Anomaly detection and confidence interval‐based replacement in decay state coefficient of ship power system DOI Creative Commons
Xingshan Chang,

Xinping Yan,

Bohua Qiu

et al.

IET Intelligent Transport Systems, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 14, 2024

Abstract The anomaly detection and predictive replacement of the degradation decay state coefficient ( D esc ) ship power system (SPS) are crucial for ensuring their operational safety maintenance efficiency. This study introduces YC3Model, a model based on dynamic triple sliding window mechanism, Gaussian process regression) to address this challenge. It combines temporal variation characteristics coefficient's original data, first‐order, second‐order differential data in both normal abnormal trend intervals. calculates three local statistical measures within each employs Z‐score method detection. combination windows reduces false positives negatives, enhancing precision For detected anomalies, regression is used prediction replacement, providing confidence intervals increase reliability predicted values. Experimental results demonstrate that YC3Model exhibits superior accuracy adaptability SPS, surpassing traditional methods across range evaluation metrics. confirms potential health monitoring offering reliable input intelligent operation (IO&M) SPS.

Language: Английский

Quantitative prediction of toxicological points of departure using two-stage machine learning models: A new approach methodology (NAM) for chemical risk assessment DOI

Vaisali Chandrasekar,

Saad Mohammad,

Omar M. Aboumarzouk

et al.

Journal of Hazardous Materials, Journal Year: 2025, Volume and Issue: 487, P. 137071 - 137071

Published: Jan. 10, 2025

Language: Английский

Citations

10
Outlier Detection Performance of a Modified Z-Score Method in Time-Series RSS Observation With Hybrid Scale Estimators DOI Creative Commons
Abdulmalik Shehu Yaro, Filip Malý, Pavel Pražák

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 12785 - 12796

Published: Jan. 1, 2024

The modified Z-score (mZ-score) method has been used to detect outliers in time series received signal strength (RSS) observations. Its performance is dependent on the scale estimator used, and each advantages disadvantages over others. One approach developing a that combines of two or more estimators through hybridization. In this paper, outlier detection mZ-score with different hybridization approaches for Sn median absolute deviation (MAD) determined analysed. Three hybrid are identified, namely weighted, maximum, average estimators. using three experimentally generated publicly available time-series RSS datasets. Based simulation results, weighted results best amongst When compared mean-shift-based (MOD) technique, k-means clustering-based density-based spatial clustering (DBSCAN) performs better little no false alarm negative detections.

Language: Английский

Citations

16
Combustion condition predictions for C2-C4 alkane and alkene fuels via machine learning methods DOI
Mingfei Chen,

Jiaying He,

Xuan Zhao

et al.

Fuel, Journal Year: 2024, Volume and Issue: 373, P. 132375 - 132375

Published: July 2, 2024

Language: Английский

Citations

1
Empirical study of outlier impact in classification context DOI
Hufsa Khan, Muhammad Tahir Rasheed, Shengli Zhang

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 256, P. 124953 - 124953

Published: Aug. 2, 2024

Language: Английский

Citations

0
Optimizing compressive strength prediction using adversarial learning and hybrid regularization DOI Creative Commons
Tamoor Aziz,

Haroon Aziz,

Srijidtra Mahapakulchai

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Aug. 7, 2024

The infrastructure industry consumes natural resources and produces construction waste, which has a detrimental impact on the environment. To mitigate these adverse effects reduce raw material consumption, waste materials can be repurposed to achieve sustainability. However, recycled deteriorate intrinsic properties of concrete. A suitable ratio aggregates produce desired compressive strength. Compiling sufficient data in civil engineering laboratories make reliable conclusions is time-consuming costly. Therefore, this research proposes novel approach for predicting strengths using limited data. generative adversarial network was employed generate synthetic Hybrid training, utilizing either conventional loss or heuristic loss, prevents model from overfitting by adaptively adjusting regularization term. Random noise multivariate normal distribution embedded heuristically into training samples capture intricate variations. Sensitivity analysis indicated that size coarse aggregate water are most significant features, aligning with their correlations. Interestingly, superplasticizer, density aggregate, absorption contributed significantly predictions despite low propounded method outperforms random forest, support vector regression, artificial neural network, adaptive boosting scoring mean squared error 7.97, root 2.82, absolute 2.13, coefficient determination 0.96. These results suggest proposed technique effectively contribute sustainable practices accurately strengths.

Language: Английский

Citations

0
Extended Representation Learning Based Neural Network Model for Outlier Detection DOI

Sidratul Muntaha,

Sohana Jahan,

Md. Anwarul Islam Bhuiyan

et al.

Journal of Artificial Intelligence Machine Learning and Neural Network, Journal Year: 2024, Volume and Issue: 46, P. 12 - 26

Published: Oct. 1, 2024

Outlier detection problems have drawn much attention in recent times for their variety of applications. An outlier is a data point that different from the rest and can be detected based on some measure. In years, Artificial Neural Networks (ANN) been used extensively finding outliers more efficiently. This method highly competitive with other methods currently use such as similarity searches, density-based approaches, clustering, distance-based linear methods, etc. this paper, we proposed an extended representation learning neural network. model follows symmetric structure like autoencoder where dimensions are initially increased original then reduced. Root mean square error to compute score. Reconstructed calculated analyzed detect possible outliers. The experimental findings documented by applying it two distinct datasets. performance compared several state-of-art approaches Rand Net, Hawkins, LOF, HiCS, Spectral. Numerical results show outperforms all these terms 5 validation scores, Accuracy (AC), Precision (P), Recall, F1 Score, AUC

Language: Английский

Citations

0
Anomaly detection and confidence interval‐based replacement in decay state coefficient of ship power system DOI Creative Commons
Xingshan Chang,

Xinping Yan,

Bohua Qiu

et al.

IET Intelligent Transport Systems, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 14, 2024

Abstract The anomaly detection and predictive replacement of the degradation decay state coefficient ( D esc ) ship power system (SPS) are crucial for ensuring their operational safety maintenance efficiency. This study introduces YC3Model, a model based on dynamic triple sliding window mechanism, Gaussian process regression) to address this challenge. It combines temporal variation characteristics coefficient's original data, first‐order, second‐order differential data in both normal abnormal trend intervals. calculates three local statistical measures within each employs Z‐score method detection. combination windows reduces false positives negatives, enhancing precision For detected anomalies, regression is used prediction replacement, providing confidence intervals increase reliability predicted values. Experimental results demonstrate that YC3Model exhibits superior accuracy adaptability SPS, surpassing traditional methods across range evaluation metrics. confirms potential health monitoring offering reliable input intelligent operation (IO&M) SPS.

Language: Английский

Citations

0