J. Cent. South Univ. (2016) 23: 2284-2295

DOI: 10.1007/s11771-016-3286-7

Fault diagnosis and fault-tolerant control of photovoltaic micro-inverter

LI Zhou(ÀîÖÛ), PENG Tao(ÅíÌÎ), ZHANG Peng-fei(ÕÅÅô·É), HAN Hua(º«»ª), YANG Jian(Ñ)

School of Information Science and Engineering, Central South University, Changsha 410083, China

Central South University Press and Springer-Verlag Berlin Heidelberg 2016

Abstract: An observer-based fault diagnosis method and a fault tolerant control for open-switch fault and current sensor fault are proposed for interleaved flyback converters of a micro-inverter system. First, based on the topology of a grid-connected micro-inverter, a mathematical model of the flyback converters is established. Second, a state observer is applied to estimate the currents online and generate corresponding residuals. The fault is diagnosed by comparing the residuals with the thresholds. Finally, a fault-tolerant control that consists of a fault-tolerant topology for the faulty switch and a simple software redundancy control for the faulty current sensor, is proposed to achieve a fault-tolerant operation. The feasibility and effectiveness of the proposed method has been verified by simulation and experimental results.

Key words: fault diagnosis; fault-tolerant control; state observer; photovoltaic micro-inverter

1 Introduction

In recent years, in pursuit of photovoltaic power generation cost reductions, system reliability enhancements, efficiency improvements and other targets, including a new type of inverter structure and micro- inverters have been introduced. Compared with conventional large PV systems, micro-inverters have a higher energy harvest, improved system-enlargement, ¡°plug and play¡± operation, and so on [1-2]. Currently, a significant amount of research has been performed on the efficiency [3-4] and power decoupling [5-6] techniques for micro-inverters. However, little research has been performed on online monitoring, fault diagnosis and reliable operation of the micro-inverter systems. Micro- inverters always work in bad environments, such as high temperatures, hot and humid conditions. A micro-inverter is usually attached to a single PV panel, so it must have a lifespan that matches the PV panel¡¯s life span, that is, 25 years [7-8]. Therefore, issues, including real-time monitoring of the operation state of the micro-inverter system, fault diagnosis and timely handling of a fault, are gradually brought into sharp focus by the demands for high reliability and long lifespan of micro-inverters.

Interleaved flyback converters are the core parts of the micro-inverter system and are responsible for most of the system functionalities, such as booster, isolation, and component maximum power point tracking (MPPT). The power switch of interleaved flyback converters is in a long-term state of high frequency modulation, bearing high voltage and large current, relatively serious heating, with large switching losses [9]. These conditions combined with uninterrupted operation for a long time may cause a series of problems, such as component ageing and parameter drift, which would eventually lead to the performance deterioration, failure rate increase and reliability degradation of the power switch [10-11]. Research shows that most of the power electronic device failures are caused by open and direct faults of power devices in the main circuit. Usually, a direct fault has a very short duration, which would quickly burn out the power device and is finally characterized by an open circuit fault [12]. When an open circuit fault occurs in one power switch of the two lines of the flyback converters, the other normal converter has to bear all of the power that originally was equally distributed in the two lines of the flyback converters. Thus, the current through the power switch would be larger in the normally operating flyback converter and the high-frequency transformer would be overloaded, so the speed of damage in a normal switch is greatly increased and the occurrence of a secondary fault is easily triggered. In addition, the two current sensors that used to detect the primary currents flowing through inductors, are magnetic elements made of ferrite which is prone to failure because it has small volume and easily broken feature. If one of the sensors malfunctions, the two lines of primary currents become unbalanced and it could directly affect the averaged control of the whole system.In particular, if the malfunction of power switch or current sensor is not handled in time, it may lead to the magnetic saturation of the high-frequency transformer after running with a fault for a long time, which would eventually cause fatal damage to the power switch and seriously affect the system performance and device service life [13-14].

