Reduce Prediction Errors Research Articles

Predicting the 305-Day Milk Yield of Holstein-Friesian Cows Depending on the Conformation Traits and Farm Using Simplified Selective Ensembles

In animal husbandry, it is of great interest to determine and control the key factors that affect the production characteristics of animals, such as milk yield. In this study, simplified selective tree-based ensembles were used for modeling and forecasting the 305-day average milk yield of Holstein-Friesian cows, depending on 12 external traits and the farm as an environmental factor. The preprocessing of the initial independent variables included their transformation into rotated principal components. The resulting dataset was divided into learning (75%) and holdout test (25%) subsamples. Initially, three diverse base models were generated using Classifiction and Regression Trees (CART) ensembles and bagging and arcing algorithms. These models were processed using the developed simplified selective algorithm based on the index of agreement. An average reduction of 30% in the number of trees of selective ensembles was obtained. Finally, by separately stacking the predictions from the non-selective and selective base models, two linear hybrid models were built. The hybrid model of the selective ensembles showed a 13.6% reduction in the test set prediction error compared to the hybrid model of the non-selective ensembles. The identified key factors determining milk yield include the farm, udder width, chest width, and stature of the animals. The proposed approach can be applied to improve the management of dairy farms.

The Engine Combustion Phasing Prediction Based on the Support Vector Regression Method

While traditional one-dimensional and three-dimensional numerical simulation techniques require a lot of tests and time, emerging Machine Learning (ML) methods can use fewer data to obtain more information to assist in engine development. Combustion phasing is an important parameter of the spark-ignition (SI) engine, which determines the emission and power performance of the engine. In the engine calibration process, it is necessary to determine the maximum brake torque timing (MBT) for different operating conditions to obtain the best engine dynamics performance. Additionally, the determination of the combustion phasing enables the Wiebe function to predict the combustion process. Existing studies have unacceptable errors in the prediction of combustion phasing parameters. This study aimed to find a solution to reduce prediction errors, which will help to improve the calibration accuracy of the engine. In this paper, we used Support Vector Regression (SVR) to reconstruct the mapping relationship between engine inputs and responses, with the hyperparametric optimization method Gray Wolf Optimization (GWO) algorithm. We chose the engine speed, load, and spark timing as engine inputs. Combustion phasing parameters were selected as engine responses. After machine learning training, we found that the prediction accuracy of the SVR model was high, and the R2 of CA10−ST, CA50, CA90, and DOC were all close to 1. The RMSE of these indicators were close to 0. Consequently, SVR can be applied to the prediction of combustion phasing in SI gasoline engines and can provide some reference for combustion phasing control.

Effects of Manufacturing-induced Operating-conditions Amplified Stresses on Fatigue-life Prediction of Micro Gas Turbine Blades – A Simulation-based Study

This study investigates the influence of residual stresses induced by the manufacturing process, on the fatigue life prediction of micro gas turbine blades manufactured by laser metal deposition additive manufacturing technique and heat-treated at different heat-treatment parameters. Finite element modelling, using commercial software Abaqus CAE ® and FE-Safe ®, was used to simulate the turbine blade and disk assembly’s operations and analyze induced stresses, as well as estimate the life cycle of the assembly. Results show internal stress build-ups, up to 400 MPa in magnitude can be induced in engineered components right from the point of additive manufacture, the effects of which can become amplified to as much as double that scale or more by the time the component is in operation. As much as between 75% to 300% fatigue life prediction error reduction can hence be attained by simply accounting for induced stress contribution of the process of its manufacture, rendering cost-saving condition-based monitoring and maintenance/overhaul decision making options.

Machine learning based photovoltaic energy prediction scheme by augmentation of on-site IoT data

This paper proposes a novel one-day-ahead photovoltaic (PV) energy prediction scheme using on-site temperatures based on ensemble of convolutional neural networks (POST-enCNN). Even meteorological information from weather station can help predict PV energy, additional information is necessary to describe on-site weather condition. Therefore, the proposed scheme takes advantage of on-site weather data along with the meteorological information to predict PV energy generation accurately. The proposed prediction process is divided into two stages. In the first stage, the day-ahead on-site temperatures are predicted based on meteorological information and the past on-site temperatures data. Specifically, incremental trend of past on-site temperatures are analyzed and transformed into a numerical values. These values are applied to ensemble of CNNs to increase prediction accuracy of the day-ahead on-site temperatures. In the second stage, the predicted day-ahead on-site temperatures are combined with meteorological information. Finally, using the long short-term memory (LSTM) based neural network, the day-ahead PV energy generation is predicted using the combined weather information. Through the simulation results, we show that the proposed POST-enCNN achieve an 27% reduction of prediction error compared to a conventional method that predicts the day-ahead PV energy generation depending solely on meteorological information.

