to ASM, a
Then the characteristic values can be extracted at
It can be seen from ASM and Fig.
Therefore, the frequency of the characteristic signal
(b) For a continuous period 2 signal
Because
Similarly, a constant C is selected to determine t
i
Then, a
Since
(c) For a continuous period k signal
Similarly, t i which is determined by a constant C satisfies
In a similar way, the characteristic signal
Similarly, the characteristic signal extracted from
According to ASM and Fig.
As shown in Fig.
According to the above analysis and theoretical proof, for periodic signals with T, the general formula of the characteristic signal is
For non-periodic signals, the general formula of the characteristic signal is
It can be seen from the definition of ASM (Eqs. (
As can be seen from the above analysis and proof of ASM, the choice of marking time points t
i
The f(t) in Fig.
According to ASM and the principle of choosing t
i, the circuit system for extracting characteristic signal has been designed in this paper. Its block diagram is shown in Fig.
Time marker module is designed to determine the marking time points t
i
The typical Chua’s system in Ref.[15] can generate a variety of periodic states and chaotic states by adjusting its internal parameters, such as period 1, period 2, period 3, period 4, …, single scroll chaos, double scroll chaos, and so on. Therefore, the output y of Chua’s system is chosen as the tested signal. The states of Chua’s system can be changed by adjusting its internal linear resistance. The digital oscilloscope is used to do the following experiments.
When Chua’s system turns into the state of period 1 by adjusting the resistance, the timing diagram of the tested signal is shown in Fig.
Adjusting R to make Chua’s system in the state of period 2, the timing diagram and frequency spectrum diagram of the tested signal are shown in Figs.
When Chua’s system is in the state of period 3, the timing diagram and frequency spectrum diagram of the tested signal are shown in Fig.
Keep on adjusting R, when Chua’s system turns into the state of period 4, it is difficult to identify the periodic state of the system by the timing diagram shown in Fig.
When Chua’s system is in single scroll chaos state, it is hard to identify the running state of the system by the timing diagram shown in Fig.
Adjust R to make Chua’s system turn into the double scroll chaos state. The running state of the system cannot be identified by the timing diagram shown in Fig.
As shown in Figs.
Now, the characteristic signals extracted by the designed system are applied to 45-degree line in Ref.[11] and SP in Ref.[12], then get Figs.
ASM is studied from theory and experiment, and it is confirmed that the characteristic signal extracted by this method can retain the frequency information of the tested signal. In order to clarify ASM, the general formulas of characteristic signals extracted by ASM are theoretically derived for the first time. In addition, ASM has the following features:
(i) ASM can automatically extract characteristic signal without an additional control signal.
(ii) It can be seen from Eqs. (
(iii) ASM not only can be applied for analyzing the single-variable nonlinear systems, but also can be applied to analyze the multivariable-nonlinear systems.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] |