Reduce Prediction Errors Research Articles

Research on cooling load estimation through optimal hybrid models based on Naive Bayes

Cooling load estimation is crucial for energy conservation in cooling systems, with applications like advanced air-conditioning control and chiller optimization. Traditional methods include energy simulation and regression analysis, but artificial intelligence outperforms them. Artificial intelligence models autonomously capture complex patterns, adapt, and scale with more data. They excel at predicting cooling loads influenced by various factors, like weather, building materials, and occupancy, leading to dynamic, responsive predictions and energy optimization. Traditional methods simplify real-world complexities, highlighting artificial intelligence’s role in precise cooling load forecasting for energy-efficient building management. This study evaluates Naive Bayes-based models for estimating building cooling load consumption. These models encompass a single model, one optimized with the Mountain Gazelle Optimizer and another optimized with the horse herd optimization algorithm. The training dataset consists of 70% of the data, which incorporates eight input variables related to the geometric and glazing characteristics of the buildings. Following the validation of 15% of the dataset, the performance of the remaining 15% is tested. Based on analysis through evaluation metrics, among the three candidate models, Naive Bayes optimized with the Mountain Gazelle Optimizer (NBMG) demonstrates remarkable accuracy and stability, reducing prediction errors by an average of 18% and 31% compared to the other two models (NB and NBHH) and achieving a maximum R2 value of 0.983 for cooling load prediction.

Implementation of a reinforcement learning system with deep q network algorithm in the amc dash mark i game

Reinforcement learning is a branch of artificial intelligence that trains algorithms using a trial-and-error system. Reinforcement learning interacts with its environment and observes the consequences of its actions in response to rewards or punishments received. Reinforcement Learning uses information from every interaction with its environment to update its knowledge. The problem identified from this research is the lack of consistency, which is not always the same for Non-Player Characters (Agents) in the process of exploring an environment (Game environment). This research uses the Software Development Life Cycle (SDLC) Waterfall model method to train Non Player Characters (Agents) in the Amc Dash Mark I Game which uses the Deep Q Network (DQN) algorithm in several stages. Training results show improvements in model performance over time. The average duration of the episode and average reward episode showed an increase of 7.75 to 24.7, while the exploration rate decreased to 0.05. This indicates that the model has experienced learning and is improving to achieve better rewards by performing fewer actions. The lower loss also shows that the model has succeeded in reducing prediction errors and improving prediction capabilities.

GCN–Informer: A Novel Framework for Mid-Term Photovoltaic Power Forecasting

Predicting photovoltaic (PV) power generation is a crucial task in the field of clean energy. Achieving high-accuracy PV power prediction requires addressing two challenges in current deep learning methods: (1) In photovoltaic power generation prediction, traditional deep learning methods often generate predictions for long sequences one by one, significantly impacting the efficiency of model predictions. As the scale of photovoltaic power stations expands and the demand for predictions increases, this sequential prediction approach may lead to slow prediction speeds, making it difficult to meet real-time prediction requirements. (2) Feature extraction is a crucial step in photovoltaic power generation prediction. However, traditional feature extraction methods often focus solely on surface features, and fail to capture the inherent relationships between various influencing factors in photovoltaic power generation data, such as light intensity, temperature, and more. To overcome these limitations, this paper proposes a mid-term PV power prediction model that combines Graph Convolutional Network (GCN) and Informer models. This fusion model leverages the multi-output capability of the Informer model to ensure the timely generation of predictions for long sequences. Additionally, it harnesses the feature extraction ability of the GCN model from nodes, utilizing graph convolutional modules to extract feature information from the ‘query’ and ‘key’ components within the attention mechanism. This approach provides more reliable feature information for mid-term PV power prediction, thereby ensuring the accuracy of long sequence predictions. Results demonstrate that the GCN–Informer model significantly reduces prediction errors while improving the precision of power generation forecasting compared to the original Informer model. Overall, this research enhances the prediction accuracy of PV power generation and contributes to advancing the field of clean energy.

Variability of intervertebral joint stiffness between specimens and spine levels.

