Predictive Maintenance Techniques Research Articles

In situ monitoring of industrial-scale chemical vapor deposition using residual gas analysis

The maintenance of industrial-scale equipment demands more effort and incurs higher costs compared to that of laboratory-scale equipment, necessitating the use of predictive maintenance techniques to mitigate unforeseen outcomes. This paper presents a demonstration of the in situ monitoring of industrial-scale low-pressure chemical vapor deposition equipment by using a residual gas analyzer (RGA). The RGA provides reactions of precursor gases by tracing the concentration and composition of residual gases within vacuum reactors. Furthermore, RGA analysis enables the prediction of the growth rate of the thin-film, providing insights into thin-film growth dynamics during the process. In situ monitoring using an RGA can be applied to diverse semiconductor fabrication processes, including thin-film growth, oxidation, reactive ion etching, and ion implantation, owing to its compatibility with vacuum processes. This study addresses the existing difficulties in the exploration of RGA monitoring techniques for industrial-scale equipment and bridges the disparity between laboratory-scale and industrial-scale processes.

A Comprehensive Survey of Predictive Maintenance Techniques for Aircraft Engines Utilizing the C-MAPSS Dataset

The application of deep learning and sophisticated machine learning techniques is driving the rapid advancement of aircraft engine prognostics and predictive maintenance. Remain ing Useful Life (RUL) of aviation engines has been the subject of numerous studies aimed at improving prediction accuracy and efficacy to improve aviation safety and maintenance plans. Innovative approaches and technologies are demonstrated by these projects, which use a variety of methodologies and datasets, including C-MAPSS and N-CMAPSS. Combining feature engi neering, ensemble learning, and deep learning models such as Restricted Boltzmann Machines (RBMs), Long Short-Term Mem ory (LSTM) networks, Convolutional Neural Networks (CNNs), and Deep Bidirectional Recurrent Neural Networks (DBRNNs) is one prominent method. Features are chosen and models are optimized using a variety of methods, including Genetic Algorithms, Recursive Feature Elimination, Lasso, and Feature Importances. In prognostic modeling, the research emphasize the need of interpretability, model adaptability, and measuring uncertainty. Additionally, in order to pinpoint important features and improve model transparency, the study investigates the use of explainable AI techniques such aggregated feature importances with cross-validation (AFICv) and Shapley additive explanation (SHAP). In order to capture prediction uncertainties, the inte gration of Gaussian Processes (GPs) and Bayesian Deep Neural Networks (DNNs) is also investigated. This provides insights into uncertainty-aware prognosis and predictive analytics for industrial assets. The development and publication of datasets such as the N-CMAPSS dataset also makes it possible to conduct more comprehensive and realistic assessments of prognostic models under real-world flight conditions, providing useful tools for benchmarking and improving machine learning algorithms in predictive maintenance. All things considered, these research projects highlight current developments and the possibility of combining cutting edge technology to improve system reliability, improve predictive maintenance techniques, and guarantee safer airline operations. Key Words: Aircraft engine, Prognostics, Predictive main tenance, Machine learning, Deep learning, Remaining Useful Life (RUL), Aviation safety, Feature engineering, Ensemble learning, Uncertainty quantification, Explainable AI, Bayesian Deep Neural Networks (DNNs), Gaussian Processes (GPs), N CMAPSS dataset, Real flight conditions, Industrial assets, System reliability, Benchmarking

Enhancing Transparency and Accountability in Predictive Maintenance with Explainable AI

