This paper proposes an efficient DC voltage control scheme in rail transit traction power systems with reversible converters for DC voltage regulation and power sharing. The proposed DC voltage control scheme is executed in a hierarchical structure based on the optimal power flow. Unlike conversional constant DC voltage control applied in traction power systems, this scheme allows the reversible converters operating in both rectifier and inverter regions and considers regenerative braking power sharing amongst multiple traction substations. The steady- state equivalent models of traction power systems including reversible converters are established, and the AC/DC sequential power flow algorithm based on Newton–Raphson method is presented. Finally, a multi-trains dynamic operation simulation is carried out with a subway line. The results prove the validity of proposed scheme and indicate that the system energy saving effect in traction power systems with reversible converters is obvious with proposed scheme.

The frequent icing of catenary in cold areas has seriously affected the normal operation of urban rail transit. Among many de-icing methods for catenary, the thermal de-icing method which uses current Joule heat to melt ice has been widely adopted due to its efficiency and safety. However, most of the thermal de-icing methods are only applicable when the line is out of service, which cannot solve the icing problem of catenary during the operation period. In order to solve this problem, this paper proposes a novel online thermal de-icing method which takes the catenary of up and down lines as the de-icing current loop. Based on the Bolsdorf theory, the de-icing and anti-icing current of catenary are calculated, and the structure and control strategy of de-icing power unit (DIPU) are designed. The influence of online de-icing on the traction power supply system is quantitatively analyzed by the electric network model. The accurate dynamic simulation model of the traction power supply system including the DIPU is established to verify the analysis conclusion. The results show that the online thermal de-icing method has limited influence on the traction power supply system, but achieves good de-icing performance in the normal operation of the line.

This paper proposes a potential distributed reactive power compensation method in dc traction power systems with reversible substations. It utilizes the entire capacity of reversible substations during night hours and the reminding capacity except for regenerative braking energy feedback during day hours. A unified dispatch scheme is developed to allocate reactive power compensation capacity for each reversible substation. During night hours the entire capacity is used to accomplish the optimal power flow objective. During day hours the excess capacity after active power production is used to improve the system reactive power factor dynamically. The different objectives throughout the day are implemented in a unified formula which avoids manually change. In addition the outage of reversible substations and the low dc voltage conditions are considered and an over-modulation suppression strategy is included in the unified dispatch scheme to adjust the reactive power allocation among reversible substations. This novel utilization method of reversible substations is able to provide significant improvements in the power factor of dc traction power systems which would otherwise require expensive additional equipment such as static synchronous compensator. Finally the effectiveness of the proposed method is verified with MATLAB/SIMULINK.

In urban rail traction power supply systems, the optimal power flow method is mostly utilized for energy optimization. Conventional power flow optimization adopts offline calculation methods to obtain the optimal strategy under given scenarios. However, this method cannot continuously realize the ideal optimization effect for the time-varying flexible traction power supply system. For this reason, this paper proposes a real-time adaptive energy optimization method to solve the problem above. Firstly, the control objective of the optimal power flow is determined based on analyzing the system power flow. Secondly, a hierarchical control system is designed to achieve adaptive optimal control for the flexible traction power supply system. Moreover, a swarm intelligence optimization algorithm-based optimal power flow calculation method is employed to realize the real-time optimal power flow. Finally, some simulations are performed to verify the effectiveness and accuracy of the proposed real-time optimal power flow. It is demonstrated that the real-time optimal power flow method makes it possible to consider the low delay and high accuracy of control commands at the same time. Compared with the conventional offline method, the proposed real-time method can reduce the input electricity consumption by 1.47%, which is valuable in practical applications.

The problem of high rail potential is common in urban rail transit lines that use running rail to return train current. At present, the research of rail potential mainly adopts modeling analysis. However, the rail potential results of the traditional model are generally far less than the actual situation, and the distribution is not accurate enough. It is difficult to predict the rail overvoltage in actual operation by the traditional model. This article explains the reason for the low accuracy of the traditional model from the perspective of the rail skin effect and corrects the rail resistance parameters of the traditional model. A calculation method of rail skin effect equivalent resistance is proposed, and a distribution correction model is established to simulate the rail potential. Compared with the traditional model, the rail potential results of the correction model are obviously increased, and the value and distribution of rail potential are closer to the actual system. The accuracy of the traditional model is improved obviously after the correction of rail resistance parameters, and the problem of inaccurate analysis and prediction of OVPD by traditional models has been solved.

Low-voltage circuit breakers (LVCBs) fault diagnosis based on artificial intelligence (AI) algorithm has always been a research hotspot and got some recent advances. However, AI algorithms usually require sufficient data to train the model, so intelligent fault diagnosis is a challenging task when lack of fault signals. To solve this problem, a fault vibration signal augmentation method based on Wasserstein distance and conditional generative adversarial networks (CGANs) is proposed in this paper. The proposed method uses Wasserstein distance to optimize the adversarial training of generator and discriminator, thus the generator can generate vibration signals under different fault conditions, which can be used to extend the training dataset. In order to verify the improvement effect of this method on the accuracy of LVCB fault diagnosis. Multiple fault classifiers are trained using generated and real fault signals, and a multi-dimensional evaluation index system is built to evaluate the classification effect. Experimental results reveal that the method can generate fault signals with high similarity and improve the accuracy of fault diagnosis.