Introduction: Musculoskeletal multibody models of the spine can be used to investigate the biomechanical behaviour of the spine. In this context, a correct characterisation of the passive mechanical properties of the intervertebral joint is crucial. The intervertebral joint stiffness, in particular, is typically derived from the literature, and the differences between individuals and spine levels are often disregarded. Methods: This study tested if an optimisation method of personalising the intervertebral joint stiffnesses was able to capture expected stiffness variation between specimens and between spine levels and if the variation between spine levels could be accurately captured using a generic scaling ratio. Multibody models of six T12 to sacrum spine specimens were created from computed tomography data. For each specimen, two models were created: one with uniform stiffnesses across spine levels, and one accounting for level dependency. Three loading conditions were simulated. The initial stiffness values were optimised to minimize the kinematic error. Results: There was a range of optimised stiffnesses across the specimens and the models with level dependent stiffnesses were less accurate than the models without. Using an optimised stiffness substantially reduced prediction errors. Discussion: The optimisation captured the expected variation between specimens, and the prediction errors demonstrated the importance of accounting for level dependency. The inaccuracy of the predicted kinematics for the level-dependent models indicated that a generic scaling ratio is not a suitable method to account for the level dependency. The variation in the optimised stiffnesses for the different loading conditions indicates personalised stiffnesses should also be considered load-specific.

Using Decomposed Error for Reproducing Implicit Understanding of Algorithms.

Reproducibility is important for having confidence in evolutionary machine learning algorithms. Although the focus of reproducibility is usually to recreate an aggregate prediction error score using fixed random seeds, this is not sufficient. Firstly, multiple runs of an algorithm, without a fixed random seed, should ideally return statistically equivalent results. Secondly, it should be confirmed whether the expected behaviour of an algorithm matches its actual behaviour, in terms of how an algorithm targets a reduction in prediction error. Confirming the behaviour of an algorithm is not possible when using a total error aggregate score. Using an error decomposition framework as a methodology for improving the reproducibility of results in evolutionary computation addresses both of these factors. By estimating decomposed error using multiple runs of an algorithm and multiple training sets, the framework provides a greater degree of certainty about the prediction error. Also, decomposing error into bias, variance due to the algorithm (internal variance), and variance due to the training data (external variance) more fully characterises evolutionary algorithms. This allows the behaviour of an algorithm to be confirmed. Applying the framework to a number of evolutionary algorithms shows that their expected behaviour can be different to their actual behaviour. Identifying a behaviour mismatch is important in terms of understanding how to further refine an algorithm as well as how to effectively apply an algorithm to a problem.

Transmission Line Icing Prediction Based on Dynamic Time Warping and Conductor Operating Parameters

Aiming to improve on the low accuracy of current transmission line icing prediction models and ignoring the objective law of icing of transmission lines, a transmission line icing prediction model considering the effect of transmission line tension on the bundle of icing thickness is proposed, based on a convolutional neural network (CNN) and bidirectional gated recurrent unit (BiGRU). Firstly, the finite element calculation model of the conductor and insulator system was established, and the change rule between transmission line tension and icing thickness was studied. Then, the convolutional neural network and bidirectional gated recurrent unit were used to construct a transmission line icing thickness prediction model The model incorporated a weighted fusion of soft−dynamic time warping (Soft−DTW) and the icing change rule as the loss function. Optimal weights were determined through the utilization of the grid search algorithm and cross−validation, contributing to an enhancement of the model’s generalization capabilities and a reduction in prediction errors. The results indicate that the proposed prediction model can consider the impact of line operating parameters, avoiding the shortcomings of prediction results conflicting with actual physical laws. Compared with traditional non−mechanical models, the proposed model showed reductions in root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) by 0.26–0.51%, 0.24–0.44%, and 5.77–13.33%, respectively, while the coefficient of determination (R2) increased by 0.07–0.13.

Amplifying Sensing Capabilities: Combining Plasmonic Resonances and Fresnel Reflections through Multivariate Analysis.