Predictive maintenance is a critical aspect of industrial operations, enabling proactive identification and mitigation of potential failures in machinery and equipment. However, the widespread adoption of AI-driven predictive maintenance solutions has been hindered by the opaque nature of many machines learning models, raising concerns about transparency, accountability, and trust. This research aims to address these challenges by developing explainable AI techniques for predictive maintenance in industrial systems. By integrating interpretability methods with advanced predictive models, we seek to enhance the transparency and interpretability of AI-driven maintenance decisions. Our proposed methodology combines state-of-the-art machine learning algorithms with local and global explainability techniques, such as LIME, SHAP, and feature importance analysis. Through extensive experiments on real-world industrial data, we evaluate the performance of our explainable AI models and demonstrate their ability to provide insightful explanations, enabling domain experts to understand the underlying reasoning and critical factors contributing to maintenance predictions. Furthermore, we explore the impact of explainable AI on improving trust, accountability, and adoption of AI systems in industrial predictive maintenance scenarios. Keywords— Predictive Maintenance, Explainable AI (XAI), Machine Learning, Interpretability, LIME, SHAP, Feature Importance, Industrial Systems, Trust in AI, Accountability.

Systematic review of predictive maintenance and digital twin technologies challenges, opportunities, and best practices.

Maintaining machines effectively continues to be a challenge for industrial organisations, which frequently employ reactive or premeditated methods. Recent research has begun to shift its attention towards the application of Predictive Maintenance (PdM) and Digital Twins (DT) principles in order to improve maintenance processes. PdM technologies have the capacity to significantly improve profitability, safety, and sustainability in various industries. Significantly, precise equipment estimation, enabled by robust supervised learning techniques, is critical to the efficacy of PdM in conjunction with DT development. This study underscores the application of PdM and DT, exploring its transformative potential across domains demanding real-time monitoring. Specifically, it delves into emerging fields in healthcare, utilities (smart water management), and agriculture (smart farm), aligning with the latest research frontiers in these areas. Employing the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) criteria, this study highlights diverse modeling techniques shaping asset lifetime evaluation within the PdM context from 34 scholarly articles. The study revealed four important findings: various PdM and DT modelling techniques, their diverse approaches, predictive outcomes, and implementation of maintenance management. These findings align with the ongoing exploration of emerging applications in healthcare, utilities (smart water management), and agriculture (smart farm). In addition, it sheds light on the critical functions of PdM and DT, emphasising their extraordinary ability to drive revolutionary change in dynamic industrial challenges. The results highlight these methodologies' flexibility and application across many industries, providing vital insights into their potential to revolutionise asset management and maintenance practice for real-time monitoring. Therefore, this systematic review provides a current and essential resource for academics, practitioners, and policymakers to refine PdM strategies and expand the applicability of DT in diverse industrial sectors.

Failure Prediction of Coal Mine Equipment Braking System Based on Digital Twin Models

The primary function of a mine hoist is the transportation of personnel and equipment, serving as a crucial link between underground and surface systems. The proper functioning of key components such as work braking and safety braking is essential for ensuring the safety of both personnel and equipment, thereby playing a critical role in the safe operation of coal mines. As coal mining operations extend to greater depths, they introduce heightened challenges for safe transportation, compounded by increased equipment loss. Consequently, there is a pressing need to enhance safety protocols to safeguard personnel and materials. Traditional maintenance and repair methods, characterized by routine equipment inspections and scheduled downtime, often fall short in addressing emerging issues promptly, leading to production delays and heightened risks for maintenance personnel. This underscores the necessity of adopting predictive maintenance strategies, leveraging digital twin models to anticipate and prevent potential faults in mine hoists. In summary, the implementation of predictive maintenance techniques grounded in digital twin technology represents a proactive and scientifically rigorous approach to ensuring the continued safe operation of mine hoists amidst the evolving challenges of deepening coal mining operations. In this study, we propose the integration of a CNN-LSTM algorithm within a digital twin framework for predicting faults in mine hoist braking systems. Utilizing software such as AMESim 2019 and MATLAB 2016b, we conduct joint simulations of the hoist braking digital twin system. Subsequently, leveraging the simulation model, we establish a fault diagnosis platform for the hoist braking system. Finally, employing the CNN-LSTM network model, we forecast failures in the mine hoist braking system. Experimental findings demonstrate the effectiveness of our proposed algorithm, achieving a prediction accuracy of 95.35%. Comparative analysis against alternative algorithms confirms the superior performance of our approach.

