Study Title Description Industry

Lucas C. Brito et al 2022 An explainable This study explores Industrial machi

Kozielski et al 2023 Contextual ExplThis study explores ePower generatio
Dintén et al 2024 Design, Buildin This paper presents Water managem
Kisten et al 2024 Explainable ArtiThis study integratesSmart agricultur
Gawde et al 2024 Explainable Pre This study addressesManufacturing
Cummins et al 2024 Explainable Pre This survey categori Cross-industry a
Su et al 2024 Fault Diagnosis This study proposes Renewable ener
Nair et al 2023 Predicting Li-I This study examines Energy storage
Khan et al 2024 Secure ExplainaThis study develops Heavy transport
Gawde et al 2024 An Explainable This study explores Manufacturing
Spangler et al 2024 Explainable, De This study develops Nuclear power
Steenwinckel et al 2021 FLAGS: A Method This study introduce Transportation
Gama et al 2024 From Fault DeteThis study introduce Transportation
Feng et al 2023 Interpretable F This study proposes aIndustrial equip
Lei et al 2024 Interpretable OThis study introduce Aerospace
Xu et al 2024 Physics-informeThis study proposes Mechanical syst
Wu et al 2024 Prediction modeThis study proposes Civil Engineering
Liao et al 2023 Remaining Usefu This study proposes Aerospace
Peters et al 2022 Sensor-integratThis study compares Mechanical syst
Alomari et Andó 2024 SHAP-based insiThis study explores Aerospace
Kim et Kim 2024 Vibration spectrThis study enhances Mb echanical syst
Jiang et al 2024 Wind turbine anThis study proposes Renewable Ener
Allal et al 2024 Wind turbine faThis study develops aRenewable Ener
Zhu et al 2025 Wind turbine fauThis study introduce Renewable Ener
Shaer et Shami 2024 WrapperRL: Rein This study proposes tIndustrial acoust
Usuga-Cadavid et al 2022 Using Deep LearThis study evaluates Various industries (aeronautics, electronics)
Awuku et al 2024 Interpretable maThis study proposes Oil and Gas
Wellsandt et al 2022 Hybrid-augmentedThis study explores tManufacturing and industrial maintenance
 Type of equipment studied Equipment Type
Mechanical systems Rotating machinery (e.g., bearings
Mechanical systems Rotating machinery (Coal crusher)
Mechanical systems Water pumps
Mechanical systems Irrigation systems and agricultural
Mechanical systems Rotating machines (e.g., motors, b
Mechanical systems Mechanical systems, industrial equ
Mechanical systems Offshore wind turbines
Electrical systems Li-ion batteries
Mechanical systems Air brake systems in heavy vehicles
Mechanical systems Rotating industrial machines (e.g.,
Mechanical systems General nuclear plant components (
Mechanical systems Train bogie monitoring systems
Mechanical systems Metro trains (air production units)
Mechanical systems Rotating machinery (bearings)
Mechanical systems Aerospace engines
Mechanical systems Rotating machinery (e.g., bearings)
Civil systems Structural materials (rocks)
Mechanical systems Aero engines
Mechanical systems Gears
Mechanical systems Jet engines
Mechanical systems Bearings
Mechanical systems Wind turbines
Mechanical systems Wind turbines
Mechanical systems Wind turbines
Mechanical systems Noisy industrial environments
ous industries (aeronautics, electronics) General industrial equipment
Mechanical systems Oil pipelines
ufacturing and industrial maintenance General industrial assets
 Classification Population Study design
Fault detection and fault diagnosis Three datasets Experimental stu
Fault detection and diagnosis Dataset from a Experimental stu
Anomaly detection and failure prediction 220,320 recordExperimental stu
Prognostics and diagnostics Two datasets: TExperimental stud
Fault classification and diagnostics Multi-fault dat Experimental stu
Anomaly detection, fault diagnosis, and prognosis 102 articles re Systematic litera
Fault detection and diagnosis SCADA dataset Experimental
f stud
Remaining useful life (RUL) prediction NASA Ames Progn Experimental stu
Fault diagnosis Scania Trucks AExperimental stud
Multi-fault diagnosis Data collected Experimental stud
Maintenance decision-making optimization Simulated dataExperimental stu
Anomaly detection and root cause analysis Sensor data strExperimental stu
Anomaly detection and fault explanation Data from air prCase study with
Fault diagnosis and condition monitoring Two independent Experimental stud
Remaining useful life (RUL) prediction NASA's C-MAPSS Experimental stu
