Prediction-based Method Research Articles

A Time Series Prediction-Based Method for Rotating Machinery Detection and Severity Assessment

Monitoring the condition of rotating machinery is critical in aerospace applications like aircraft engines and helicopter rotors. Faults in these components can lead to catastrophic outcomes, making early detection essential. This paper proposes a novel approach using vibration signals and time series prediction methods for fault detection in rotating aerospace machinery. By extracting relevant features from vibration signals and using prediction models, fault severity can be effectively quantified. Our experimental results show that the proposed method has potential in early fault detection and is applicable to various types of bearing faults and the different statuses of these faults under complex running conditions, achieving very good generalization ability.

'What would my peers say?' Comparing the opinion-based method with the prediction-based method in Continuing Medical Education course evaluation.

Although medical courses are frequently evaluated via surveys with Likert scales ranging from "strongly agree" to "strongly disagree," low response rates limit their utility. In undergraduate medical education, a new method with students predicting what their peers would say, required fewer respondents to obtain similar results. However, this prediction-based method lacks validation for continuing medical education (CME), which typically targets a more heterogeneous group than medical students. In this study, 597 participants of a large CME course were randomly assigned to either express personal opinions on a five-point Likert scale (opinion-based method; n = 300) or to predict the percentage of their peers choosing each Likert scale option (prediction-based method; n = 297). For each question, we calculated the minimum numbers of respondents needed for stable average results using an iterative algorithm. We compared mean scores and the distribution of scores between both methods. The overall response rate was 47%. The prediction-based method required fewer respondents than the opinion-based method for similar average responses. Mean response scores were similar in both groups for most questions, but prediction-based outcomes resulted in fewer extreme responses (strongly agree/disagree). We validated the prediction-based method in evaluating CME. We also provide practical considerations for applying this method.

Fast grading method based on data driven capacity prediction for high-efficient lithium-ion battery manufacturing

With the large-scale expansion of the battery market, the cost optimization of battery manufacturing has become a focus of attention. Among the complex production process of the battery, capacity grading requires a full discharge to measure the capacity and results in high cost. This study proposes a fast grading method in which the batteries are half discharged and graded according to the capacity predicted by a neural network. The prediction-based method takes half the time and saves about 37 % of the energy consumption. Twenty-three features are extracted as the initial features. The collinear features are screened-out, and three feature reduction methods are compared. Permutation importance can effectively clarify the nonlinear relationship and determine the critical features. The root-mean-square error of the testing set is 0.18 %. The method is feasible and has the potential for further evolution. The electrochemical mechanism involved in the features is analyzed by simulation. Under the guidance of the mechanism, the method can be transferred to other batteries.

A Novel Data Management Scheme in Cloud for Micromachines

In cyber-physical systems (CPS), micromachines are typically deployed across a wide range of applications, including smart industry, smart healthcare, and smart cities. Providing on-premises resources for the storage and processing of huge data collected by such CPS applications is crucial. The cloud provides scalable storage and computation resources, typically through a cluster of virtual machines (VMs) with big data tools such as Hadoop MapReduce. In such a distributed environment, job latency and makespan are highly affected by excessive non-local executions due to various heterogeneities (hardware, VM, performance, and workload level). Existing approaches handle one or more of these heterogeneities; however, they do not account for the varying performance of storage disks. In this paper, we propose a prediction-based method for placing data blocks in virtual clusters to minimize the number of non-local executions. This is accomplished by applying a linear regression algorithm to determine the performance of disk storage on each physical machine hosting a virtual cluster. This allows us to place data blocks and execute map tasks where the data blocks are located. Furthermore, map tasks are scheduled based on VM performance to reduce job latency and makespan. We simulated our ideas and compared them with the existing schedulers in the Hadoop framework. The results show that the proposed method improves MapReduce performance in terms of job latency and makespan by minimizing non-local executions compared to other methods taken for evaluation.

Crossband Filtering for Weighted Prediction Error-Based Speech Dereverberation

Weighted prediction error (WPE) is a linear prediction-based method extensively used to predict and attenuate the late reverberation component of an observed speech signal. This paper introduces an extended version of the WPE method to enhance the modeling accuracy in the time–frequency domain by incorporating crossband filters. Two approaches to extending the WPE while considering crossband filters are proposed and investigated. The first approach improves the model’s accuracy. However, it increases the computational complexity, while the second approach maintains the same computational complexity as the conventional WPE while still achieving improved accuracy and comparable performance to the first approach. To validate the effectiveness of the proposed methods, extensive simulations are conducted. The experimental results demonstrate that both methods outperform the conventional WPE regarding dereverberation performance. These findings highlight the potential of incorporating crossband filters in improving the accuracy and efficacy of the WPE method for dereverberation tasks.

