Bridges, pipelines and roads can be engineering objects. As a result, the basis of the material for each of them is different, as well as certain parameters of their strength. The main defects considered in the work are corrosion and cracks. Based on the statistical analysis of these parameters and machine learning classifiers, it is possible to predict the defect or not. Due to the insufficient amount of known data, the formation of the model requires more complex dependencies. The described methodology can be used to complement the diagnosis of changes in the state of surface defects of engineering structures. The advanced model includes the set parameters of the components of the main mixture, the presence of protection of materials and selected main ones that affect the rate of change in the state of damage to objects. The most accurate prediction 96% for the collected data was found in the Decision Tree Classifier. The accuracy of the forecasting of the collected parameters makes it possible to assess the criticality of the system’s condition and the planning of restoration works in the studied areas. Thus, in such monitoring or diagnostic systems, it is possible to predict and track a year of defects in structures. The obtained results can be useful for improving the durability and checking the condition of the structures under study.