Underground power quality disturbance identification system based on improved deep learning algorithm

• Main Authors: HU LIJIA, WANG YU, CHEN YEHUI
• Format: Patent
• Language: Chinese; English
• Published: 29.07.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; MEASURING; MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS; TESTING

The invention relates to an underground power quality disturbance identification system based on an improved deep learning algorithm, and the method comprises the following steps: S1, collecting data through a mutual inductor, obtaining nine kinds of single disturbance signals and composite power quality disturbance signals through a sampling board card, and uploading the data to an upper computer through an industrial standard RS-485 communication interface by the collection board card; s2, in an upper computer, determining an optimal feature layer and freezing parameters of a pre-trained deep convolutional neural network model Alex Net, and performing model migration; s3, after the model is migrated, replacing a Softmax classifier of the migrated model with an SVM (Support Vector Machine), and obtaining a new full connection layer in the model; and S4, performing fine adjustment on the parameters of the latest model by using the single disturbance signal and the composite power quality disturbance signal af