Characteristics Of Time Series Research Articles

Fractal Analysis of Mining Wastewater Time Series Parameters: Balkhash Urban Region and Sayak Ore District

The population life and health quality are significantly reduced due to water resources pollution caused by heavy metals, especially in urban agglomerations located close to metal ore mining and processing facilities. The greatest environmental pollution occurs during the extraction of Cu, Zn, and Pb. In this study, a fractal R/S analysis of wastewater discharge indicators time series from a metal ore mining facility located in the Sayak ore district in the Republic of Kazakhstan (turbidity, electrical conductivity, flow magnitude, and pH level) was carried out. A sharp increase in the flow rate was recorded from 10 to 15 July 2024 and an increase in the electrical conductivity from 4 to 26 July 2024. However, the latest type of indicator assessment does not exceed the critical level for life. The presence of electrical conductivity indicators time series long-term memory and persistence was also recorded (the Hurst exponent for the electrical conductivity time series is fixed in the 0.56 to 0.59 range and does not go below the threshold value for randomness according to the Anis-Lloyd formula). Thus, the value-changing process is controlled and stable, and minor changes in turbidity indicate that these releases do not significantly harm the environment. Despite this, the results obtained do not allow for a comprehensive analysis of the state of releases as the data from all deposits is not available. Therefore, due to the time constraints of the data provided for analysis, it is difficult to fully assess the impact of specific metal ore mining facilities on the environmental safety of the Balkhash urban region. In addition, many studies indicate very high risks of chronic diseases for the population living in this region. The findings of this study enable us to conclude that the application of fractal analysis and the calculation of fractal characteristics for time series of emissions can serve as an indicator of the environmental status within the given area. This information can be used by environmental services to build reliable environmental pollution monitoring systems.

Pressure and Temperature Prediction of Oil Pipeline Networks Based on a Mechanism-Data Hybrid Driven Method

To ensure the operational safety of oil transportation stations, it is crucial to predict the impact of pressure and temperature before crude oil enters the pipeline network. Accurate predictions enable the assessment of the pipeline’s load-bearing capacity and the prevention of potential safety incidents. Most existing studies primarily focus on describing and modeling the mechanisms of the oil flow process. However, monitoring data can be skewed by factors such as instrument aging and pipeline friction, leading to inaccurate predictions when relying solely on mechanistic or data-driven approaches. To address these limitations, this paper proposes a Temporal-Spatial Three-stream Temporal Convolutional Network (TS-TTCN) model that integrates mechanistic knowledge with data-driven methods. Building upon Temporal Convolutional Networks (TCN), the TS-TTCN model synthesizes mechanistic insights into the oil transport process to establish a hybrid driving mechanism. In the temporal dimension, it incorporates real-time operating parameters and applies temporal convolution techniques to capture the time-series characteristics of the oil transportation pipeline network. In the spatial dimension, it constructs a directed topological map based on the pipeline network’s node structure to characterize spatial features. Data analysis and experimental results show that the Three-stream Temporal Convolutional Network (TTCN) model, which uses a Tanh activation function, achieves an error rate below 5%. By analyzing and validating real-time data from the Dongying oil transportation station, the proposed hybrid model proves to be more stable, reliable, and accurate under varying operating conditions.

Application and Empirical Analysis of Fuzzy Neural Networks in Mining Social Media Users’ Behavioral Characteristics and Formulating Accurate Online Marketing Strategies

In the current digital social environment, social media platforms have become an important position for user behavior insights and precision marketing. User behavioral data on social media contain rich information, but they are often fuzzy, uncertain and highly complex. Fuzzy neural network (FNN), as an advanced model combining fuzzy logic and neural network theory, provides a powerful tool for processing and analyzing social media user behavioral features. This study is dedicated to exploring the application of fuzzy neural networks in social media user behavior analysis and their key role in the design of accurate online marketing strategies. We construct and optimize a fuzzy neural network model by meticulously classifying and quantifying user behavioral features, including behavioral frequency features, content topic features, social interaction features, and time series features, as well as applying fuzzy set theory to deal with fuzzy features such as emotional states. Through empirical analysis, we will show how fuzzy neural networks can reveal the intrinsic laws behind user behaviors, and how these insights can be used to design and implement precise online marketing strategies to improve advertising effectiveness, user engagement, and brand loyalty.

