|
The operation mode of the emergency generator system is divided into a common mode and a standby mode, that is, a continuous operation mode in an area without mains electricity, and a standby mode in an area with mains electricity. When the emergency generator is abnormal in the power system, the generator set starts supply power to the main equipment. Most of the causes will be natural disasters and a small number of human error factors. How to make the emergency generators start normally in an emergency state and continuouly operate and supply power, a set of maintenance mechanism should be established for the equipment to make the power supply stable. This thesis proposes the importance of emergency generator system, various mechanical and electrical maintenance, inspection records, inspection and measurement, maintenance spare parts, data collection, etc.. The formation of mechanical and electrical maintenance practices at work to provide important equipment for generator stability operation and maintenance cases, fault analysis when three-phase current imbalance is abnormally tripped, and use the scene to obtain the operating data of the triplex emergency generators at that time, applying big data analysis combined with artificial intelligence neural network to cause three phase current is unbalanced, but the pre-alarm and failure analysis of the trip off setting are not reached to facilitate discrimination. The collected data are classified into six parameters: maximum value of three-phase current, minimum value of three-phase current, average value of three-phase current, difference change of R-phase current value, difference change of T-phase current value and change of S-phase current value are input variables. The variation rate of the total difference of three-phase current as the output variable, and a three-phase current imbalance based on a feed forward back propagation neural network (Feed Forward Back Propagation Neural Network) is established. Failure analysis model, computer simulation analysis and example tests verify the practicability and effectiveness of the method proposed in this thesis, and the results of the study. It can be used as a reference for rapid fault analysis and distinguish when the emergency generator is running in the future.
Keywords: Emergency Generator、Big data analysis 、Artificial Neural Networks、Fault analysis
|