The main problem oil and gas leakage in pipeline detection is that the pipeline distance is too long, and the effect of traditional detection mode is not good. A detection method of oil and gas leakage in pipeline using conjugate gradient echo state network(ESN) based on optical fiber technology is proposed. Firstly, the real-time data of long-distance oil and gas pipeline is monitored and collected by optical fiber detection technology, and the dimension compression method of unmarked data is used to reduce the dimension and simplify the data by combining the data series method and rough set algorithm. Secondly, ESN is selected to analyze and process the dimension reduced oil and gas pipeline leakage data. The weight of ESN is optimized preliminarily, and then the preliminary weight results are further optimized by using the semi positive qualitative of Hessian matrix to realize the recognition ability of ESN to the leakage data of oil and gas pipelines which is based on regularization. Finally, through the simulation of the measured data, the proposed leakage monitoring model of oil and gas pipeline has higher recognition accuracy and calculation efficiency.