Design And Implementation Of Mechanical Fault Diagnosis System Based On Fft Of Cloud Computing

FFT is an important method for frequency domain analysis of vibration signals in mechanical fault diagnosis,and it is also an important means for fault detection.In mechanical fault diagnosis,FFT operations up to 106 points are often needed.Parallelization of FFT is a research hotspot and development trend in order to improve operation efficiency.The existing FFT algorithm based on MapReduce has some shortcomings in reducing system I/O and computational complexity.It is inefficient and difficult to meet the real-time processing requirements of mechanical fault diagnosis for massive vibration data.In order to solve the above problems,this paper proposes a parallel FFT algorithm based on Spark parallel computing framework,taking advantage of Spark memory computing and sparse Fourier transform,and designs and implements a mechanical fault diagnosis system based on this algorithm.The main work and achievements of this paper are as follows:(1)The principle of FFT algorithm is deeply studied and the feasibility of its parallelization is analyzed.The existing FFT parallel algorithm based on MapReduce is studied and analyzed,and the existing problems and shortcomings are found.(2)A parallel FFT algorithm based on Spark is proposed and verified by experiments.By calculating the FFT of part of the time domain signal,the algorithm can get the spectrum which can represent the frequency domain characteristics of the whole signal.At the same time,it can make full use of the memory by RDD conversion and reduce the I/O overhead of the system.The experimental results show that when the data reaches 4*108,the speed of FFT parallel algorithm based on Spark is nearly 40 times faster than that of FFT parallel algorithm based on MapReduce in the range of error allowance.At the same time,compared with the FFT experiment results of the traditional method,the calculation error of large frequency points is smaller.(3)Design and implement a mechanical fault diagnosis system for wind turbines,which includes four modules:user management,data preprocessing,fault diagnosis and data storage.The above modules are designed in brief and in detail,and the system is implemented by programming in Java language and Spring MVC architecture.The system provides a visual interface to support the monitoring and fault diagnosis of mechanical operation.(4)By testing the system,including function test and load test,it proves that the system can effectively and timely complete the mechanical fault diagnosis work.