Table of Contents Author Guidelines Submit a Manuscript
Journal of Applied Mathematics
Volume 2014, Article ID 823937, 12 pages
http://dx.doi.org/10.1155/2014/823937
Research Article

A Method to Determine Oscillation Emergence Bifurcation in Time-Delayed LTI System with Single Lag

Key Laboratory of Smart Grid of Ministry of Education, Tianjin University, Tianjin 300072, China

Received 23 January 2014; Accepted 21 May 2014; Published 3 July 2014

Academic Editor: Xinghuo Yu

Copyright © 2014 Yu Xiaodan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

One type of bifurcation named oscillation emergence bifurcation (OEB) found in time-delayed linear time invariant (abbr. LTI) systems is fully studied. The definition of OEB is initially put forward according to the eigenvalue variation. It is revealed that a real eigenvalue splits into a pair of conjugated complex eigenvalues when an OEB occurs, which means the number of the system eigenvalues will increase by one and a new oscillation mode will emerge. Next, a method to determine OEB bifurcation in the time-delayed LTI system with single lag is developed based on Lambert W function. A one-dimensional (1-dim) time-delayed system is firstly employed to explain the mechanism of OEB bifurcation. Then, methods to determine the OEB bifurcation in 1-dim, 2-dim, and high-dimension time-delayed LTI systems are derived. Finally, simulation results validate the correctness and effectiveness of the presented method. Since OEB bifurcation occurs with a new oscillation mode emerging, work of this paper is useful to explore the complex phenomena and the stability of time-delayed dynamic systems.

1. Introduction

In the nature, system trend is not only dependent on the current operating point but also subject to its previous conditions. This phenomenon is called time delay [1, 2]. As a main cause to the control deterioration and system instability, time delays exist widespread in the industry of communication [3], biology [4], mechanism [5], chemistry [6], society [7], and power utility [810]. Therefore, it is an important issue and frequent topic to evaluate the impact of time delays on the stability of dynamic systems.

Many methods to analyze the stability of time-delayed dynamic system have been proposed in the past literatures. For example, [1013] applied Lyapunov theory to derive various stability criteria for time-delayed systems. To seek the suitable Lyapunov function (or Lyapunov functional) and to construct the corresponding criteria were the two key points of such methods. Linear matrix inequality (LMI) technique, Lyapunov-Razumikhin theory, and Lyapunov-Krasovskii theory were usually used in the derivation. Just as we know, characteristic equation of time-delayed system has transcendental element, which is very difficult to handle in the analysis. In order to surmount this problem, [14, 15] utilized Rekasius substitution method and [16] used Pade approximation method to approximate the transcendental elements and transfer the characteristic equation into a polynomial, which was used to calculate the system critical eigenvalues. Reference [9, 17] treated the stability region method as a tool to determine the small signal stability region of time-delayed dynamic system. All points in the stability region were small signal stable, and the boundary of the stability region was composed by some bifurcation points, such as saddle node bifurcation, Hopf bifurcation, and singularity induced bifurcation. Reference [1822] applied bifurcation theory to analyze some complex phenomena existing in the time-delayed systems, such as Hopf bifurcation, period doubling bifurcation, and chaos.

Time-delayed system is a complex dynamic system, in which many interesting and intricate dynamic behaviors exist. For example, Figure 1 from [23] demonstrated that a real eigenvalue split into a pair of conjugated complex eigenvalues at point with increase of time delay in a time-delayed power system. In this paper, this phenomenon is called oscillation emergence bifurcation (OEB). And, we apply the theory of bifurcation analysis and Lambert function to propose a mechanism to the occurrence of this bifurcation. As an available tool, Lambert function [24, 25] has been utilized to analyze some complex behaviors in nonlinear system including stability analysis of time-delayed system [2629]. It is applied to derive a general method to determine OEB bifurcation in this paper.

823937.fig.001
Figure 1: A real eigenvalue splits into a pair of conjugated eigenvalues.

The contents of this paper are organized into six sections. Following this introduction, definition of OEB bifurcation is put forward in Section 2 and some of its features are also discussed in this section. In Section 3, Lambert function is introduced and a mechanism of the OEB’s occurrence is proposed. Section 4 presents a method to determine OEB bifurcation in time-delayed LTI system with single lag based on the Lambert function. In Section 5, the presented approach is demonstrated by numerical simulations. Finally, Section 6 concludes and summarizes this paper.