Screening of Machine Learning Techniques on Predictive Maintenance: a Scoping Review

Predictive maintenance (PdM) is a set of actions and techniques to early detect failures and defects on machines before they occur, and the usage of machine learning and deep learning techniques in predictive maintenance has increased during the last years. Even with this increase of the literature, there is still a gap concerning the application of such techniques for PdM in the industry, as there are no clear guidelines about which information to use for a PdM system, how to process the information, and what machine learning techniques should be used in order to obtain acceptable results. This scoping review is performed in order to observe the current status on the use of Machine Learning and Deep Learning in predictive maintenance in academia and provide answer to the questions related to these guidelines. For this purpose, a literature review of the last five years is carried out, using those articles that cover information about sources of information used for PdM, the treatment given to such data and the machine learning (ML) methods or techniques used. The Web of Science: Core Collection database is used as a source of information, specifically the Science Citation Index Expanded (SCIE). The review shows that there are different information sources used for machine learning and deep learning in PdM, depending on the origin of the data and the availability of it, and as well whether the data sets are private or public. Also, we can observe that data used for both training and making predictions does not only use traditional pre-processing techniques, but that about one fifth of the articles even propose new techniques in this field. Additionally, we compare a wide range of techniques and algorithms which are used in Deep Learning -being ANN the most used- and in Machine Learning, being SVM the most used algorithm, closely followed by Random Forest. Based on the results, we provide indications about how to apply ML for PdM in industry. Keywords: machine learning, predictive maintenance, artificial intelligence, deep learning, data processing, data collection

Run-to-Failure Relay Dataset for Predictive Maintenance Research With Machine Learning

Predictive maintenance (PdM) offers enormous potential for improvement, both economically and ecologically. The extended use of components can improve sustainability, minimize the occurrence of component failures and associated safety risks for people and the environment, and ultimately increase the profitability of plants. PdM is therefore a future key technology that requires sensing, communication of collected data, and analysis of data to implement. In this work, the phoenix contact relay (PCR) dataset is provided and used to demonstrate that there is a need for research into new PdM techniques. The PCR dataset presented includes life data from 546 relays with a total of more than 106 million switching cycles collected over a period of several years at a sampling rate of 10 kHz. Manufacturers and loads were varied to ensure practical relevance. This sets PCR apart from previously published datasets in terms of number of measurements, complexity of measurements, scope of units tested, and practicality, providing the opportunity to design and evaluate new methods for PdM.

Optimizing energy efficiency in data center cooling towers through predictive maintenance and project management

Optimizing energy efficiency in data center cooling towers is crucial for reducing operational costs and environmental impact. This review explores the integration of predictive maintenance and project management to achieve this goal. By leveraging predictive maintenance techniques, data center operators can anticipate and address potential issues before they lead to costly downtime or inefficiencies. Project management plays a key role in coordinating these efforts, ensuring that maintenance activities are carried out efficiently and effectively. Predictive maintenance relies on data analytics and machine learning algorithms to monitor the condition of cooling towers in real-time. By analyzing data such as temperature, pressure, and flow rates, these algorithms can detect anomalies and predict potential failures. This proactive approach allows data center operators to schedule maintenance activities during planned downtime, minimizing disruptions to operations. Project management practices, such as Agile or Waterfall methodologies, are essential for coordinating predictive maintenance efforts. Project managers oversee the planning, execution, and monitoring of maintenance activities, ensuring that they are completed on time and within budget. They also facilitate communication between stakeholders, including maintenance teams, data analysts, and management, to ensure that everyone is aligned and working towards the same goals. By integrating predictive maintenance with project management, data center operators can achieve significant improvements in energy efficiency. By addressing maintenance issues proactively, operators can reduce energy consumption, extend the lifespan of cooling equipment, and minimize downtime. Additionally, project management practices ensure that these efforts are coordinated and effective, maximizing the benefits of predictive maintenance. In conclusion, optimizing energy efficiency in data center cooling towers requires a holistic approach that combines predictive maintenance and project management. By leveraging predictive maintenance techniques and project management practices, data center operators can achieve significant improvements in energy efficiency, reduce operational costs, and enhance environmental sustainability.