Fault detection approaches can be divided into three approaches: the analytical model-based approach, the knowledge-based approach, and the signal processing- based approach [15]. The knowledge-based approach requires substantial time to build a fault diagnosis expert system. The signal processing-based approach can only make a rough judgment of the fault range; in most cases, it is difficult to locate the fault directly. The analytical model-based approach is the most direct and effective approach for obtaining an accurate mathematical model and is also the most mature approach. In general, the analytical model-based approach can be classified into three approaches: an observer-based approach, equivalent space approach and parameter estimation approach. Observer-based fault detection is to estimate the output of the system from the measurement by using an observer in a deterministic setting. Then, the output estimation error is used as a residual and is compared with a predefined threshold. If the residual is greater than the threshold, there is a failure [16]. This approach can provide real time monitoring of the working state of the system to detect faults in a timely manner. It possesses the advantages of simple implementation and strong commonality. Recently the observer-based approach for fault diagnosis has been widely used in all types of power electronic systems. In Ref. [17], a sliding mode observer is used to estimate the converter arm currents and the cell capacitor voltages in modular multilevel converters for fault detection of the IGBTs. In Ref. [18], the adaptive observer is used to estimate the rotor resistance and phase current in an induction motor system for the fault diagnosis of the speed sensor, dc side voltage sensor and current sensor. In Ref. [19], the Luenberger observer is applied to estimate the DC-link voltage, which is used for fault diagnosis of a voltage sensor. However, this approach to detection and diagnose the faults of power switch and current sensor in the interleaved flyback converters of a grid-connected micro-inverter has not been reported to date.

In a system with high reliability requirements, even if a fault occurs in some part of the system, it is very important that the operation of the system is uninterrupted and the safety of the operating personnel is absolutely ensured [20]. Hence, it is necessary to add a fault-tolerant capability for the interleaved flyback converters to ensure the high reliability requirement of the micro-inverter system. Currently, many kinds of fault-tolerant control have been widely applied to various power electronic systems [21-23]. But there has not been enough attention on the micro-inverter system.

This work proposes a diagnostic method and fault- tolerant scheme for a fault of power switch and current sensor in the interleaved flyback converters of a micro- inverter system. At first, the nonlinearity of the flyback converter model is linearized by using small signal analysis, which is presented in Section 2. Then, in Section 3, a state observer is applied to estimate the two lines of currents flowing through the inductors, which is known to give a good performance with a fast response and high reliability. The fault of a converter is diagnosed by comparing the thresholds with two residuals. The proposed fault-tolerant control which consists of a fault- tolerant topology for the faulty power switch and a simple software redundancy control for the faulty current sensor, is described in Section 4. The simulation and experimental results in Section 5 are used to show the validity and feasibility of the proposed method.

2 Modeling of interleaved flyback converters

Figure 1 shows the structure of the interleaved flyback micro-inverter. It contains three parts: the interleaved flyback converters, full bridge inverter and output filter. The interleaved flyback converters have two lines of flyback converters with the same structure and parameters, which are used to convert the DC power from the solar panel to a half-sine wave, which has a double-power frequency. The full bridge inverter switches the half-sine wave into an AC signal that has a power frequency. The filter is used to suppress the EMI/EMC noise and provide impedance between the inverter output and grid. CS1 and CS2 are used to detect the currents flowing through the primary inductors. S_pv1 and S_pv2 are the power switches of the interleaved flyback converters. Under normal operation, the S_pv1 and S_pv2 switches are complementary.

Assume that the grid voltage u_ac is a half-wave, rectified voltage with the same the root-mean-square(RMS) value as the AC grid. The simplified interleaved flyback converters connected to an equivalent grid without considering the process of full bridge inverter in Fig. 1 are shown in Fig. 2.

Fig. 1 Structure of interleaved flyback micro-inverter

Fig. 2 Model of interleaved flyback converters

According to Fig.2, the interleaved flyback converters have four states, corresponding to four energy storage elements L_m1, L_m2, L_f and C_f. Selecting i_Lm1, i_Lm2, i_ac, u_f to state variables, which are the primary currents flowing through inductor L_m1 and L_m2, the secondary side current flowing through filtering inductor L_f, the secondary side filter capacitor C_f voltage, respectively. According to Kirchhoff¡¯s voltage law and current law (KVL and KCL), the differential equations of the converters over one-switching cycle are formulated as follows:

  (1)

where u_pv is the dc voltage of the photovoltaic panels; d₁ and d₂ are the duty ratios of switch S_pv1 and S_pv2;R_p is the high-frequency transformer internal resistance, L_m1 and L_m2 are the inductances of the two high-frequency transformers; R_f, L_f and C_f are the output filter resistance, inductance and capacitance, respectively; R_on is the resistance of S_pv1 and S_pv2; and N is the high-frequency transformer turn ratio of the secondary side to the primary side.It is easy to see that Eq. (1) is a fourth order nonliear state equation.