A signal prediction-based method for motor imagery EEG classification

Motor imagery (MI)-based brain-computer interface (BCI) usually uses EEG signals recorded from numerous scalp electrodes for accurate MI classification, which requires long preparation time and high cost, hindering its practical application. A novel method based on signal prediction was proposed in this study to achieve high MI classification accuracy by using EEG recorded from only a small number of electrodes. The signal prediction model was built by using regression technique to estimate the full-channel EEG signals (e.g., recorded from 19, 22, and 29 electrodes) from the few-channel EEG signals on the central region, and then the predicted full-channel EEG signals were used for MI feature extraction and classification. The proposed methods based on multiple linear regression, ridge regression and extreme learning machine regression were compared with the traditional methods that directly used recorded few-channel and full-channel EEG signals, respectively, on three public EEG datasets. The proposed method significantly outperformed the traditional method that directly used few-channel EEG, and was comparable to or even superior to the traditional method that directly used full-channel EEG. The proposed method can accurately predict full-channel EEG and achieve high classification accuracy with only a small number of electrodes with fixed locations, which has great potential in building MI-based BCIs that meet the requirements of practical applications.

Research on real-time identification method of model parameters for the photovoltaic array

Real-time identification study of model parameters has a facilitating effect on fault diagnosis and health state evaluation in photovoltaic (PV). Two effective methods for the real-time identification of PV array model parameters are proposed. Firstly, a PV array modeling method is proposed. This method is used to verify the accuracy of real-time model parameter identification. Then, an effective preprocessing method for the measured current–voltage (I–V) curves is proposed. The preprocessing method can improve the data quality of measured I–V curves, which helps improve the accuracy of the model parameter extraction. Next, the gorilla troops optimizer (GTO) is used to extract parameters of measured I–V curves. The extracted historical model parameters are the data sources for the real-time model parameter identification. Finally, the real-time parameter identification methods based on the time series prediction and the irradiance–temperature (G–T) grid searching are proposed. The root mean square error (RMSE) between the calculated I–V and measured I–V curves is one of the critical evaluation metrics. The RMSE of the time series prediction-based method ranges from 0.01A to 0.1A. The RMSE of the G–T grid searching-based method is around 0.01A. These two methods can complement each other and have good application prospects.

Early Detection and Diagnosis of Chronic Kidney Disease Based on Selected Predominant Features.

In numerous perilous cases, a quick medical decision is needed for the early detection of chronic diseases to avoid austere consequences that may be fatal. Chronic kidney disease (CKD) is a prevalent disease that presents a variety of challenges, including soaring costs for intervention, urgency, and, more importantly, difficulty in early detection of the disease. The current study carries out a prediction-based method that helps in detecting and diagnosing CKD patients which enables a fast and accurate decision-making process at the early stage. A combination of preprocessing and feature selection methods was developed; additionally, several prediction models, such as K-nearest neighbor (KNN), support vector machine (SVM), random forest (RF), and bagging, were trained based on the processed dataset. The performance evaluation shows higher reliability of all models in terms of accuracy, precision, sensitivity, F-measure, specificity, and area under the curve (AUC) score. Specifically, KNN outperformed with an accuracy of 99.50%, sensitivity of 99.2%, precision of 100%, specificity of 98.7%, and F-measure and AUC score of 99.6%. The experimental results of KNN show the best fitted model compared to the existing state-of-the-art methods. Moreover, the reduced feature set proves that just a few clinical tests are enough to detect CKD, resulting in diagnosis cost reduction.

