Study On Metro Stray Current Corrosion Behavior And Prediction Model Under The Influence Of Multiple Factors

With the continuous improvement of China’s urbanization process,urban traffic congestion has become a major factor restricting urban development.In order to effectively solve this problem,urban rail transit system such as subway and light rail has gradually become the main support of urban transportation due to their large capacity,environmental protection,and space saving.In the process of subway operation,a phenomenon of current leakage that is widely present in the main structure of the subway system and the surrounding underground environment will occur,which is so-called subway stray current.Its essence is due to the degradation of the insulation performance of the subway DC traction reflux system.The subway stray current will lead to a serious electrochemical corrosion effect on the buried metal pipelines around the subway line,causing the thining of the wall of buried metal pipeline,and in servere cases,it can lead to the perforation of the buried metal pipeline,resulting in the leakage of dangerous transportation media such as natural gas,gas and avaiation fuel.This may induce explosions,fires and other safety accidents.Therefore,studying the electrochemical corrosion of buried metal pipelines caused by stray current cannot only improve the reliability of subway systems,but also reduce the safety risk of urban buried pipeline transportation systems.The subway system is the future development direction of urban public transportation is the basis for a deep understanding of stray current corrosion evolution.Due to the complex characteristics of the underground environment,common monitoring methods are difficult to apply,and exploring new corrosion assessment and prediction methods shows important engineering practical significance for judging the current corrosion status and future development trends of buried pipelines.Therefore,in-depth study of the corrosion evolution law and formation mechanism of subway stray current,and exploration of corrosion evaluation and prediction methods based on it,is a key issue to realize the safe operation of the subway system and reliable service of the buried transportation system.The research is funded by the Natural Science Foundation of China,combined with the actual working conditions of buried pipelines under the action of stray current corrosion,using electrochemical epxeriments as the basic research approach,and at the same time using signal processing methods,machine learning algorithms and other means to carry out in-depth development research on the electrochemical corrosion behavior and prediction model of underground pipelines under the influence of multiple factors by subway stray current.The research works mainly includes:（1）The framework of the subway DC traction system and negative reflux system is analyzed,and the reasons for the formation of subway stray current are summarized.The environmental factors of subway stray current corrosion are discussed,and the macro characteristics of subway stray current corrosion are obtained through the stray current corrosion experiment of buried steel bars,which lays the foundation for further electrochemical epxeriments.The simulated stray current corrosion experiment is designed,and the chemical composition and preparation method of Q235A sample,solution preparation process and method,electrochemical experiment system,and content and specific steps of electrochemical measurements are introduced.And the experiment for stray current corrosion of Q235 steel in soil electrolyte is designed.（2）The results of simulated acceleration experiment and coupon experiment are analyzed.Firstly,the stray current corrosion polarization characteristics in NaCl solution,NaCl-Na₂SO₄ solution and NaCl-Na₂SO₄-NaHCO₃ solution are discussed separately,and the influence of different external factors on corrosion current density and linear polarization resistance is analyzed.Secondly,the electrochemical impedance spectroscopy（EIS）characteristics in NaCl solution,NaCl-Na₂SO₄ solution and NaCl-Na₂SO₄-NaHCO₃ solution were analyzed repectively,including the characteristics of Nyquist diagram and Bode diagram,fitted equivalent electrical circuit,the variation of equivalent circuit electrical parameters with different external factors and its corresponding relationship with electrochemical corrosion evolution law,and the effect law of different external factors on the EIS,are studied.Meanwhile,the corrosion surface morphology and the interface structure between the corrosion product and metal substrate under different external factors are analyzed.The mechanism of electrochemical corrosion behavior and evolution regularity under stray current corrosion are explored,and different corrosion stages are divided.The experimental results of simulated corrosion in the soil electrolyte are researched.Through the comparison of experimental results between solution and soil electrolyte,results in solution electrolyte are verified to be able to simulate stray current corrosion in the engineering environment.（3）The effect of DC drift elimination parameters on electrochemical noise（EN）signals in time-domain and frequency-domain is studied.Then the chaotic characteristics of EN signals under simulated stray current corrosion are explored,and the characteristics of EN signals under simulated stray current corrosion are analyzed in time-domain and frequency-domain.The influence of different external factors on the characteristic parameters of EN signals in frequency-domain is discussed.Based on the wavelet transform results of the EN signal,a corrosion rate evaluation method under stray current corrosion is proposed.The variation of energy under different decomposition layers is studied with varying external factors.Taking the sum of the logarithmic energy from first to sixth decomposition layer as the basic parameter,the correlation between it and corrosion rate is studied with varying external factors.（4）In view of the inherent shortcoming of the QPSO algorithm,an improved QPSO algorithm named LWQPSO is proposed,which combines the average best position and Levi flight.The effect of performance improvement of the designed algorithm is verified by standard test functions:Ackley,Griewank,Bohachevsky1 and Bohachevsky2.Then,the prediction algorithm flow and framework of proposed artificial neural network optimized by LWQPSO is designed,and a corrosion current density prediction data set is established based on the measurement results of simulated accerelated experiments.Therefore,a prediction model for corrosion current density based on LWQPSO-NN is constructed and neural network training is conducted.On the basis of the prediction results of the LWQPSO-NN algorithm,the impact of the population size,maximum iterations,and linear and non-linear descent strategy of contraction-expansion coefficient on the average prediction accuracy and its distribution is analyzed,and accuracy sensitivity of different parameters is determined.The accuracy and stability of the LWQPSO-NN algorithm are compared with BPNN,QPSO-NN and WQPSO-NN.Finally,the performance advantage of LWQPSO-NN is proved to be effective in the prediction problem of corrosion current density induced by stray current.This dissertation has 121 figures,42 tables and 166 references.