Combining Crop Modeling with Remote Sensing Data Using a Particle Filtering Technique to Produce Real-Time Forecasts of Winter Wheat Yields under Uncertain Boundary Conditions

Within-season crop yield forecasting at national and regional levels is crucial to ensure food security. Yet, forecasting is a challenge because of incomplete knowledge about the heterogeneity of factors determining crop growth, above all management and cultivars. This motivates us to propose a method for early forecasting of winter wheat yields in low-information systems regarding crop management and cultivars, and uncertain weather condition. The study was performed in two contrasting regions in southwest Germany, Kraichgau and Swabian Jura. We used in-season green leaf area index (LAI) as a proxy for end-of-season grain yield. We applied PILOTE, a simple and computationally inexpensive semi-empirical radiative transfer model to produce yield forecasts and assimilated LAI data measured in-situ and sensed by satellites (Landsat and Sentinel-2). To assimilate the LAI data into the PILOTE model, we used the particle filtering method. Both weather and sowing data were treated as random variables, acknowledging principal sources of uncertainties to yield forecasting. As such, we used the stochastic weather generator MarkSim® GCM to produce an ensemble of uncertain meteorological boundary conditions until the end of the season. Sowing dates were assumed normally distributed. To evaluate the performance of the data assimilation scheme, we set up the PILOTE model without data assimilation, treating weather data and sowing dates as random variables (baseline Monte Carlo simulation). Data assimilation increased the accuracy and precision of LAI simulation. Increasing the number of assimilation times decreased the mean absolute error (MAE) of LAI prediction from satellite data by ~1 to 0.2 m2/m2. Yield prediction was improved by data assimilation as compared to the baseline Monte Carlo simulation in both regions. Yield prediction by assimilating satellite-derived LAI showed similar statistics as assimilating the LAI data measured in-situ. The error in yield prediction by assimilating satellite-derived LAI was 7% in Kraichgau and 4% in Swabian Jura, whereas the yield prediction error by Monte Carlo simulation was 10 percent in both regions. Overall, we conclude that assimilating even noisy LAI data before anthesis substantially improves forecasting of winter wheat grain yield by reducing prediction errors caused by uncertainties in weather data, incomplete knowledge about management, and model calibration uncertainty.

A novel feedback correction-adaptive Kalman filtering method for the whole-life-cycle state of charge and closed-circuit voltage prediction of lithium-ion batteries based on the second-order electrical equivalent circuit model

Accurate state of charge (SOC) and closed-circuit voltage (CCV) prediction is essential for lithium-ion batteries and their model performance. In this study, a novel feedback correction-adaptive Kalman filtering (FC-AKF) method is proposed for the online battery state co-prediction, which is adaptive to the whole-life-cycle of the lithium-ion battery based on the improved second-order equivalent circuit model (SO-ECM). For the feedback correction strategy, the optimized iterative state initialization is conducted using the uncertainty covariance matrix of the prior three-time points with the convergence of the updating process. The experimental results show that the SOC prediction error of the proposed FC-AKF method is 0.0099% and 0.975% compared with the ampere-hour integral method under the dynamic stress test (DST) and the Beijing bus dynamic stress test (BBDST) working conditions, respectively. Also, the CCV traction by the SO-ECM is 0.80 V and has fast initial convergence and quick prediction error reduction characteristics. The constructed iterative calculation model promotes the accurate SOC and CCV co-prediction effect, improving the safety and longevity of lithium-ion batteries with high precision and fast convergence advantages.

