In order to maintain the integrity and safety of pipeline systems, structural health monitoring (SHM) is essential for continuous assessment. This study introduces a novel artificial intelligence (AI)-augmented active acoustic sensing technique for uninterrupted pipeline defect detection and localization. Conventional inspection methods, such as in-line inspection (ILI), are hindered by their accessibility, infrequent application, time-intensive data analysis, and the necessity for defect-specific tools. These conventional methods typically rely on intricate transducer/receiver configurations and traditional wave processing techniques, necessitating exact knowledge of pipeline geometry. In this study, we utilize judicious combinations of numerical and experimental tests to detect defects. The method aims to reduce the effects of wave dispersion and streamline the detection process. Finite element analysis (FEA) is a numerical technique used to investigate wave propagation dynamics within the pipe wall. Experimentally, an axial piezo transducer is used to introduce acoustic energy into the pipe wall, while multiple acoustic emission sensors are placed throughout the pipe test section that serve as distributed receivers. To simulate damage, a circumferential notch-type defect is created on the external surface of the pipe. The initial acoustic stimulation employs sinusoidal wave packets of 50 kHz and 70 kHz. However, these commonly used frequencies in defect assessment are known to attenuate quickly, posing challenges in accurately pinpointing defects. To address this, the natural frequencies of the system are identified, and two frequencies close to these resonant frequencies chosen as excitation sources. This innovative approach demonstrates effectiveness in both defect detection and localization tasks, showcasing its potential as a more reliable method. Acknowledging the velocity of longitudinal modes within the selected frequency range, we determine an optimal signal duration to counteract end pipe reflections. Each response collected from the pipe is subject to power spectral analysis, with the root mean square error (RMSE) computed against baseline references. The AI component of this method, anchored by a neural network model, not only enhances performance and adaptability but also, as shown in our study, achieves higher accuracies in both defect detection and localization tasks when contrasting conventional frequencies with those near the resonant frequencies of the system. The model is trained using diverse data, including signals from various locations and those with added noise. To manage variations in signal-to-noise ratio and wave propagation attenuation, Z-score normalization is applied to preprocess the RMSE data. Additionally, the study investigates the application of a highly sensitive Lock-in Amplifier, from which directly generated features, along with indirectly computed redundant features, introduce a novel approach in feature-based defect detection and localization tasks. These features are integrated into the AI component, enhancing the sensitivity of signal measurement. By achieving reasonable accuracy in detection and localization, this method proves crucial in scenarios where signals are completely obscured by noise, necessitating a more extended experimental setup. Thus, this AI-powered active acoustic sensing technique effectively differentiates between intact and damaged pipes and locates the defect within the range of two sensors, demonstrating its viability for practical use in pipeline integrity management. Furthermore, phase response analysis successfully pinpoints the exact location of the defect with reasonable accuracy, despite the presence of multiple reflections.