Skip to main content
Log in

Stochastic optimal control of MDOF nonlinear systems under combined harmonic and wide-band noise excitations

  • Original Paper
  • Published:
Nonlinear Dynamics Aims and scope Submit manuscript

Abstract

A strategy for stochastic optimal control of multi-degrees-of-freedom (MDOF) strongly nonlinear systems under combined harmonic and wide-band noise excitations is proposed. First, a stochastic averaging procedure is developed to obtain the partially averaged Itô stochastic differential equations for weakly controlled strongly nonlinear systems under combined harmonic and wide-band noise excitations. Then, the dynamical programming equation for stochastic optimal control problem is derived and solved to yield the optimal control law. Finally, the responses of optimally controlled MDOF nonlinear systems are predicted by solving the Fokker–Planck–Kolmogorov equation associated with the fully averaged Itô stochastic differential equations. As an example, the stochastic optimal control of two coupled Duffing oscillators under combined harmonic and wide-band noise excitations is worked out in detail to illustrate the strategy. The effectiveness of the proposed control strategy is verified using the results from Monte Carlo simulation.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10

Similar content being viewed by others

References

  1. Zhu, W.Q., Ying, Z.G.: Optimal nonlinear feedback control of quasi Hamiltonian systems. Sci. Chin. Ser. A 42, 1213–1219 (1999)

    Article  MATH  MathSciNet  Google Scholar 

  2. Zhu, W.Q., Ying, Z.G., Soong, T.T.: An optimal nonlinear feedback control strategy for stochastically excited structural systems. Nonlinear Dyn. 24, 31–51 (2001)

    Article  MATH  MathSciNet  Google Scholar 

  3. Zhu, W.Q., Yang, Y.Q.: Stochastic averaging of quasi non-integrable-Hamiltonian systems. J. Appl. Mech. (ASME) 64, 157–164 (1997)

    Article  MATH  MathSciNet  Google Scholar 

  4. Zhu, W.Q., Huang, Z.L., Yang, Y.Q.: Stochastic averaging of quasi integrable Hamiltonian systems. J. Appl. Mech. (ASME) 64, 975–984 (1997)

    Article  MATH  MathSciNet  Google Scholar 

  5. Zhu, W.Q., Huang, Z.L., Suzuki, Y.: Stochastic averaging and Lyapunov exponent of quasi partially integrable Hamiltonian systems. Int. J. Non-Linear Mech. 37, 419–437 (2002)

    Article  MATH  MathSciNet  Google Scholar 

  6. Chen, L.C., Zhu, W.Q: Stochastic averaging of strongly nonlinear oscillators with small fractional derivative damping under combined harmonic and white noise excitation. Nonlinear Dyn. 56, 231–241 (2009)

  7. Fleming, W.H., Rishel, R.W.: Deterministic and Stochastic Optimal Control. Springer, New York (1975)

    Book  MATH  Google Scholar 

  8. Fleming, W.H., Soner, H.M.: Controlled Markov Processes and Viscosity Solutions. Springer, New York (1992)

    Google Scholar 

  9. Yong, J.M., Zhou, X.Y.: Stochastic Control, Hamiltonian Systems and HJB Equations. Springer, New York (1999)

    Google Scholar 

  10. Zhu, W.Q., Huang, Z.L., Deng, M.L.: Feedback minimization of first-passage failure of quasi non-integrable Hamiltonian systems. Int. J. Nonlinear Mech. 37, 1057–1071 (2002)

    Article  MATH  MathSciNet  Google Scholar 

  11. Zhu, W.Q., Huang, Z.L., Deng, M.L.: First-passage failure and its feedback minimization of quasi partially integrable Hamiltonian systems. Int. J. Nonlinear Mech. 38, 1133–1148 (2003)

    Article  MATH  MathSciNet  Google Scholar 

  12. Zhu, W.Q., Deng, M.L., Huang, Z.L.: Optimal bounded control of first-passage failure of quasi-integrable hamiltonian systems with wide-band random excitation. Nonlinear Dyn. 33, 187–207 (2003)

  13. Zhu, W.Q., Huang, Z.L.: Feedback stabilization of quasi-integrable Hamiltonian system. J. Appl. Mech. (ASME) 70, 129–136 (2003)

  14. Zhu, W.Q., Ying, Z.G., Ni, Y.Q., Ko, J.M.: Optimal nonlinear stochastic control of hysteretic systems. J. Eng. Mech. (ASCE) 126, 1027–1032 (2000)

    Article  Google Scholar 

  15. Ying, Z.G., Zhu, W.Q.: A stochastic optimal control strategy for partially observable nonlinear quasi-Hamiltonian systems. J. Sound Vib. 310, 184–196 (2008)

    Article  Google Scholar 

  16. Ying, Z.G., Zhu, W.Q., Soong, T.T.: A stochastic optimal semi-active control strategy for ER/MR dampers. J. Sound Vib. 259, 45–62 (2003)

    Article  MATH  MathSciNet  Google Scholar 

  17. Sri Namachchivaya, N.: Almost sure stability of dynamical systems under combined harmonic and stochastic excitations. J. Sound Vib. 151, 77–90 (1991)