Enhancing water quality prediction with advanced machine learning techniques: An extreme gradient boosting model based on long short-term memory and autoencoder

Achieving carbon neutrality in the pulp and paper industry necessitates effectively recycling pulp and papermaking wastewater, where continuous monitoring of effluent quality indices is crucial. This study suggests a novel machine learning-based model named LSTMAE-XGBOOST that integrates the feature extraction capabilities of autoencoder, the sequential feature learning capabilities of LSTM, and the high prediction accuracy of XGBOOST. This model is capable of extracting the complex relationships, non-Gaussian characteristics, and time series features from the papermaking wastewater data, and it demonstrates superior predictive performance. Compared to traditional machine learning models, the proposed model exhibits higher prediction accuracy. Specifically, when contrasted with partial least squares regression, LSTMAE-XGBOOST achieves a 40% increase in R2 and a 35% reduction in RMSE. Further comparative assessments against other machine learning-based hybrid models with similar structures confirm the superiority of integrating LSTM and XGBOOST within the hybrid model approach. This study contributes a compelling methodology for modeling effluent quality indices, offering significant implications for environmental management in the pulp and paper industry.

An adaptive network with consecutive and intertwined slices for real-world time-series forecasting

Real-world systems are chiefly comprised of multiple sensors and devices, whose instantaneous data and upcoming tendencies are demanded by the decision makers to formulate future schemes. Accuracy and timeliness are both of paramount significance. Thereby, although the deep forecasting models are promising, the space and resources are often limited in reality, rendering their adaptability requisite. Unfortunately, the majority of current deep forecasting models are far from adaptive and acquire heavy hyper-parameter tuning processes to obtain the satisfactory performances. Moreover, their strategies to reduce the model complexities sometimes bring extra hyper-parameters. To tackle these issues, we propose AdaNS: an Adaptive Network with consecutive and intertwined Slices. The data-driven features of time-series in frequency domain are used in AdaNS to adaptively determine the model structure, as well as hyper-parameters, and categorize variables. In accordance with these frequency-based features, input sequences are first sliced in a consecutive way to extract the universal features and then in an intertwined way to extract the local features, for hierarchical and comprehensive forecasting. Extensive experiments show that AdaNS achieves state-of-the-art performances on fourteen benchmarks, virtually without any hyper-parameter tuning. The source code is available at https://github.com/OrigamiSL/AdaNS.

A method for recognizing abnormal behaviors of personnel at petroleum stations based on GTB-ResNet

PurposeThe abnormal behaviors of staff at petroleum stations pose significant safety hazards. Addressing the challenges of high parameter counts, lengthy training periods and low recognition rates in existing 3D ResNet behavior recognition models, this paper proposes GTB-ResNet, a network designed to detect abnormal behaviors in petroleum station staff.Design/methodology/approachFirstly, to mitigate the issues of excessive parameters and computational complexity in 3D ResNet, a lightweight residual convolution module called the Ghost residual module (GhostNet) is introduced in the feature extraction network. Ghost convolution replaces standard convolution, reducing model parameters while preserving multi-scale feature extraction capabilities. Secondly, to enhance the model's focus on salient features amidst wide surveillance ranges and small target objects, the triplet attention mechanism module is integrated to facilitate spatial and channel information interaction. Lastly, to address the challenge of short time-series features leading to misjudgments in similar actions, a bidirectional gated recurrent network is added to the feature extraction backbone network. This ensures the extraction of key long time-series features, thereby improving feature extraction accuracy.FindingsThe experimental setup encompasses four behavior types: illegal phone answering, smoking, falling (abnormal) and touching the face (normal), comprising a total of 892 videos. Experimental results showcase GTB-ResNet achieving a recognition accuracy of 96.7% with a model parameter count of 4.46 M and a computational complexity of 3.898 G. This represents a 4.4% improvement over 3D ResNet, with reductions of 90.4% in parameters and 61.5% in computational complexity.Originality/valueSpecifically designed for edge devices in oil stations, the 3D ResNet network is tailored for real-time action prediction. To address the challenges posed by the large number of parameters in 3D ResNet networks and the difficulties in deployment on edge devices, a lightweight residual module based on ghost convolution is developed. Additionally, to tackle the issue of low detection accuracy of behaviors amidst the noisy environment of petroleum stations, a triple attention mechanism is introduced during feature extraction to enhance focus on salient features. Moreover, to overcome the potential for misjudgments arising from the similarity of actions, a Bi-GRU model is introduced to enhance the extraction of key long-term features.

