Font Size: a A A

IoT-based Multirate Fault Sampling And Troubleshooting System For Induction Furnace

Posted on:2019-04-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y C YuFull Text:PDF
GTID:2348330545993383Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
Induction furnace is widely used in the metal heating.To make sure that the induction furnace is functioning well,it is necessary to monitor some signals of the circuit.It is inefficient that the technicians need to combine the signals with the experience to troubleshoot the fault on site when fault occurs.To solve this problem,this paper designed an IoT-based multirate fault sampling and troubleshooting system for induction furnace.The system can be divided into two relatively independent parts,the multirate fault sampling system and the fault troubleshooting system.The multirate fault sampling system is based on the architecture of Internet of Things(IoT).It consists of three parts,including a fault data sampling module installed in the electrical cabinet of the induction furnace,a server module deployed in the wide area network,and a user terminal module.For the multirate fault sampling system,a sampling method is designed to sample mixed signals of high and low rates in the limited time based on the multi-channel analog-digital converter.This method can sample mixed signals of high and low rates in the limited time simultaneously with reasonable configuration of the ADC channels.It solves the need of sampling on a single data sampling chip while improving the cost and resource utilization.The user terminal can connect to the on-site fault data sampling module through the server with access to fault data and on-site status data.The fault troubleshooting system classifies faults based on the faults data.First,a large number of faults data are marked as a data set of multi-class supervised learning.And then two kinds of classification are used.One is to use LSTM neural network classifier,which considers the fault sequence data as input data directly.The other is to obtain fault waveforms energy eigenvectors using wavelet decomposition as a start,and then to classify the faults using random forest classifier with energy eigenvectors as input data.Both methods succeeded in acceptable results,which verifies the feasibility of the fault troubleshooting system.Composed of the two parts,an IoT-based multirate fault sampling and troubleshooting system for induction furnace realized the function of remotely monitoring and troubleshooting the faults of the induction furnace,and paves the way for the automatic monitoring system.
Keywords/Search Tags:IoT, Multirate Sampling, Troubleshooting, LSTM, Wavelet Decomposition, Random Forest
PDF Full Text Request
Related items