Abstract
The accurate and robust speed prediction of high-speed trains (HSTs) is a challenging task in the automatic train operation (ATO) because HSTs operate in an open-air situation with much noise and many uncertainties. This paper contributes to the development of robust speed prediction methods for the ATO of HSTs based on improved echo state networks (ESNs). Firstly, an adaptive temporal scale selection approach is introduced to improve the accuracy and efficiency of ESN modeling, due to the importance of proper temporal scale selection in relation to the sound prediction performance of ESNs. Also, a random weight scaling mechanism is employed to enhance the feasibility and robustness of the proposed method, as the learning ability of ESNs lies in the constrained random connection weights. Furthermore, several conditions for the stability of the closed-loop control system are given. Our experiment results demonstrate that the proposed method successfully achieves sound performance in terms of speed prediction accuracy, efficiency and robustness.
Similar content being viewed by others
References
Yang H, Fu YT, Wang DH (2018) Multi-ANFIS model based synchronous tracking control of high-speed electric multiple unit. IEEE Trans Fuzzy Syst 26(3):1472–1484
Dong HR, Ning B, Cai BG, Hou ZS (2010) Automatic train control system development and simulation for high-speed railways. IEEE Circ Syst Mag 10(2):6–18
Fu YT, Yang H, Wang DH (2017) Real-time optimal control of tracking running for high-speed electric multiple unit. Inf Sci 376:202–215
Sun HQ, Hou ZS, Li DY (2012) Coordinated iterative learning control schemes for train trajectory tracking with overspeed protection. IEEE Trans Autom Sci Eng 10(2):323–333
Yang H, Fu YT, Zhang KP, Li ZQ (2014) Speed tracking control using an ANFIS model for high-speed electric multiple unit. Control Eng Pract 23:57–65
Ji HH, Hou ZS, Zhang RK (2015) Adaptive iterative learning control for high-speed trains with unknown speed delays and input saturations. IEEE Trans Autom Sci Eng 13(1):260–273
Song Q, Song YD, Cai WC (2011) Adaptive backstepping control of train systems with traction/braking dynamics and uncertain resistive forces. Vehicle Syst Dyn 49(9):1441–1454
Song Q, Song YD, Tang T, Ning B (2011) Computationally inexpensive tracking control of high-speed trains with traction/braking saturation. IEEE Trans Intell Transp Syst 12(4):1116–1125
Feng XS (2011) Optimization of target speeds of high-speed railway trains for traction energy saving and transport efficiency improvement. Energy Policy 39(12):7658–7665
Yang H, Liu HE, Fu YT (2015) Multi-objective operation optimization for electric multiple unit-based on speed restriction mutation. Neurocomputing 169:383–391
Raghunathan RS, Kim HD, Setoguchi T (2002) Aerodynamics of high-speed railway train. Prog Aerosp Sci 38(6–7):469–514
Baker C (2010) The flow around high speed trains. J Wind Eng Ind Aerod 98(6–7):277–298
Song Q, Song YD (2011) Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures. IEEE Trans Neural Netw Learn Syst 22(12):2250–2261
Gao RZ, Wang YJ, Lai JF, Gao H (2016) Neuro-adaptive fault-tolerant control of high speed trains under traction-braking failures using self-structuring neural networks. Inf Sci 367:449–462
Yang H, Zhang KP, Liu HE (2016) Online regulation of high speed train trajectory control based on TS fuzzy bilinear model. IEEE Trans Intell Transp Syst 17(6):1496–1508
Jaeger H, Haas H (2004) Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless communication. Science 304(5667):78–80
Xing K, Wang Y, Zhu Q, Zhou H (2012) Modeling and control of McKibben artificial muscle enhanced with echo state networks. Control Eng Pract 20(5):477–488
Li Q, Zhou W, Xia XH (2018) Estimate and characterize PV power at demand-side hybrid system. Appl Energy 218:66–77
Jaeger H (2001) The “echo state” approach to analysing and training recurrent neural networks-with an erratum note, Bonn, Germany: German National Research Center for Information Technology GMD Technical Report, vol 34, p 13
Jaeger H, LukosEvicIus M, Popovici D, Siewert U (2007) Optimization and applications of echo state networks with leaky-integrator neurons. Neural Netw 20(3):335–352
McDermott PL, Wikle CK (2017) An ensemble quadratic echo state network for nonlinear spatio-temporal forecasting. Stat 6(1):315–330
Buonomano DV, Merzenich MM (1995) Temporal information transformed into a spatial code by a neural network with realistic properties. Science 267(5200):1028–1030
Reshef DN, Reshef YA, Finucane HK et al (2011) Detecting novel associations in large data sets. Science 334(6062):1518–1524
Bomberger JD, Seborg DE (1998) Determination of model order for NARX models directly from input–output data. J Process Control 8(5–6):459–468