Selection Criteria for Evaluating Predictive Maintenance Techniques for Rotating Machinery using the Analytic Hierarchical Process (AHP)

A primary industry objective is to ensure that machinery remains operational through effective management. Predictive maintenance plays a significant role in monitoring the working condition of machinery. The goal of this study is to establish the criteria for evaluating predictive maintenance techniques for rotating machines utilizing the Analytic Hierarchy Process (AHP). To achieve this target, survey data were collected from questionnaires and interviews with 20 experts, which had at least 20 years of professional experience. The Analytic Hierarchy Process (AHP) was utilized for criteria selection. The findings disclosed that predictive maintenance factors for rotating machines were ranked as follows: vibration analysis (45.5%), acoustic analysis (22.7%), oil analysis (22.4%), infrared thermography (5.8%), and wear particle analysis (3.6%). The Consistency Ratio (CR) was determined to be less than 10%, indicating a high level of consensus among the experts. Given the elevated importance attributed to vibration analysis, it can be concluded that the latter is the primary criterion for selecting predictive maintenance techniques for rotating machines.

Simulation-based digital twin of a high-speed turbomachine (fan) used in avionics cooling

In this contribution, the simulation-based digital twin of a high-speed turbomachine, i.e. a fan with a nominal speed of 16500 rpm, that has been developed for the aerospace industry with multiple uses in various platforms, and the corresponding findings are presented. The relevant simulation-based digital twin is created by first running various multi-physics engineering simulations, i.e. computational solid mechanics, computational fluid dynamics, and low-frequency electromagnetic analyses with Ansys software. Subsequent dynamic reduced order models (ROM) are formed and integrated together to create the digital twin that are bidirectionally connected with the asset. The digital twin is also validated via experimental data that are obtained with various setups for several parameters read at discrete locations via sensors, based on fidelity comparisons both between experiments and simulations, and between simulations and ROMs. Several operational scenarios including those with stress tests are run and the asset is checked against fatigue and thermal requirements for the rotor-stator assembly and the motor circuit respectively. This is done with an inhouse program that incrementally reads time, temperature, and rotational speed data and in which the relevant fatigue and thermal criteria are defined. This program is also used to simulate and analyse what-if scenarios. It is seen that vibration fatigue dominates all possible sources of failure in most cases. As a novel aspect, the fatigue induced by starts and stops of the fan is expressed as equivalent operating hours (EOH) depending on time and temperature parameters of the corresponding scenarios, as opposed to be taken as a constant value for all scenarios. Using this novelty, it is also conceptually demonstrated that simulation-based digital twins can play a significant role in operation and maintenance (O&M) when combined with conventional data-driven predictive maintenance techniques, not only for O&M teams but also asset owners and original equipment manufacturers (OEMs) particularly with the use of what-if analyses. Approaches to simulation-based digital twins in context with multi-physics and ROM fidelity are also discussed based on several digital twin maturity models.

