Identification of a Managed River Reach by a Bayesian Approach

Abstract

This paper considers the problem of identification, and more particularly of time-delay estimation, of a river reach managed to produce hydroelectric power. Difficulties lie in the obligation to use data collected during a combined feedback/feedforward control carried out by a human operator. We propose a Bayesian identification method, nonsupervised and simple to implement, estimating jointly the time-delay and a finite impulse response (FIR). It is based on the detection of an abrupt change in the FIR at a time equal to the time-delay. Experimental results show the effectiveness of the proposed method to estimate the river reach time-delay from data collected in imposed experimental conditions.

Generalized expectation–maximization approach to LPV process identification with randomly missing output data

“…The prior information that the impulse response (IR) of the local process tends to vary smoothly is employed to construct the prior distribution of the local model parameters p(θ m ) by considering the second derivatives of the parameters [19]. Here, the second derivative of the IR coefficient of the mth local model is assumed to follow the stationary Gaussian distribution with zero mean and variance α −1 m and the precision parameter α m is imposed to follow the Gamma distribution α m ∼ Gamma(c m , d m ).…”

“…In practice, the second derivative is approximately calculated by second difference. Define the matrix D as [19] Since no prior information of the noise variance σ 2 is available, p(σ 2 ) is set to be uniformly distributed.…”

“…(19) cannot be optimized directly due to the existence of unmeasurable noise-free output (or model output) x m k in ϕ m k . Here, the auxiliary model principle is used to cope with this problem [10] and the auxiliary model is constructed based on the estimated parameters in the sth…”

Model-Free Based Water Level Control for Hydroelectric Power Plants

“…Automatic water level control in open channels remains a challenge for researchers and engineers, as demonstrated by an abundant literature (see, e.g., Cantoni, Weyer, Li, Ooi, Mareels & Ryan (2007); Dang Van Mien (1999); Dumur, Libaux & Boucher (2001); Georges & Litrico (2002); de Halleux, Prieur, Coron, d'Andréa-Novel, Bastin, (2003); Litrico & Fromion (2009); Rabbani, Dorchies, Malaterre, Bayen & Litrico (2009); Thomassin, Bastogne & Richard (2009); Zhuan & Xia (2007)). The excellent review by Zhuan & Xia (2007), which is devoted to channel flow control methodologies, insists on the inherent complexity of this type of problems.…”

A probabilistic indirect adaptive control for systems with input‐dependent noise

“…The most general approach for solving the noise adaptive filtering was the Bayesian approach 1, 3, 4. Several applications of the Bayesian method have been discussed in the literature, examples are: the control of the basic oxygen furnace 5, traffic flow forecasting 6 and the estimation of the river reach time delay 7. However, for complex control problems it is insufficient to estimate parameters uncertainty only.…”