Bayesian system identification of a nonlinear dynamical system using a novel variant of Simulated Annealing. (February 2015)

• Record Type: Journal Article
• Authors: Green, P.L.
• Is Part Of: Mechanical systems and signal processing. Volume 52/53(2015)
• Page Start: 133
• Page End: 146
• Publication Date: 2015-02
• Subjects: Bayesian model updating -- Nonlinear system identification -- Markov chain Monte Carlo -- Simulated Annealing -- Deviance Information Criterion; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2014.07.010
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: This work details the Bayesian identification of a nonlinear dynamical system using a novel MCMC algorithm: 'Data Annealing'. Data Annealing is similar to Simulated Annealing in that it allows the Markov chain to easily clear 'local traps' in the target distribution. To achieve this, training data is fed into the likelihood such that its influence over the posterior is introduced gradually - this allows the annealing procedure to be conducted with reduced computational expense. Additionally, Data Annealing uses a proposal distribution which allows it to conduct a local search accompanied by occasional long jumps, reducing the chance that it will become stuck in local traps. Here it is used to identify an experimental nonlinear system. The resulting Markov chains are used to approximate the covariance matrices of the parameters in a set of competing models before the issue of model selection is tackled using the Deviance Information Criterion. Abstract : Highlights: Bayesian system identification of a nonlinear dynamical system using a novel MCMC algorithm. The new algorithm – Data Annealing – is computationally cheap and requires little manual tuning. Data Annealing is applied to the system identification of a nonlinear dynamical system. This process is repeated using several competing model structures. The issue of model selection is addressed using the Deviance Information Criterion.