Feil B, Abonyi J, Szeifert F (2004) Model order selection of nonlinear input–output models-a clustering based approach. J Process Control 14(6):593–602
Wang DH, Li M (2017) Stochastic configuration networks: fundamentals and algorithms. IEEE Trans Cybern 47(10):3466–3479
Wang DH, Liu HE, Yang H (2020) Improved echo state networks with scaling random weight matrices, Technical Report LTU-2020, 18 May 2020
Buehner M, Young P (2006) A tighter bound for the echo state property. IEEE Trans Neural Netw Learn Syst 17(3):820–824
Zhang B, Miller DJ, Wang Y (2012) Nonlinear system modeling with random matrices: echo state networks revisited. IEEE Trans Neural Netw Learn Syst 23(1):175–182
Ozturk MC, Xu D, Príncipe JC (2007) Analysis and design of echo state networks. Neural Comput 19(1):111–138
Strauss T, Wustlich W, Labahn R (2012) Design strategies for weight matrices of echo state networks. Neural Comput 24(12):3246–3276
Funahashi KI, Nakamura Y (1993) Approximation of dynamical systems by continuous time recurrent neural networks. Neural Netw 6(6):801–806
LukosEvicIus M, Jaeger H (2009) Reservoir computing approaches to recurrent neural network training. Comput Sci Rev 3(3):127–149
El Hihi S, Bengio Y (1996) Hierarchical recurrent neural networks for long-term dependencies, In: Advances in neural information processing systems, pp 493–499
Gath I, Geva AB (1989) Unsupervised optimal fuzzy clustering. IEEE Trans Pattern Anal 11(7):773–780
Clerc M, Kennedy J (2002) The particle swarm-explosion, stability, and convergence in a multi-dimensional complex space. IEEE Trans Evolut Comput 6(1):58–73
Hu M, Wu T, Weir JD (2013) An adaptive particle swarm optimization with multiple adaptive methods. IEEE Trans Evolut Comput 17(5):705–720
Jiang M, Luo YP, Yang SY (2007) Stochastic convergence analysis and parameter selection of the standard particle swarm optimization algorithm. Inf Process Lett 102(1):8–16
Wang SY (1993) Existence of a Pareto equilibrium. J Optim Theory Appl 79(2):373–384
Valerio D, Trujillo JJ, Rivero M, Machado JT, Baleanu D (2013) Fractional calculus: a survey of useful formulas. Eur Phys J Spec Top 222(8):1827–1846
Rakkiyappan R, Cao J, Velmurugan G (2015) Existence and uniform stability analysis of fractional-order complex-valued neural networks with time delays. IEEE Trans Neural Netw Learn Syst 26(1):84–97
Pahnehkolaei SM, Alfi A, Machado JT (2017) Uniform stability of fractional-order leaky integrator echo state neural network with multiple time delays. Inf Sci 418:703–716
Acknowledgements
This work was supported in part by the National Natural Science Foundation of China under Grant 61673172, 61663013, 61733005, by State Key Laboratory of Synthetical Automation for Process Industries under Grant PAL-N201801.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A
Appendix A
Proof
Suppose \(\mathbf{{x}}(t) = [{x_1}(t),{x_2}(t), \ldots ,{x_n}(t)]\), \({\mathbf{{x}}^\mathrm{{o}}}(t) = [{x^o}_1(t),{x^o}_2(t), \ldots ,{x^o}_n(t)]\) are two solutions of the system defined in (33), \({x_i}(t) = \xi (t):[ - {t_d},0] \rightarrow \mathbf{{R}}\) and \({x_i}^ \circ (t) = \eta (t):[ - {t_d},0] \rightarrow \mathbf{{R}}\), \(\mathrm{{i = 1,2,}}\ldots \mathrm{{,n}}\), where \(\xi (t)\) and \(\eta (t)\) denote two different initial conditions for the system. Besides, let \(\gamma \) and \(\sigma \) be two time measuring scales satisfying that \(\gamma - \phi _1\sigma = t - k\tau \) and \(\gamma - \phi _2\sigma = t-\mu \tau \) with \(\phi _1, \phi _2 \in \mathbf{Z} \). According to Definition 1 and (33), we have
Here, let us set \(\alpha \in (0,1)\). Then, from Assumption 1, Definition 2 and the properties of fractional operators [41, 42], we have
where e is the system tracking error. Define \(\theta = \gamma - {\phi _2}\sigma \), thus (38) can be rewritten as follows:
Let \(\beta = t - \theta - {\phi _2}\sigma \) and extend the integrals of (38), then (39) can be rewritten as
From Euler’s gamma function,
then (40) can be rewritten as
Since \({e^{ - {\phi _2}\sigma }} \in (0,1]\), it derives from (42) that
It follows that
thus (43) can be rewritten as
Then, we have
Subsequently, we can obtain
Using (34), we get
Thus, for any \(\varepsilon > 0\), there is
such that if \(\left\| {{\varvec{\xi }}(t) - {\varvec{\eta }}(t)} \right\| \le \delta \), then
holds. Therefore, according to Definition 3, the system is uniformly stable and it converges to the equilibrium point with arbitrary precision. The proof is finished.
Rights and permissions
About this article
Cite this article
Liu, H., Yang, H. & Wang, D. Robust speed prediction of high-speed trains based on improved echo state networks. Neural Comput & Applic 33, 2351–2367 (2021). https://doi.org/10.1007/s00521-020-05096-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-020-05096-y