Daily power demand prediction for buildings at a large scale using a hybrid of physics-based model and generative adversarial network

Power demand prediction for buildings at a large scale is required for power grid operation. The bottom-up prediction method using physics-based models is popular, but has some limitations such as a heavy workload on model creation and long computing time. Top-down methods based on data driven models are fast, but less accurate. Considering the similarity of power demand patterns of single buildings and the superiority of generative adversarial network (GAN), this paper proposes a new method (E-GAN), which combines a physics-based model (EnergyPlus) and a data-driven model (GAN), to predict the daily power demand for buildings at a large scale. The new E-GAN method selects a small number of typical buildings and utilizes EnergyPlus models to predict their power demands. Utilizing the prediction for those typical buildings, the GAN then is adopted to forecast the power demands of a large number of buildings. To verify the proposed method, the E-GAN is used to predict 24-hour power demands for a set of residential buildings. The results show that (1) 4.3% of physics-based models in each building category are required to ensure the prediction accuracy; (2) compared with the physics-based model, the E-GAN can predict power demand accurately with only 5% error (measured by mean absolute percentage error, MAPE) while using only approximately 9% of the computing time; and (3) compared with data-driven models (e.g., support vector regression, extreme learning machine, and polynomial regression model), E-GAN demonstrates at least 60% reduction in prediction error measured by MAPE.

Investigation of the relationship between f0f2 and the TEC using single station neural network models

The relationship between the ionospheric critical plasma frequencies (f0F2) and GNSS-TEC (Global Navigation Satellite System –Total Electron Content) measurements was investigated using an artificial neural network method. About 20 pairs of ionosonde-GNSS receiver stations from 2000 to 2016 were used. Results from this work indicate that the relationship between f0F2 and TEC is mostly affected by the seasons, followed by the level of solar activity, and then the local time. Geomagnetic activity was the least significant of the factors investigated. The relationship between f0F2 and TEC was also shown to exhibit spatial variation; the variation is less conspicuous for closely located stations. Single station models predicted the f0F2 more accurately at their particular localities and clearly overestimated values of the f0F2 ionosonde observations when used at different localities. This finding indicates that model predictions are better (in terms of reduced prediction errors) for the stations for which they are developed than for a different station. Our result visibly point out that models developed for a particular station cannot be effectively applied in another station located farther apart in space. The new approach described in this study represents an important contribution in space weather prediction.

Towards robust reduction of nonlinear errors in dynamic spectrum spectroscopy for effective noninvasive optical detection of blood components

Most conventional noninvasive optical methods for blood component detection suffer from individual subjects’ variability interferences. Dynamic spectrum (DS) is a spectroscopy technique that can eliminate these limitations and improve the accuracy of blood component assessment. The DS method is an adaptive technique that can extract blood information from multiple and different spectral wavelengths of the pulsating arterial part of the photoplethysmography (PPG) signal. However, high concentrations of human blood have strong scattering properties that result in a nonlinear effect during spectral analysis. There is still a paucity of robust techniques to address this nonlinear effect in DS and noninvasive optical detection of blood components. This investigation proposed a novel optimized extended Kalman filter (O-EKF) technique and radial basis function neural network (RBF-NN) to address the nonlinear effect in spectroscopy measurements. The O-EKF and RBF-NN were considered nonlinear algorithms with the adaptive ability to approximate the nonlinear relationship between spectral absorbance and concentration. Hemoglobin samples with concentration levels between 120 g/L and 175 g/L acquired from 231 volunteer subjects were used for sample prediction and spectrum accuracy evaluation. The spectrum extraction from single positioning obtained a correlation coefficient of prediction (Rp) = 0.721 and a root mean square error of prediction (RMSEp) = 10.023 with the partial least square (PLS) model. Further evaluation with the original dual-position spectrum extraction without optimized filtering recorded Rp = 0.636 and RMSEp = 10.630. The results of the dual-position spectrum extraction with O-EKF and the RBF-NN model proved to be effective with a significant increase in Rp = 0.976 and a decrease in RMSEp = 8.132. The results represent a 35% increase in hemoglobin prediction and 31% reduction in prediction error when compared to spectrum detection in single-position. The results demonstrate that nonlinearity from scattering can be reduced through multi-position spectrum detection and optimized filtering. The present study could provide a robust alternative to eliminate nonlinearity and thus improve the linear concentration detection of blood components.