Assume that the converters work in the quiescent point Q. In this steady condition, the duty ratios of the point are D₁ and D₂, the dc input voltage is V_pv, the output voltage is V_ac, and the primary magnetic inductance currents, second side filter inductance current and filter capacitor voltage are I_Lm1, I_Lm2, I_ac and V_f, respectively. During quiescent operation, the following steady-state conditions are satisfied:

       (2)

where

Draw a tiny perturbation at the quiescent point Q of the state variables and input variables, namely,

                          (3)

By substituting Eq. (3) into Eq. (1), we get

 (4)

Using steady-state conditions Eq. (2) and supposing that the second order is approximately zero, Eq. (4) simplifies to

 (5)

where R₁=D₁(R_on+R_p), and R₂=D₂(R_on+ R_p).

By rearranging Eq. (5), the linearized small signal model of the single flyback converter is described as:

                               (6)

where and are the state variables, input control parameters and output measured parameters, respectively; are state matrix, control matrix, output matrix. In addition to SUB>SUB>SUB>SUB>and

3 Fault diagnostic method

According to Eq. (6), a state observer to estimate the current of the converters is defined as

                    (7)

where ¡°£Þ¡± indicates the estimated value; H is the observer gain matrix.

Figure 3 shows the block diagram of the state observer.

Fig. 3 Block diagram of state observer

Using the observer to obtain estimation the state estimation error can be calculated as and

                          (8)

By subtracting Eq. (7) from Eq. (6), we get:

                       (9)

Combining Eqs. (8) and (9), we get:

                                                                        (10)

where H=[H₁H₂H₃ ]^T;.

Select appropriate H₁, H₂, H₃ and H₄ to make sure the estimation error approaches zero and quickly tracks the actual value [23].

The current estimation error, r, is calculated as follows:

                       (11)

where andare the RMS values of the estimation and v₁ and v₂ are the direct proportion parameters, which can increase the degree of freedom of the system design to improve the ability of residual generation and further improves the sensitivity to the fault signal and robustness to interference.

Set thresholds J_th1+ and J_th1-, J_th2+ and J_th2- following the condtions:

                             (12)

When the residuals the system is trouble-free, that is, flag equals 0, and continues to monitor. When the residuals to avoid instantaneous signal interference, only if the fault time lasts more than ¡÷t=T/10, the system is judged to have a fault and the fault location can be transferred for further analysis.

Under the premise of a fault is happened, the specific location approach is as follows:

If the fault occurred in the power switches , otherwise, in the current sensors.

When the fault occurred in the power switch S_pv1 of the first flyback converter; when the fault occurred in the power switch S_pv2 of the second flyback converter; when the fault occurred in the current sensor CS1 of the first flyback converter; when the fault occurred in the current sensor CS2 of the second flyback converter.

4 Fault-tolerant control

4.1 Fault-tolerant topology for faulty power switch

The reconfigurable micro-inverter structure with fault-tolerant capability for the faulty power switch is shown in Fig. 4, which incorporates three auxiliary devices, V₁, V₂ and S_pv3. The auxiliary devices, V₁ and V₂, which are silicon-controlled rectifiers (SCRs), are used to modify the micro-inverter structure after a fault occurrence by connecting the primary port of the high frequency transformer and another auxiliary device S_pv3 that has the same parameters as S_pv1 and S_pv2. Under normal operating conditions, all of the auxiliary devices, V₁, V₂ and S_pv3, are in the off state. The micro-inverter with an S_pv1 open fault has to trigger V₁, S_pv3 and make both of them conductive to form a new circuit channel instead of the faulty branch. The operation of switch S_pv3 is the same as the normal switch S_pv1 and remains complementary with S_pv2.

4.2 Fault-tolerant control for faulty current sensor

When the current sensor fault is detected, a fault-tolerant control is shown in Fig. 5. In this case, thefaulty measured current is replaced by the estimated one for the averaged control.