2. Definition of OEB Bifurcation

In this paper, all studies are based on the following time-delayed LTI system with single lag: where ; is the vector of the state variables; is the constant of time delay; is the system trajectory in the range of .

It is well known that the following equation is the corresponding characteristic equation of system : where is the eigenvalue to be determined. The eigenvalue spectrum of system is denoted by : Theoretically, it can be calculated from (2) directly when is given. Further, denote the set of the real eigenvalue by and the set of the complex eigenvalue with nonzero imaginary by : where , are the eigenvalue numbers of and , respectively. Then the following (5) holds: where is the eigenvalue number of .

It can be found that when , system degenerates into the following LTI differential equation: where . And, the number of eigenvalues is ; that is, . When increases, the values of , , and will change accordingly. Therefore, they can be considered as functions of as follows: , , and are three functions of to yield the numbers of real eigenvalues, complex eigenvalues, and whole eigenvalues of system .

Definition 1 (definition of oscillation emergence bifurcation). Presume that there is a value of time delay   for system of (1). Its has real eigenvalues and has complex eigenvalues with . And, after , the numbers of and turn to and . If the following equations hold: there is an oscillation emergence bifurcation at .

According to Definition 1, suppose there is a real eigenvalue in at . It will split into a pair of conjugated eigenvalues , as shown in Figure 2. It is manifested that the number of will decrease by one, the number of will increase by two, and number of will increase by one after an OEB bifurcation occurs. At the same time, a new oscillation mode subject to , will emerge after OEB bifurcation.

823937.fig.002
Figure 2: Illustration of OEB bifurcation.

3. Lambert Function and OEB Bifurcation

Our aim is to derive a general method based on Lambert function to determine OEB bifurcation in system defined by (1). A brief introduction to Lambert function is firstly given.

The following function is called Lambert function: where . Its inverse function is expressed as

Lambert function is not injective and the relation is multivalued (except at 0). It has many solution branches in complex plane [25], which can be denoted by , . If the attention is restricted to real-valued , then the relation is defined only for . For , there are two possible real values of as shown in Figure 3. We denote the branch satisfying by (solid line in Figure 3), and the branch satisfying by (dashed line in Figure 3). is referred to the principal branch of Lambert function. It increases from to , while branch decreases from to . They encounter at point as shown in Figure 3, and the value of point is given as follows:

823937.fig.003
Figure 3: and branches of Lambert function.

Supposed that in (9) and (10) is a system parameter and it varies with another parameter :

Substitute (12) into (10). It can be found that is also a function of :

Then, the curve can be plotted. An example is shown in Figure 4, where curve crosses the dotted line of at some points which can be divided into two categories:

823937.fig.004
Figure 4: Curve of .

If is on the interval of , that is, the intervals above the line with , (13) has real solutions. While if is on the interval of , that is, the grey intervals with , (13) only has complex solutions with nonzero imaginary.

Recalling the definition of OEB, we can find that is just a set of OEB bifurcations of the system expressed by (13). That is, at any point of , a real eigenvalue will turn into a pair of conjugated complex eigenvalues with the increment of .

Here the following typical transcendental item in one-dimensional (1-dim) delayed-system is treated as an example for further discussion to Figure 4: where , and . The value of can be derived from (15) based on Lambert function as follows: where

Figure 5 depicts the curve of with , . It is evident that point with in the figure is an OEB bifurcation of system (15) according to Definition 1 and Figure 4.

823937.fig.005
Figure 5: Curve of .

From Figures 4 and 5, it is asserted that the following formula holds at the OEB bifurcation points when : And, if has complex values on one side of the OEB bifurcation point, the above condition will be modified as follows: where is the real part of . For a time-delayed system, if its characteristic equation can be transformed into a suitable form, conditions of (18) or (19) may be adopted as a criterion to determine the OEB bifurcation.

4. An Approach to Determine OEB Bifurcation

In this section, we aim to derive a general method to transform the characteristic equation of time-delayed system into a form that can be handled by Lambert function so as to determine the occurrence of OEB bifurcation. Methods suitable for one-dimensional (1-dim) and two-dimensional (2-dim) time-delayed LTI systems are first derived and then they are expanded to the high-dimensional systems.

