Yongxiang Liu

A Prediction Method of Charging Station Capacity Based on Deep Learning

with the national new energy strategy periodically adjusting and continuously advancing, the construction of electric vehicles charging infrastructure getting increasingly closing, the problem of power supply capacity prediction that the social users in some region use it to charge their electric vehicles is the key issue that affects the power grid construction. In this paper, a forecasting method that based on machine learning is proposed to solve the problem.

The Interoperability Test Study of Electric Vehicle A.C. Charging Spot

The interoperability test between electric vehicle and charging equipment plays an important role on the popularization and application of electric vehicles. The principle of control pilot circuit of charging equipment is introduced firstly. Based on new national standard and the new charging interoperability test specification, the interoperability testing device of A.C. charging spot is designed and the device is used to test the interoperability test items. The test results show that the following problems exist in the interoperability test of the charging spot: (1) the potential of power supply exceeds the allowable range of ± 0.8V. (2) Charging connection control timing does not meet the standard requirements. (3) When the charging spot is under abnormal condition, it cant disconnect the AC power supply circuit within the specified time.

Research on Kalman Filter Prediction Method Based on Decision Tree Analysis

The electric vehicle parameters collected by timesharing lease management platform have been accumulated increasingly, which has laid a firm foundation for large data analysis. Pretreatment technology plays an important role in the process of data acquisition. In particular, there is a risk of intervention in the process of transmitting monitoring data between two platforms. Therefore, this paper puts forward the prediction method of decision tree analysis based on Kalman filter by adopting the intervention probability curve of monitoring data and the quantitative relationship between the covariance ratio, putting the intervention probability curve interval and measurements as input to establish a regression relationship, and adding the object decision tree into the traditional Kalman filtering algorithm. This method can reduce the intervention effects, and achieve filtering and predicting function for the data collected by the main operating platform, which lays a solid technical support for big data analysis efficiently.