A proposed approach to the classification of bearing condition using wavelets and random forests

Background: Intensive Care Unit (ICU) has the highest mortality rate in the world. ICU has special equipment that leads to the hospital's most costly parts. The length of stay in the ICU is a special issue, and reducing this time is a practical approach. We aimed to use artificial intelligence to help early and timely diagnosis of the disease to help with health. Methods: We designed a rule-based intelligent system to predict the length of stay and the mortality rate of trauma patients in ICU. A neuro-Fuzzy and eight machine learning models were used to predict the mortality rate in trauma patients in ICU. The performances of these techniques were evaluated with accuracy, sensitivity, specificity, and area under the ROC curve. Decision-Table was used to predict the length of stay in trauma patients in ICU. For comparison, eight machine learning models were used. The method is compared based on Mean absolute error and relative absolute error (%). Results: Neuro-Fuzzy expert system and Decision-Table showed better results than other techniques. Accuracy, sensitivity, specificity, and ROC Area of Nero-Fuzzy are 83.6735, 0.9744, 0.3000, 0.8379, and 1, respectively. The mean absolute error and Relative absolute error (%) of the Decision-Table model are 4.5426 and 65.4391, respectively. Conclusion: Neuro-Fuzzy expert system with the highest level of accuracy and a Decision-Table with the lowest Mean absolute error, which are rule-based models, are the best models. Therefore, these models are recommended as a valuable tool for prediction parameters of ICU as well as medical decision-making.

show abstract

“…Better neural network and SVM performance ( 26 ). Machine learning techniques were used to predict the length of stay and mortality ( 28 ).…”

Section: Discussionmentioning

confidence: 99%

A Rule Based Intelligent Software to Predict Length of Stay and the Mortality Rate in Trauma Patients in the Intensive Care Unit

Montazeri¹,

Ahmadinejad²,

Montazeri³

et al. 2023

ijph

View full text Add to dashboard Cite

show abstract

“…Compared with a single decision tree, the training speed of the Random Forest is relatively fast, and it is easy to make the parallel method. Moreover, the accuracy can still be maintained even though a large part of the feature is missing [17]. However, due to its many decision trees, a large amount of memory may be required on larger projects.…”

Section: Random Forestmentioning

confidence: 99%

A study on the crucial indicators for breast cancer detection using machine learning algorithm

2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Since breast cancer is the most serious disease affecting women, early detection comes as a priority. The Wisconsin Breast Cancer Dataset (WBCD), which was retrieved from the UCI database, has been applied in numerous studies in recent years to help with the definitive diagnosis. Machine learning (ML) algorithms, such as K-Nearest Neighbor (KNN), Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), Naive Bayes (NB), Support Vector Machine (SVM), and Neural Network (NN) can be used to attain the upshot. Although these algorithms make predictions well, advantages cannot overshadow drawbacks because the outcome is circumstantial by the peculiar dataset itself and cannot draw a direct conclusion reflecting the deeper issue. To implement ML skills to figure out the factors that influence the prediction most in a statistical dimension, this paper uses the dataset above, compares five methods, and chooses three best classifiers: KNN, RF, and SVM. After selection, the author eliminates every single variable each time to get the accuracy, and compares them with the full model’s accuracy. Having controlled variables, it can be informed that Clump Thickness and Bare Nuclei are the factors that matter most.

show abstract

“…Vakharia et al [10] calculated bearing signal features through the weighted gain method, and the Random Forest (RF) is used to classify the faults. Ferrenc and Lutovac [11] used wavelet decomposition to get the 18 dimensional characteristics of fault signals and used RF to classify them.…”

Section: Introductionmentioning

confidence: 99%

Bearing Intelligent Fault Diagnosis Based on Wavelet Transform and Convolutional Neural Network

Guo

Liu

et al. 2020

Shock and Vibration

View full text Add to dashboard Cite

As one of the important parts of modern mechanical equipment, the accurate real-time diagnosis of rolling bearing is particularly important. Traditional fault diagnosis methods have some disadvantages, such as low diagnostic accuracy and difficult fault feature extraction. In this paper, a method combining Wavelet transform (WT) and Deformable Convolutional Neural Network (D-CNN) is proposed to realize accurate real-time fault diagnosis of end-to-end rolling bearing. The vibration signal of rolling bearing is taken as the monitoring target. Firstly, the Orthogonal Matching Pursuit (OMP) algorithm is used to remove the harmonic signal and retain the impact signal and noise. Secondly, the time-frequency map of the signal is obtained by time-frequency transform using Wavelet analysis. Finally, the D-CNN is used for feature extraction and classification. The experimental results show that the accuracy of the method can reach 99.9% under various fault modes, and it can accurately identify the fault of rolling bearing.

show abstract

A proposed approach to the classification of bearing condition using wavelets and random forests

Cited by 5 publications

References 9 publications

A Rule Based Intelligent Software to Predict Length of Stay and the Mortality Rate in Trauma Patients in the Intensive Care Unit

A Rule Based Intelligent Software to Predict Length of Stay and the Mortality Rate in Trauma Patients in the Intensive Care Unit

A study on the crucial indicators for breast cancer detection using machine learning algorithm

Bearing Intelligent Fault Diagnosis Based on Wavelet Transform and Convolutional Neural Network

Contact Info

Product

Resources

About