Multivariate analysis applied in biosensing greatly improves analytical performance by extracting relevant information or bypassing confounding factors such as nonlinear responses or experimental errors and noise. Plasmonic sensors based on various light coupling mechanisms have shown impressive performance in biosensing by detecting dielectric changes with high sensitivity. In this study, gold nanodiscs are used as metasurface in a Kretschmann setup, and a variety of features from the reflectance curve are used by machine learning to improve sensing performance. The nanostructures of the metasurface generate new plasmonic features, apart from the typical resonance that occurs in the classical Kretschmann mode of a gold thin film, related to the evanescent field beyond total internal reflection. When the engineered metasurface is integrated into a microfluidic chamber, the device provides additional spectral features generated by Fresnel reflections at all dielectric interfaces. The increased number of features results in greatly improved detection. Here, multivariate analysis enhances analytical sensitivity and sensor resolution by 200% and more than 20%, respectively, and reduces prediction errors by almost 40% compared to a standard plasmonic sensor. The combination of plasmonic metasurfaces and Fresnel reflections thus offers the possibility of improving sensing capabilities even in commonly available setups.

Research on the Remaining Life Prediction Method of Rolling Bearings Based on Multi-Feature Fusion

Rolling bearings are one of the most important and indispensable components of a mechanical system, and an accurate prediction of their remaining life is essential to ensuring the reliable operation of a mechanical system. In order to effectively utilize the large amount of data collected simultaneously by multiple sensors during equipment monitoring and to solve the problem that global feature information cannot be fully extracted during the feature extraction process, this research presents a technique for forecasting the remaining lifespan of rolling bearings by integrating many features. Firstly, a parallel multi-branch feature learning network is constructed using TCN, LSTM, and Transformer, and a parallel multi-scale attention mechanism is designed to capture both local and global dependencies, enabling adaptive weighted fusing of output features from the three feature extractors. Secondly, the shallow features obtained by the parallel feature extractor are residually connected with the deep features through the attention mechanism to improve the efficiency of utilizing the information of the front and back features. Ultimately, the combined characteristics produce the forecasted findings for the RUL of the bearing using the fully connected layer, and RUL prediction studies were performed with the PHM 2012 bearing dataset and the XJTU-SY bearing accelerated life test dataset, and the experimental results demonstrate that the suggested method can effectively forecast the RUL of various types of bearings with reduced prediction errors.

Tropical cyclone trajectory based on satellite remote sensing prediction and time attention mechanism ConvLSTM model

The accurate and timely prediction of tropical cyclones is of paramount importance in mitigating the impact of these catastrophic meteorological events. Presently, methods for predicting tropical cyclones based on satellite remote sensing images encounter notable challenges, including the inadequate extraction of three-dimensional spatial features and limitations in long-term forecasting. As a response to these challenges, this study introduces the Temporal Attention Mechanism ConvLSTM (TAM-CL) model, designed to conduct thorough spatiotemporal feature extraction on three-dimensional atmospheric reanalysis data of tropical cyclones. By leveraging ConvLSTM with three-dimensional convolution kernels, our model enhances the extraction of three-dimensional spatiotemporal features. Furthermore, an attention mechanism is integrated to bolster long-term prediction accuracy by emphasizing crucial temporal nodes. In the evaluation of tropical cyclone track and intensity forecasts across 24, 48, and 72 h, TAM-CL demonstrates a notable reduction in prediction errors, thereby underscoring its efficacy in forecasting both cyclone tracks and intensities. This contributes to an effective exploration of the application of deep networks in conjunction with atmospheric reanalysis data.

A deep learning combined prediction model for prediction of ship motion attitude in real conditions

ABSTRACT This paper presents the TRSA-EMD-GAN ship motion attitude prediction model, which utilizes self-attention and generative adversarial networks (GAN) to accurately predict ship motion attitudes. The TRSA mechanism based on time residuals is incorporated into the model to capture the different influences of various attitude points on the prediction and their temporal relationships by using a time mask. Moreover, the model employs a variation mode decomposition generative adversarial network (VMD-GAN) for ship motion attitude prediction through feature fusion. In the VMD-GAN model, VMD is combined with a GRU neural network as the generator, while a convolutional neural network serves as the discriminator. Simulation experiments confirm the effectiveness of the TRSA-VMD-GAN model in predicting ship motion attitudes, resulting in reduced prediction errors and improved accuracy and efficiency.

