Shouke Wei

Human action recognition (HAR) plays an important role in social interaction in various fields. This study proposes a light-weight skeleton and two-layer bidirectional LSTM-based Seq2Seq model (SB2_Seq2Seq) for HAR to trade off recognition accuracy, users’ privacy and computer resource usage. An experiment was conducted to compare the proposed SB2_Seq2Seq with other skeleton-based Seq2Seq models and non-skeleton RGB video frame-based LSTM, CNN and seq2seq models. The UCF50 dataset was used for model evaluation, where 60%, 20% and 20% for model training, validation and testing, respectively. The experimental results show that the proposed model achieves 93.54% accuracy with 0.0214 Mean Square Error (MSE), suggesting that the proposed model outperforms all the other models. Besides, it also shows that the proposed model achieves state-of-the-art accuracy compared with state-of-the-arts methods in literature.

With the increasing use of wireless sensor networks in water quality monitoring, an enormous amount of streaming data is generated by widely deployed sensors. However, the current batch mode used for data analysis can no longer meet the diverse combination of monitoring indicators and the requirement for timely analysis results on an all-weather basis. To overcome these challenges and analyze a large amount of water quality data quickly and accurately, we propose a stream-DBSCAN distributed stream processing clustering model. First, real-time data streams are processed using the distributed stream computing framework Flink. Then, the DBSCAN clustering algorithm is applied to cluster each dataset as a different dimension of the cluster. Finally, the time distribution characteristics of the data in the same cluster are analyzed to identify the water quality variation rules. The system can extract data noise points and identify sudden deterioration of water quality. We tested the model using datasets on three water quality indices, pH, ammonia nitrogen (NH₄N), and turbidity, in the Yantai Menlou Reservoir from May to August 2019. The results demonstrate that the system can efficiently and quickly perform cluster analysis on streaming data. By analyzing the clustering results, we found that the daily variation of water quality and sudden pollution events in the Menlou Reservoir are consistent with the actual situation.

This paper developed wavelet decomposition and Seq2Seq hybrid models (W-Seq2Seq) to predict water quality. Four Seq2Seq models, namely one-layer unidirectional model (Uni1), one-layer bidirectional model (Bi1), two-layers unidirectionalmodel (Uni2) and two-layers bidirectional model (Bi2) were proposed in this study. Daubechies5 (db5) wavelet was used to decompose datasets into low frequency series and high frequency series, and these low frequency signals were used as the inputs of the proposed models. Data series of four water quality indices (pH, NH₄-N, Conductivity and Turbidity) collected from the Menlou Reservoir in China were used for model training, validation and testing. The main results show that all the four models have very good performances to fit the historical datasets during the training process. Whereas, the comparison testing results suggest that Bi2 are superior to the other three models in terms of prediction accuracy and generalization ability, which can significantly improve the prediction accuracy of water quality data with high complexity.

Large scale real-time water quality monitoring system usually produces vast amounts of high frequency data, and it is difficult for traditional water quality monitoring system to process such large and high frequency data generated by wireless sensor network. A real-time processing and early warning system framework is proposed to solve this problem, Apache Storm is used as the big data processing platform, and Kafka message queue is applied to classify the sample data into several data streams so as to reserve the time series data property of a sensor. In storm platform, Daubechies Wavelet is used to decompose the data series to obtain the trend of the series, then Long Short Term Memory Network (LSTM) model is used to model and predict the trend of the data. This paper provides a detailed description concerning the distribution mechanism of aggregated data in Storm, data storage format in HBase, the process of wavelet decomposition, model training and the application of mode for prediction. The application results in Xin’an River in Yantai City reveal that the prosed system framework has a very good ability to model big data with high prediction accuracy and robust processing capability.

Due to rapid socioeconomic development, continuous population growth and urbanization, the world is facing a severe shortage of fresh water, particularly in arid and semi‐arid regions. A lack of water will put pressure on agricultural production, water pollution, as well as eco‐environmental degradation. Traditional water resources assessment mainly focused on blue water, ignoring green water. Therefore, analysis of spatiotemporal distribution of blue and green water resources in arid and semi‐arid regions is of great significance for water resources planning and management, especially for harmonizing agricultural water use and eco‐environmental water requirements. This study applied the Soil and Water Assessment Tool (SWAT) model and the Sequential Uncertainty Fitting algorithm (SUFI‐2) to calibrate and validate the SWAT model based on river discharges in the Wei River, the largest tributary of the Yellow River in China. Uncertainty analysis was also performed to quantify the blue and green water resources availability at different spatial scales. The results showed that most parts of the Wei River basin (WRB) experienced a decrease in blue water resources during the recent 50 years with a minimum value in the 1990s. The decrease is particularly significant in the most southern part of the WRB (the Guanzhong Plain), one of the most important grain production bases in China. Variations of green water flow and green water storage were relatively small both on spatial and temporal dimensions. This study provides strategic information for optimal utilization of water resources in arid and semi‐arid river basin. Copyright © 2014 John Wiley & Sons, Ltd.

A wavelet-neural network (WNN) hybrid modelling approach for monthly river flow estimation and prediction is developed. This approach integrates discrete wavelet multi-resolution decomposition and a back-propagation (BP) feed-forward multilayer perceptron (FFML) artificial neural network (ANN). The Levenberg-Marquardt (LM) algorithm and the Bayesian regularization (BR) algorithm were employed to perform the network modelling. Monthly flow data from three gauges in the Weihe River in China were used for network training and testing for 48-month-ahead prediction. The comparison of results of the WNN hybrid model with those of the single ANN model show that the former is able to significantly increase the prediction accuracy. Editor D. Koutsoyiannis; Associate editor H. Aksoy Citation Wei, S., Yang, H., Song, J.X., Abbaspour, K., and Xu, Z.X., 2013. A wavelet-neural network hybrid modelling approach for estimating and predicting river monthly flows. Hydrological Sciences Journal, 58 (2), 374–389.