| dc.description.abstract | Faults in industrial equipment lead to production interruptions, losses and, consequently, loss of competitiveness with those that mitigate these problems. The development and implementation of solutions that seek to detect and identify faults in this equipment are, mostly indispensable to minimize production losses and potential risks to the health of people working in these environments. Present in these machines, the engines induction motors, which coupled in a bearing system, create the necessary movement to carry out the objective activity of the machine. These moving parts, with the use or with any adversity, at some point they will collapse, showing the fault, so it is necessary to monitor them in real time to predict and avoid them. The current state-of-the-art presents some works that consider the vibration energy in the spectrum, but the construction of a system that monitors in real time and can be used in several devices, no works were found in this sense. This work presents a solution that integrates software and hardware for solve the aforementioned problem, including in real time and that learns the behaviour equipment, suggesting regions of alert and danger, serving as a tool for taking decisions. For the development, the physical quantities of vibration were used, together with signal processing techniques and machine learning. After development and implementation, the solution was used in a case study in companies from different branches of industry, which obtained very positive results. The solution was able to store, process and make available the vibration and temperature data in real time, where the system detected in advance in one case, and not the other, despite the advanced state of wear of the equipment, according to an ISO 10816-1 standard, demonstrate that the equipment was in severe condition. | en |
