This study deals with the health monitoring of fiber-reinforced composite panels using ultrasonic guided waves and a class of algorithms originally tested in the vibration (lower-acoustic) domain based on attractor based state space analysis involving auto- or cross-prediction methods. The examined components are representative of the wing skin-to-spar joints of unmanned aerial vehicles. The guided wave propagation problem is first studied analytically by a model that accounts for all relevant temperature-dependent parameters of a pitch/catch system on a carbon-reinforced epoxy composite plate. Elements included in the model are actuator/plate and plate/sensor interactions through shear lag behavior, piezoelectric and dielectric permittivity properties of the transducers and Lamb wave dispersion properties of the substrate plate. The model is used to predict the So response spectra in the composite plate for the temperature range of 233 to 313 K, which is typical of normal aircraft operational variability. Different types of bond discontinuities that include various disbond sizes and poorly cured adhesive are then examined experimentally over the range of temperatures using time series prediction algorithms combined with ultrasonic chaotic excitations. An active excitation signal is imparted to the structure through a macrofiber composite (MFC) patch on one side of the bonded joint and sensed using an equivalent MFC patch on the opposite side of the joint. There is an MFC actuator/sensor pair for each bond condition to be identified.