Composite GW baseline enabled large range temperature compensation method based on singular value decomposition and ADALINE network

Abstract

Guided wave (GW) based structural health monitoring (SHM) technology has been widely researched and applied in many engineering fields, especially in the aerospace field. However, environmental and operational conditions (EOCs) of structure like temperature variation may cause significant influence on GW signals and reduce the monitoring accuracy and reliability, which has become a major factor that hinders this technology from real applications. This paper proposes for the first time a singular value decomposition (SVD) and adaptive filter linear neural (ADALINE) network based data-driven method to compensate temperature variation caused influence on GW signal. By extracting singular vector matrices from baseline signals to train the network, the traditional ADALINE method’s dependence on signal linear correlation is eliminated. The composite baseline set composed of the generated weight matrices and one selected stored baseline signal can achieve accurate temperature compensation over a large range temperature. In order to verify the feasibility of the proposed method, experimental validations are performed on a composite skin structure under −10 °C ∼ 50 °C temperature environment. By using a stored baseline signal at −10 °C and the corresponding weights, compensation signals are accurately generated at different temperatures, in which the maximum temperature interval reaches 60 °C. The maximum errors in amplitude and phase of the compensation signal are less than −38 dB and −50 dB, which also shows a great potential for larger temperature interval of compensation.