Fig. 4 Fault-tolerant topology for faulty power switch

Fig. 5 Fault-tolerant control for faulty current sensor

4.3 Fault-tolerant control for the system

Figure 6 shows the flowchart of the proposed diagnosis approach and fault-tolerant control for the faults of power switch and current sensor. If a fault occurs, the absolute values of the two residuals become higher than the minor threshold value at first. After a detection time of ¡÷t=T/10, if the absolute values of the two residuals do not exceed the larger threshold, current sensor fault is diagnosed. Next, according to the relationship between the two residuals and the minor, thresholds determine the faulty current sensor location. Therefore, it is necessary to substitute the corresponding the estimated current for the faulty measured current to maintain normal operations.

5 Simulations

Simulations based on a Matlab/Simulink environment were performed to verify the proposed diagnosis and fault-tolerant control approaches. Table 1 shows the simulation parameters of the micro-inverter system, which are set on the basis of circuit parameters of 200 W micro-inverter device in our lab.In this work, J_th1+, J_th1-, J_th2+ and J_th2- are chosen as 0.7, -0.7, 0.5 and -0.5, respectively.

Fig. 6 Flowchart of proposed diagnosis approach and fault- tolerant control

Table 1 Simulation parameters

5.1 Power-switch failure

Figure 7 shows the estimation performance of the state observer in one line flyback converter (the other line with same result) under normal operation conditions. It can be clearly seen that the estimated currents do track the actual currents well, the residual is very small, and the fault flag remains zero under this circumstance.

Set the faulty time of the power switch S_{pv1 in the} first flyback converter to t₁=0.023 s. Then, the current inthe first converter becomes zero, and the absolute values of the two residuals are more than the minor threshold after a period. Among the detection time ¡÷t=T/10, the absolute values of the two residuals exceed the larger threshold. At t₂=0.027 s, the system detects a fault occurrence, and the fault flag1 is changed from 0 to 1. Figure 8 shows the estimation performance of the state observer for the S_pv1 open fault operation condition.

Fig. 7 Estimation performance of state observer for normal operations:

After diagnosing that S_pv1 has a fault, it is necessary to shut off the drive signal of switch S_pv1, and trigger V₁ and S_pv3 to be conductive to form a new circuit channel instead of the faulty branch. The operation of switch S_pv3 is the same as the normal switch S_pv1 and remains complementary with S_pv2.

Figure 9 illustrates the transient responses from the normal operating mode to the fault-tolerant mode after a fault occurrence at switch S_pv1. From the top, the actual current of the first flyback converter, the actual current of the second flyback converter, and the fault flags are sequentially displayed. Until t₃=0.038 s, the currents of the two flyback converters return to normal. It takes a total time of 0.015 s from a failure to running again after the fault tolerant control.

Fig. 8 Operation condition in the case of S_pv1 open fault:

Fig. 9 Transient responses from normal operating mode to fault- tolerant mode after S_pv1 open fault occurrence:

5.2 Current-sensor failure

Set the faulty time of the current sensor CS1 in the first flyback converter to t₁=0.02 s. Then, the current in the first converter becomes bigger, and the absolute values of the two residuals are more than the minor threshold after a period. Among the detection time ¡÷t= T/10, the absolute values of the two residuals do not exceed the larger threshold. At t₂=0.0285 s, the system detects a fault occurrence, and the fault flag 3 is changed from 0 to 1. Figrue 10 shows the estimation performance of the state observer for the current sensor CS1 fault operation condition.

After diagnosing that CS1 is a fault, it is necessary to substitute the estimated current for the faulty measured current i_Lm1 to maintain normal operations. Figure 11 illustrates the transient responses from the normal operating mode to the fault-tolerant mode after a fault occurrence at CS1.

6 Experimental results

A prototype for the proposed micro-inverter is built in the lab for experimental verification, as shown in Fig. 12. The parameter settings are the same as those in the simulation. The MOSFEFs used in the interleaved flyback converters are IRFS4321, the two current sensors used to detect the two lines of primary currents are B82801B305A125, and the microprocessor of the micro- inverter is a STM32F207VC digital signal processor.

Fig. 10 Operation condition in the case of current sensor CS1 fault:

Fig. 11 Transient responses from normal operating mode to fault-tolerant mode after CS1 fault occurrence:

Fig. 12 Prototype of interleaved flyback micro-inverter built in laboratory

6.1 Power-switch failure

Figure 13 illustrates the primary currents of the two flyback converters, the output current for normal operation. The two primary currents are complementary in phase, and their amplitudes are almost the same.