Amphis: Managing Reconfigurable Processor Architectures With Generative Adversarial Learning

Dynamic resources management in reconfigurable processors often manifests as a hard online decision-making task, which should yield premier solutions that must meet Quality-of-Service (QoS) requirements while maximizing the system’s efficiency. Most prior works rely on a hard-to-train predictor to model the complicated relationships between processor configurations and performance. To decide the proper resource allocation, the predictor needs to tentatively evaluate a group of possible configurations, and then decide the best configuration for the workload. This tedious process has an expensive runtime overhead for resource configuration in processors. Besides, prior works focus on improving the prediction accuracy, however, higher performance prediction cannot guarantee a good system outcome. Inspired by recent advances in adversarial learning, we present a generative adversarial network (GAN)-based framework, Amphis, which can directly generate the on-demand processor configuration for any scheduled-in application. By evaluating Amphis on a reconfigurable processor with 18 different workloads, our results demonstrate that the GAN-based method provides tremendous overhead reduction (up to 90%) compared to the SOTA prediction-based method WNNM while providing higher resource utilization.

Frequency regulation of off‐grid system with battery energy storage system using deep Q‐network

This paper proposes a model-free decision algorithm for battery energy storage system (BESS) charging/discharging using deep reinforcement learning (DRL) to regulate off-grid frequency fluctuation. This method is novel since the frequency regulation problem is cast in an off-grid system to a deep Q-network framework, which avoids directly solving a specific optimization model or model-based control. The advantage of the proposed method is that the agents learn to comprehensively make short-term forecasts of demand and PV output, analyse trends in forecast errors, and create control signals within the framework of DRL. A reward function with two criteria, frequency target value and frequency constraint value, guides the agent to suppress the frequency fluctuation by adjusting the weight coefficient. The effectiveness of the proposed method is verified using computational simulations with the off-grid system model and actual power consumption data. The proposed method enables controlling the BESS to satisfy the balance of supply and demand on the off-grid, thus realizing a control method that does not require predictive information. Moreover, the frequency control can be achieved more accurately in a short computation time than the prediction-based method using the seasonal auto-regressive integrated moving average.

Prediction of the void formation in no-flow underfill process using machine learning-based algorithm

This paper investigated the variation variables on the void formation in the no-flow underfill process. A total of 96 distinct no-flow underfill cases were numerically simulated to determine the void formations for different combinations of underfill parameters. The numerical model has been well validated with the experiment, for which the discrepancies are less than 4 %. It was found that the void formation rate increases with the chip placement speed however it decreases with the increase in bump pitch. The highest chip placement speed of 14 mm/s produces 4–6 % meanwhile, the low chip placement speed (2–5 mm/s) produces around 2–3.5 % of void formation. A supervised machine learning (ML) approach is implemented using the validated numerical results to build a prediction-based method for no-flow underfill process. Three ML methods were trained and tested. Neural networks regression was recommended compared to other methods since it has the highest R-squared value and lowest error: 0.95159 and 0.15885, respectively. Chip placement speed obtained the highest score of 1.791641 for permutation feature in ML, which indicates the most significant variable affecting the package's void formation. The obtained result will aid the industry to choose the most favourable no-flow underfill process parameters regarding the void formation issue, thus increasing the reliability of the package.

Video Anomaly Detection Based on Optical Flow Feature Enhanced Spatio–Temporal Feature Network FusionNet-LSTM-G

Video anomaly detection refers to the rapid identification and localization of anomalous events during video surveillance. It has become a challenging task due to the dependence of anomalous event definitions on specific scenes and the scarcity, diversity and non-exhaustiveness of anomalous events. This manuscript proposes a robust frame prediction-based method for video anomaly detection, it can solve the problems of underutilized spatio-temporal sequence information and difficulty in capturing subtle motions. Firstly, a spatial-temporal feature fusion network is constructed using FusionNet and LSTM. Secondly, feature optimization is performed on this spatio-temporal feature fusion network using optical flow feature enhancement. To further improve the robustness of the model, patch GAN is added to FusionNet-LSTM for adversarial training, and the objective function in training is optimized with apparent constraints and image similarity constraints to approach the real frame. Finally, to reduce the sequence error, a PNSR-based mean normalized anomaly score calculation is used in the manuscript. Experiments show that the average accuracy of the proposed method for anomalous event detection on UCSD Ped2 and CUHK Avenue datasets is 96.9% and 86.1% respectively, which is 1.5% and 1.0% better than the benchmark model U-Net, and the model has low computational complexity where FLOP is 6.196G and the number of parameters is 6.829M. It can meet the application requirements of real-time video anomaly detection and has great value for security video surveillance.

