| In this paper,a system is designed that can receive pipeline data close to the pipeline transportation site and carry out data denoising and pipeline leakage detection in real time based on the idea of edge computing.The system avoids the process of data transmission through multi-layer networks,reduces data transmission links,reduces data transmission time,and improves the reliability of data transmission.Data collection and processing directly at the edge side of the network not only relieve the pressure on the data processing center,but also improve the real-time performance of pipeline leakage detection.The system is designed based on the Raspberry Pi platform,with strong portability,low cost,and high inspection efficiency.First of all,the application scenarios are analyzed for the pipeline transportation site,and specific goals are given to the functional requirements and performance requirements of the system based on the characteristics of the application scenarios.According to the problems existing in the field and the functional requirements of the system,the idea and scheme of the overall design of the system are given.Then,according to the overall design scheme,the data transmission network and the technology types as well as hardware equipment required by the edge computing experimental platform are compared and selected.Finally,the selected hardware devices are deployed,and the debugging and configuration process of the devices are given.Secondly,a signal denoising model based on the combination of variational mode decomposition(VMD)and improved Canberra distance(CD)is proposed.The model first uses VMD to decompose the original signal into multiple intrinsic mode functions(IMF).Then the probability density function(pdf)of each IMF,the original signal and the improved CD values between them are calculated.The effective IMFs are selected according to the increment of the improved CD value obtained by adjacent IMF.The remaining noisy IMFs are denoised by wavelet,and the two components are reconstructed to achieve signal denoising.The simulation results show that the proposed algorithm can better preserve the signal features and has a good denoising effect.Thirdly,deep learning is used as the leakage detection method of the system,and 1dimension Convolutional Neural Networks(1D-CNN)is selected to judge the operating conditions of the pipeline and detect abnormal situations.Then,several experiments and tests are carried out for the structure and parameter size of the 1D-CNN model,and the accuracy and running speed of the model are analyzed and compared.Finally,the network structure and parameters suitable for the system are selected.Finally,the system hardware is integrated with the algorithm as a whole,and the functions of the system are debugged,and the stability of the system network transmission and the overall running speed are tested,and the results are analyzed.The results show that the data transmission network connection of the system is stable,and the system can complete long-distance data interaction.The overall running speed of the system is fast,and the system has good real-time performance,which meets the performance requirements of the initial design.In this thesis,a pipeline leakage detection system based on Raspberry PI platform and Lo Ra communication technology is designed with edge computing as the guiding ideology.A improved CD-VMD denoising algorithm is proposed,and the effectiveness of improved CD-VMD denoising algorithm for practical pipeline signals is verified by simulation experiments.It is implemented in Python environment and combined with Raspberry PI platform.The 1D-CNN model is used in pipeline leakage detection,and the appropriate structure and parameters are selected by analyzing the experimental results,and the trained network is transplanted into the Raspberry Pi platform.The system combines an embedded platform,signal denoising algorithm and pipeline leakage detection technology,which improves the accuracy and practicability of the system,and provides a new application idea for pipeline leakage detection. |
PDF Full Text Request