Single-Channel Sleep EEG Classification Method Based on LSTM and Hidden Markov Model

Background: The single-channel sleep EEG has the advantages of convenient collection, high-cost performance, and easy daily use, and it has been widely used in the classification of sleep stages. Methods: This paper proposes a single-channel sleep EEG classification method based on long short-term memory and a hidden Markov model (LSTM-HMM). First, the single-channel EEG is decomposed using wavelet transform (WT), and multi-domain features are extracted from the component signals to characterize the EEG characteristics fully. Considering the temporal nature of sleep stage changes, this paper uses a multi-step time series as the input for the model. After that, the multi-step time series features are input into the LSTM. Finally, the HMM improves the classification results, and the final prediction results are obtained. Results: A complete experiment was conducted on the Sleep-EDFx dataset. The results show that the proposed method can extract deep information from EEG and make full use of the sleep stage transition rule. The proposed method shows the best performance in single-channel sleep EEG classification; the accuracy, macro average F1 score, and kappa are 82.71%, 0.75, and 0.76, respectively. Conclusions: The proposed method can realize single-channel sleep EEG classification and provide a reference for other EEG classifications.

Space-Based Mapping of Pre- and Post-Hurricane Mangrove Canopy Heights Using Machine Learning with Multi-Sensor Observations

Coastal mangrove forests provide numerous ecosystem services, which can be disrupted by natural disturbances, mainly hurricanes. Canopy height (CH) is a key parameter for estimating carbon storage. Airborne Light Detection and Ranging (LiDAR) is widely viewed as the most accurate method for estimating CH but data are often limited in spatial coverage and are not readily available for rapid impact assessment after hurricane events. Hence, we evaluated the use of systematically acquired space-based Synthetic Aperture Radar (SAR) and optical observations with airborne LiDAR to predict CH across expansive mangrove areas in South Florida that were severely impacted by Category 3 Hurricane Irma in 2017. We used pre- and post-Irma LiDAR-derived canopy height models (CHMs) to train Random Forest regression models that used features of Sentinel-1 SAR time series, Landsat-8 optical, and classified mangrove maps. We evaluated (1) spatial transfer learning to predict regional CH for both time periods and (2) temporal transfer learning coupled with species-specific error correction models to predict post-Irma CH using models trained by pre-Irma data. Model performance of SAR and optical data differed with time period and across height classes. For spatial transfer, SAR data models achieved higher accuracy than optical models for post-Irma, while the opposite was the case for the pre-Irma period. For temporal transfer, SAR models were more accurate for tall trees (>10 m) but optical models were more accurate for short trees. By fusing data of both sensors, spatial and temporal transfer learning achieved the root mean square errors (RMSEs) of 1.9 m and 1.7 m, respectively, for absolute CH. Predicted CH losses were comparable with LiDAR-derived reference values across height and species classes. Spatial and temporal transfer learning techniques applied to readily available spaceborne satellite data can enable conservation managers to assess the impacts of disturbances on regional coastal ecosystems efficiently and within a practical timeframe after a disturbance event.

Feasibility Analysis for Predicting Indian Ocean Bigeye Tuna (Thunnus obesus) Fishing Grounds Based on Temporal Characteristics of FY-3 Microwave Radiation Imager Data

Efficient and accurate fishery forecasting is of great significance in ensuring the efficiency of fishery operations. This paper proposes a fishery forecasting method using a brightness temperature (TB) time series spatial feature extraction and fusion model. Using Indian Ocean bigeye tuna fishery data from 2009 to 2021 as a reference, this paper discusses the feasibility of fishery forecasting using FY-3 Microwave Radiation Imager (MWRI) Level 1 TB data. For this paper, we designed a deep learning network model for radiometer TB time series feature extraction (TimeTB-FishNet) based on the Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), and Attention mechanism. After expanding the dimensions of TB features, the model uses them together with spatiotemporal feature factors (year, month, longitude, and latitude) as features. By adding the GRU and Attention to the CNN, the CNN-GRU-Attention model architecture is established and can extract deep time series spatial features from the data to achieve the best results. In the model validation experiments, the TimeTB-FishNet model performed optimally, with a coefficient of determination (R2) of 0.6643. In the generalization experiments, the R2 also reached 0.6261, with a root mean square error (RMSE) of 46.6031 kg/1000 hook. When the sea surface height (SSH) was introduced, the R2 further reached 0.6463, with a lower RMSE of 45.1318 kg/1000 hook. The experimental results show that the proposed method and model are feasible and effective. The proposed model can directly use enhanced radiometer TB data without relying on lagging ocean environmental product data, performing deep temporal and spatial feature extraction for fishery forecasting. This method can provide a reference for the fishing of bigeye tuna in the Indian Ocean.

