194 publications from this institution
High-speed railway (HSR) is a key transport mode for achieving carbon reduction targets and promoting sustainable regional economic development due to its fast, efficient, and low-carbon nature. Accurate wind speed forecasting (WSF) is vital for HSR systems, as it provides future wind conditions that are critical for ensuring safe train operations. Numerous WSF schemes based on deep learning have been proposed. However, accurately forecasting strong wind events remains challenging due to the complex and dynamic nature of wind. In this study, we propose a novel hybrid network architecture, MHSETCN-LSTM, for forecasting strong wind. The MHSETCN-LSTM integrates temporal convolutional networks (TCNs) and long short-term memory networks (LSTMs) to capture both short-term fluctuations and long-term trends in wind behavior. The multi-head squeeze-and-excitation (MHSE) attention mechanism dynamically recalibrates the importance of different aspects of the input sequence, allowing the model to focus on critical time steps, particularly when abrupt wind events occur. In addition to wind speed, we introduce wind direction (WD) to characterize wind behavior due to its impact on the aerodynamic forces acting on trains. To maintain the periodicity of WD, we employ a triangular transform to predict the sine and cosine values of WD, improving the reliability of predictions. Massive experiments are conducted to evaluate the effectiveness of the proposed method based on real-world wind data collected from sensors along the Beijing–Baotou railway. Experimental results demonstrated that our model outperforms state-of-the-art solutions for WSF, achieving a mean-squared error (MSE) of 0.0393, a root-mean-squared error (RMSE) of 0.1982, and a coefficient of determination (R2) of 99.59%. These experimental results validate the efficacy of our proposed model in enhancing the resilience and sustainability of railway infrastructure.Furthermore, the model can be utilized in other wind-sensitive sectors, such as highways, ports, and offshore wind operations. This will further promote the achievement of Sustainable Development Goal 9.
Small-target detection in sea clutter is a key challenge in marine radar surveillance, crucial for maritime safety and target identification. This study addresses the challenge of weak feature representation in one-dimensional (1D) sea clutter time-series analysis and suboptimal detection performance for sea surface small targets. A novel dual-feature image detection method incorporating an improved mobile vision transformer (MobileViT) network is proposed to overcome these limitations. The method converts 1D sea clutter signals into two-dimensional (2D) fused images by means of a Gramian angular difference field (GADF) and recurrence plot (RP), enhancing the model’s key-information extraction. The improved MobileViT architecture enhances detection capabilities through multi-scale feature fusion with local–global information interaction, integration of coordinate attention (CA) for directional spatial feature enhancement, and replacement of ReLU6 with SiLU activation in MobileNetV2 (MV2) modules to boost nonlinear representation. Experimental results on the IPIX dataset demonstrate that dual-feature images outperform single-feature images in detection under a 10−3 constant false-alarm rate (FAR) condition. The improved MobileViT attains 98.6% detection accuracy across all polarization modes, significantly surpassing other advanced methods. This study provides a new paradigm for time-series radar signal analysis through image-based deep learning fusion.
Wind speed prediction (WSP) provides future wind information and is crucial for ensuring the safety of high-speed railway systems (HSRs). However, the accurate prediction of wind speed (WS) remains a challenge due to the nonstationary and nonlinearity of wind patterns. To address this issue, a novel artificial-intelligence-based WSP model (EE-VMD-TCGRU) is proposed in this paper. EE-VMD-TCGRU combines energy-entropy-guided variational mode decomposition (EE-VMD) with a customized hybrid network, TCGRU, that incorporates a novel loss function: the Gaussian kernel mean square error (GMSE). Initially, the raw WS sequence is decomposed into various frequency-band components using EE-VMD. TCGRU is then applied for each decomposed component to capture both long-term trends and short-term fluctuations. Furthermore, a novel loss function, GMSE, is introduced to the training of TCGRU to analyze the WS’s nonlinear patterns and improve prediction accuracy. Experiments conducted on real-world WS data from the Beijing–Baotou railway demonstrate that EE-VMD-TCGRU outperforms benchmark models, achieving a mean absolute error (MAE) of 0.4986, a mean square error (MSE) of 0.4962, a root mean square error (RMSE) of 0.7044, and a coefficient of determination (R2) of 94.58%. These results prove the efficacy of EE-VMD-TCGRU in ensuring train operation safety under strong wind environments.