IoT-Enabled predictive maintenance for sustainable transportation fleets

This study explores the use of Internet of Things (IoT) based predictive maintenance techniques for sustainable transportation fleets. It utilizes various datasets to enhance operational efficiency and reduce environmental consequences. An examination of the fleet data uncovers interesting findings: the average mileage of the fleet is about 28,400 miles, indicating that different vehicles have been used to different extents. Notably, vehicle 002 stands out with the greatest mileage of 32,000 miles. Varying sensor measurements reveal discrepancies in tire pressure, brake pad thickness, and oil levels, suggesting different patterns of wear across the fleet. The historical maintenance data highlight the differences in maintenance intervals among automobiles. Based on predictive maintenance analysis, it is projected that vehicle 001 will need its next oil change after covering 27,000 miles, which is an increase of 2,000 miles compared to its last service. Percentage change study demonstrates the ever-changing nature of maintenance needs, highlighting the need of customized maintenance interventions that are specifically designed for each vehicle's unique characteristics. The combination of these discoveries clarifies the potential of IoT-enabled predictive maintenance in customizing tailored maintenance plans, increasing fleet efficiency, and reducing environmental impact. This research offers practical insights for adopting proactive maintenance techniques, promoting sustainability, and improving operational efficiency in transportation fleets.

Assets Performance Management systems for hydroelectric power plants—A survey

This paper presents a survey about several works in the field of predictive maintenance for hydroelectric power plants related to Assets Performance Management systems. It discusses about how to make a system study to better understand the structure of the Hydropower Plants and its components and how to define the most significant items to the maintenance. Also, it presents predictive maintenance techniques in terms of failure detection, diagnosis and prognostics. The research shows that data-driven methods are increasingly used by Hydropower Plants. With the recent development of artificial intelligence methods, it is possible to discover unknown patterns in sensor data and learn existing relationships in the data that may not be easy for a human to discover. In this way, some hybrid methods, which mix concepts such as the transformation of time series data into the frequency domain, statistical modeling and machine learning, are being developed to perform the tasks of APM systems. The work also shows the possibility of developing methods focusing on reinforcement learning or explainable artificial intelligence for the industrial sector. Such advances can count as a contribution to the innovative nature of the electricity sector, which tends to become increasingly sustainable in order to mitigate the environmental impacts caused by production.

MACHINE LEARNING INNOVATIONS IN PREDICTIVE MAINTENANCE: A COMPREHENSIVE REVIEW OF APPLICATIONS IN THE MINING SECTOR

In recent years, machine learning (ML) has burgeoned as a transformative tool, particularly within predictive maintenance applications. The mining sector, characterized by its heavy machinery and capital-intensive equipment, stands to benefit immensely from advancements in predictive maintenance techniques. This comprehensive review delves into the recent innovations in ML-driven predictive maintenance and their significant applications within the mining industry. Drawing from an array of case studies and empirical analyses, this paper underscores the tangible operational efficiencies and cost-saving benefits brought about by these ML methodologies. Furthermore, it offers critical insights into the challenges, best practices, and the potential future trajectory of this intersection of machine learning and mining operations.

The Optimization of a Model for Predicting the Remaining Useful Life and Fault Diagnosis of Landing Gear

With the development of next-generation airplanes, the complexity of equipment has increased rapidly, and traditional maintenance solutions have become cost-intensive and time-consuming. Therefore, the main objective of this study is to adopt predictive maintenance techniques in daily maintenance in order to reduce manpower, time, and the cost of maintenance, as well as increase aircraft availability. The landing gear system is an important component of an aircraft. Wear and tear on the parts of the landing gear may result in oscillations during take-off and landing rolling and even affect the safety of the fuselage in severe cases. This study acquires vibration signals from the flight data recorder and uses prognostic and health management technology to evaluate the health indicators (HI) of the landing gear. The HI is used to monitor the health status and predict the remaining useful life (RUL). The RUL prediction model is optimized through hyperparameter optimization and using the random search algorithm. Using the RUL prediction model, the health status of the landing gear can be monitored, and adaptive maintenance can be carried out. After the optimization of the RUL prediction model, the root-mean-square errors of the three RUL prediction models, that is, the autoregressive model, Gaussian process regression, and the autoregressive integrated moving average, decreased by 45.69%, 55.18%, and 1.34%, respectively. In addition, the XGBoost algorithm is applied to simultaneously output multiple fault types. This model provides a more realistic representation of the actual conditions under which an aircraft might exhibit multiple faults. With an optimal fault diagnosis model, when an anomaly is detected in the landing gear, the faulty part can be quickly diagnosed, thus enabling faster and more adaptive maintenance. The optimized multi-fault diagnosis model proposed in this study achieves average accuracy, a precision rate, a recall rate, and an F1 score of more than 96.8% for twenty types of faults.

