Smart sensor network organization: sensor data fusion and industrial fault traceability
MetadataShow full metadata
The industrial environment usually contains multiple motors that are supplied through a common power bus. The power-line acts as a good conducting environment for signals to travel through the power-line network. In effect, this influences other motors with noisy signals that may indicate a fault condition. Further complexity arises when signals are generated by motors with different power ratings, a different slip speed and more than one source of fault signals. This sort of complexity and mixed signals from multiple sources makes them difficult to measure and precisely correlate to a given machine or fault. Generally, an industrial power-line network consists of different sizes of induction motor from small to large, which together can have a considerable combined influence on the overall system’s operation. The combined effect of all these induction motors can have a strong impact on power-line network permanence. In this thesis, the concept of cross evaluation of motor fault signals is considered to be signal propagation manifesting into healthy signal. Different concepts relating to propagation and manifestation of faults will be discussed and analysed. Initially, a systematic technique was employed to analyse the influence of the fault electric current signals of different motors within a power-line network. Further analysis analysed the attenuation ratio of electrical signals that leads toward a technical framework which evaluates the strength of signal propagation over a power-line network. The diagnostic process was demonstrated at individual sensing points to estimate the strength of propagated signals and identify fault points. This proved very helpful in maximizing the different independent observations. A sample industrial distributed motor network was simulated, to observe the behavior of a distributed power-line network in the presence of fault components. The multi-motor dynamic simulation model was developed, to compare the results with the test-bed practical results, to validate the acquired data. A number of case scenario experiments was done to verify the simulation results and validate the accuracy of these results. In this research, analytical results present significant improvements in describing the interference of faulty signals amongst motors running parallel to the power-line network. Some shortcomings were observed while implementing the strategy of distributed fault diagnosis, including false identification of similar types of fault symptom in power-line network and failure of the diagnosis system due to interference from non-linear noisy signals travelling within multi motor network. Some of these complications are supposed to be solvable by using an efficient and proper knowledge-based numerical technique. Furthermore, the focus of this research was also to develop a wireless node-level feature extraction technique for data fusion, using MCSA at end node-level. Decision-level fusion was implemented at the node coordinator for efficient fault diagnosis. In conclusion, this research does not claim to provide a complete solution to cover all types of fault diagnosis in electric drives. But it is a fitting attempt to provide a more reliable industry solution for motor fault diagnosis.