Techno-economic evaluation of stand-alone energy supply to a health clinic considering pandemic diseases (COVID-19) challenge

The increase in the number of patients in health care centers boosts electricity consumption. Such a load jump also adversely affects the energy supply, in particular, in rural off-grid systems. To overcome the mentioned challenges, some innovative and practical approaches with available optimization tools should be employed. This study addresses the possibility of developing a challenge prediction-based method for optimizing a reliable and affordable hybrid renewable energy system (HRES) to aid with energy challenges associated with pandemic conditions. The results indicate that the PV/diesel/battery hybrid system with a maximum energy cost of $0.141/kWh and a renewable fraction of more than 50% can meet demand even during the most severe load jumps. Furthermore, the fuel constraints during pandemic years can increase the energy costs up to 2.5 times, and the required photovoltaic installation capacity by about four times. Due to the 20.1% PV output boost, Vertical-axis tracking systems are recommended in areas with limited PV installation space. It is concluded that, by considering the likely effects of pandemics, the supplied energy cost to the rural health clinic equipment and water treatment loads would be between $0.113–0.200/kWh.

KfreqGAN: Unsupervised detection of sequence anomaly with adversarial learning and frequency domain information

Sequence anomaly detection in time series is of critical importance to wide applications ranging from finance, healthcare to IT system monitoring. Most current researches use the reconstruction-based deep learning algorithms to solve the problem. In this article, we aim to use a prediction-based method to detect sequence anomalies in univariate time series, because the latter methods can detect anomalies using historical information revealing normal patterns in time series whereas the former methods simply consider current sequences. However, it is challenging because there exists both uncertainty in the future and performance deterioration under long detection horizon. To tackle the challenges, we propose an unsupervised algorithm called KfreqGAN, which is based on adversarially trained sequence predictor. The adversarial learning architecture helps the model make accurate predictions for future sequences. In addition, auxiliary information from frequency domain is used to help the model capture the characteristics of time series for achieving satisfactory predictions. We conduct extensive experiments on two public-available datasets, with results demonstrating the effectiveness of the proposed algorithm and its superiority to baseline algorithms.

A hybrid recommender system based-on link prediction for movie baskets analysis

Over the past decade, recommendation systems have been one of the most sought after by various researchers. Basket analysis of online systems’ customers and recommending attractive products (movies) to them is very important. Providing an attractive and favorite movie to the customer will increase the sales rate and ultimately improve the system. Various methods have been proposed so far to analyze customer baskets and offer entertaining movies but each of the proposed methods has challenges, such as lack of accuracy and high error of recommendations. In this paper, a link prediction-based method is used to meet the challenges of other methods. The proposed method in this paper consists of four phases: (1) Running the CBRS that in this phase, all users are clustered using Density-based spatial clustering of applications with noise algorithm (DBScan), and classification of new users using Deep Neural Network (DNN) algorithm. (2) Collaborative Recommender System (CRS) Based on Hybrid Similarity Criterion through which similarities are calculated based on a threshold (lambda) between the new user and the users in the selected category. Similarity criteria are determined based on age, gender, and occupation. The collaborative recommender system extracts users who are the most similar to the new user. Then, the higher-rated movie services are suggested to the new user based on the adjacency matrix. (3) Running improved Friendlink algorithm on the dataset to calculate the similarity between users who are connected through the link. (4) This phase is related to the combination of collaborative recommender system’s output and improved Friendlink algorithm. The results show that the Mean Squared Error (MSE) of the proposed model has decreased respectively 8.59%, 8.67%, 8.45% and 8.15% compared to the basic models such as Naive Bayes, multi-attribute decision tree and randomized algorithm. In addition, Mean Absolute Error (MAE) of the proposed method decreased by 4.5% compared to SVD and approximately 4.4% compared to ApproSVD and Root Mean Squared Error (RMSE) of the proposed method decreased by 6.05 % compared to SVD and approximately 6.02 % compared to ApproSVD.