4.1. 1-Dim System

The following 1-dim system is considered in this subsection: where , , and . The characteristic equation of (20) is where . It can be rewritten as According to (15), can be solved from (22) based on Lambert function: where

Similar to Figure 5, the OEB bifurcation can be determined from (24) via condition (18).

4.2. 2-Dim System

Let us consider the following 2-dim system: where , , ; . Assume are two eigenvalues of the following matrix:

It means that there are two negative and real eigenvalues at , so system given by (25) is stable without time delay. Supposed that is an eigenvalue of (25) at and the following equation holds: System characteristic equation can be derived as follows: With , , , , , and .

In order to determine the OEB bifurcation in the system, we reorganize (28) into the following form: where . And, it can be expanded as follows:

Comparing (30) with (28), the following equations hold: And, the following result can be obtained: when and there is no real solution for (31). That is (28) cannot be converted into the form of (29) in real domain. In order to use the criterion given by (18) or (19), the following two scenarios are considered, respectively.

(i) Consider . Equation (28) can be converted into the form of (29) in real domain. Equation (29) can be rewritten as follows in order to use the condition given by (18) or (19): where is a temporary variable.

Via the first formula in (35), the following equation holds: According to (15) and (16), with , , , has the following form: where

Via the second formula in (35), the following equation holds:

Similarly, having defined , , , the following result can be obtained: where

Equations (37) and (40) can be utilized to solve the values of and when is given. They can be further taken into (38) and (41) to get the values of and . And if any one of the following equations holds, there is an OEB bifurcation: Obviously, if is a complex eigenvalue, defined by (34) and , given by (38) and (41) are all complex values. They are all real values if is a real eigenvalue. Hence, condition (19) is suitable for this case.

(ii) Consider . Since (28) cannot be converted into the form of (29) in real domain at this condition, it is needed to find a new method for converting (28) into the following form: where . Also, its expansion is given as follows: Comparing it with (28), the following equations hold:

Hence, To guarantee that , , , , are real numbers, the following relationship must hold:

Since can be selected arbitrarily, it is set as follows:

Substituting (53) into (51) yields

Since , it must be true that . So and are two real numbers. According to (49)–(51), , , are all real numbers. In order to use condition (19), (44) is written as

It can be indicated that the second formula in (57) is equivalent to the one in (35). Thus, it can generate the same results given by (40), (41), and condition (43). The new conditions will be derived from the first formula in (57) as follows:

Suppose that there exists a real variable such that

Inserting it into (59) yields

Defining , , , eigenvalue is expressed as where

Defining , , , the similar result from (61) is obtained: where

Equations (40), (62), and (64) can be used to solve the values of , , and when is given. They can be further taken into (41), (63), and (65) to get the values of , , and . As a conclusion, any of the following conditions hold, and there is an OEB bifurcation in the system:

4.3. High-Dimensional Systems

Consider system given by (1) with . Its characteristic equation has the following general form: where and is a polynomial of : Some general forms of the characteristic equation for system are listed in Table 1. is the number of coefficients in the general form. For example, there are five coefficients which are in the characteristic equation for , and there are nine coefficients in the characteristic equation with .

tab1
Table 1: General forms of the characteristic equation for time-delayed system.

(i) 3-Dim Case (). Take 3-dim system as an example to explain the principle of proposed method. General form of the characteristic equation for is given as below:

It can be converted into the following form so as to utilize the previous results: where is described by (28). And, , , are variables to be determined. Setting , (70) can be written as Via the first formula in (71), It is obvious that is a typical 2-dim case discussed above. Hence, the result of Section 4.2 can be directly applied to (72).

Further, via the second formula of (71), where , . The result of 1-dim system in Section 4.1 can be applied to (73). And the following result can also be derived:

From the above discussion, it can be asserted that a 3-dim system can be transformed to a combination of a 1-dim system and a 2-dim system. Therefore, conclusions in Sections 4.1 and 4.2 can be applied. For simplification, it is denoted as

As a conclusion, any of the following conditions hold, and there is an OEB bifurcation in the system: where (76) and (77) are concluded from (72) with conditions expressed by (42)-(43) when (or conditions expressed by (66) when ).

(ii) Case of High Dimension with . Similar to 3-dim case, the characteristic equation of high-dimensional system can also be converted into lower ones so as to apply the previous derived results. For example, for ,

Setting , (79) can be rewritten as

The first formula is a 3-dim case, and the second is a 2-dim case. Hence, And, the cases discussed for 2-dim and 1-dim cases can be applied to determine OEB bifurcation in 4-dim system.