Fig. 13 Experimental waveform of primary currents of two flyback converters, and output current under normal operation:

Figure 14 shows the actual currents of the two flyback converters and the output current for the S_pv1 open-fault operation condition. It can be clearly seen that when the fault occurs, the faulty branch¡¯s current becomes zero, the other current amplitude nearly doubles,the output current amplitude remains the same but the waveform quality gets worse, and its distortion rate becomes bigger.

Fig. 14 Experimental waveform of primary currents of two flyback converters, and output current under S_pv1 open-fault operation condition:

Figure 15 illustrates the transient responses from the normal operating mode to the S_pv1 open-fault operating mode. As seen, from a failure to diagnosis of the fault, and then disconnecting the faulty switch and triggering the corresponding redundancy switches, it takes one and a half period grid cycles to restore normal operation. Under normal circumstances and after the fault tolerance,the distortion rate of the output current is 3.18%, while during fault condition, it is 10.92%.

Fig. 15 Experimental waveform of fault-tolerant control in the case of S_pv1 open-fault:

6.2 Current-sensor failure

Figure 16 shows the actual currents of the two flyback converters and the output current for the current sensor CS1 fault operation condition. It can be clearly seen that when the fault occurs, the faulty branch¡¯s current increases, the other current decreases, the output current amplitude remains the same but the waveform quality gets worse, and its distortion rate becomes bigger.

Figure 17 shows the performance of the fault- tolerant control for the micro-inverter under the current sensor CS1 outage. It can be clearly seen that the fault-tolerant control performances are good under the current sensor fault as normal condition.

7 Conclusions

1) The open-switch fault case and current sensor fault case are analyzed in the interleaved flyback converters of a micro-inverter system, a fault diagnosis method and fault-tolerant control are proposed.

2) If an open-switch fault or current sensor fault occurs in the micro-inverter system, the performance and reliability can be degraded. The fault can be diagnosed by comparing the thresholds with two residuals obtained from the actual currents and estimated currents that are measured by the state observer. The proposed fault- tolerant control consists of a fault-tolerant topology for the faulty power switch and a simple software redundancy control for the faulty current sensor, which improves the reliability of the whole system and the output current distortion rate is also decreased. Furthermore, the probability of a secondary fault is greatly reduced.

Fig. 16 Experimental waveform of primary currents of two flyback converters, and output current under current sensor CS1 fault operation condition:

Fig. 17 Experimental waveform of fault-tolerant control in case of current sensor CS1 faulty:

3) The proposed fault diagnosis method and fault-tolerant control are simple and easy to implement. The feasibility and effectiveness of the proposed method have been verified by the simulation and experimental results.

4) In general, the proposed method can detect faults in real time, and handle it timely in the event of a failure, which not only greatly improves the reliability of the system, but also prolongs the service lifespan of the micro-inverter to a certain extent.

References

[1] LI Q, WOLFS P. A review of the single phase photovoltaic module integrated converter topologies with three different DC link configurations [J]. IEEE Transactions on Power Electronics, 2008, 23(3): 1320-1333.

[2] STETZ T, MARTEN F, BRAUN M. Improved low voltage grid-integration of photovoltaic systems in germany [J]. IEEE Transactions on Sustain Energy, 2013, 4(2): 534-542.

[3] ZHANG Z, HE X F, LIU Y F. An optimal control method for photovoltaic grid-tied-interleaved flyback microinverters to achieve high efficiency in wide load range [J]. IEEE Transactions on Power Electronics, 2013, 28(11): 5074-5087.

[4] KIM Y H, JANG J W, SHIN S C, WON C Y. Weighted-efficiency enhancement control for a photovoltaic ac module interleaved flyback inverter using a synchronous rectifier [J]. IEEE Transactions on Power Electronics, 2014, 29(12): 6481-6493.

[5] HU Hai-bing, HARB S, KUTKUT N H, SHEN Z J, BATARSEH I. A single-stage microinverter without using eletrolytic capacitors [J]. IEEE Transactions on Power Electronics, 2013, 28(6): 2677-2687.

[6] KIM J G, KIM K D, NOH Y S, JUNG Y C, WON C Y. Analysis and design of a three-port flyback inverter using an active power decoupling method to minimize input capacitance [J]. Journal of Power Electronics, 2013, 13(4): 558-568.