Temporal information sharing-based multivariate dynamic mode decomposition

This paper introduces temporal information shared multi-variable dynamic mode decomposition (TIMDMD), a novel data-driven algorithm for multi-variable modal decomposition. TIMDMD leverages joint singular value decomposition to share temporal information across variables, resulting in multi-variable rather than single-variable optimization. The algorithm effectively addresses several common issues with traditional DMD approaches, such as inconsistent physical interpretations, a lack of phase consistency between variables, and the mixing of frequency components in the reconstructed flow field. To demonstrate its efficacy, TIMDMD is applied to the analysis of wake flows behind a circular cylinder and a pitching airfoil. The results highlight TIMDMD's ability to align modal indices across variables, correct phase relationships, reduce prediction errors, and improve the clarity of frequency components in the reconstructed flow field.

Rain-t: Daily Rainfall Predictive Model Using 6-Gene Genetic Expression for Historical Data-Based Forecasting

Extreme weather conditions such as heavy rainfalls have been wreaking havoc not only in urban areas but also in an entire watershed. The development of a flood management plan and flood mitigating structures to alleviate the impacts of flooding is very crucial because it needs intensive and continuous historical data. However, missing data due to equipment failure that gathers the rainfall data could be a problem. Rainfall data is not only useful in designing flood mitigating structures but also in planning our day-to-day activities ahead of time. To address this problem, this paper proposes a predictive model which able to forecast in a short lead-time and predict missing data within the dataset. In this paper, three predictive models will be compared namely recurrent neural network, Gaussian processing regression, and the proposed 6-gene genetic expression-based predictive modeling (MGGP). 29-year 24-hour cumulative rainfall data which were sourced in PAGASA Tacloban city weather station, Philippines, was used. The data were cleaned by removing negative values. Two datasets were created, the first (RFDS1) dataset which makes use of three indices (year, month, and days), and the second (RFDS2) dataset which was orchestrated and transformed to increase correlation and reduce prediction errors which had an additional two datasets (ave(t-1,t-2),t-1). Each method used three and five time-based indices. The result shows an erratic behavior of the model from three methods that used the RFDS1, while RFDS2 had a more stable predictive model. This shows that the data orchestration and transformation greatly improved the correlation and reduced errors. However, MGGP showed the best results among the three methods.

Tunnel deformation prediction during construction: An explainable hybrid model considering temporal and static factors

This paper presents a novel hybrid model designed for predicting mountain tunnel deformation during construction, incorporating both temporal and static factors. Utilizing a boosting ensemble technique, the model effectively integrates a bidirectional Long Short-Term Memory (Bi-LSTM) network—acclaimed for its proficiency with time-series data—with the Light Gradient Boosting Machine (Light GBM) model—recognized for its adept handling of tabular data. In the prediction procedure, the Bi-LSTM and Light GBM modules are engaged to process time-dependent and static factors, respectively. The model’s performance was evaluated against seven other established machine learning models using data from the Liangwangshan Tunnel, whose outcomes demonstrated the superiority of our model in terms of its local and overall reduction in prediction errors. By introducing static factors associated with each monitoring section via the Light GBM, our model offers a robust solution to the issue of time delayed prediction—a challenge inadequately addressed in time series prediction. Finally, we used the SHAP (SHapley Additive exPlanations) method to interpret the hybrid model’s decision-making mechanisms. The findings reveal a significant correlation between the prediction rules employed by our model and the fundamental principles of tunnel engineering and physical mechanics, thereby underlining its reliability and potential applicability in practical scenarios.

