Research on Pick-up Feature of ESP Recording Charts Based on Bitmap Images

During the oil production, in order to monitor the working conditions of an electrical submersible pump （ESP）,an electrical current recorder is used to monitor the electric motor current of an ESP. The recorder charts indicate various working conditions of the ESP. Subtle malfunctions or abnormal problems of the ESP can be detected and further analyzed from various features of these current curves on the recording charts. Presently, these current charts are manually read and analyzed in oil fields. In this paper, a diagnosis expert system is presented for automatically analyzing these current recording charts and identifying the working condition of the ESP. This expert system includes an open knowledge base, which can be updated or enriched according to the identified features of the current curves on the recording charts, and a condition monitoring and failure pattern recognition method, which is called ＂pick-up method of feature of the recording chart＂, and can be correctly applied in most cases. It has been shown that this expert system can effectively improve the accuracy and efficiency of failure diagnosis and working condition monitoring of ESPs.

The Influence of Training Quality on the Performance of Neural Networks

This paper analyses the error sources in neural network prediction. The relationship between prediction error and quality of training sets is revealed. The influence of quality of training sets on the performance of an artificial neural network(ANN) applied in time series prediction is discussed. A numerical criterion called degree of consistency(DCT) defined from the statistical point of view for evaluating quality of training sets is introduced. Some simulation results and corresponding suggestions are presented along with the new criterion in order to properly select the training sets for neural network training.

Learning from the crowd:Road infrastructure monitoring system

The condition of the road infrastructure has severe impacts on the road safety, driving comfort, and on the rolling resistance. Therefore, the road infrastructure must be moni- tored comprehensively and in regular intervals to identify damaged road segments and road hazards. Methods have been developed to comprehensively and automatically digitize the road infrastructure and estimate the road quality, which are based on vehicle sensors and a supervised machine learning classification. Since different types of vehicles have various suspension systems with different response functions, one classifier cannot be taken over to other vehicles. Usually, a high amount of time is needed to acquire training data for each individual vehicle and classifier. To address this problem, the methods to collect training data automatically for new vehicles based on the comparison of trajectories of untrained and trained vehicles have been developed. The results show that the method based on a k-dimensional tree and Euclidean distance performs best and is robust in transferring the information of the road surface from one vehicle to another. Furthermore, this method offers the possibility to merge the output and road infrastructure information from multiple vehicles to enable a more robust and precise prediction of the ground truth.