A prediction-based method to estimate student learning outcome: Impact of response rate and gender differences on evaluation results

Background Low response rates threaten the reliability and validity of student evaluations of teaching. Previous research has shown that asking students to predict how satisfied their fellow students were with a course produces reliable results at lower response rates. The aim of this study was to investigate whether this prediction-based method can also be used to evaluate student learning outcome. Methods Before and after a cardiorespiratory module, 128 fourth-year medical students provided self-assessments and predictions of performance on 27 specific learning objectives and took formative tests on the respective contents. Pre-post performance gain was compared across all three modalities. Results Formative exam results indicated a performance gain of 63.0%. Self-assessed and prediction-based performance gains were identical (67.8%) but both slightly overestimated actual performance gain. Irrespective of the method used, a response rate of 20% was sufficient to produce reliable results. Compared to male students, females greatly overestimated their peers’ performance which led to inflated performance gain values. Conclusions Student self-assessments and predictions are equally valid sources of learning outcome measures, and low response rates are sufficient to produce stable results. When using a prediction-based approach, a tendency to overestimate learning outcome in female students needs to be taken into account.

A new reversible data hiding in transform domain

Reversible watermarking is a technique permitting lossless data hiding. In such a method, the lossless recovering of both watermark and host image is essential. For some applications, such as medical imaging and military systems, it is so vital not only to recover the host image exactly but also to increase security. To obtain these goals, a new reversible watermarking scheme is presented. Since embedding in a transform domain improves security, the proposed method uses Reversible Walsh-Hadamard Transform (RWHT) to commute the host image. Afterward, Singular Value Decomposition (SVD) technique is performed on the transformed image for watermark embedding. For a full recovery, additional information is encoded using Quick Response (QR) code, which is embedded by a prediction-based method. To evaluate the performance of the proposed method, a set of comparative experiments is done. The obtained results confirm the effectiveness of the proposed method in both visual quality and capacity.

Anomaly detection based on machine learning in IoT-based vertical plant wall for indoor climate control

Indoor climate is closely related to human health, comfort and productivity. Vertical plant wall systems, embedded with sensors and actuators, have become a promising application for indoor climate control. In this study, we explore the possibility of applying machine learning based anomaly detection methods to vertical plant wall systems so as to enhance the automation and improve the intelligence to realize predictive maintenance of the indoor climate. Two categories of anomalies, namely point anomalies and contextual anomalies are researched. Prediction-based and pattern recognition-based methods are investigated and applied to indoor climate anomaly detection. The results show that neural network-based models, specifically the autoencoder (AE) and the long short-term memory encoder decoder (LSTM-ED) model surpass the others in terms of detecting point anomalies and contextual anomalies, respectively, therefore can be deployed into vertical plant walls systems in industrial practice. Based on the results, a new data cleaning method is proposed and a prediction-based method is deployed to the cloud in practice as a proof-of-concept. This study showcases the advancements in machine learning and Internet of things can be fully utilized by researches on building environment to accelerate the solution development.

LUISA: Decoupling the Frequency Model From the Context Model in Prediction-Based Compression

AbstractPrediction-based compression methods, like prediction by partial matching, achieve a remarkable compression ratio, especially for texts written in natural language. However, they are not efficient in terms of speed. Part of the problem concerns the usage of dynamic entropy encoding, which is considerably slower than the static alternatives. In this paper, we propose a prediction-based compression method that decouples the context model from the frequency model. The separation allows static entropy encoding to be used without a significant overhead in the meta-data embedded in the compressed data. The result is a reasonably efficient algorithm that is particularly suited for small textual files, as the experiments show. We also show it is relatively easy to built strategies designed to handle specific cases, like the compression of files whose symbols are only locally frequent.

Seismic data interpolation using deep internal learning

Seismic data interpolation is a meaningful research topic in the field of seismic data processing. In this paper, we propose deep internal learning for interpolating regularly sampled aliased seismic data, to improve the upsampling accuracy of regularly sampled aliased seismic data. The proposed algorithm, contrary to previous deep external learning-based seismic interpolation relying on prior training for vast external seismic data, exploits the characteristics of the field data itself, based on the feature similarity between the regularly missing and remaining samples. Internal learning generates training samples solely from the currently remaining regularly undersampled seismic data, and then trains a simple convolutional neural network using the training set. Finally, the trained model is used to upsample the current seismic traces regularly with high accuracy, and can adapt itself intelligently to different field data for the upsampling requirement. This enables seismic data antialiasing interpolation on regularly sampled seismic data with a small sample in the case of insufficient data. The performance of the proposed deep internal learning is assessed using synthetic and field data, respectively. Moreover, the comparison of the proposed deep internal learning with a classic prediction-based interpolation method and deep external learning-based seismic interpolation validates the effectiveness of the proposed algorithm.