Fault diagnosis method of automobile rolling bearing based on transfer learning and improved DenseNet

Aiming at the problems caused by ignoring the time series characteristics, the scarcity of labeled data and the long diagnosis time in the fault diagnosis of one-dimensional vibration signals of automobile bearings, a new method combining improved DenseNet and transfer learning is proposed in this study. This method uses Recurrent Plot (RP) technology to convert one-dimensional vibration data into two-dimensional images to fully tap the potential value of time series. By optimizing the DenseNet network structure, the fault features are extracted effectively. Lightweight network design and MobileViT Attention mechanism are used to reduce the number of parameters and improve computing efficiency. With the help of transfer learning technology, the fault features in the source domain are transferred to the target domain, which solves the problem of cross-condition diagnosis and greatly reduces the diagnosis time. The experimental results show that the proposed method can improve the accuracy of fault identification and diagnosis efficiency, and achieve accurate classification.

ASK-HAR: Attention-Based Multi-Core Selective Kernel Convolution Network for Human Activity Recognition

Human Activity Recognition (HAR) is an increasingly popular field of study aimed at automatically identifying and categorizing human movements and activities using various tracking devices and measurements, such as sensors and cameras. Smartphones have emerged as a prevalent sensor modality for HAR, offering abundant data on an individual’s movements through GPS, accelerometers, and gyroscopes. In conventional convolutional neural networks (CNNs), the artificial neurons within each feature layer typically possess identical receptive fields (RFs). This paper introduces a novel model, ASK-HAR (Attention-based Multi-Core Selective Kernel Convolution Network for HAR), which enhances HAR by performing kernel selection between multiple branches with different RFs using attention mechanisms. The selective kernel mechanism is leveraged to optimize HAR performance. Additionally, the CBAM attention module is employed for time series feature extraction and activity recognition within the overall framework. Extensive experiments conducted on benchmark HAR datasets, including UCI-HAR, USC-HAD, WISDM, PAMAP2, and DSADS, demonstrate that the ASK-HAR model consistently achieves high accuracy across all datasets.

Hybrid Deep Learning Model for Vegetable Price Forecasting Based on Principal Component Analysis and Attention Mechanism

Abstract With the aim of enhancing the accuracy of current models for forecasting vegetable prices and improving market structures, this study focuses on the prices of bell peppers at the Nanhuanqiao Market in Suzhou. In this paper, we propose a hybrid Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) model for vegetable price forecasting based on Principal Component Analysis (PCA) and Attention Mechanism (ATT). Initially, we utilized the Pearson correlation coefficient to filter out the factors impacting prices. Then, we applied PCA to reduce dimensionality, extracting key price features. Next, we captured local sequence patterns with CNN, while handling time-series features with GRU. Finally, these outputs were integrated via ATT to generate the final prediction. Our results indicate that the hybrid CNN-GRU model, enhanced by PCA and ATT, achieved a Root Mean Square Error (RMSE) as low as 0.16. This performance is 11.11%, 11.11%, and 15.79% better than that of the PCA-CNN, PCA-GRU, and CNN-GRU-ATT models, respectively. Furthermore, in order to prove the effectiveness of our proposed model, the proposed model is compared with the state-of-the-art models and classical machine learning algorithms under the same dataset, the results indicate that our proposed hybrid deep learning model based on PCA and ATT shows the best performance. Consequently, our model offers a valuable reference for vegetable price prediction.

Cargo Volume Forecasting for Logistics Sorting Centers Based on Random Forest Model

With the rapid development of logistics industry, logistics sorting center, as a key node in e-commerce logistics network, its cargo volume prediction and personnel scheduling strategy are of great significance to improve operational efficiency and reduce costs. The purpose of this study is to deeply analyze the cargo volume data of 57 sorting centers through machine learning method, establish an accurate cargo volume prediction model, and optimize the personnel scheduling strategy based on the prediction results. In this study, we preprocess the historical cargo data, including missing value filling, time series conversion and feature construction. Then, the random forest model was used to carry out fitting analysis on the cargo volume data of different sorting centers. By comparing the evaluation indexes such as mean square error (MSE), root mean square error (RMSE) and mean absolute error (MAE), the superiority of random forest model compared with neural network, support vector regression and linear regression models was verified. This study not only improves the accuracy of cargo volume forecast, but also realizes the reasonable allocation of human resources, and provides scientific decision support for the operation and management of logistics sorting center.