Further, the following recursion formula for high-dimensional systems with can be obtained:

Then, a higher dimension system can be converted into two lower dimension systems so as to determine the occurrence of OEB bifurcation accordingly.

5. Case Studies

5.1. 1-Dim System

Consider the following 1-dim system:

According to (24), the curve can be depicted in Figure 6 and the corresponding eigenvalue locus is depicted in Figure 7. For a pair of conjugated eigenvalues, only the one above the real axis is showed in the figures in this paper.

823937.fig.006
Figure 6: Curve of .
823937.fig.007
Figure 7: Curve of an eigenvalue.

Similar to Figure 5, point with in Figure 6 is an OEB bifurcation of system described by (83). The variation of the system eigenvalue in Figure 7 verifies the analysis above. That is, the real eigenvalue splits into a pair of conjugated complex eigenvalues at the OEB bifurcation with .

In order to demonstrate the impact of OEB bifurcation to the system defined by (83), its unit step responses with different values are shown in Figure 8. It can be found that the system trajectory converges monotonously in the range of and converges oscillatedly in the range of . It is manifested that there is OEB bifurcation at , which leads to a new oscillation mode and increases the number of system eigenvalues by one.

fig8
Figure 8: Simulation results with various values.
5.2. 2-Dim System

The following two cases are employed to validate the proposed method.

(i) Case A. Consider where is defined by (26).

It is obvious that the conditions described by (42)-(43) are suitable for this case.

Via the tracing method proposed by [23], we can obtain the system eigenvalue locus. Figure 9 depicts the curve of the second eigenvalue with increasing (the first one is always on the real axis, so it is not shown in the figure). There is an OEB bifurcation at point with . According to (38) and (41), the curves of and can be depicted in Figure 10. Solid line represents that the corresponding , , is real, while dotted line represents that it is complex. Some results nearby point are listed in Table 2. It can be found that condition (42) holds at point with and it indicates the occurrence of OEB bifurcation precisely.

tab2
Table 2: values near point in Figure 10.
823937.fig.009
Figure 9: Locus of the second eigenvalue with increasing.
823937.fig.0010
Figure 10: Curves of and .

(ii) Case B. Consider where is defined by (26).

It can be asserted that the conditions described by (66) are appropriate for this case. Similar to Case A, the eigenvalue locus with increasing can be obtained via the tracing method expressed by [23]. The eigenvalue locus is shown in Figure 11, while real and imaginary parts of the eigenvalues are shown in Figure 12. Note that two real eigenvalues at the beginning move close to each other on the real axis. They encounter at point and then turn to a pair of conjugated complex eigenvalues. According to Definition 1, there is no OEB bifurcation in the system since the number of eigenvalues does not increase here.

823937.fig.0011
Figure 11: Eigenvalue locus of Case B.
fig12
Figure 12: Real and imaginary parts of two eigenvalues.

Figure 13 plots the curves of , , and . Similarly, solid and dotted lines are used to represent the real and complex values of , . It is obvious that any condition identified by (66) does not hold here, so there is no OEB bifurcation in the system.

823937.fig.0013
Figure 13: Curves of , , and .

Case studies above demonstrate that conditions expressed by (42)-(43) or (66) are correct. And, the method proposed can correctly indicate the occurrence of OEB bifurcation in the 2-dim time-delayed system.

5.3. 3-Dim System

Consider a 3-dim system defined by (1) with , as follows: It can be found that the system have three real eigenvalues with : where .

Similar to Section 5.2, we can get the eigenvalue locus shown in Figure 14, where the first and second eigenvalues always move on the real axis, while the third eigenvalue splits into a pair of conjugated complex eigenvalues at point with so as to yield an OEB bifurcation.

823937.fig.0014
Figure 14: Eigenvalue locus with time delay increasing.

Table 3 gives the coefficients in (69) and (70). According to the definition of by (76)–(78), we can depict the curves of , , and in Figure 15. The curve (red line) hits the horizontal line of at and condition (78) holds which indicate the occurrence of the OEB bifurcation.

tab3
Table 3: Coefficients of given by (69) and (70).
823937.fig.0015
Figure 15: Curves of , , and .