RoBoost-PLS2-R: An extension of RoBoost-PLSR method for multi-response

Recently, a novel robust PLSR method was developed to address the problem of outliers in the data. In this paper, an extension of this method, called RoBoost-PLS2-R is proposed to predict multi-response variables. Robustness and efficiency of this new approach have been validated on two simulated data sets and one real data set containing different outlier scenarios. Its performance was also compared with reference methods (PLS2-R and RSIMPLS) for predicting multi-response variables. Results confirm that RoBoost-PLS2-R greatly reduces prediction errors when data contain outliers. Prediction performances of RoBoost-PLS2-R are close to the optimal model (PLS2-R) calibrated without outliers and also to RSIMPLS method. This method seems to be a reliable and a competitive robust regression tool for predicting multi-response variables.

Analyzing the Association between Pattern and Returns Using Goodman–Kruskal Prediction Error Reduction Index (λ)

For selecting and interpreting appropriate behaviour of proportion between buy/neutral/sell patterns and high/moderate/low returns, the prediction error reduction index is a very useful tool. It is operationally interpretable in terms of the proportional reduction in error of estimation. We first obtain the buy/sell pattern using an Optimal Band. The analysis of the association between patterns and returns is based on the Goodman–Kruskal prediction error reduction index ( λ ). Empirical analysis suggests that the prediction of returns from patterns is more impressive or of less error as compared to the prediction of patterns from returns. We demonstrated the prediction index for Index NIFTY 50, BANK-NIFTY, and NIFTY-IT of NSE (National Stock Exchange), for the period 2010–2020.

Probabilistic electric load forecasting through Bayesian Mixture Density Networks

This work presents a novel approach to address a challenging and still unsolved problem of neural network based load forecasting systems, that despite the significant results reached in terms of prediction error reduction, still lack suitable indications regarding sample-wise trustworthiness of their predictions. The present approach is framed on Bayesian Mixture Density Networks, enhancing the mapping capabilities of neural networks by integrated predictive distributions, and encompassing both aleatoric and epistemic uncertainty sources. An end-to-end training method is developed, aimed to discover the latent functional relation to conditioning variables, characterize the inherent load stochasticity, and convey parameters uncertainty in a unique framework. To achieve reliable and computationally scalable estimators, both Mean Field variational inference and deep ensembles are integrated. Experiments have been performed on short-term load forecasting tasks at both regional and fine-grained household scale, to investigate heterogeneous operating conditions. Different architectural configurations are compared, showing by Continuous Ranked Probability Score based tests that significant performance improvements are achieved by integrating flexible aleatoric uncertainty patterns and multi-modalities in the parameters posterior space.

Tumor Classification of Gene Expression Data by Fuzzy Hybrid Twin SVM

A new classification model, the fuzzy hybrid twin support vector machine (FHTWSVM), is proposed by combining the fuzzy TWSVM and the hypersphere SVM. The hypersphere SVM is utilized for generating the hyperspheres for the positive and negative class with the smallest possible radius, so that the hyperspheres can contain as many samples as possible. The samples which the hyperspheres cover form a new sample set. Furthermore a distance-based fuzzy function is utilized to calculate the fuzzy factors for the samples. Finally FHTWSVM is used to train all samples with the parameters optimized by grid search. This method can maximize intra-class clustering for noise removal and reduce the influence of outliers. To demonstrate the superiority of the performance of FHTWSVM over other classifiers, e.g., KNN, RF, Bayesian, TWSVM, AdaBoost and XGBoost, a series of experiments is conducted using eight gene expression datasets. The evaluation results show that the proposed approach can improve the classification performance as well as reduce prediction errors for the datasets.

QoS prediction for service selection and recommendation with a deep latent features autoencoder

The number of services on the Internet has increased rapidly in recent years. This makes it increasingly difficult for users to find the right services from a large number of the functionally equivalent candidate. In many cases, the number of services invoked by a user is quite limited, resulting in a large number of missing QoS values and sparseness of data. Consequently, predicting QoS values of the services is important for users to find the exact service among many functionally similar services. However, improving the accuracy of QoS prediction is still a problem. Despite the successful results of the proposed QoS prediction methods, there are still a set of issues that should be addressed, such as Sparsity and Overfitting. To address these issues and improve prediction accuracy. In this paper, we propose a novel framework for predicting QoS values and reduce prediction error. This framework named auto-encoder for neighbor features (Auto-NF) consists of three steps. In the first step, we propose an extended similarity computation method based on Euclidean distance to compute the similarity between users and find similar neighbors. In the second step, we form clusters of similar neighbors and partition the initial matrix into sub-matrices based on these clusters to reduce the data sparsity problem. In the third step, we propose a simple neural network autoencoder that can learn deep features and select an ideal number of latent factors to reduce the overfitting phenomenon. To validate and evaluate our method, we conduct a series of experiments use a real QoS dataset with different data densities. The experimental results demonstrate that our method achieves higher prediction accuracy compared to existing methods.