A study of short-term wind power segmentation forecasting method considering weather on ramp segments

The short-term fluctuation of wind power can affect its prediction accuracy. Thus, a short-term segmentation prediction method of wind power based on ramp segment division is proposed. A time-series trend extraction method based on moving average iteration is proposed on the full-time period to analyze the real-time change characteristics of power time-series initially; secondly, a ramp segment extraction method based on its definition and identification technique is proposed based on the results of the trend extraction; and a segmentation prediction scheme is proposed to lean the power prediction under different time-series: the LightGBM-LSTM is proposed for the non-ramping segment using point prediction, and the CNN-BiGRU-KDE is proposed for probabilistic prediction of ramp segments. From the results, this ramp segment definition and identification technique can effectively identify the ramp process of wind power, which makes up for the misidentification and omission of the classical climbing event definition; meanwhile, the segment prediction scheme not only meets the prediction accuracy requirements of the non-ramping segment, but also provides the effective robust information for the prediction of the ramping period, which offers reliable reference information for the actual wind farms. In particular, it is well adapted to wind power prediction under extreme working conditions caused by ramping weather, which is a useful addition to short-term wind power prediction research.

EMA fault diagnosis method based on multi-information fusion of GRU and improved attention mechanism

Aiming at the problems of time series feature loss and fault information loss in traditional electromechanical servo system (EMA) fault diagnosis methods based on machine learning and deep learning, a multi-information fusion EMA fault diagnosis method based on gated recurrent unit (GRU) and improved attention mechanism is proposed. Firstly, the collected different sensor signals are divided into different channels, and the time series features of each channel signal are extracted by GRU. Then, the self-attention mechanism is introduced to further distinguish the important relationship between different time points of the signal. The multi-channel attention mechanism is further introduced to adaptively fuse the features of different channels, and finally the fault diagnosis is realized through the classifier. The experimental results based on the test bench data set show that: compared with the single sensor model, the diagnosis rate is improved 10%; compared with the model without the introduction of the attention mechanism, the diagnosis rate is improved 5.2%; compared with the classic machine learning, deep learning and the improved algorithm based on deep learning in the past two years, the diagnosis rate of the diagnosis model designed in this paper is 98.5%above, and the diagnosis effect is the best.

Medium- and Long-Term Distributed Photovoltaic Power Prediction Based on Multiple Time Series Feature and Multiple-Model Fusion

Distributed photovoltaic power stations have advantages such as local direct power supply and reduced transmission energy consumption, and whose demands are constantly being developed. Conducting research on medium- and long-term distributed photovoltaic prediction will have significant value for applications such as the electricity trade market, power grid operation, and the planning of new power stations. Due to characteristics such as long time dependence, disperse power stations, and strong randomness, making accurate and stable predictions becomes very difficult. In this research, we propose a multiple time series feature and multiple-model fusion-based ensemble learning model for medium- and long-term distributed photovoltaic power prediction (M2E-DPV). Considering the wave influence and the differences in distributions in different areas of photovoltaic power, multiple feature combinations are designed to increase feature expression ability and adaptability. Based on the boost ensemble learning model, trained on a single model of different time scale features, the optimal scoring strategy is used for multiple model fusion in the rolling prediction process, and finally, time-segmented probabilistic correction is performed. The experiment results show the effectiveness of the M2E-DPV under multiple feature combinations and multi-model fusion strategies. The average MAPE, R2, and ACC indicators are 0.15, 0.96, and 0.91, respectively. Compared with other methods, there is a significant improvement, indicating that the prediction ability of the model framework proposed in this paper is advanced and robust.

A Monitoring Device and Grade Prediction System for Grain Mildew.

Mildew infestation is a significant cause of loss during grain storage. The growth and metabolism of mildew leads to changes in gas composition and temperature within granaries. Recent advances in sensor technology and machine learning enable the prediction of grain mildew during storage. Current research primarily focuses on predicting mildew occurrence or grading using simple machine learning methods, without in-depth exploration of the time series characteristics of mildew process data. A monitoring device was designed and developed to capture high-quality microenvironment parameters and image data during a simulated mildew process experiment. Using the "Yongyou 15" rice varieties from Zhejiang Province, five simulation experiments were conducted under varying temperature and humidity conditions between January and May 2023. Mildew grades were defined through manual analysis to construct a multimodal dataset for the rice mildew process. This study proposes a combined model (CNN-LSTM-A) that integrates convolutional neural networks (CNN), long short-term memory (LSTM) networks, and attention mechanisms to predict the mildew grade of stored rice. The proposed model was compared with LSTM, CNN-LSTM, and LSTM-Attention models. The results indicate that the proposed model outperforms the others, achieving a prediction accuracy of 98%. The model demonstrates superior accuracy and more stable performance. The generalization performance of the prediction model was evaluated using four experimental datasets with varying storage temperature and humidity conditions. The results show that the model achieves optimal prediction stability when the training set contains similar storage temperatures, with prediction accuracy exceeding 99.8%. This indicates that the model can effectively predict the mildew grades in rice under varying environmental conditions, demonstrating significant potential for grain mildew prediction and early warning systems.

