The process of extracting modal parameters using vibration response data from aerospace, civil and mechanical structures is well-established and many techniques exist to cater for the availability of spatial and temporal data. These techniques need extensive interaction with an expert user to guide them towards an acceptable set of solutions and are not adequate for structural health monitoring which fundamentally requires an automated process. Research into automated algorithms for the extraction and tracking of modal parameters started to gather momentum recently due to advances in technology and computing. Currently there is a lack of automated procedures due to the difficulty of replacing the interactions of an expert user with software algorithms and those that have been proposed have not yet been widely adopted. In this thesis, we propose a new automated method to track resonant frequencies for the purpose of detecting change. The method uses wavelet decomposition, principal component analysis, spectrum estimation and adaptive filtering. The aim is to identify resonant frequencies and then to monitor their magnitudes and frequencies in an automated fashion without user interaction for the detection of change in performance. The proposed method is validated on several benchmark problems widely studied in the literature, one simulated and four experimental. It is shown that using the new method it is possible to detect all the data cases for these benchmark structures because they produce changes in the resonant frequencies or in their magnitudes. The new method is also compared with an existing automated method called frequency domain decomposition (FDD) and it is shown that for the benchmark problems considered in this thesis the frequency tracking performance of the new method is superior.