Powdery Mildew Blumeria graminis (PMBG) is one of the most dangerous diseases for winter wheat plants, causing damage to all above-ground plant organs. The main aim of this study is to develop and validate machine learning (ML) models with explainable AI (XAI) capabilities for accurate risk prediction of PMBG in winter wheat crops at the pre-symptomatic stage. Multiple heterogeneous ML classifiers with XAI for PMBG risk prediction have been developed in this study. The weather data used in this study were collected from two regions in Ukraine and included hourly air temperature, solar radiation, leaf wetness duration and other measurements of soil and climatic parameters. Several different feature selection approaches were leveraged to retrieve the most salient features. The multistack of ML models has been used to find the best-performing pipeline, which achieved an accuracy of 82 %. Further, diverse XAI methods such as Shapley Additive Values (SHAP), ELI5, Anchor and Local Interpretable Model-agnostic Explanations (LIME) have been applied to understand the model predictions. The precision, recall, f1-score and AUC obtained were 85%, 82%, 82% and 72 % respectively. As a result a decision support system has been developed to predict the risk of wheat powdery mildew using soil and climatic parameters monitoring, ML, and XAI. This study provides the holistic risk prediction of PMBG for the enhancement of wheat stress resistance during the full cycle of its cultivation in open-field conditions.