Use of a novel autoassociative neural network for nonlinear steady‐state data reconciliation

Du, Yang-Guang; Hodouin, D.; Thibault, Jules

doi:10.1002/aic.690430714

Cited by 17 publications

(7 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In network training, the input and the target vectors presented to the network are identical and the objective function, consisting of the mean squared errors between the network outputs and its inputs, is minimized. After successful training, an AANN can be regarded as a filter used to perform steady‐state data reconciliation 10, 11…”

Section: Aann‐based Ddr Algorithmmentioning

confidence: 99%

Autoassociative neural networks for robust dynamic data reconciliation

2007

Self Cite

View full text Add to dashboard Cite

in Wiley InterScience (www.interscience.wiley.com).Reliable estimation of process variables for plant monitoring and control is an important topic that has been studied extensively. The Kalman filter has often been used and has acquired an enviable reputation. However, use of the Kalman filter suffers from two restrictive conditions: (1) it requires state-space models and (2) it has to be tuned online to achieve its best performance. Recently, an alternative methodology based on dynamic data reconciliation has been proposed to overcome the first restriction. Although the approach of dynamic data reconciliation can incorporate any form of model, it involves online optimization that may require long computation time for complex systems. This report explores a new methodology based on a combination of autoassociative neural networks (AANNs) and dynamic data reconciliation that overcomes the need for online tuning, as required by the Kalman filter, as well as online optimization as required by conventional dynamic data reconciliation methods. Simulation examples of a distillation column demonstrate that the AANN-based dynamic data reconciliation approach is capable of effectively attenuating measurement noise and is robust to changes of the noise level in plant measurements and the loss of measurements.

show abstract

Section: Aann‐based Ddr Algorithmmentioning

confidence: 99%

Autoassociative neural networks for robust dynamic data reconciliation

2007

Self Cite

View full text Add to dashboard Cite

show abstract

“…The most advantage of neural network is that (1) we do not need to know the physical mechanism of a target system, (2) any system regardless of linear or nonlinear can be modeled, and (3) it has robustness; especially the application of the AANN was remarkable [17], [18]. The AANN does auto-association function.…”

Section: Neural Network Model With Pcamentioning

confidence: 99%

“…These methodologies have the selection technique of input parameters, special training technique, and network structure to optimize training time and accuracy. Especially the application of the auto-associative neural network (AANN) was useful from the viewpoint of the combination of the neural networks and principal component analysis (PCA) [17], [18].…”

Section: Introductionmentioning

confidence: 99%

Comparative study on state analysis of BOP in NPPs

Heo

Chang

2003

IEEE Trans. Nucl. Sci.

View full text Add to dashboard Cite

Under the deregulation environment, the performance enhancement of balance of plant (BOP) or turbine cycle in nuclear power plants is being highlighted. To analyze performance level of turbine cycle, we use the performance test procedures provided from an authorized institution such as American Society of Mechanical Engineers. However, through plant investigation, it was proved that the requirements of the performance test procedures about the reliability and quantity of sensors was difficult to be satisfied. As a solution of this weakness, a state analysis method that is the expanded concept of signal validation, was proposed on the basis of the statistical and the neural network approaches. Authors recommended the statistical linear regression model by analyzing correlation among turbine cycle parameters as a reference state analysis method. Its advantage is that its derivation is not heuristic, it is possible to calculate model uncertainty, and it is easy to apply to an actual plant. The error of the statistical linear regression model is below 3% under normal as well as abnormal system states. Additionally the auto-associative neural network (AANN) model was recommended since the statistical model is impossible to apply to the validation of all of the sensors and is sensitive to the outlier that is the signal located out of statistical distribution. Because there are a lot of sensors need to be validated in turbine cycle, principal component analysis (PCA) and wavelet analysis (WA) were applied as a pre-processor for the reduction of input dimension and for the enhancement of training accuracy, and the results were compared. Both are effective to reduce the number of the input and output dimension, but WA is superior to PCA in the compensation of the outliers. The outlier localization capability of WA enhanced the robustness of the AANN. In case of the AANN with WA, the training accuracy of the model was nearly 100% and the trained neural network restored the degraded signals to the values within 3% of the true signals.Index Terms-Autoassociative neural network, principal component analysis, state analysis, turbine cycle, wavelet analysis.

show abstract

“…Reconciliation techniques are useful to calculate and display performance indices, such as concentrate grade and recovery, which, due to their better reliability, may improve manual or automatic process performance optimisation (Bazin & Franklin, 1996;Hodouin, Bazin, & Makni, 1997). Reconciliation methods based on artificial neural networks have also been proposed, but they do not offer either the same rigorous statistical and physical background or the same analytical tools for the evaluation of the results reliability (Du, Hodouin, & Thibault, 1997b;Aldrich & van Deventer, 1994).…”

Section: Data Reconciliationmentioning

confidence: 99%

State of the art and challenges in mineral processing control

Hodouin

Jämsa-Jounela²,

Carvalho

et al. 2000

IFAC Proceedings Volumes

View full text Add to dashboard Cite

The objective of process control in the mineral industry is to optimise the recovery of the valuable minerals, while maintaining the quality of the concentrates delivered to the metal extraction plants. The paper presents a survey of the control approaches for ore size reduction and mineral separation processes. The present limitations of the measurement instrumentation are discussed, as well as the methods to upgrade the information delivered by the sensors. In practice, the overall economic optimisation goal must be hierarchically decomposed into simpler control problems. Model-based and AI methods are reviewed, mainly for grinding and flotation processes, and classified as mature, active or emerging.

show abstract

Use of a novel autoassociative neural network for nonlinear steady‐state data reconciliation

Cited by 17 publications

References 36 publications

Autoassociative neural networks for robust dynamic data reconciliation

Autoassociative neural networks for robust dynamic data reconciliation

Comparative study on state analysis of BOP in NPPs

State of the art and challenges in mineral processing control

Contact Info

Product

Resources

About