Wind Power Short-Term Time-Series Prediction Using an Ensemble of Neural Networks

Short-term wind power forecasting has difficult problems due to the very large variety of speeds of the wind, which is a key factor in producing energy. From the point of view of the whole country, an important problem is predicting the total impact of wind power’s contribution to the country’s energy demands for succeeding days. Accordingly, efficient planning of classical power sources may be made for the next day. This paper will investigate this direction of research. Based on historical data, a few neural network predictors will be combined into an ensemble that is responsible for the next day’s wind power generation. The problem is difficult since wind farms are distributed in large regions of the country, where different wind conditions exist. Moreover, the information on wind speed is not available. This paper proposes and compares different structures of an ensemble combined from three neural networks. The best accuracy has been obtained with the application of an MLP combiner. The results of numerical experiments have shown a significant reduction in prediction errors compared to the naïve approach. The improvement in results with this naïve solution is close to two in the one-day-ahead prediction task.

Learning consistent representations with temporal and causal enhancement for knowledge tracing

Knowledge tracing is a crucial component of intelligent educational systems and deep learning technologies have significantly propelled its advancement. However, most existing models suffer from the incomplete data modeling problem of knowledge tracing, leading to inconsistent representations of students’ actual knowledge states. This paper proposes a new approach called Temporal- and Causal-enhanced Knowledge Tracing to improve the consistency of students’ knowledge state representations in intelligent educational systems. In particular, our method introduces a causal self-attention mechanism based on front-door adjustment theory, which improves interaction representation and reduces prediction errors caused by dataset bias. To effectively integrate the features of interval and response times into our model, we further use forget and input gates to simulate knowledge forgetting and acquisition, respectively. This maintains consistent learning behavior representations and improves model predictions. The results of our extensive experiments on the three datasets demonstrate that our method outperforms previous knowledge-tracing methods in predicting student scores.

Multiview Deep Learning-based Efficient Medical Data Management for Survival Time Forecasting.

In recent years, data-driven remote medical management has received much attention, especially in application of survival time forecasting. By monitoring the physical characteristics indexes of patients, intelligent algorithms can be deployed to implement efficient healthcare management. However, such pure medical data-driven scenes generally lack multimedia information, which brings challenge to analysis tasks. To deal with this issue, this paper introduces the idea of ensemble deep learning to enhance feature representation ability, thus enhancing knowledge discovery in remote healthcare management. Therefore, a multiview deep learning-based efficient medical data management framework for survival time forecasting is proposed in this paper, which is named as "MDL-MDM" for short. Firstly, basic monitoring data for body indexes of patients is encoded, which serves as the data foundation for forecasting tasks. Then, three different neural network models, convolution neural network, graph attention network, and graph convolution network, are selected to build a hybrid computing framework. Their combination can bring a multiview feature learning framework to realize an efficient medical data management framework. In addition, experiments are conducted on a realistic medical dataset about cancer patients in the US. Results show that the proposal can predict survival time with 1% to 2% reduction in prediction error.

CTFNet: Long-Sequence Time-Series Forecasting Based on Convolution and Time-Frequency Analysis.

Although current time-series forecasting methods have significantly improved the state-of-the-art (SOTA) results for long-sequence time-series forecasting (LSTF), they still have difficulty in capturing and extracting the features and dependencies of long-term sequences and suffer from information utilization bottlenecks and high-computational complexity. To address these issues, a lightweight single-hidden layer feedforward neural network (SLFN) combining convolution mapping and time-frequency decomposition called CTFNet is proposed with three distinctive characteristics. First, time-domain (TD) feature mining-in this article, a method for extracting the long-term correlation of horizontal TD features based on matrix factorization is proposed, which can effectively capture the interdependence among different sample points of a long time series. Second, multitask frequency-domain (FD) feature mining-this can effectively extract different frequency feature information of time-series data from the FD and minimize the loss of data features. Integrating multiscale dilated convolutions, simultaneously focusing on both global and local context feature dependencies at the sequence level, and mining the long-term dependencies of the multiscale frequency information and the spatial dependencies among the different scale frequency information, break the bottleneck of data utilization, and ensure the integrity of feature extraction. Third, highly efficient-the CTFNet model has a short training time and fast inference speed. Our empirical studies with nine benchmark datasets show that compared with state-of-the-art methods, CTFNet can reduce prediction error by 64.7% and 53.7% for multivariate and univariate time series, respectively.

