Abstract
Abstract
Identification of dynamical and hybrid systems using trajectory data is an important way to construct models for complex systems where derivation from first principles is too difficult. In this paper, we study the identification problem for switched dynamical systems with polynomial ODEs. This is a difficult problem as it combines estimating coefficients for nonlinear dynamics and determining boundaries between modes. We propose two different algorithms for this problem, depending on whether to perform prior segmentation of trajectories. For methods with prior segmentation, we present a heuristic segmentation algorithm and a way to classify themodes using clustering. Formethods without prior segmentation, we extend identification techniques for piecewise affine models to our problem. To estimate derivatives along the given trajectories, we use Linear MultistepMethods. Finally, we propose a way to evaluate an identified model by computing a relative difference between the predicted and actual derivatives. Based on this evaluation method, we perform experiments on five switched dynamical systems with different parameters, for a total of twenty cases. We also compare with three baseline methods: clustering with DBSCAN, standard optimization methods in SciPy and identification of ARX models in Matlab, as well as with state-of-the-art identification method for piecewise affine models. The experiments show that our two methods perform better across a wide range of situations.
- [ACZ+20] An J, Chen M, Zhan B, Zhan N, Zhang M (2020) Learning one-clock timed automata. In Proceedings of 26th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 444–462. SpringerGoogle Scholar
- [AL89] Algorithms for the optimal identification of segment neighborhoodsBulletin of Mathematical Biology1989511395497890210.1016/S0092-8240(89)80047-3Google ScholarCross Ref
- [Ang87] Learning regular sets from queries and counterexamplesInformation and computation19877528710691636010.1016/0890-5401(87)90052-6Google ScholarDigital Library
- [AS14] Alur R, Singhania N (2014) Precise piecewise affine models from input-output data. In Proceedings of the 14th International Conference on Embedded Software, pp. 3(1)–3(10)Google Scholar
- [AWZ+ss] Learning real-time automata2021SCIENCE CHINA Information SciencesIn pressGoogle Scholar
- [BDG+20] Mining shape expressions from positive examplesIEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems202039113809382010.1109/TCAD.2020.3012240Google ScholarCross Ref
- [BG08] Butcher JC, Goodwin N (2008) Numerical methods for ordinary differential equations, volume 2. Wiley Online LibraryGoogle Scholar
- [BGPV05] A bounded-error approach to piecewise affine system identificationIEEE Transactions on Automatic Control2005501015671580217187410.1109/TAC.2005.856667Google ScholarCross Ref
- [BHKL09] Bollig B, Habermehl P, Kern C, Leucker M (2009) Angluin-style learning of NFA. In Proceedings of the 21st International Joint Conference on Artificial Intelligence, pp. 1004–1009Google Scholar
- [BIB11] Baptista R, Ishihara JY, Borges GA (2011) Split and merge algorithm for identification of piecewise affine systems. In Proceedings of the 2011 American Control Conference, pp. 2018–2023Google Scholar
- [BPK16] Discovering governing equations from data by sparse identification of nonlinear dynamical systemsProceedings of the National Academy of Sciences20161131539323937349408110.1073/pnas.1517384113Google ScholarCross Ref
- [BVVB05] Borges J, Verdult V, Verhaegen M, Botto MA (2005) A switching detection method based on projected subspace classification. In Proceedings of the 44th IEEE Conference on Decision and Control, pp. 344–349Google Scholar
- [DD17] Drews S, DAntoni L (2017) Learning symbolic automata. In Proceedings of the 23rd International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 173–189. SpringerGoogle Scholar
- [Eyk74] System Identification and State Estimation1974Wiley0709.93520Google Scholar
- [FCC+08] Farzan A, Chen Y-F, Clarke EM, Tsay Y-K, Wang B-Y (2008) Extending automated compositional verification to the full class of omega-regular languages. In Proceedings of the 14th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 2–17. SpringerGoogle Scholar
- [FMLM03] A clustering technique for the identification of piecewise affine systemsAutomatica2003392205217213694210.1016/S0005-1098(02)00224-8Google ScholarDigital Library
- [GAG+00] Physiobank, physiotoolkit, and physionet: Components of a new research resource for complex physiologic signalsCirculation202010123e215e220Google Scholar
- [GPV12] A survey on switched and piecewise affine system identificationIFAC Proceedings Volumes2012451634435510.3182/20120711-3-BE-2027.00332Google ScholarCross Ref
- [Hen96] Henzinger TA (1996) The theory of hybrid automata. In Proceedings of the 11th Annual IEEE Symposium on Logic in Computer Science, pp. 278–292. IEEE Computer SocietyGoogle Scholar
- [HV05] Hashambhoy Y, Vidal R (2005) Recursive identification of switched ARX models with unknown number of models and unknown orders. In Proceedings of the 44th IEEE Conference on Decision and Control, pp. 6115–6121. IEEEGoogle Scholar
- [KD19] Keller R, Du Q (2019) Discovery of dynamics using linear multistep methodsGoogle Scholar
- [LB08] Lauer F, Bloch G (2008) Switched and piecewise nonlinear hybrid system identification. In Proceedings of the 11th International Workshop on Hybrid Systems: Computation and Control, pp. 330–343. SpringerGoogle Scholar
- [LBG18] Lamrani I, Banerjee A, Gupta SKS (2018) Hymn: mining linear hybrid automata from input output traces of cyber-physical systems. In Proceedings of the 2018 IEEE Industrial Cyber-Physical Systems, pp. 264–269. IEEEGoogle Scholar