It should be mentioned that the transformation from (69) to (70) is not unique. Table 4 shows two sets of coefficients in (70). The corresponding , , and curves are illustrated, respectively, by Figures 16 and 17. It is notable that in Figure 16 or Figure 17, the curve (black line) collides with the horizontal line of at . The condition of (76) holds at this point so as to indicate occurrence of the OEB bifurcation.

tab4
Table 4: Two sets of coefficients in (70).
823937.fig.0016
Figure 16: Curves of , , and with set I of Table 4.
823937.fig.0017
Figure 17: Curves of , , and with set II of Table 4.

From the case study above, it is verified that conditions expressed by (76)–(78) can indicate the occurrence of OEB bifurcation in the 3-dim time-delayed LTI system.

6. Conclusions

In this paper, a method based on Lambert function theory is presented to determine the oscillation emergence bifurcation (OEB) in time-delayed LTI system with single lag. Conditions of OEB’s occurrence in 1-dim, 2-dim, 3-dim, and higher dimension systems are derived, respectively. And, case studies confirm the correctness and preciseness of the proposed method. Work of this paper is beneficial to explore the complex dynamic phenomena existing in the time-delayed system. However, the proposed method is only valid for the time-delayed system with single lag. How to expand it to the multiple time-delayed dynamic systems will be further discussed in the future.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgment

This work is supported by the Natural Science Foundation of China under Grant nos. 51277128 and 51377117.

