The deep-sea environment in which offshore booster stations are located is complex and variable, and is highly susceptible to natural factors such as waves and fog. A single sensor is easily affected by external factors, resulting in abnormal or lost data, making it difficult to fully capture the uncertainty in the actual inspection process, leading to problems such as cruise trajectory deviation from expectations and significant root mean square error. Research on unmanned intelligent inspection methods for offshore booster stations based on multi-sensor fusion technology. Firstly, the observation data is denoised using the forgetting factor, and the optimal weight is assigned based on the real-time deviation of each sensor through dynamic weighting method to improve the fusion accuracy. Subsequently, the trustworthiness of sensor data is evaluated through error matrix and confidence threshold, abnormal data is eliminated, and root mean square error is reduced. Finally, based on the correlation function and weight factor calculation, the multi-sensor data is weighted and combined to output a global state estimation value, completing intelligent inspection. According to the experimental results, the coordinates of points A, B, and C inspected by this method are consistent with the expected inspection trajectory, and the root mean square error is ultimately reduced to 0.01, which can provide a more solid guarantee for the safe and stable operation of offshore energy facilities.