    Article  Google Scholar 

  18. Cai, G.O., Lin, Y.K.: Nonlinearly damped systems under simultaneous broad-band and harmonic excitations. Nonlinear Dyn. 6, 163–177 (1994)

  19. Huang, Z.L., Zhu, W.Q.: Exact stationary solutions of averaged equations of stochastically and harmonically excited mdof quasi-linear systems with internal and/or external resonances. J. Sound Vib. 204, 249–258 (1997)

  20. Huang, Z.L., Zhu, W.Q., Suzuki, Y.: Stochastic averaging of strongly non-linear oscillators under combined harmonic and white-noise excitations. J. Sound Vib. 238, 233–256 (2000)

    Article  MATH  MathSciNet  Google Scholar 

  21. Zhu, W.Q., Wu, Y.J.: First-passage time of Duffing oscillator under combined harmonic and white-noise excitations. Nonlinear Dyn. 32, 291–305 (2003)

    Article  MATH  Google Scholar 

  22. Huang, Z.L., Zhu, W.Q.: Stochastic averaging of quasi-integrable Hamiltonian systems under combined harmonic and white noise excitations. Int. J. Non-Linear Mech. 39, 1421–1434 (2004)

    Article  MATH  MathSciNet  Google Scholar 

  23. Wu, Y.J., Zhu, W.Q.: Stochastic averaging of strongly nonlinear oscillators under combined harmonic and wide-band noise excitations. J. Vib. Acoust. 130(5), 051004 (2008)

    Article  Google Scholar 

  24. Wu, Y.J., Feng, C.S., Huan, R.H.: Optimal bounded control for stationary response of strongly nonlinear oscillators under combined harmonic and wide-band noise excitations. Math. Probl. Eng. 2011, p. 21 (2011). doi:10.1155/2011/635823

  25. Xu, Z., Cheung, Y.K.: Averaging method using generalized harmonic functions for strongly nonlinear oscillators. J. Sound Vib. 174, 563–576 (1994)

    Article  MATH  MathSciNet  Google Scholar 

Download references

Acknowledgments

The research reported in the present paper was supported by the National Natural Science Foundation of China under Grant Nos. 10932009, 11072212, 11272279, 11321202, 11372271, 51175474.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to W. Q. Zhu.

Appendices

Appendix 1: The averaged drift and diffusion coefficients

The coefficients in Eq. (52) are as follows:

$$\begin{aligned} \overline{{m}}_i&= \overline{{H}}_i (\mathbf{A})+F_i (\mathbf{A}) \quad (i=1,2),\end{aligned}$$
(62)
$$\begin{aligned} \overline{{m}}_3&= \overline{{H}}_3 (\mathbf{A})+\Omega _{11} -b_{01} ({A}_1 )\nonumber \\&\quad +\frac{E_{11} \cos \Gamma [2b_{01} (A_1 )-b_{21} (A_1 )]}{[4(\alpha _1 A_1 ^{2}+\omega _1^2 )]} \end{aligned}$$
(63)
$$\begin{aligned} F_1 (A_1 ,\Gamma )&= \frac{E_1 \sin \Gamma \left[ 2b_{01} (A_1 )\!-\!b_{21} (A_1 )\right] }{[4(\alpha _1 A_1 ^{2}\!+\!\omega _{10}^2 )]} \nonumber \\&-\,\frac{A_1 \left[ \beta _{10} (16\omega _1^2 \!+\!10\alpha _1 A_1^2 )\!+\!A_1^2 \beta _{11} (4\omega _1^2 \!+\!3\alpha A_1^2 )\!+\!A_2^2 \beta _{12} (16\omega _1^2 \!+\!10\alpha A_1^2 )\right] }{[32(\omega _1^2 \!+\!\alpha _1 A_1^2 )]} \\ F_2 (A_2 )&= -\frac{A_2 \left[ \beta _{20} (16\omega _2^2 +10\alpha _2 A_2^2 )+A_1^2 \beta _{21} (16\omega _2^2 +10\alpha _2 A_2^2 )+A_2^2 \beta _{22} (4\omega _2^2 +3\alpha _2 A_2^2 )\right] }{[32(\omega _2^2 +\alpha _2 A_2^2 )]} \end{aligned}$$
$$\begin{aligned} \overline{{H}}_i (A_i )&= \overline{{m}}_{i1} +\overline{{m}}_{i2} +\overline{{m}}_{i3} +\overline{{m}}_{i4} \quad (i=1,2),\\ \overline{{H}}_3 (A_1 )&= \overline{{m}}_{31} +\overline{{m}}_{32} +\overline{{m}}_{33} +\overline{{m}}_{34} ,\\ \overline{{m}}_{i1}&= \overline{{m}}_{i11} S_{i1} (\omega _i (A_i))+\overline{{m}}_{i13} S_{i1} (3\omega _i (A_i))\\&+\, \overline{{m}}_{i15} S_{i1} (5\omega _i (A_i))+\overline{{m}}_{i17} S_{i1} (7\omega _i (A_i )),\\ \overline{{m}}_{i3}&= \overline{{m}}_{i31} S_{i1} (\omega _i (A_i ))+\overline{{m}}_{i33} S_{i1} (3\omega _i (A_i ))\\&+\,\overline{{m}}_{i35} S_{i1} (5\omega _i (A_i ))+\overline{{m}}_{i37} S_{i1} (7\omega _i (A_i )),\\ \overline{{m}}_{i2}&= \overline{{m}}_{i22} S_{i2} (2\omega _i (A_i ))+\overline{{m}}_{i24} S_{i2} (4\omega _i (A_i ))\\&+\,\overline{{m}}_{i26} S_{i2} (6\omega _i (A_i ))+\overline{{m}}_{i28} S_{i2} (8\omega _i (A_i )),\\ \overline{{m}}_{i4}&= \overline{{m}}_{i42} S_{i2} (2\omega _i (A_i ))+\overline{{m}}_{i44} S_{i2} (4\omega _i (A_i ))\\&+\,\overline{{m}}_{i46} S_{i2} (6\omega _i (A_i ))+\overline{{m}}_{i48} S_{i2} (8\omega _i (A_i )),\\ \overline{{m}}_{31}&= \overline{{m}}_{311} I_1 (\omega _i (A_i ))+\overline{{m}}_{313} I_1 (3\omega _i (A_i ))\\&+\,\overline{{m}}_{315} I_1 (5\omega _i (A_i ))+\overline{{m}}_{317} I_1 (7\omega _i (A_i )),\\ \overline{{m}}_{33}&= \overline{{m}}_{331} I_1 (\omega _i (A_i ))+\overline{{m}}_{333} I_1 (3\omega _i (A_i ))\\&+\,\overline{{m}}_{335} I_1 (5\omega _i (A_i ))+\overline{{m}}_{337} I_1 (7\omega _i (A_i )),\\ \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{32}&= \overline{{m}}_{322} I_2 (2\omega _i (A_i ))+\overline{{m}}_{324} I_2 (4\omega _i (A_i ))\\&+\,\overline{{m}}_{326} I_2 (6\omega _i (A_i ))+\overline{{m}}_{328} I_2 (8\omega _i (A_i )),\\ \overline{{m}}_{34}&= \overline{{m}}_{342} I_2 (2\omega _i (A_i ))+\overline{{m}}_{344} I_2 (4\omega _i (A_i ))\\&+\,\overline{{m}}_{346} I_2 (6\omega _i (A_i ))+\overline{{m}}_{348} I_2 (8\omega _i (A_i )),\\ I_i (\omega )&= \frac{D_{1i} }{\omega _{1i} }\frac{\omega }{\omega ^{2}+\omega _{1i}^2 }\quad { (i=1,2)} \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{i11}&= \pi \left[ b_{2i} (A_i )-2b_{0i} (A_i )\right] \frac{\left\{ 2\alpha A_i \left[ 2b_{0i} (A_i )-b_{2i} (A_i )\right] +(A_i^2 \alpha _i +\omega _i^2 )\left[ db_{2i} (A_i )/dA_i -2(db_{0i} (A_i )/dA_i )\right] \right\} }{\left[ 8(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] }\\ \overline{{m}}_{i13}&= \pi [b_{2i} (A_i )-b_{4i} (A_i )] \frac{\left\{ 2\alpha A_i [b_{4i} (A_i )-b_{2i} (A_i )]+(A_i^2 \alpha _i +\omega _i^2 )\left[ db_{2i} (A_i )/dA_i -2(db_{4i} (A_i )/dA_i )\right] \right\} }{[8(A_i^2 \alpha _i +\omega _i^2 )^{3}]} \\ \overline{{m}}_{i15}&= \pi \left[ b_{4i} (A_i )-2b_{6i} (A_i )\right] \frac{\left\{ 2\alpha A_i \left[ b_{6i} (A_i )-b_{4i} (A_i )\right] +(A_i^2 \alpha _i +\omega _i^2 )\left[ db_{4i} (A_i )/dA_i -2(db_{6i} (A_i )/dA_i )\right] \right\} }{\left[ 8(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] } \\ \overline{{m}}_{i17}&= \frac{\pi b_{6i} (A_i )\left\{ -2\alpha A_i b_{6i} (A_i )+(A_i^2 \alpha _i +\omega _i^2 )(db_{6i} (A_i )/dA_i )\right\} }{\left[ 8(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] }\\ \overline{{m}}_{i22}&= \pi A_i \left[ 2b_{0i} (A_i )-b_{4i} (A_i )\right] \frac{\left\{ \left[ 2b_{0i} (A_i )-b_{4i} (A_i )\right] (A_i^2 \alpha _i -\omega _i^2 )-A_i (A_i^2 \alpha _i -\omega _i^2 )\left[ 2db_{0i} (A_i )/dA_i -(db_{4i} (A_i )/dA_i )\right] \right\} }{\left[ 32(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] } \\ \overline{{m}}_{i24}&= \pi A_i \left[ b_{2i} (A_i )-b_{6i} (A_i )\right] \frac{\left\{ \left[ b_{6i} (A_i )-b_{2i} (A_i )\right] (A_i^2 \alpha _i -\omega _i^2 )-A_i (A_i^2 \alpha _i -\omega _i^2 )\left[ db_{2i} (A_i )/dA_i -(db_{6i} (A_i )/dA_i )\right] \right\} }{\left[ 32(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] } \\ \overline{{m}}_{i26}&= \frac{\pi A_i b_{4i} (A_i )\left\{ b_{4i} (A_i )(A_i^2 \alpha _i -\omega _i^2 )-A_i (A_i^2 \alpha _i -\omega _i^2 )(db_{4i} (A_i )/dA_i )\right\} }{\left[ 32(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] }\\ \overline{{m}}_{i28}&= \frac{\pi A_i b_{6i} (A_i )\left\{ b_{6i} (A_i )(A_i^2 \alpha _i -\omega _i^2 )-A_i (A_i^2 \alpha _i -\omega _i^2 )(db_{6i} (A_i )/dA_i )\right\} }{\left[ 32(A_i^2 \alpha _i +\omega _i^2 )^{3}\right] } \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{i31}&= \frac{-\pi [b_{2i}^2 (A_i )-4b_{0i}^2 (A_i )]}{[8A_i (A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{i33}&= \frac{-3\pi [b_{4i}^2 (A_i )-b_{2i}^2 (A_i )]}{[8A_i (A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{i35}&= \frac{-5\pi [b_{6i}^2 (A_i )-b_{4i}^2 (A_i )]}{[8A_i (A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{i37}&= \frac{7\pi b_{6i}^2 (A_i )}{[8A_i (A_i^2 \alpha _i +\omega _i^2 )^{2}]}, \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{i42}&= \frac{\pi A_i [2b_{0i} (A_i )-b_{4i} (A_i )][2b_{0i} (A_i )+2b_{2i} (A_i )+b_{4i} (A_i )]}{[16(A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{i44}&= \frac{\pi A_i [b_{2i} (A_i )-b_{6i} (A_i )][b_{2i} (A_i )+2b_{4i} (A_i )+b_{6i} (A_i )]}{[8(A_i^2 \alpha _i +\omega _i^2 )^{2}]}, \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{i46}&= \frac{3\pi A_i b_{4i} (A_i )[b_{4i} (A_i )+2b_{6i} (A_i )]}{[16(A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{i48}&= \frac{\pi A_i b_{6i}^2 (A_i )}{[4(A_i^2 \alpha _i +\omega _i^2 )^{2}]},\\ \overline{{m}}_{311}&= \frac{[2b_{01} (A_1 )+b_{21} (A_1 )]^{2}}{[8A_1^2 (A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{313}&= \frac{3[b_{21} (A_1 )+b_{41} (A_1 )]^{2}}{[8A_1^2 (A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \overline{{m}}_{315}&= \frac{5[b_{41} (A_1 )+b_{61} (A_1 )]^{2}}{[8A_1^2 (A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \overline{{m}}_{317}&= \frac{7[b_{61} (A_1 )]^{2}}{[8A_1^2 (A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \overline{{m}}_{322}&= \frac{[2b_{01} (A_1 )+2b_{21} (A_1 )+b_{41} (A_1 )]^{2}}{[16(A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \overline{{m}}_{324}&= \frac{[b_{21} (A_1 )+2b_{41} (A_1 )+b_{61} (A_1 )]^{2}}{[8(A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{326}&= \frac{3[b_{41} (A_1 )+2b_{61} (A_1 )]^{2}}{[16(A_1^2 \alpha _1 +\omega _1^2 )^{2}]},\\ \overline{{m}}_{328}&= \frac{[b_{61} (A_1 )]^{2}}{[4(A_1^2 \alpha _1 +\omega _1^2 )^{2}]}, \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{331}&= \left[ 2b_{01} (A_1 )-b_{21} (A_1 )\right] \frac{\left\{ -\left[ b_{21} (A_1 )+2b_{01} (A_1 )\right] (3A_1^2 \alpha _1 +\omega _1^2 )+A_1 (\alpha _1 A_1^2 +\omega _1^2 )\left[ 2(db_{01} (A_1 )/dA_1 )+(db_{21} (A_1 )/dA_1 )\right] \right\} }{\left[ 8(A_1^2 \alpha _1 +\omega _1^2 )^{3}\right] }, \\ \overline{{m}}_{333}&= \left[ b_{21} (A_1 )-b_{41} (A_1 )\right] \frac{\left\{ -\left[ b_{21} (A_1 )+b_{41} (A_1 )\right] (3A_1^2 \alpha _1 +\omega _1^2 )+A_1 (\alpha _1 A_1^2 +\omega _1^2 )\left[ (db_{21} (A_1 )/dA_1 )+(db_{41} (A_1 )/dA_1 )\right] \right\} }{\left[ 8(A_1^2 \alpha _1 +\omega _1^2 )^{3}\right] }, \\ \overline{{m}}_{335}&= \left[ b_{41} (A_1 )-b_{61} (A_1 )\right] \frac{\left\{ -\left[ b_{41} (A_1 )+b_{61} (A_1 )\right] (3A_1^2 \alpha _1 +\omega _1^2 )+A_1 (\alpha _1 A_1^2 +\omega _1^2 )\left[ (db_{41} (A_1 )/dA_1 )+(db_{61} (A_1 )/dA_1 )\right] \right\} }{\left[ 8(A_1^2 \alpha _1 +\omega _1^2 )^{3}\right] }, \\ \overline{{m}}_{337}&= \frac{b_{61} (A_1 )\left\{ -b_{61} (A_1 )(3A_1^2 \alpha _1 +\omega _1^2 )+A_1 (\alpha _1 A_1^2 +\omega _1^2 )\left[ (db_{61} (A_1 )/dA_1 )\right] \right\} }{\left[ 8(A_1^2 \alpha _1 +\omega _1^2 )^{3}\right] } \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{342}&= \big \{-2A_1 \alpha _1 \left[ 2b_{01} (A_1 )+2b_{21} (A_1 )+b_{41} (A_1 )\right] \\&+\,(A_1^2 \alpha _1 +\omega _1^2 )\Bigg [\frac{2db_{01} (A_1 )}{dA_1 }+\frac{2db_{21} (A_1 )}{dA_1 }\\&+\frac{2db_{41} (A_1 )}{dA_1 }\Bigg ] \frac{A_1 [(2b_{01} (A_1 )-b_{41} (A_1 )\big \}}{[32(A_1^2 \alpha _1 +\omega _1^2 )^{3}]} \\ \overline{{m}}_{344}&= \big \{-2A_1 \alpha _1 [b_{21} (A_1 )+2b_{41} (A_1 )+b_{61} (A_1 )]\\&+\,(A_1^2 \alpha _1 +\omega _1^2 )\Bigg [\frac{2db_{21} (A_1 )}{dA_1 }+\frac{2db_{41} (A_1 )}{dA_1 }\\&+\,\frac{2db_{61} (A_1 )}{dA_1 }\Bigg ] \frac{A_1 [(2b_{21} (A_1 )-b_{61} (A_1 )\big \}}{[32(A_1^2 \alpha _1 +\omega _1^2 )^{3}]} \end{aligned}$$
$$\begin{aligned} \overline{{m}}_{346}&= Ab_{41} (A_1 ) \times \frac{\left\{ -2A_1 \alpha _1 [b_{41} (A_1 )+2b_{61} (A_1 )]+(A_1^2 \alpha _1 +\omega _1^2 )[(db_{41} (A_1 )/dA_1 +(2db_{61} (A_1 )/dA_1 )]\right\} }{[32(A_1^2 \alpha _1 +\omega _1^2 )^{3}]}, \\ \overline{{m}}_{348}&= \frac{A_1 b_{61} (A_1 )\left\{ -2A_1 \alpha _1 b_{61} (A_1 )+(A_1^2 \alpha _1 +\omega _1^2 )db_{61} (A_1 )/dA_1 )\right\} }{[32(A_1^2 \alpha _1 +\omega _1^2 )^{3}]}, \end{aligned}$$
$$\begin{aligned} \overline{{b}}_{ii}&= \overline{{b}}_{ii1} +\overline{{b}}_{ii2} , \quad (i=1,2)\end{aligned}$$
(64)
$$\begin{aligned} \overline{{b}}_{33}&= \overline{{b}}_{331} +\overline{{b}}_{332} , \end{aligned}$$
(65)
$$\begin{aligned}&\overline{{b}}_{ij} =0, \quad i\ne j,\\ \overline{{b}}_{ii1}&= \overline{{b}}_{ii11} S_1 (\omega _i (A_i ))+\overline{{b}}_{ii13} S_1 (3\omega _i (A_i))\\&\quad +\,\overline{{b}}_{ii15} S_1 (5\omega _i (A_i ))+\overline{{b}}_{ii17} S_1 (7\omega _i (A_i )),\\ \overline{{b}}_{ii2}&= \overline{{b}}_{ii22} S_2 (\omega _i (A_i ))+\overline{{b}}_{ii24} S_2 (4\omega _i (A_i))\\&\quad +\,\overline{{b}}_{ii26} S_1 (6\omega _i (A_i ))+\overline{{b}}_{ii28} S_1 (8\omega _i (A_i )),\\ \overline{{b}}_{331}&= \overline{{b}}_{3311} S_1 (\omega _1 (A_1 ))+\overline{{b}}_{3313} S_1 (3\omega _1 (A_1))\\&\quad +\,\overline{{b}}_{3315} S_1 (5\omega _1 (A_1 ))+\overline{{b}}_{3317} S_1 (7\omega _1 (A_1 )),\\ \overline{{b}}_{333}&= \overline{{b}}_{3330} S_2 (0)\!+\!\overline{{b}}_{3332} S_2 (2\omega _1 (A_1 ))\!+\!\overline{{b}}_{3334} S_2 (4\omega _1 (A_1))\\&\quad +\,\overline{{b}}_{3336} S_2 (6\omega _1 (A_1 ))+\overline{{b}}_{3338} S_2 (8\omega _1 (A_1 )),\\ \end{aligned}$$
$$\begin{aligned} b_{ii11}&= \frac{\pi [b_{2i} (A_i )-2b_{0i} (A_i )]^{2}}{[4(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii13}&= \frac{\pi [b_{2i} (A_i )-2b_{4i} (A_i )]^{2}}{[4(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii15}&= \frac{\pi [b_{4i} (A_i )-b_{6i} (A_i )]^{2}}{[4(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii17}&= \frac{\pi b_{6i}^2 (A_i )}{[4(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii22}&= \frac{\pi A_i^2 [b_{4i} (A_i )-2b_{0i} (A_i )]^{2}}{[16(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii24}&= \frac{\pi A_i^2 [b_{2i} (A_i )-b_{6i} (A_i )]^{2}}{[16(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii26}&= \frac{\pi A_i^2 b_{4i}^2 (A_i )}{[16(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{ii28}&= \frac{\pi A_i^2 b_{6i}^2 (A_i )}{[16(\alpha _i A_i^2 +\omega _{0i}^2 )^{2}]},\\ b_{3311}&= \frac{\pi [2b_{02} (A_1 )+b_{22} (A_1 )]^{2}}{[4A_1^2 (\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3313}&= \frac{\pi [b_{21} (A_1 )+b_{41} (A_1 )]^{2}}{[4A_1^2 (\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},,\\ b_{3315}&= \frac{\pi [b_{41} (A_1 )+b_{61} (A_1 )]^{2}}{[4A_1^2 (\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3317}&= \frac{\pi b_{61} ^{2}(A_1 )}{[4A_1^2 (\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3320}&= \frac{\pi [2b_{01} (A_1 )+b_{21} (A_1 )]^{2}}{[8(\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3322}&= \frac{\pi [2b_{01} (A_1 )+2b_{21} (A_1 )+b_{41} (A_1 )]^{2}}{[16(\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3324}&= \frac{\pi [b_{21} (A_1 )+2b_{41} (A_1 )+b_{61} (A_1 )]^{2}}{[16(\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3326}&= \frac{\pi [b_{41} (A_1 )+2b_{61} (A_1 )]^{2}}{[16(\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]},\\ b_{3328}&= \frac{\pi b_{61}^2 (A_1 )}{[16(\alpha _1 A_1^2 +\omega _{01}^2 )^{2}]}. \end{aligned}$$