AER-HYBRITECH: Averaging Ensemble Regression with Hybrid Encoding and Enhanced Feature Selection Technique for Predictive Maintenance

Predictive maintenance is critical to modern industrial operations, preventing unexpected equipment failures and minimizing downtime. Existing methods often encounter challenges related to data preprocessing, missing data imputation, and feature selection. This paper presents "AER-HYBRITECH," a novel approach that addresses these challenges and enhances the predictive maintenance process. Traditional methods overlook the intricate relationships within the data, resulting in suboptimal predictive performance. To bridge this gap, the proposed AER-HYBRITECH algorithm is introduced. AER-HYBRITECH stands out in several ways. Firstly, it utilizes a hybrid encoding technique that converts categorical data into a more informative numerical representation by incorporating the average values of label-encoded data and its frequency, leading to improved feature utilization. Furthermore, it introduces the AER-MDI (Averaging Ensemble Regression-based Missing Data Imputation) technique, which combines M5P, REPTree, and linear regression models to impute missing data, ensuring a more complete dataset. The algorithm also implements Min-Max normalization to scale numeric features, making them compatible for further analysis. One of the key innovations of AER-HYBRITECH is its enhanced hybrid feature selection (EHFS) approach. The AER-HYBRITECH algorithm transforms and preprocesses the data and ensures that predictive maintenance models are built on a solid foundation, resulting in more accurate predictions and reduced maintenance costs.

A Comprehensive Review of Real-time Monitoring and Predictive Maintenance Techniques: Revolutionizing Natural Fibre Composite Materials Maintenance with IoT

Integrating the Internet of Things (IoT) and natural fiber-reinforced polymer composites (NFPCs) can revolutionize monitoring and maintaining composites. By incorporating sensors and wireless communication technology into the composites, real-time monitoring and predictive maintenance can be achieved. This review provides a comprehensive overview of the current state-of-the-art in the use of IoT for real-time monitoring and predictive maintenance of NFPCs. This paper covers the various types of sensors used, IoT networks and protocols employed, and data analysis techniques to detect potential issues and predict failures. This paper also highlights the benefits and challenges of using IoT for composite maintenance and this technology’s future directions and potential applications. This review provides valuable insights for researchers, engineers, and practitioners in composites, the IoT, and predictive maintenance.

An explainable deep learning approach for detection and isolation of sensor and machine faults in predictive maintenance paradigm

The predictive health maintenance techniques identify the machine faults by analyzing the data collected by low-cost sensors assuming that sensors are free from any faults. However, aging and environmental condition cause sensors also be faulty, leading to incorrect interpretations of the collected data and subsequently resulting in erroneous machine health predictions. To mitigate this problem, this paper proposes a hybrid model that can differentiate between sensor and system faults. The data used for training the model is collected from a power system hardware setup by experimental procedures. A convolutional neural network (CNN) model is used to extract optimized features from the raw data automatically, which are then fed to the eXtreme Gradient Boosting (XGBoost) model for sensor and machine fault isolation with an overall accuracy of 98.15%. The data having sensor fault was then fed to a deep autoencoder, which eliminated the sensor fault components and reconstructed the data with an average root mean square error of 0.0576. Thereafter, the corrected signal was used to detect the system fault using the hybrid CNN-XGBoost model with 99.77% accuracy. Therefore, by isolating the sensor faults, the proposed technique establishes better confidence in predictive maintenance. Further, explainable AI has been utilized to interpret the model prediction in human-understandable terms in order to increase trustworthiness.