On account of current algorithm and parameter design difficulties and low detection accuracy in feature extractions of small target detections in sea clutter environment, this paper proposes a correspondingly improved four feature extraction method by FAST. After the short-time Fourier transform is applied, a time–frequency distribution spectrogram of original data is generated. Candidate feature points (CFP) are first extracted by FAST algorithm, and then a four-feature extraction is implemented with FAST and DBSCAN combined. The feature distinction is enhanced through a feature optimization. Upon the construction of the four-dimensional feature vectors, XGBoost classifier algorithm classifies and detects these feature vectors. The genetic algorithm optimizes the hyperparameters in XGBoost and updates the decision threshold in real time to control the detection method’s false alarm rate. The IPIX dataset is employed for experimental verification. Verification results confirm that this proposed detection method has better performance than several other currently used detection methods. The detection performance is improved by 7% and 13.8% when observation time is set at 0.512 s and 1.024 s, respectively.
Data loss or distortion causes adverse effects on the accuracy and stability of the thunderstorm point charge localization. To solve this problem, we propose a data complementary method based on the atmospheric electric field apparatus array group. The electric field component measurement model of the atmospheric electric field apparatus is established, and the orientation parameters of the thunderstorm point charge are defined. Based on the mirror method, the thunderstorm point charge coordinates are obtained by using the potential distribution formulas. To test the validity of the basic algorithm, the electric field component measurement error and the localization accuracy are studied. Besides the azimuth angle and the elevation angle, the localization parameters also include the distance from the apparatus to the thunderstorm cloud. Based on a primary electric field apparatus, we establish the array group of apparatuses. Based on this, the data measured by each apparatus is complementarily processed to regain the thunderstorm point charge position. The results show that, compared with the radar map data, this method can accurately reflect the location of the thunderstorm point charge, and has a better localization effect. Additionally, several observation results during thunderstorm weather have been presented.
The initial distribution of cement particles within a square element is obtained by introducing periodic boundary conditions.In the two-dimensional simulation of the cement hydration,all points on the cement surface are assumed to have an equal reaction rate and each cement particle is represented by three concentric circles.By introducing three parameters,the mutual interference between neighbouring cement particles is quantified and the incremental relationships among various phase constituents are established.The validity of the proposed numerical method is verified by comparing its results with the exiting test results.The effect of curing temperature on the degree of hydration is analyzed.The area fractions of capillary pores,unhydrated cement and hydrated products as well as the variations of the two-point probability function of capillary pores and the capillary perimeter per unit area with hydration time are quantitatively analyzed by means of the proposed method.Some useful conclusions are drawn.
In addition to being sensitive to humidity, humidity sensors with moisture sensitive elements are also sensitive to ambient temperature. The fusion of temperature and humidity data is an effective way to improve the accuracy of humidity sensors. In view of the problem of insufficient adaptive ability and poor universality in the current compensation algorithm, a piecewise processing of measured error at different temperatures by using multiple linear regression is proposed in this paper. The least squares method and back propagation (BP) neural network improved by a genetic simulated annealing algorithm (GSA-BP) were used to compensate the measured humidity data of different temperature ranges. The efficiency of the GSA-BP algorithm was tested, and the compensation function model was established. The compensation accuracy was also compared with the accuracies obtained by other methods. The experimental results show that the adaptive segmentation compensation method can significantly improve the measured error of the humidity sensor over a wide temperature range.
To reduce the negative effect on sound source localization when the source is at an extreme angle and improve localization precision and stability, a theoretical model of a three-plane five-element microphone array is established, using time-delay values to judge the sound source’s quadrant position. Corresponding judgment criteria were proposed, solving the problem in which a single-plane array easily blurs the measured position. Based on sound source geometric localization, a formula for the sound source azimuth calculation of a single-plane five-element microphone array was derived. The sinusoids and cosines of two elevation angles based on two single-plane arrays were introduced into the sound source spherical coordinates as composite weighted coefficients, and a sound source localization fusion algorithm based on a three-plane five-element microphone array was proposed. The relationship between the time-delay estimation error, elevation angle, horizontal angle, and microphone array localization performance was discussed, and the precision and stability of ranging and direction finding were analyzed. The results show that the measurement precision of the distance from the sound source to the array center and the horizontal angle are improved one to threefold, and the measurement precision of the elevation angle is improved one to twofold. Although there is a small error, the overall performance of the sound source localization is stable, reflecting the advantages of the fusion algorithm.
