Abstract
Today, artificial intelligence (AI) provides support for different powerful data science and optimization techniques in China. Our goal was to collect research works and articles that describe novel methods and applications of AI. This covers a wide spectrum of applications, ranging from evaluation systems and data discrimination to forecasting models.
For example, in the article entitled “An Empirical Study on the Artificial Intelligent Writing Evaluation System in China CET,” Xiaoxia Lu for the problems of the artificial intelligent writing evaluation system in China College English writing. There are automated composition evaluation services on the Internet for both English teachers and students to reduce teachers' workload. These automated systems can directly determine the English writing level of students and help them improve their writing skills. Juku AWE is one of the most commonly used systems among colleges and universities in China. An empirical study was conducted on the use of Juku AWE in English courses in colleges. The results showed that use of Juku AWE helped students with their English writing skills effectively. Both teachers and students had positive attitudes about the use of Juku AWE in terms of immediate and clear feedback, time-saving, and arousing interest in English writing.
In the article entitled “Abnormal Data Region Discrimination and Cross Monitoring-Points Historical Correlation Repair of Water Intake Data,” Xue Huifeng et al. have proposed an Abnormal Data Region Discrimination (ADRD) algorithm and the Cross Monitoring-Points Historical Correlation Repair (CMHCR) method to discriminate and repair abnormal data in order to solve the problems of abnormal values that existed in water intake monitoring data and a centralized uploaded report. The characteristics of abnormal data distribution were analyzed, and the ADRD algorithm was proposed. ADRD uses the relationship between 0 values and abnormal large values, as well as the ratio of abnormal large values to expectation, in order to distinguish the abnormal data domain. The correlation between the monitoring data of current detection points and historical data of different detection points was analyzed. The results showed that the data of current monitoring points and historical data obtained from the same points did not completely match to give the maximum correlation.
In the article entitled “Multiple Targets Tracking with Big-Data-Based Measurement for EBPSK Transceiver,” Yao Yu et al. considered the extended binary phase shift keying (EBSPK) transmit–receive system as a high-resolution radar tracking system. The target kinematic states were estimated from a time series of target range and velocity measurements. Such measurements usually provide a huge amount of data. In this paper, a big data-based tracking strategy was developed, incorporating Doppler measurements as part of the data association procedure, and the advantages of the proposed technique were evaluated. First, the principle of the EBPSK transceiver was introduced. In the proposed system, the range-spread target was denoted by the target impulse response (TIR). Second, an efficient big data association scheme was proposed, which utilized both target range and target velocity measurements to track multiple linear targets. When target velocity measurements are not incorporated into target kinematic state estimation, a nonlinear filter bank is not necessary when combining target velocity measurements. Finally, a great enhancement in the tracking performance of the big Doppler data association method with multiple linear targets combined with probabilistic data association scheme was demonstrated by simulation.
In the article entitled “An Improved Particle Filtering Algorithm Using Different Correlation Coefficients for Non-linear System State Estimation,” Qingxu Meng et al. proposed an improved particle filtering algorithm, combining different rank correlation techniques to solve the shortcomings of degeneracy. By simulating the iteration operation in MATLAB software, it was revealed that the proposed algorithm had greater accuracy than SIR, GSPF, and GMSPPF in Gaussian mixture noise.
Finally, in the article entitled “Mid-term power load forecasting model based on KPCA-PSOBP,” Zhao Liu et al. proposed a forecasting model to improve the accuracy of mid-term power load forecasting by combing Kernel Principal Component Analysis (KPCA) with Back Propagation Neural Network (BPNN). Using the data provided by European Network on Intelligent Technologies to test the model, the Mean Absolute Percent Error (MAPE) of load forecasting model was calculated to be only 1.39%. Therefore, the feasibility and validity of the model was proved.
We hope you enjoy reading the papers published in this special issue as much as we did!