Existing methods for diagnosing fault in hydraulic component face significant challenges when dealing with dynamic conditions. This paper focuses on the study of leakage fault within hydraulic directional valve and proposes a fault diagnosis method based on Bayesian theory. Firstly, this method calculates signal characteristics. Then it selects fault features based on correlation analysis and introduces principal component analysis to construct leakage features. Finally, by utilizing Bayesian model and Markov chain Monte Carlo sampling iteration to estimate the parameters of the leakage model, the diagnosis of leakage faults in hydraulic directional valve is achieved. Experimental results demonstrate that compared to existing fault diagnosis models, this method exhibits higher robustness and accuracy under dynamic pressure conditions, while also avoiding the drawbacks of deep learning methods that rely on large amounts of fault data.