Measurement correction method for force sensor used in dynamic pressure calibration based on artificial neural network optimized by genetic algorithm
- 4 October 2017
- journal article
- Published by IOP Publishing in Metrologia
- Vol. 54 (6) , 810-820
Abstract
We present an optimization algorithm to obtain low-uncertainty dynamic pressure measurements from a force-transducer-based device. In this paper, the advantages and disadvantages of the methods that are commonly used to measure the propellant powder gas pressure, the applicable scope of dynamic pressure calibration devices, and the shortcomings of the traditional comparison calibration method based on the drop-weight device are firstly analysed in detail. Then, a dynamic calibration method for measuring pressure using a force sensor based on a drop-weight device is introduced. This method can effectively save time when many pressure sensors are calibrated simultaneously and extend the life of expensive reference sensors. However, the force sensor is installed between the drop-weight and the hammerhead by transition pieces through the connection mode of bolt fastening, which causes adverse effects such as additional pretightening and inertia forces. To solve these effects, the influence mechanisms of the pretightening force, the inertia force and other influence factors on the force measurement are theoretically analysed. Then a measurement correction method for the force measurement is proposed based on an artificial neural network optimized by a genetic algorithm. The training and testing data sets are obtained from calibration tests, and the selection criteria for the key parameters of the correction model is discussed. The evaluation results for the test data show that the correction model can effectively improve the force measurement accuracy of the force sensor. Compared with the traditional high-accuracy comparison calibration method, the percentage difference of the impact-force-based measurement is less than 0.6% and the relative uncertainty of the corrected force value is 1.95%, which can meet the requirements of engineering applications.Keywords
Funding Information
- National Defense Basic Scientific Research Project (JSJL2014606B003)
This publication has 18 references indexed in Scilit:
- Linking dynamic to static pressure by laser interferometryMetrologia, 2013
- Experimental techniques for ballistic pressure measurements and recent development in means of calibrationJournal of Physics: Conference Series, 2013
- Application of artificial neural network–genetic algorithm (ANN–GA) to correlation of density in nanofluidsFluid Phase Equilibria, 2012
- Dynamic calibration method of high-pressure transducer based on quasi-δ function excitation sourceMeasurement, 2012
- Frequency analysis of a step dynamic pressure calibratorReview of Scientific Instruments, 2012
- Artificial neural network modeling to predict the high temperature flow behavior of an AZ81 magnesium alloyMaterials & Design (1980-2015), 2012
- Deformation Prediction of Landslide Based on Improved Back-propagation Neural NetworkCognitive Computation, 2012
- Hydraulic square-wave pressure generator with a specific rotating valveMeasurement, 2009
- Square-wave pressure generator using a novel rotating valveMetrologia, 1999
- Identification of Nonlinear Dynamic Systems Using Neural NetworksJournal of Applied Mechanics, 1993