Time-Dependent Visiting Trip Planning with Crowd Density Prediction Based on Internet of Things Localization

This paper proposes a time-dependent visiting trip planning (TVTP) framework to find the most efficient visiting order and plan the fastest moving paths based on Internet of Things (IoT) localization. The proposed TVTP framework consists of a deep learning based crowd density prediction model and a time-dependent visiting trip planning algorithm. In the developed prediction model, densely connected convolutional networks are explored with spatiotemporal data fusion to further reduce prediction errors. In the designed planning algorithm, visitors are guided to multiple target places at feasible time points to minimize total moving time based on predicted future crowd densities. According to our review of relevant research, this is the first framework that integrates deep learning for crowd density prediction with time-dependent planning for a multi-target visiting trip, which can precisely estimate future moving times based on predicted crowd densities and efficiently plan the optimal visiting order and guiding paths with the shortest total moving time to visit all target places. The open crowd dataset is adopted to evaluate the performance of existing works and TVTP. Experimental results show that our framework outperforms existing methods and can accurately predict the future crowd density of indoor people as well as significantly reduce the total moving time in the planned multi-target visiting trip.

Generating Synthetic Data to Reduce Prediction Error of Energy Consumption

Renewable and nonrenewable energy sources are widely incorporated for solar and wind energy that produces electricity without increasing carbon dioxide emissions. Energy industries worldwide are trying hard to predict future energy consumption that could eliminate over or under contracting energy resources and unnecessary financing. Machine learning techniques for predicting energy are the trending solution to overcome the challenges faced by energy companies. The basic need for machine learning algorithms to be trained for accurate prediction requires a considerable amount of data. Another critical factor is balancing the data for enhanced prediction. Data Augmentation is a technique used for increasing the data available for training. Synthetic data are the generation of new data which can be trained to improve the accuracy of prediction models. In this paper, we propose a model that takes time series energy consumption data as input, pre-processes the data, and then uses multiple augmentation techniques and generative adversarial networks to generate synthetic data which when combined with the original data, reduces energy consumption prediction error. We propose TGAN-skip-Improved-WGAN-GP to generate synthetic energy consumption time series tabular data. We modify TGAN with skip connections, then improve WGAN-GP by defining a consistency term, and finally use the architecture of improved WGAN-GP for training TGAN-skip. We used various evaluation metrics and visual representation to compare the performance of our proposed model. We also measured prediction accuracy along with mean and maximum error generated while predicting with different variations of augmented and synthetic data with original data. The mode collapse problem could be handled by TGAN-skip-Improved-WGAN-GP model and it also converged faster than existing GAN models for synthetic data generation. The experiment result shows that our proposed technique of combining synthetic data with original data could significantly reduce the prediction error rate and increase the prediction accuracy of energy consumption.

Rethinking Satellite Data Merging: From Averaging to SNR Optimization

Merging of multiple satellite datasets is a simple yet effective way to reduce prediction error. However, most merging methods for satellite data today are based on weighted averaging first proposed in 1969 for economic forecasting, which does not provide optimal outcomes when applied to satellite data. If our aim is to produce a merged data product that minimizes the prediction errors against a prediction target, there is no reason to insist that the merged product be an average of the parent datasets. A more disciplined approach based on mathematical optimization would be to minimize prediction errors. However, formulating merging as an optimization problem is insufficient by itself as the statistics needed for optimization, e.g., signal-to-noise ratio (SNR) of parent products, are often unavailable in practice and must be estimated jointly. In this article, we address both of these problems for data merging. We first formulate optimal merging of satellite data as a SNR optimization (SNR-opt), and propose an estimation method to jointly estimate the required SNRs. This SNR-based approach has a natural interpretation as a multi-input single-output Wiener filter. Through extensive experimental validation on three global-scale satellite-derived soil moisture and land surface temperature products, we demonstrate that our SNR-opt significantly improves merging over the existing weighted averaging scheme.

