Fault Feature Recognition Based On Three Phase Current&Voltage Integrated Information

Intelligent substation is the extension and expansion of digital substation.It's the substation which has achieved station-wised informatization,automation and interaction.However,the immaturity of data sampling technology brings potential threat to the reliability of relay protection.Considering the unpredictability of the information errors occurred during the operation and the complexity of its effect on the relay protection,screening the wrong information at the terminal of secondary information is necessary.For relay protection,distinguishing fault feature from information errors feature is even more critical since they are quite similar.With the three phase voltage¤t integrated information taken as input,an online fault feature recognition method is proposed in this paper to distinguish three different situations-no fault&no info errors,no fault&info errors and indeed transmission line short circuit faults.The work of the paper can be concluded as:1.Research on the possible information error situations and their possible effect on relay protection.Such as small mutual-transformer,abnormal A/D sampling,frame loss and flying spot.2.Research on the theory and implementation of deep learning algorithm-convolution neural network.The implementation of the theory on function fitting and image recognition is discussed,thus expanding the CNN theory to the information error recognition for relay protection.3.Based on the algorithm platform and simulation software,an online wrong message recognition method is proposed.The method can recognize the wrong message from three phase current&voltage integrated information,providing reliable data for relay protection.4.Based on three phase current&voltage sampled value information,transmission line short circuit fault phase selection is implemented with the modified CNN.The deep-learning algorithm-CNN is originally applied in the area of intelligent substation wrong message recognition,a fault feature recognition method based on three phase current&voltage information is proposed.The protection information errors situations and fault situations can be effectively and accurately distinguished,thus preventing protection malfunction due to information errors.With the accumulation of intelligent substation operation experience,newly discovered information error situations can be added to the network through training.