The invention provides a system and method for monitoring the pipeline leakage based on high-and-low-frequency hybrid detection. The system comprises a head end ultrahigh frequency sensor, an end ultrahigh frequency sensor, a head end acceleration sensor, an end acceleration sensor, a head end lower computer, an end lower computer, a head end upper computer and an end upper computer; a method based on the system is also provided, the method comprises the following steps of: acquiring high-frequency and low-frequency mixed historical data and classifying the working conditions of the high-frequency and low-frequency mixed historical data, performing preprocessing, multi-information domain analysis processing and feature extraction, processing by using a covariance kernel slice inverse regression algorithm to obtain a feature vector subspace, acquiring high-frequency and low-frequency mixed real-time data, obtaining the feature vector subspace according to the steps of acquiring the historical data, classifying the feature vector subspace by using a linear discriminant analysis, calculating the similarity, obtaining the real-time data classification result, and if the leakage exists,calculating the leakage position, providing a reliable data basis for the troubleshooting. The accuracy of the leakage feature value is ensured, the false alarm is reduced to the maximum extent, andmore accurate judgment is made on the leakage of a small flow.