Fault diagnosis for MSF dynamic states using neural networks

This work outlines the development of a fault diagnostic system to supervise a MSF desalination plant during dynamic states (e.g., start-up and changes of operative conditions) using artificial neural networks (ANNs). This diagnostic system processes the plant data to determine whether the process state during a dynamic state is normal or not. In the last case, the diagnostic system determines the cause of the abnormal process state. The diagnostic system has an ANN for each potential fault (equipment malfunctions or operator mistakes). Every ANN processes the plant data looking for symptoms of their respective faults. At a given time, the result reported by an ANN is an index between 0 and 1. This number represents the security about how the corresponding fault is affecting the plant. The higher this value, the higher the security of the affirmation. The structure of each ANN is simpler than that reported in the literature; however, the performance is better. These results were obtained due to a careful selection of the diagnostic system output and the use of a special training method. This training method calculates an appropriate value for the output of each ANN instead to set it at 0 or 1 only. The new value of the output does not depend on the fault provoking the inputs but only on the degree of matching between the observed evolution and the expected one for the fault corresponding to each ANN. Finally, a dynamic simulator was used to evaluate the performance of the diagnostic system.

Fig. 10. Diagnostics for the MSF plant. A diagnostic system to supervise a desalination MSF plant during dynamic states based on ANNs was introduced. The structure of each ANN was simpler than those presented in previous works; however, system performance was much better. The reason for this result is the new training method developed throughout this work and its background theory. Further studies will be guided towards the implementation of a computer program that will be able to combine this system with a qualitative one [9,10] to generate appropriate explanations for each diagnostic. This is an important feature since they help the operator to make correct decisions.