Physical information-guided multidirectional gated recurrent unit network fusing attention to solve the Black-Scholes equation

Reasonable option pricing is crucial in the financial derivatives market. Finding analytical solutions for the Black-Scholes (BS) equation, particularly for American options or with fluctuating volatility and interest rates, is challenging. BS equations exhibit strong time-series characteristics, with asset prices typically adhering to geometric Brownian motion. To address the BS equations, we propose a sequence-to-sequence model guided by physical information (PI), called PiMGA. The PiMGA fuses a multidirectional gated recurrent unit (GRU) network with an attention module, where multidirectional GRU enhances the coding performance of the input sequences and the attention module balances the feature weights of the hidden variables. Prior physical knowledge in BS equations is jointly used as a constraint, forming the penalty function for objective optimization. This allows PiMGA to serve as an efficient approximation function in the learning paradigm of physically informed machine learning to solve BS equations. BS equations with various complexities illustrate the accuracy and feasibility of PiMGA for numerical solutions. Furthermore, the out-of-distribution generalization ability of PiMGA is verified by predicting the Nasdaq 100 index.

Assessing groundwater level modelling using a 1-D convolutional neural network (CNN): linking model performances to geospatial and time series features

Abstract. Groundwater level (GWL) forecasting with machine learning has been widely studied due to its generally accurate results and low input data requirements. Furthermore, machine learning models for this purpose can be set up and trained quickly compared to the effort required for process-based numerical models. Despite demonstrating high performance at specific locations, applying the same model architecture to multiple sites across a regional area can lead to varying accuracies. The reasons behind this discrepancy in model performance have been scarcely examined in previous studies. Here, we explore the relationship between model performance and the geospatial and time series features of the sites. Using precipitation (P) and temperature (T) as predictors, we model monthly groundwater levels at approximately 500 observation wells in Lower Saxony, Germany, applying a 1-D convolutional neural network (CNN) with a fixed architecture and hyperparameters tuned for each time series individually. The GWL observations range from 21 to 71 years, resulting in variable test and training dataset time ranges. The performances are evaluated against selected geospatial characteristics (e.g. land cover, distance to waterworks, and leaf area index) and time series features (e.g. autocorrelation, flat spots, and number of peaks) using Pearson correlation coefficients. Results indicate that model performance is negatively influenced at sites near waterworks and densely vegetated areas. Longer subsequences of GWL measurements above or below the mean negatively impact the model accuracy. Besides, GWL time series containing more irregular patterns and with a higher number of peaks might lead to higher model performances, possibly due to a closer link with precipitation dynamics. As deep learning models are known to be black-box models missing the understanding of physical processes, our work provides new insights into how geospatial and time series features link to the input–output relationship of a GWL forecasting model.

Rapid diagnosis of proportional valve pilot stage stiction fault via dynamic model-based temporal features

As the key flow control elements in modern mechatronic systems, electrohydraulic proportional valves determine the total system operating status. Due to effects such as mechanical wear, micro scale particles may accumulate in the oil and eventually cause stiction in the valve spool. Therefore, early prognosis of valve faults via spool displacement sensing is critical for efficient and economical operations. However, the spool stiction is difficult to measure due to the fully enclosed design of valves, excluding the possible sensor installation options. Traditionally, the sensor-less fault diagnosis is done by machine learning to classify the features extracted using frequency domain methods, which can be time consuming and the trained model is computationally too heavy for the onboard controller. In this paper, a light-weight spool stiction diagnosis method is proposed, where time-series features extracted from coil current signals as well as spool dynamics can be used to quantitively determine the stiction severeness. A hardware-in-the-loop test bed is designed to perform valve spool stiction simulating experiments, and a piloted operated solenoid proportional valve is selected as the validation target. A three-layer regression neural network is established as the baseline. Initial results show the proposed model can detect spool stiction at various degrees, which accuracies are on par with the baseline neural network but nearly three orders of magnitude faster.