Occupational health and safety is closely related to the detection and diagnosis of equipment faults. In the era of the fourth industrial revolution, it is essential to automate fault detection and diagnosis so that a warning system can be developed otherwise appropriate action can be taken before any human injury or damage occurs. Globally there are many established and practiced methods to identify and rely on injuries due to device malfunctions, but there are many limitations in low- and middle-income countries. Occupational health and safety culture is not available due to Inadequate infrastructure, lack of skilled manpower, financial crisis and others. Developing and maintaining automated fault detection and diagnosis systems is not easy in the context of LDCs. Considering the context of low and middle-income countries, in this study, implementation of a fault detection and diagnosis system by developing a data acquisition system with some new setup and designing a feature extraction and reduction algorithm to predict faults continuously with minimal dataset have been introduced. The outcome and process have been validated mathematically and statistically. To overcome the computational complexity and all other limitations, some rigorous mathematical frameworks are used to design the whole study and justify the performance of the developed system. Cost-effectiveness and superiority compared to other systems are discussed and future directions of this work are clearly stated so that others can use these methods or extend this development to use the system in their context.
The visualization of the whole work:
The newly developed feature extraction and reduction algorithm is visualized below:
The folders "BEARING", "MISALIGNMENT" and "Structural_Losseness" contain the Excel files. The signals of faulty data are stored in those files.