Neural-network-based method of calibration and measurand reconstruction for a high-pressure measuring system

Massicotte, Daniel; Legendre, S.; Barwicz, A.

doi:10.1109/19.744175

Cited by 17 publications

(9 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consequently, even a small number of training data sets render good detection accuracy compared to standard matrix inversion method [10]. However, for the high nonlinearity and cross-sensitivity system, a large number of training data sets are required to obtain good detection accuracy [11].…”

Section: Resultsmentioning

confidence: 99%

Application of an artificial neural network for simultaneous measurement of bending curvature and temperature with long period fiber gratings

Sun

Chan

Tan

et al. 2007

Sensors and Actuators A: Physical

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

Application of an artificial neural network for simultaneous measurement of bending curvature and temperature with long period fiber gratings

Sun

Chan

Tan

et al. 2007

Sensors and Actuators A: Physical

View full text Add to dashboard Cite

“…Static approaches based on the current value of the signal of interest i y(in) only [16], [17], [18], or also on the measurement of a dis-where ci and c<* are the optimization parameters, k(.,.) is a turbance signal [19], [20], [21], as well as dynamic compensa-suitable kernel function, C > 0 and vi > 0 are further paramtion approaches [22], [23] have been proposed in the literature.…”

Section: Sensor Compensationmentioning

confidence: 99%

Uncertainty-Complexity Trade-Offs for Sensor Compensation Design

Gubian

Marconato

Boni

et al. 2007

2007 IEEE International Workshop on Advanced Methods for Uncertainty Estimation in Measurement

View full text Add to dashboard Cite

In this work we focus on the design of reduced-complexity unit and the transceiver is an almost mandatory solution. Thus, sensor compensation modules based on learning-from-examples tech-system complexity optimization becomes a major concern at niques. A Multi-Objective Optimization design framework is pro-design time. The design of an SVR-based compensation alposed, where system complexity and compensation uncertainty are gorithm that performs as accurately as possible while keeping considered as two conflicting costs to be jointly minimized. In addi-a low implementation cost is a problem that can be tackled in tion, suitable statistical techniques are applied to cope with the vari-a Multi-Objective Optimization (MOO) framework [5], once ability in the uncertainty estimation arising from the limited avail-precise indexes for compensation accuracy (or uncertainty) and ability of data at design time. Experimental results on a synthetic for complexity are defined.benchmark are provided to show the validity of the proposed methodology.In this paper, a synthetic dataset for sensor compensation Keywords -sensor compensation, support vector machines, multi-[6] will be used as a case study in order to introduce the MOO objective optimization design framework [7]. The classic mean squared error (mse) will be used to evaluate compensation uncertainty, while a spe-I. INTRODUCTION cific PIC microcontroller SVR implementation will be used asRecently, measurement systems such as Wireless Sensor target architecture on which an implementation complexity inNetworks (WSNs) are increasingly gaining popularity as one dex will be defined. Design space exploration is speeded up of the most interesting examples of ubiquitous sensing tech-by employing the successful technique of Genetic Algorithms nologies [1]. Thanks to their ability to sense the environment, (GAs). In this framework, results will be shown and insights process collected data and exchange information via wireless on the inherent trade-offs will be given. communication among neighbor nodes or with the outside world, WSNs can in fact be viewed as a promising solution for a wide and heterogeneous range of real world applications.As a second contribution of the work, we will focus our atOne of the most critical issues that needs to be dealt with tention on the estimation uncertainty that arises from the scarse in the design of sensor networks is related to the intrinsic be-availability of labeled data. At design time, mse is estimated havior of the sensor itself. In fact, the dynamic characteristic on a sequence of sensor data of which the desired values are of the sensor can be altered by non-linearities and memory dis-known (i.e. the labels). Labeled data usually come from a torsions, whose effects can be neutralized only by employing calibration bank or from some other time consuming measurespecific compensation strategies. Usually, sensor compensa-ment process, and cannot be produced in large quantity at low tion techniques based on inverse modeling criteria are consid...

show abstract

“…Classical dynamic error correction algorithms are usually characterized by high complexity of numerical operations, in particular in the case of describing the transducer dynamics by means of higher order differential equations. ANN as "universal approximators" [12,13,14] have been widely used for transducer static error correction [15,16,17,18], in particular for transducer and measuring instrument calibration [19,20,21]. Nevertheless, in the field of realtime dynamic error correction, solutions using DSP [22,23,24], FPGA technique [25] and analog circuits [26,27] are dominant.…”

Section: Introductionmentioning

confidence: 99%

Neural Network Based Real-time Correction of Transducer Dynamic Errors

Roj¹

2013

Measurement Science Review

View full text Add to dashboard Cite

In order to carry out real-time dynamic error correction of transducers described by a linear differential equation, a novel recurrent neural network was developed. The network structure is based on solving this equation with respect to the input quantity when using the state variables. It is shown that such a real-time correction can be carried out using simple linear perceptrons. Due to the use of a neural technique, knowledge of the dynamic parameters of the transducer is not necessary. Theoretical considerations are illustrated by the results of simulation studies performed for the modeled second order transducer. The most important properties of the neural dynamic error correction, when emphasizing the fundamental advantages and disadvantages, are discussed.

show abstract

Neural-network-based method of calibration and measurand reconstruction for a high-pressure measuring system

Cited by 17 publications

References 13 publications

Application of an artificial neural network for simultaneous measurement of bending curvature and temperature with long period fiber gratings

Application of an artificial neural network for simultaneous measurement of bending curvature and temperature with long period fiber gratings

Uncertainty-Complexity Trade-Offs for Sensor Compensation Design

Neural Network Based Real-time Correction of Transducer Dynamic Errors

Contact Info

Product

Resources

About