A Data Driven Algorithm for Compensating the Errors of Strain Gauges in Decision Support of Ship Hull Structures

Abstract

AbstractThe smart structure is one of the major frontiers of scientific and technological innovations in the area of engineering structures, and structure health monitoring (SHM) is the foundation of smart structures as it provides the vital perceptible information about the state of the structure itself and also that of the environment. For engineering structures, strain gauges are the most important sensors used to monitor the state of the structure, which, however, and including the fiber Bragg grating strain gauges used for long-term monitoring, are prone to the problems of zero-drift and temperature drift. Such problems not only result in misleading information about the state of the structure for the SHM but also lead to incorrect recommendations or decisions for the control and operation of smart structures. In this paper, the method for compensating the errors of strain gauges is investigated. To circumvent the problem of possible changes in error patterns of strain gauges over time, a BP neural network based data driven algorithm is devised so as to improve the accuracy of error compensation by means of updating utilizing the monitoring data of the structure in service. The algorithm is devised and validated for the structure health monitoring a ship, and based on which a decision support system is realized. For other types of engineering structure, data driven algorithm for error compensation can be established in the same manner to realize the structure health monitoring and decision support system.KeywordsStrain gaugeError compensationData driven algorithmNeural networkStructure health monitoring