A Deep Learning Model with a Self-Attention Mechanism for Leg Joint Angle Estimation across Varied Locomotion Modes.

Conventional trajectory planning for lower limb assistive devices usually relies on a finite-state strategy, which pre-defines fixed trajectory types for specific gait events and activities. The advancement of deep learning enables walking assistive devices to better adapt to varied terrains for diverse users by learning movement patterns from gait data. Using a self-attention mechanism, a temporal deep learning model is developed in this study to continuously generate lower limb joint angle trajectories for an ankle and knee across various activities. Additional analyses, including using Fast Fourier Transform and paired t-tests, are conducted to demonstrate the benefits of the proposed attention model architecture over the existing methods. Transfer learning has also been performed to prove the importance of data diversity. Under a 10-fold leave-one-out testing scheme, the observed attention model errors are 11.50% (±2.37%) and 9.31% (±1.56%) NRMSE for ankle and knee angle estimation, respectively, which are small in comparison to other studies. Statistical analysis using the paired t-test reveals that the proposed attention model appears superior to the baseline model in terms of reduced prediction error. The attention model also produces smoother outputs, which is crucial for safety and comfort. Transfer learning has been shown to effectively reduce model errors and noise, showing the importance of including diverse datasets. The suggested joint angle trajectory generator has the potential to seamlessly switch between different locomotion tasks, thereby mitigating the problem of detecting activity transitions encountered by the traditional finite-state strategy. This data-driven trajectory generation method can also reduce the burden on personalization, as traditional devices rely on prosthetists to experimentally tune many parameters for individuals with diverse gait patterns.

Optimization of Depression Level Detection Based on Noninvasive Parameters with Artificial Neural Network

We optimized the measurement of depression levels by developing an instrument that processes four physical parameters non-invasively with an Artificial Neural Network (ANN). High-accuracy detection instruments help establish a diagnosis to realize early and appropriate prevention and treatment of depression. Depression among students tends to increase, which has an impact on reducing productivity and even attempting suicide. It can reduce the quality and quantity of the productive age generation. The selection of non-invasive physical parameters, namely heart rate, hours of sleep, respiratory rate, and blood oxygen, considers practical aspects for the examiner and the subject examined. This research uses an exploratory method to develop an ANN-based tool for detecting student depression levels. The ANN-based Depression Detector (dANN) input layer harvests data from measurements of the four parameters. Then, the ANN hidden layer processes and studies the data in depth to present five levels of depression in the output layer. Learning results using Loss Graphs show that dANN reduces prediction errors or training iterations at Epochs greater than 130 or offers high accuracy after reaching Epoch 130. DANN prioritized high accuracy in practical measurements. dANN has the potential to be further developed by integrating psychological parameter data, considering that depression is multifactorial. The result of classification reported a precision of 1.0. This felicitating value presents the success of learning and training on dANN. Furthermore, dANN will be built on a web basis to improve performance by implementing an expert system.

A framework for developing data-driven correction factors for solar PV systems

Correcting simulated solar photovoltaic (PV) output poses challenges due to the limited availability of measured PV output data. This study introduces a framework for developing correction factors capable of adjusting bias errors in hourly simulated PV output across various levels of global horizontal irradiance (GHI). GHI-dependent correction factors are developed for each PV plant, with hourly simulated PV output validated against the measured output for 37 PV plants in South Korea. Performance evaluation using U95, a measure of model uncertainty, reveals a significant reduction (by up to 0.24) in prediction errors. The reduction is largely attributed to reductions of nMBE s (by up to 0.33) and partly to reductions of nRMSE s (by up to 0.11), demonstrating mitigation of both random and bias errors. The framework exhibits a promising reduction in forecasting errors for monthly energy yields and performance ratios. Given that the proposed framework requires a short length of training data (<4 months), its versatility allows for adoption by existing software packages relying on physical PV modeling, offering potential enhancements in forecasting accuracy for practical applications.