Soil texture prediction through stratification of a regional soil spectral library

Knowing the spatial distribution of soil texture, which is a physical property, is essential to support agricultural and environmental decision making. Soil texture can be estimated using visible, near infrared, and shortwave infrared (Vis-NIR-SWIR) spectroscopy. However, the performance of spectroscopic models is variable because of soil heterogeneity. Currently, few studies address the effects of soil sample variability on the performance of the models, especially for larger spectral libraries that include soils that are more heterogeneous. Therefore, the objectives of this study were to: i) apply Vis-based color parameters on the stratification of a regional soil spectral library; ii) evaluate the performance of the predictive models generated from the spectral library stratification; iii) compare the performance of stratified models (SMs) and the model without stratification (WSM), and iv) explain possible changes in prediction accuracy based on the SMs. Thus, a regional soil spectral library with 1 535 samples from the State of Santa Catarina, Brazil was used. Soil reflectance data were obtained by Vis-NIR-SWIR spectroscopy in the laboratory using a spectroradiometer covering the 350–2 500 nm spectral range. Sand, silt, and clay fractions were determined using the pipette method. Twenty-two components of color parameters were derived from the Vis spectrum using the colorimetric models. A cubist regression algorithm was used to assess the accuracy of the applicability of the initial models (SMs and WSM) and of the validation between the clusters. Fractional order derivatives (FODs) at 0.5, 1.5, and 2 intervals were used to explain possible changes in the performance of the SMs. The SMs with higher contents of clay and iron oxides obtained the highest accuracy, and the most important spectral bands were identified, mainly in the 480–550 and 850–900 nm ranges and the 1 400, 1 900, and 2 200 nm bands. Therefore, stratification of soil spectral libraries is a good strategy to improve regional assessments of soil resources, reducing prediction errors in the qualitative determination of soil properties.

Cross-Attention Fusion Based Spatial-Temporal Multi-Graph Convolutional Network for Traffic Flow Prediction.

Accurate traffic flow prediction is essential to building a smart transportation city. Existing research mainly uses a given single-graph structure as a model, only considers local and static spatial dependencies, and ignores the impact of dynamic spatio-temporal data diversity. To fully capture the characteristics of spatio-temporal data diversity, this paper proposes a cross-Attention Fusion Based Spatial-Temporal Multi-Graph Convolutional Network (CAFMGCN) model for traffic flow prediction. First, introduce GCN to model the historical traffic data’s three-time attributes (current, daily, and weekly) to extract time features. Second, consider the relationship between distance and traffic flow, constructing adjacency, connectivity, and regional similarity graphs to capture dynamic spatial topology information. To make full use of global information, a cross-attention mechanism is introduced to fuse temporal and spatial features separately to reduce prediction errors. Finally, the CAFMGCN model is evaluated, and the experimental results show that the prediction of this model is more accurate and effective than the baseline of other models.

Data-Driven Remaining Useful Life Prediction for Lithium-Ion Batteries Using Multi-Charging Profile Framework: A Recurrent Neural Network Approach

Remaining Useful Life (RUL) prediction for lithium-ion batteries has received increasing attention as it evaluates the reliability of batteries to determine the advent of failure and mitigate battery risks. The accurate prediction of RUL can ensure safe operation and prevent risk failure and unwanted catastrophic occurrence of the battery storage system. However, precise prediction for RUL is challenging due to the battery capacity degradation and performance variation under temperature and aging impacts. Therefore, this paper proposes the Multi-Channel Input (MCI) profile with the Recurrent Neural Network (RNN) algorithm to predict RUL for lithium-ion batteries under the various combinations of datasets. Two methodologies, namely the Single-Channel Input (SCI) profile and the MCI profile, are implemented, and their results are analyzed. The verification of the proposed model is carried out by combining various datasets provided by NASA. The experimental results suggest that the MCI profile-based method demonstrates better prediction results than the SCI profile-based method with a significant reduction in prediction error with regard to various evaluation metrics. Additionally, the comparative analysis has illustrated that the proposed RNN method significantly outperforms the Feed Forward Neural Network (FFNN), Back Propagation Neural Network (BPNN), Function Fitting Neural Network (FNN), and Cascade Forward Neural Network (CFNN) under different battery datasets.