Health Prognostics Classification with Autoencoders for Predictive Maintenance of HVAC Systems

Buildings’ heating, ventilation, and air-conditioning (HVAC) systems account for significant global energy use. Proper maintenance can minimize their environmental footprint and enhance the quality of the indoor environment. The adoption of Internet of Things (IoT) sensors integrated into HVAC systems has paved the way for predictive maintenance (PdM) grounded in real-time operational metrics. However, HVAC systems without such sensors cannot leverage the advantages of current data-driven PdM techniques. This work introduces a novel data-driven framework, the health prognostics classification with autoencoders (HPC-AE), designed specifically for PdM. It utilizes solely HVAC power consumption and outside temperature readings for its operations, both of which are readily obtainable. The primary objective of the HPC-AE framework is to facilitate PdM through a health prognostic approach. The HPC-AE framework utilizes an autoencoder for feature enrichment and then applies an artificial neural network to classify the daily health condition of an HVAC system. A multi-objective evaluation metric is employed to ensure optimal performance of the autoencoder within this framework. This metric evaluates the autoencoder’s proficiency in reducing reconstruction discrepancies in standard data conditions and its capability to differentiate between standard and degraded data scenarios. The HPC-AE framework is validated in two HVAC fault scenarios, including a clogged air filter and air duct leakage. The experimental results show that compared to methods used in similar studies, HPC-AE exhibits a 5.7% and 2.1% increase in the F1 score for the clogged air filter and duct leakage scenarios.

Linear regression and R-squared correlation analysis on major nuclear online plant cooling system

The primary cooling system is an integral part of a nuclear reactor that maintains reactor operational safety. It is essential to investigate the effects of the cooling system parameter before implementing predictive maintenance techniques in the reactor monitoring system. This paper presents a linear regression and R-squared correlation analysis of the nuclear plant cooling system parameter in the TRIGA PUSPATI Reactor in Malaysia. This research examines the primary cooling system's temperature, conductivity, and flow rate in maintaining the nuclear reactor. Data collection on the primary coolant system has been analyzed, and correlation analysis has been derived using linear regression and R-squared analysis. The result displays the correlation matrix for all sensors in the primary cooling system. The R-squared value for TT5 versus TT2 is 89%, TT5 versus TT3 is 94%, and TT5 against TT4 is 66% which shows an excellent correlation to the linear regression. However, the conductivity sensor CT1 does not correlate with other sensors in the system. The flow rate sensor FT1 positively correlates with the temperature sensor but does not correlate with the conductivity sensor. This finding can help to better develop the predictive maintenance strategy for the reactor monitoring program.

Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm

Machines can make an appropriate operating and maintenance choice when defects are accurately and promptly predicted. Researchers choose data-driven predictive maintenance techniques for making the prediction more quickly and economically than alternative methods. To maintain secure and dependable manufacturing operations, the production equipment needs predictive maintenance. It is crucial to have an efficient predictive maintenance model for preventing unexpected shutdown by defects during production. The majority of related research focuses on early fault warnings but ignores how different issues differ in severity. Making the right maintenance policy is extremely important for a manufacturing organization because maintenance affects employee safety, economy, reliability, and availability. To solve these difficulties in the automotive sector, an efficient predictive maintenance strategy with deep structured architecture is offered to predict faults in the machines and increase the lifespan of the equipment. Firstly, the required data is collected from external sources for the predictive maintenance system. The garnered data is given to the pre-processing section and the transformation stage. Secondly, the transformed data is subjected to the autoencoder for selecting the optimal encoded vectors with the Henry Gas Solubility Search and Rescue Optimization (HGSSRO). The selected features are given to deep hybrid learning with Serial Cascaded Deep Learning (SCDL) to predict the occurrence of faults. Here, the autoencoder obtains the encoded vectors from the optimally selected features. Then, the optimal encoded vectors are subjected to the LSTM to acquire the features. The extracted features are given to DNN for the final prediction of failures. In this hybrid SCDL, the variables are optimized by the same HGSSRO to improve the prediction performance. The efficacy of the developed deep learning-aided predictive system is validated ed by comparing the conventional prediction models.