Abstarct: Wireless sensor networks (WSNs) have attracted special attention due to their wide applications in various fields such as environmental monitoring, intelligent transportation, medicine, and military. However, due to their dependence on scattered sensors and limited resources, these networks are prone to various errors, including data errors, which can negatively affect the accuracy and reliability of the system. Therefore, fast and accurate fault detection in these networks has been raised as a major challenge. In this study, an effective method for data error detection in wireless sensor networks baxsed on support vector machines (SVMs) optimized with the Marine Predators Algorithm (MPA) is presented. In the proposed method, two key hyperparameters of the SVM model, namely C and sigma, are tuned by the MPA optimization algorithm. The optimization process is designed to select the best values for these hyperparameters by increasing the accuracy of the model on the training data set. Then, the optimized model is evaluated for fault detection on the test data. The results obtained show that the proposed method has higher accuracy compared to similar methods, and its average accuracy reaches 99.67%. These results indicate that the combination of the marine predator algorithm with the support vector machine, as an effective approach in optimizing fault detection models, has a high potential in improving the accuracy and efficiency of diagnostic systems.