Research On Stray Current Corrosion Mechanism And Corrosion Prediction Method Of Urban Rail Transit

The urban rail transit system provides great convenience to people's lives,but the system also brings many problems in application.The system adopts DC traction power supply.Since the track is not absolutely insulated from the ground,part of the current leaks to the ground,thereby forming stray currents.Stray current will corrode the buried pipeline near the track,reduce the strength of the buried pipeline,and threaten the safe operation of the subway.Therefore,it is necessary to study the corrosion mechanism of stray current.By building a stray current simulation experiment platform,the influence of stray current on the corrosion of buried pipelines is analyzed,and the corresponding corrosion prediction model is established.According to the experimental results,the model achieves a good prediction effect.The specific research content is as follows:(1)In order to collect stray current data and analyze the corrosion mechanism of stray current,build a stray current simulation experiment platform.Set up a certain number of electrode pads on the experimental platform,by changing the p H,humidity and power supply voltage value of the soil,and record the potential value under different conditions.Through the obtained data,the distribution of stray current in the soil and the corrosion mechanism are analyzed.The results show that there is a strong correlation between stray current and soil p H,humidity,and power supply voltage,and the stray current leaks seriously near the positive and negative poles of the power supply.Corrosion protection in this area should be strengthened.Based on the potential offset value,the soil potential gradient is calculated,and the corrosion risk assessment is carried out according to the hazardous index of the soil potential gradient.(2)Provide a corrosion prediction model based on SVM and BP neural network.According to the potential value collected in the experiment,the Pearson correlation coefficient is used to quantitatively analyze the correlation between the polarization potential value and the experimental environment,and the Corrosion prediction model is established based on the SVM and BP network.The vertical distance,horizontal distance,and input voltage are used as input characteristics to predict the offset value of the potential in the soil.The experimental results show that both SVM and BP network can effectively predict the potential offset value,and the prediction accuracy of BP network is better than SVM.Therefore,the prediction model based on BP network provides a certain reference value for subway stray current corrosion protection.(3)Provide a prediction model based on ensemble learning.Due to the complicated data acquisition methods of stray current experiments,it is difficult to obtain a large number of data samples.The BP neural network requires a large amount of data for training,and a small amount of data is likely to cause over fitting of the network and degrade the performance of the model.Therefore,a stray current corrosion prediction method based on ensemble learning is proposed.This method achieves the prediction effect of strong learners by integrating multiple weak learners.The experimental results show that when the Stacking model chooses XGBoost,GBDT,and RF as the base model and linear regression as the meta-model,the model has the best prediction accuracy and good generalization ability.