Appendix 2: An approximate solution of final dynamical programming equation

For ergodic control of 2-DOF Duffing system, the final dynamical programming equation is of the form:

$$\begin{aligned} \eta _c&= L_1 (\mathbf{H},\Gamma )+\tilde{m}_1 (\mathbf{H},\Gamma )\frac{\partial V}{\partial H_1 }+\tilde{m}_2 (\mathbf{H})\frac{\partial V}{\partial H_2 }\nonumber \\&\quad +\,\tilde{m}_3 (\mathbf{H},\Gamma )\frac{\partial V}{\partial \Gamma }-\frac{m_{11} (\mathbf{H})}{4R_1 }\left( {\frac{\partial V}{\partial H_1 }} \right) ^{2} \nonumber \\&\quad -\,\frac{m_{22} (\mathbf{H})}{4R_2 }\left( {\frac{\partial V}{\partial H_2 }} \right) ^{2}-\frac{m_{33} (\mathbf{H})}{4R_1 }\left( {\frac{\partial V}{\partial \Gamma }} \right) ^{2}\nonumber \\&\quad +\,\frac{b_{11} }{2}\frac{\partial ^{2}V}{\partial H_1^2 }+\frac{b_{22} }{2}\frac{\partial ^{2}V}{\partial H_2^2 }+\frac{b_{33} }{2}\frac{\partial ^{2}V}{\partial \Gamma ^{2}} \end{aligned}$$
(66)

Suppose that \(L_1 (\mathbf{H},\Gamma )\) can be expanded into Fourier series with respect to\(\Gamma \),

$$\begin{aligned} L_1 (\mathbf{H},\Gamma )&= L_{10} (\mathbf{H})+\sum _{i=1}^\infty (L_{1i}^c (\mathbf{H})\cos i\Gamma \nonumber \\&\quad +\, L_{1i}^s (\mathbf{H})\sin i\Gamma ) \end{aligned}$$
(67)

Then the solution of Eq. (66) can be assumed of the form of the following Fourier series:

$$\begin{aligned} V(\mathbf{H},\Gamma )=V_0 (\mathbf{H})+\sum _{i=1}^\infty {(V_i^c (\mathbf{H})\cos i\Gamma +V_i^s (\mathbf{H})\sin i\Gamma )}\nonumber \\ \end{aligned}$$
(68)

Note that coefficients \(\tilde{m}_1 , \tilde{m}_2 \) and \(\tilde{m}_3 \) in Eq. (54) are of the form

$$\begin{aligned} \tilde{m}_1 (\mathbf{H},\Gamma )&= m_{10} (\mathbf{H})+m_{11}^s (\mathbf{H})\sin \Gamma \nonumber \\ \tilde{m}_2 (\mathbf{H})&= m_{20} (\mathbf{H})\nonumber \\ \tilde{m}_3 (\mathbf{H},\Gamma )&= m_{30} (\mathbf{H})+m_{31}^c (\mathbf{H})\cos \Gamma \end{aligned}$$
(69)

and \(m_{i0} (i=1,2,3),m_{11}^s ,m_{31}^c \) are all independent of \(\Gamma \). Substituting Eqs. (67)–(69), the following series of equation can be obtained:

$$\begin{aligned}&\eta _c =L_{10} (\mathbf{H})+m_{10} (\mathbf{H})\frac{\partial V_0 }{\partial H_1 }+m_{20} (\mathbf{H})\frac{\partial V_0 }{\partial H_2 }\nonumber \\&\quad -\,\frac{1}{2}m_{11}^s (\mathbf{H})\frac{\partial V_1^c }{\partial H_1 }+\frac{1}{2}m_{31}^c (\mathbf{H})V_1^s\nonumber \\&\quad -\,\frac{m_{11} (\mathbf{H})}{4R_1 }\!\left\{ {\left( \!{\frac{\partial V_0 }{\partial H_1 }}\!\right) ^{2}\!+\!\frac{1}{2}\sum _{i=1}^\infty {\left[ \!\!{\left( \!{\frac{\partial V_i^c }{\partial H_1 }} \right) ^{2}\!+\! \left( \!{\frac{\partial V_i^s }{\partial H_1 }} \right) ^{2}}\!\right] }}\!\right\} \nonumber \\&\quad -\,\frac{m_{22} (\mathbf{H})}{4R_2 }\!\left\{ {\left( \! {\frac{\partial V_0 }{\partial H_2 }} \!\right) ^{2}\!+\!\frac{1}{2}\sum _{i=1}^\infty {\!\left[ {\!\left( \! {\frac{\partial V_i^c }{\partial H_2 }} \!\right) ^{2}\!+\!\left( \! {\frac{\partial V_i^s }{\partial H_2 }} \!\right) ^{2}} \!\right] } }\!\right\} \nonumber \\&\quad -\,\frac{m_{33} (\mathbf{H})}{8R_1 }\!\left[ {\sum _{i=1}^\infty {i^{2}\!\left( \!{V_i^{c^{2}}\!+\!V_i^{s^{2}} } \right) } }\!\right] \!+\!\frac{b_{11} }{2}\frac{\partial ^{2}V_0 }{\partial H_1^2 }\!+\!\frac{b_{22} }{2}\frac{\partial ^{2}V_0 }{\partial H_2^2 }\nonumber \\\end{aligned}$$
(70)
$$\begin{aligned}&0=L_{11}^c (\mathbf{H})+m_{10} (\mathbf{H})\frac{\partial V_1^c }{\partial H_1 }+m_{20} (\mathbf{H})\frac{\partial V_1^c }{\partial H_2 }\nonumber \\&\quad +\,m_{30} (\mathbf{H})V_1^s (\mathbf{H})+\frac{1}{2}m_{11}^s (\mathbf{H})\frac{\partial V_2^c }{\partial H_1 }+m_{31}^c (\mathbf{H})V_2^s \nonumber \\&\quad -\,\frac{m_{11} (\mathbf{H})}{4R_1 }\!\left[ {2\frac{\partial V_0 }{\partial H_1 }\frac{\partial V_1^c }{\partial H_1 }\!+\!\!\sum _{i=1}^\infty {\left( {\frac{\partial V_i^c }{\partial H_1 }\frac{\partial V_{i+1}^c }{\partial H_1 }+\!\frac{\partial V_i^s }{\partial H_1 }\frac{\partial V_{i+1}^s }{\partial H_1 }} \right) }}\! \right] \nonumber \\&\quad -\,\frac{m_{22} (\mathbf{H})}{4R_2 }\!\left[ {2\frac{\partial V_0 }{\partial H_2 }\frac{\partial V_1^c }{\partial H_2 }\!+\!\!\sum _{i=1}^\infty {\!\left( {\frac{\partial V_i^c }{\partial H_2 }\frac{\partial V_{i+1}^c }{\partial H_2 }\!+\!\frac{\partial V_i^s }{\partial H_2 }\frac{\partial V_{i+1}^s }{\partial H_2 }} \right) }} \!\right] \nonumber \\&\quad -\,\frac{m_{33} (\mathbf{H})}{4R_1 }\left[ {\sum _{i=1}^\infty {i(i+1)\left( {V_i^c V_{i+1}^c +V_i^s V_{i+1}^s } \right) } } \right] \nonumber \\&\quad +\,\frac{b_{11} }{2}\frac{\partial ^{2}V_1^c }{\partial H_1^2 }+\frac{b_{22} }{2}\frac{\partial ^{2}V_1^c }{\partial H_2^2 }-\frac{b_{33} }{2}V_1^c \end{aligned}$$
(71)
$$\begin{aligned}&0=L_{11}^c (\mathbf{H})+m_{10} (\mathbf{H})\frac{\partial V_1^s }{\partial H_1 }+m_{20} (\mathbf{H})\frac{\partial V_1^s }{\partial H_2 }\nonumber \\&\quad -\, m_{30} (\mathbf{H})V_1^c (\mathbf{H})+m_{11}^s (\mathbf{H})\frac{\partial V_0 }{\partial H_1 }\nonumber \\&+\frac{1}{2}m_{11}^s (\mathbf{H})\frac{\partial V_2^c }{\partial H_1 }-m_{31}^c (\mathbf{H})V_2^c \nonumber \\&\quad -\,\frac{m_{11} (\mathbf{H})}{4R_1 }\!\left[ \!{2\frac{\partial V_0 }{\partial H_1 }\frac{\partial V_1^s }{\partial H_1 }\!+\!\!\sum _{i=1}^\infty {\left( {\frac{\partial V_i^c }{\partial H_1 }\frac{\partial V_{i+1}^s }{\partial H_1 }+\frac{\partial V_i^s }{\partial H_1 }\frac{\partial V_{i+1}^c }{\partial H_1 }} \!\right) } } \!\!\right] \nonumber \\&\quad -\,\frac{m_{22} (\mathbf{H})}{4R_2 }\!\left[ \!{2\frac{\partial V_0 }{\partial H_2 }\frac{\partial V_1^s }{\partial H_2 }\!+\!\!\sum _{i=1}^\infty {\left( {\frac{\partial V_i^c }{\partial H_2 }\frac{\partial V_{i+1}^s }{\partial H_2 }+\!\frac{\partial V_i^s }{\partial H_2 }\frac{\partial V_{i+1}^c }{\partial H_2 }} \right) } }\! \!\right] \nonumber \\&\quad -\,\frac{m_{33} (\mathbf{H})}{4R_1 }\left[ {\sum _{i=1}^\infty {i(i+1)\left( {V_i^s V_{i+1}^c +V_i^c V_{i+1}^s } \right) } } \right] \nonumber \\&\quad +\,\frac{b_{11} }{2}\frac{\partial ^{2}V_1^s }{\partial H_1^2 }+\frac{b_{22} }{2}\frac{\partial ^{2}V_1^s }{\partial H_2^2 }-\frac{b_{33} }{2}V_1^s \end{aligned}$$
(72)