References

  1. J. K. Hale, Theory of Functional Differential Equations, World Publishing Corporation, Beijing, China, 2003.
  2. L. Dugard and E. I. Verriest, Stability and Robust Control of Time Delay Systems, Springer, New York, NY, USA, 1997.
  3. F. S. Gentile, J. L. Moiola, and E. E. Paolini, “Nonlinear dynamics of internet congestion control: a frequency-domain approach,” Communications in Nonlinear Science and Numerical Simulation, vol. 19, no. 4, pp. 1113–1127, 2014. View at Publisher · View at Google Scholar
  4. T. Ito, H. Ohashi, T. Tamaki, and T. Yamaguchi, “Mathematical modeling of molecular recognition by an ion-gating membrane oscillator,” Journal of Membrane Science, vol. 448, pp. 231–239, 2013. View at Publisher · View at Google Scholar · View at Scopus
  5. Y. Yu, H. Tang, X. Han, and Q. Bi, “Bursting mechanism in a time-delayed oscillator with slowly varying external forcing,” Communications in Nonlinear Science and Numerical Simulation, vol. 19, no. 4, pp. 1175–1184, 2014. View at Publisher · View at Google Scholar · View at MathSciNet
  6. S. Xu and J. Bao, “Distributed control of plant-wide chemical processes with uncertain time-delays,” Chemical Engineering Science, vol. 84, pp. 512–532, 2012. View at Publisher · View at Google Scholar · View at Scopus
  7. S. N. Rader, M. L. Reagan, B. Janoiko, and J. E. Johnson, “Human- in-the-loop operations over time delay: NASA analog missions lessons learned,” in Proceedings of the 43rd International Conference on Environmental Systems, pp. 1–6, Vail, Colo, USA, 2013.
  8. H. Jia, Y. Mu, and Y. Qi, “A statistical model to determine the capacity of battery-supercapacitor hybrid energy storage system in autonomous microgrid,” International Journal of Electrical Power & Energy Systems, vol. 54, pp. 516–524, 2014. View at Publisher · View at Google Scholar · View at Scopus
  9. H. Jia, X. Yu, Y. Yu, and C. Wang, “Power system small signal stability region with time delay,” International Journal of Electrical Power and Energy Systems, vol. 30, no. 1, pp. 16–22, 2008. View at Publisher · View at Google Scholar · View at Scopus
  10. X. Yu, H. Jia, and C. Wang, “CTDAE & CTODE models and their applications to power system stability analysis with time delays,” Science China Technological Sciences, vol. 56, no. 5, pp. 1213–1223, 2013. View at Publisher · View at Google Scholar · View at Scopus
  11. Y. He, M. Wu, and J. She, “Delay-dependent exponential stability of delayed neural networks with time-varying delay,” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 53, no. 7, pp. 553–557, 2006. View at Publisher · View at Google Scholar · View at Scopus
  12. J. YiLang, J. Tao, J. HongJie, and D. ChaoYu, “A novel LMI criterion for power system stability with multiple time-delays,” Science China Technological Sciences, vol. 57, no. 7, pp. 1392–1400, 2014. View at Publisher · View at Google Scholar
  13. R. H. Gielen, M. Lazar, and S. V. Rakovic, “Necessary and sufficient Razumikhin-type conditions for stability of delay difference equations,” IEEE Transactions on Automatic Control, vol. 58, no. 10, pp. 2637–2642, 2013. View at Publisher · View at Google Scholar · View at MathSciNet
  14. N. Olgac and R. Sipahi, “An exact method for the stability analysis of time-delayed linear time-invariant (LTI) systems,” IEEE Transactions on Automatic Control, vol. 47, no. 5, pp. 793–797, 2002. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  15. Q. Sun, H. An, H. Jia, and X. Yu, “An improved power system stability criterion with multiple time delays,” in Proceedings of the IEEE General Meeting, vol. 1–8, pp. 786–792, Calgary Alberta, Calif, USA, July 2009.
  16. F. Rossi, P. Colaneri, and R. Shorten, “Padé discretization for linear systems with polyhedral Lyapunov functions,” IEEE Transactions on Automatic Control, vol. 56, no. 11, pp. 2717–2722, 2011. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  17. J. Hongjie, Y. Xiaodan, W. Chengshan et al., “Study on power system extended small signal stability region (DE-SSSR) in time delay space,” in Proceedings of the 15th IEEE Mediterranean Electrotechnical Conference (MELECON '10), vol. 1, pp. 1569–1574, Valletta, Malta, April 2010. View at Publisher · View at Google Scholar · View at Scopus
  18. Y. Ding, W. Jiang, and P. Yu, “Double Hopf bifurcation in delayed van der Pol-duffing equation,” International Journal of Bifurcation and Chaos in Applied Sciences and Engineering, vol. 23, no. 1, Article ID 1350014, 15 pages, 2013. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  19. X. Xu, H. Y. Hu, and H. L. Wang, “Stability, bifurcation and chaos of a delayed oscillator with negative damping and delayed feedback control,” Nonlinear Dynamics: An International Journal of Nonlinear Dynamics and Chaos in Engineering Systems, vol. 49, no. 1-2, pp. 117–129, 2007. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  20. F.-X. Wu, “Stability and bifurcation of ring-structured genetic regulatory networks with time delays,” IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 59, no. 6, pp. 1312–1320, 2012. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  21. T. Co, “Relay-stabilization and bifurcations of unstable SISO processes with time delay,” IEEE Transactions on Automatic Control, vol. 55, no. 5, pp. 1131–1141, 2010. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  22. S. Guo and W. Jiang, “Global stability and Hopf bifurcation for Gause-type predator-prey system,” Journal of Applied Mathematics, vol. 2012, Article ID 260798, 17 pages, 2012. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  23. X. Yu, H. Jia, and C. Wang, “An eigenvalue spectrum tracing algorithm and its application in time delayed power systems,” Automation of Electric Power Systems, vol. 36, no. 24, pp. 10–38, 2012. View at Google Scholar · View at Scopus
  24. D. J. Jeffrey, D. E. G. Hare, and R. M. Corless, “Unwinding the branches of the Lambert W function,” The Mathematical Scientist, vol. 21, no. 1, pp. 1–7, 1996. View at Google Scholar · View at Zentralblatt MATH · View at MathSciNet
  25. R. M. Corless, G. H. Gonnet, D. E. G. Hare, and D. E. Knuth, “On the Lambert W function,” Advances in Computational Mathematics, vol. 5, no. 1, pp. 329–359, 1996. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  26. E. Jarlebring and T. Damm, “The Lambert W function and the spectrum of some multidimensional time-delay systems,” Automatica, vol. 43, no. 12, pp. 2124–2128, 2007. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  27. C. Hwang and Y. Cheng, “A note on the use of the Lambert W function in the stability analysis of time-delay systems,” Automatica, vol. 41, no. 11, pp. 1979–1985, 2005. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  28. Y. Sun and A. G. Ulsoy, “Solution of a system of linear delay differential equations using the matrix lambert function,” in Proceedings of the 26th American Control Conference, pp. 2433–2438, Minneapolis, Minn, USA, June 2006. View at Scopus
  29. S. Yi, S. Yu, and J. H. Kim, “Analysis of neural networks with time-delays using the Lambert W function,” in Proceedings of the American Control Conference (ACC '11), pp. 3221–3226, San Francisco, Calif, USA, July 2011. View at Scopus