| With the acceleration of the process of smart power grid and the continuous expansion of power grid scale,higher requirements are put forward for the holographic sensing and intelligent operation and maintenance ability of main equipment status.Power transformer,especially large transformer,is one of the key equipment in power grid,its reliability directly affects the stability of power grid operation.However,with the rapid increase in the number of transformers in operation and the gradual growth of operation years,the inherent design structure and process quality of transformers have gradually caused environmental adaptability problems.Among them,the transformer safety accidents caused by mechanical failures of winding and iron core are as high as 55.6%.When faults occur,vibration and noise will be generated at the fault site under the action of internal electromagnetic field.This noise contains a large amount of transformer state information,and the existing fault detection means have the disadvantages of large error,long cycle,low positioning accuracy,so using the sound signal generated when the transformer fault to detect the transformer state has the advantages of noncontact,fast and accurate positioning.This thesis expounds the location,cause and influence of transformer faults,studies the generation mechanism of transformer winding faults and iron core faults,and develops a voice print detection system based on microphone array in view of the characteristics that the main frequencies of sound signals generated by two kinds of faults are concentrated on the fundamental frequency and the double frequency of 100Hz.Compared with the traditional voice print detection system,the positioning and position are not intuitive and lack of interpretability.The detection system developed in this thesis has the advantages of sound source visualization.Specific work is as follows:(1)Research on transformer fault signal and establishment of simulation system for microphone array sound source location.This thesis studies the generation and characteristics of sound signals of transformer winding faults and iron core faults,focuses on the analysis of the frequency of sound signals of two kinds of faults,and determines the boundary distance between near and far sound fields according to Rayleigh distance and practical application experience.At the same time,considering the possibility of sound reverberation in the propagation process,The calculation method of total sound energy of direct sound field and reverb sound field is analyzed.The advantages and disadvantages of sound source location algorithm are compared,and beamforming is selected as the sound source location algorithm in this thesis.The influence of different microphone topologies on sound source location was analyzed by using the positioning algorithm.The array topology of the system was determined according to the actual scene requirements,and the 32-element cross array structure was determined.Finally,according to the research,the establishment of microphone cluster model and the simulation of location algorithm are realized.(2)Hardware design of McSound detection system.According to the study of transformer fault signal characteristics and array model,the overall hardware requirements of the system were determined,including 32-channel sound signal sampling frequency,multi-channel simultaneous acquisition,image acquisition frequency,network port transmission,etc.According to the overall hardware requirements,the selection of main control chip,network port chip,microphone sensor and camera was determined.The hardware system consists of 32 channels sound signal acquisition module,image acquisition module,network port communication module and main control system.Finally,the hardware construction of the detection system is completed.(3)Program and software design of Mike wind pattern detection system.According to the research,the functional requirements of the system are determined,including 32-channel sound signal acquisition program,image acquisition program,network port communication program and main control program,etc.The software program design adopts the hierarchical and TopDown programming idea,and the simulation experiment of each module program is carried out to determine the correctness of the program.Realize the system multi-channel sound signal acquisition,image data acquisition,data transmission,data coding and other program functions.(4)System performance test and verification.A test platform was built for the microphone array voice print detection system designed in this thesis,and a test scheme was formulated.The test scheme was divided into hardware function test and sound source location function test.The hardware function test mainly completed the network port communication function test,voice data acquisition function test and image data acquisition function test,which verified that the network port communication of the system had good stability.The microphone acquisition module has a high resolution to the analog sound source frequency,and the image acquisition program can transmit the real-time picture correctly.The test of sound source location function is divided into the test of sound frequency and the test of analog sound source location.The experimental results show that the microphone array vowels detection system can accurately locate the analog fault sound source of different frequencies,and can realize the function of sound source visualization by superimposing with the real-time picture.Based on the actual engineering requirements,this thesis independently developed a microphone array voice print detection system with functions such as multi-channel sampling,fast speed and sound source location visualization for specific transformer fault sound signals in Shandong University,and collected real-time sound signals at the transformer site and conducted on-site location tests for simulated sound source signals.The simulated sound source location test has a broad application prospect in the field of state sensing of transformer equipment. |
PDF Full Text Request