The higher harmonic terms in the solution can be neglected. An approximated solution of Eq. (66)

$$\begin{aligned} V(\mathbf{H},\Gamma )\!=\!V_0 (\mathbf{H})\!+\!V_1^c (\mathbf{H})\cos \Gamma \!+\! V_1^s (\mathbf{H})\sin \Gamma \end{aligned}$$
(73)

To evaluate \(V_0 (\mathbf{H}), V_1^c (\mathbf{H})\) and \(V_1^s (\mathbf{H})\) from the coupled differential equations (70)–(72), the Fourier coefficients are further expanded into Taylor series:

$$\begin{aligned} m_{10}&= m_{101} H_1 +m_{102} H_1^3 +m_{103} H_2^2 H_1 +\cdots \nonumber \\ m_{20}&= m_{201} H_2 +m_{202} H_1^2 H_2 +m_{203} H_2^3 +\cdots \nonumber \\ m_{30}&= m_{300} +m_{302} H_1^2 +\cdots \nonumber \\ m_{11}^s&= m_{111}^s +m_{113}^s H_1^2 +\cdots \nonumber \\ m_{31}^c&= m_{311}^c H_1^{-1} +m_{313}^c H_1 +\cdots \nonumber \\ m_{11}&= m_{110} +m_{112} H_1^2 +\cdots \nonumber \\ m_{22}&= m_{220} +m_{222} H_2^2 +\cdots \nonumber \\ m_{33}&= m_{330} H_1^{-2} +m_{332} +\cdots \nonumber \\ b_{11}&= b_{110} +b_{112} H_1^2 +\cdots \nonumber \\ b_{22}&= b_{220} +b_{222} H_2^2 +\cdots \nonumber \\ b_{33} :&= b_{330} H_1^{-2} +b_{332} +\cdots \end{aligned}$$
(74)
$$\begin{aligned} L_{10}&= L_{100} H_1^4 +L_{101} H_1^3 +(L_{102} H_2^2 +L_{103} )H_1^2\nonumber \\&+\, L_{104} H_1 +L_{105} H_2^4 +L_{106} H_2^2 +L_{107} +\cdots \nonumber \\ L_{11}^c&= L_{200} H_1^4 +(L_{201} H_2^2 +L_{202} )H_1^2 +L_{203} +\cdots \nonumber \\ L_{11}^s&= L_{300} H_1^4 +L_{301} H_1^3 +(L_{302} H_2^2\nonumber \\&+\, L_{303} )H_1^2 +L_{304} H_1 +L_{305} +\cdots \end{aligned}$$
(75)
$$\begin{aligned} V_0&= V_{011} H_1 +V_{012} H_2 +V_{021} H_1^2\nonumber \\&+V_{022} H_2^2 +V_{023} H_1 H_2 +\cdots \nonumber \\ V_1^c&= V_{11}^c H_1 +V_{12}^c H_2 +V_{13}^c H_1^2\nonumber \\&+V_{14}^c H_2^2 +V_{15}^c H_1 H_2 +\cdots \nonumber \\ V_1^s&= V_{11}^s H_1 +V_{12}^s H_2 +V_{13}^s H_1^2\nonumber \\&+V_{14}^s H_2^2 +V_{15}^s H_1 H_2 +\cdots \end{aligned}$$
(76)

Substituting Eqs. (74)–(76) into Eqs. (70)–(72)and letting the coefficients of the same power of \(H_1 ,H_2 \) vanish yield the coefficients of polynomials \(V_0,V_1^c \) and \(V_1^s \).

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hu, R.C., Zhu, W.Q. Stochastic optimal control of MDOF nonlinear systems under combined harmonic and wide-band noise excitations. Nonlinear Dyn 79, 1115–1129 (2015). https://doi.org/10.1007/s11071-014-1727-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11071-014-1727-0

Keywords

Navigation