[7] PETRONE G, SPAGNUOLO G, TEODORESCU R, VEERACHARY M, VITELLI M. Reliability issues in photovoltaic power processing systems [J]. IEEE Transactions on Industrial Electronics, 2008, 55(7): 2569-2580.

[8] ZHANG P, WANG Y, XIAO W, LI W. Reliability evaluation of grid connected photovoltaic power systems [J]. IEEE Transactions on Sustainable Energy, 2012, 3(3): 379-389.

[9] BAGNOLI P E, CASAROSA C E, DALLAGO E, NARDONI M. Thermal resistance analysis by induced transient(TRAIT) method for power electronic devices thermal characterization I. Fundamentals and theory [J]. IEEE Transactions on Power Electronics, 1998, 13(6): 1208-1219.

[10] IM W S, KIM J S, KIM J M, LEE D C, LEE K B. Diagnosis methods for IGBT open switch fault applied to 3-phase AC/DC PWM converter [J]. Journal of Power Electronics, 2012, 12(1): 120-127.

[11] MAHMOUD S, EHSAN J, POURE P. Open- and short-circuit switch fault diagnosis for nonisolated DC¨CDC converters using field programmable gate array [J]. IEEE Transactions on Industrial Electronics, 2013, 60(9): 4136-4146.

[12] LU B, SHARMA S K. A literature review of IGBT fault diagnostic and protection methods for power inverters [J]. IEEE Transactions on Industry Applications, 2009, 45(5): 1770-1777.

[13] EISSA M M. New differential busbar characteristic based on high frequencies extracted from faulted signal during current transformer saturation [J]. Generation, Transmission & Distribution, IET, 2014, 8(4): 619-628.

[14] HAMIDREZAM, MEHDIS, HOOSHANGJR, ALIAKBAR N. Reconstruction based approach to sensor fault diagnosis using auto-associative neural networks [J]. Journal of Central South University, 2014, 21(6):2273-2281.

[15] FRANK P M. Fault diagnosis in dynamic systems using analytical and knowledge-based redundancy-a survey and some new result [J]. Automatica, 1990, 26(3): 459-474.

[16] DING S X. Model-based fault diagnosis techniques design schemes, algorithms, and tools [M]. Berlin: Springer-Verlag, 2008: 23.

[17] SHUAI Shao, WHEELER P W, CLARE J C, WASTON A J. Fault detection for modular multilevel converters based on sliding mode observer [J]. IEEE Transactions on Power Electronics, 2013, 28(11): 4867-4872.

[18] CAMPOS-DELGADO D U, ESPINOZA-TREJO D R. An observer- based diagnosis scheme for single and simultaneous open-switch faults in induction motor drives [J]. IEEE Transactions on Industrial Electronics, 2011, 28(2): 671-679.

[19] NAJAFABADI T A, SALMASI F R, PARVIZ J M. Detection and isolation of speed-, DC-link voltage-, and current-sensor faults based on an adaptive observer in induction-motor drives [J]. IEEE Transactions on Industrial Electronics, 2011, 58(5): 1662-1672.

[20] KHIEM N D, LIU Tian-hua, CHEN Der-fa, HUNG J Y. Improvement of matrix converter drive reliability by online fault detection and a fault-tolerant switching strategy [J]. IEEE Transactions on Industrial Electronics, 2012, 59(1): 244-256.

[21] KHWAN-ON S, de LILLO L, EMPRINGHAM L, WHEELER P. Fault-tolerant matrix converter motor drives with fault detection of open switch faults [J]. IEEE Transactions on Industrial Electronics, 2012, 59(1): 257-268.

[22] RIBEIRO E, CARDOSO A J M, BOCCALETTI C. Fault-tolerant strategy for a photovoltaic DC¨CDC converter [J]. IEEE Transactions on Power Electronics, 2013, 28(6): 3008-3018.

[23] KIM S C, NGUYEN T H, LEE D C, LEE K B, KIM J M. Fault tolerant control of DC-Link voltage sensor for three-phase AC/DC/AC PWM converters [J]. Journal of Power Electronics, 2014, 14(4): 695-703.

(Edited by FANG Jing-hua)

Foundation item: Project(2012AA051601) supported by the High-Tech Research and Development Program of China

Received date: 2015-05-26; Accepted date: 2015-11-15

Corresponding author: PENG Tao, Professor, PhD; Tel: +86-13607336206; E-mail: pandtao@csu.edu.cn