Improving fault detection and diagnosis (FDD) in systems with incomplete features is a prevalent challenge in industrial applications. In this context, two distinct approaches (k-means clustering and k-nearest neighbor imputation KNN) for addressing Missing Data (MD) were presented and compared. The evaluation was conducted with introducing of MD, ranging from a 10% to 50%. The proposals aim to enhance the monitoring and diagnosis of the system through the utilization of principal component analysis (PCA) for feature extraction an fault detection.