Solenoid valve features detection for predictive maintenance

In this elaborate a system for measuring, storing and processing voltage and current quantities in a solenoid valve has been developed. This system will be used in a bigger project that has the goal of doing a sensor-less monitoring of a solenoid valves to do predictive maintenance on these components. In the implementation, a microcontroller was used for performing the measurements and the LwIP library was used for a lightweight implementation of A UDP/IP communication for exporting the data in almost-real-time. Through the use of a Raspberry Pi, a first data manipulation of the data is performed and a secure communication with the AWS platform has been implemented. Using Amazon Cloud-based applications the data are stored and finally processed. The result of the complete data processing is the computation of the linked flux in the coil of the valve and then the detection the features that characterize the shape of the transients of the flux and the current during a complete switching cycle. The collected and processed quantities in this system are obtained from the valve just using the microcontroller and simple electronics without adding any sensors. The system has been used for collecting data during a first endurance test on a solenoid valve. From the results analysis it was possible to observe how the detected features evolved during the valve aging. Some evaluations of the temperature effects on the measurements and of possible improvement of the accuracy of the detection are done. The topics presented in this thesis are useful to understand which information are obtainable from a solenoid valve without the usage of additional sensors and how to retrieve, process and store them. This information can be used for doing predictive maintenance on those components by finding a relation between the evolution of the detected features and the decline of the health status of the valve itself.