- [LBV10] Lauer F, Bloch G, Vidal R (2010) Nonlinear hybrid system identification with kernel models. In Proceedings of the 49th IEEE Conference on Decision and Control, pp. 696–701. IEEEGoogle Scholar
- [LCZL17] Li Y, Chen Y-F, Zhang L, Liu D (2017) A novel learning algorithm for büchi automata based on family of dfas and classification trees. In Proceedings of the 23rd International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 208–226, Berlin, Heidelberg, Springer Berlin HeidelbergGoogle Scholar
- [Lju99] System Identification: Theory for the User1999Upper Saddle River, New JerseyPrentice Hall0615.93004Google ScholarDigital Library
- [Lju15] System identification: An overview2015In Encyclopedia of Systems and ControlSpringerGoogle Scholar
- [Lor63] Deterministic nonperiodic flowJournal of Atmospheric Sciences1963202130148402143410.1175/1520-0469(1963)020<0130:DNF>2.0.CO;2Google ScholarCross Ref
- [MKY11] Mueen A, Keogh E, Young N (2011) Logical-shapelets: an expressive primitive for time series classification. In Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 1154–1162Google Scholar
- [MM14] Maler O, Mens I-E (2014) Learning regular languages over large alphabets. In Proceedings of the 20th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 485–499. SpringerGoogle Scholar
- [Moo58] Gedanken-experiments on sequential machinesAnnals of Mathematics19583412915378059Google Scholar
- [MRBF15] Medhat R, Ramesh S, Bonakdarpour B, Fischmeister S (2015) A framework for mining hybrid automata from input/output traces. In Proceedings of the 2015 International Conference on Embedded Software, pp. 177–186. IEEEGoogle Scholar
- [NQF+19] Ničković D, Qin X, Ferrère T, Mateis C, Deshmukh J (2019) Shape expressions for specifying and extracting signal features. In Proceedings of the 19th International Conference on Runtime Verification, pp. 292–309. SpringerGoogle Scholar
- [NSV+12] Niggemann O, Stein , Vodencarevic A, Maier A, Büning HK (2012) Learning behavior models for hybrid timed systems. In Proceedings of the 26th AAAI Conference on Artificial Intelligence. AAAI PressGoogle Scholar
- [OG92] Oncina J, Garcia P (1992) Identifying regular languages in polynomial time. In Advances in Structural and Syntactic Pattern Recognition, Volumn 5 of Series in Machine Perception and Artificial Intelligence, pp. 99–108. World ScientificGoogle Scholar
- [OLB10] Segmentation of arx-models using sum-of-norms regularizationAutomatica201046611071111287719510.1016/j.automatica.2010.03.013Google ScholarDigital Library
- [Oza16] Ozay N (2016) An exact and efficient algorithm for segmentation of ARX models. In Proceedings of the 2016 American Control Conference, pp. 38–41. IEEEGoogle Scholar
- [Pen55] A generalized inverse for matricesMathematical Proceedings of the Cambridge Philosophical Society195551340641310.1017/S0305004100030401Google ScholarCross Ref
- [PJFV07] Paoletti S, Juloski ALj (2007) Giancarlo Ferrari-Trecate, and René Vidal. Identification of hybrid systems a tutorial. European journal of control, 13(2-3):242–260Google Scholar
- [POTN03] Peng H, Ozaki T, Toyoda Y, Nakano K (2003) Nonlinear system modelling using the rbf neural network-based regressive model. IFAC Proceedings Volumes, 36(16):333–338, Proceedings of the 13th IFAC Symposium on System IdentificationGoogle Scholar
- [SG09] Shahbaz M, Groz R (2009) Inferring mealy machines. In Proceedings of the 16th International Symposium on Formal Methods, pp. 207–222. SpringerGoogle Scholar
- [SHSZ19] Soto MG, Henzinger TA, Schilling C, Zeleznik L (2019) Membership-based synthesis of linear hybrid automata. In Proceedings of the 31st International Conference on Computer Aided Verification, pp. 297–314. SpringerGoogle Scholar
- [TAB+19] Tappler M, Aichernig BK, Bacci G, Eichlseder M, Larsen KG (2019) -based learning of Markov decision processes. In Proceedings of the 23rd International Symposium on Formal Methods, pp. 651–669Google Scholar
- [Vaa17] Model learningCommunications of the ACM2017602869510.1145/2967606Google ScholarDigital Library
- [VdWW11] The efficiency of identifying timed automata and the power of clocksInformation and Computation20112093606625275281410.1016/j.ic.2010.11.023Google ScholarDigital Library
- [VdWW12] Efficiently identifying deterministic real-time automata from labeled dataMachine Learning2012863295333289752610.1007/s10994-011-5265-4Google ScholarDigital Library
- [XPTP20] DBN based SD-ARX model for nonlinear time series prediction and analysisApplied Intelligence202050124586460110.1007/s10489-020-01804-2Google ScholarDigital Library
Recommendations
Two classes of implicit-explicit multistep methods for nonlinear stiff initial-value problems
The initial value problems of nonlinear ordinary differential equations which contain stiff and nonstiff terms often arise from many applications. In order to reduce the computation cost, implicit-explicit (IMEX) methods are often applied to these ...
Numerical Integration of Differential Algebraic Systems and Invariant Manifolds
AbstractThe dynamics of a differential algebraic equation takes place on a lower dimensional manifold in phase space. Applying a numerical integration scheme, it is natural to ask if and how this geometric property is preserved by the discrete dynamical ...
Universal theory of dynamical chaos in nonlinear dissipative systems of differential equations
The article presents a new universal theory of dynamical chaos in nonlinear dissipative systems of differential equations, including autonomous and nonautonomous ordinary differential equations (ODE), partial differential equations, and delay ...
Comments