The paint shop pump monitoring project was implemented in a multinational car production company. As a solution, introduction of real-time monitoring of the condition and performance of pumps, detection of gradual quality degradation of the painting process, detection of a gradual increase in the oscillation of painting parameters and prediction of outages in the production process was proposed and implemented. Saving ca 30% of maintenance costs as well as the reduction of personnel necessary for the pump maintenance have been obtained. The customer also received an intuitive graphical overview of the performance and condition of all pumps, allowing him to monitor the condition of each pump individually. The whole solution is based on the AIpump algorithm, implemented On Premise. AIpump can be deployed as a service on various types of pumps from different manufacturers, which can enable predictive maintenance worldwide as either an On Premise solution or a Cloud solution.

Paint quality is the first visual element in selecting and purchasing a product. The quality of paint is affected by many factors such as temperature, humidity, applied media, inclination of the paint surface etc. To solve this multiparametric problem we also used the methods and techniques of Data Mining.

Predictive maintenance is not new, in some form it has been available for many years. However, prior to the particular service, it is necessary to obtain detailed information on the process in order to identify the changes preceding the upcoming failure.

The project focused on reducing the amount of failures and errors in the process of car body srewing and included the predictive maintenance of the screwing head.