The basic information about the system

T he system is designed as a support in predicting water pump failures. The system is linked to the water supply system and its SCADA management system. The WPDSS uses daily pumps measurements in its prediction, conducts a prediction of possible pump failures on a weekly basis. The prediction system can be used to predict failure, regardless of the type and type of water pump. WSDDD work is based on several machine learning techniques. The WPDSS system is based on the implementation of several machine learning techniques. Prediction System has no restrictions on the size of water supply system and number of water pumps in the system.
The prediction framework has continual learning approach in its work, in order to increase the prediction accuracy.

System features:

• Adaptability of the prediction system to the size of the water supply system,
• Continual learning and improving the accuracy of the prediction,
• High prediction accuracy,
• The prediction system is adapted to work in the Internet and the cloud environment.

The WPDSS framework in its prediction process performs several activities in parallel (as shown in the picture). A prediction is performed on a weekly basis. The PDSS system daily records various measurements related to the operation of water pumps. Daily data is used to form a weekly review of each pump's performance. For the continuous monitoring of the pump operation, for a longer period, for each pump a system forms two-week case record. At the end of each week, the prediction system performs its work assessment and performs correction of internal parameters, if necessary.

Internal processes of prediction system