Fault diagnosis Two experimenta Experimental stud
Feature importance and predictive modeling 442 sets of roc Experimental stud
Remaining Useful Life (RUL) prediction NASA's C-MAPSS Experimental stud
Wear detection and condition monitoring A test rig with Experimental stu
Feature importance, interaction analysis, robustness assessment C-MAPSS dataset Experimental stud
Fault detection and diagnosis Bearing datasetExperimental stud
Anomaly detection and localization SCADA datasetsExperimental stud
Fault detection and identification SCADA data collExperimental stud
Fault detection and localization SCADA datasetsExperimental stu
Feature selection and interpretability enhancement MNE dataset com Experimental stud
Severity prediction, breakdown duration prediction, cause prediction Three datasets Experimental stu
Failure cause prediction Pipeline incide Experimental stu
Predictive maintenance process enhancement Conceptual andConceptual study
 Intervention Methods Comparaison Outcome Context Results Limitations Key findings
Implementation Foeature extraComparative analy EffectivenessIndustrial c High accuraSHAP has a highThe methodology
Proposal of contOutlier detecBasic versus exte Improved clarAddressing Local conteExtended repreContextual exten
Application of d Comparison of Transformer-GRUHybrid
o TransfAimed to imp Reduced traHigh computatio Transformer-base
Development of D a L models (e.Performance comp High accuracyAddressing m ConvolutionFocused on twoThe XAI-PdM mode
Integration of XASVM, KNN, Dec Random Forest achEnhanced inteThe study f LIME and SH High computatio Combining multip
Application of P CategorizatioXAI methods like SInsights into The transit XAI methods High computatio Adoption of XAI
Proposed CTSANData-driven
m aCTSAN model outp Improved faulThe study adThe CTSANLimited
m to two The interpretabl
Evaluation of RUExtreme DeepXDFM demonstrate XDFM improved Focuses on eSignificant Limited generalXDFM, enhanced w
Implementation Roandom Forest Random Forest achHigh accuracyFocused on Ei fficient f Restricted to AIncorporating XA
Application of FFFFT for featuMulti-sensor dataAchieved highThe study aiThe combina High computatio Multi-sensor fus
Development of IDntegration oDRL outperformed Significant c Aimed at add DRL demonst Simulated envirThe combination
Integration of s Matrix Profil FLAGS combines daFLAGS improve Focused on Epnhanced an High computatio Integrating data
Integration of L LSTM-AE trainAMRules with Cheb Significant i Addresses thThe systemRelies on high- Neuro-symbolic a
Developed an inteHilbert TransCompared local inImproved faulAddresses thThe fault f Applicable primIntegrating FL wi
Development of Operational The proposed meth Achieved lowe Addresses ch Improved RU The methods rel Combining domain
Development of Physics-InforPIPDN and MS-PIPAchieved overAimed at add Improved faHigh computatio Integrating phys
Development of Baase learnersStacking models sStacking-LR mFocused on Apchieved roRequires extensThe stacking ens
Development of PAINNs with seAttnPINN outperfo Enhanced RULFocuses on aThe propose Relies on high- Integrating PINN
Used in-situ MEM Standard metr Standard metrics The regressioThe study aiRegression-Results are specMachine learning
SHAP analysis onSHAP-based fe XGBRegressor andIdentified cri Addresses cEnhanced inLimited to C-MA SHAP-based meth
Implemented Grad Short-Time FoResNet-50 with STAchieved 99.7Addresses ch Enhanced diLimited to laboCombining Grad-C
Developed DIGNN, Correlation-bDIGNN was compar Achieved earlAimed at add Significant Relies on high- Graph-based appr
Developed a two-Feature selecThe two-tier syst Achieved 95%Addresses ch Enhanced fa The study is limTwo-tier systems
Developed the DG Dynamic graph Compared againstDG-VNF achiev The study foDG-VNF deliThe approach isIntegrating dyna
Developed Wrappe Time-series dWrapperRL outperf Reduced featu Addresses ch WrapperRLValidation
a was WrapperRL effect
Implemented stat Transformer-b Compared NLP modeCamemBERT Addresses
and thCamemBERT Thea study was lTransformer-base
Implemented LightNLP for feat Compared perform LightGBM with Aims to addrEnhanced pr The study is re Combining CSL, N
Developed a fram Natural languCompared traditioEnhanced usab Addresses thDIAs improvPrimarily a concDIAs can bridge
 Comments Problem Methodology AI Tools Used Explainable AI Techniques Used
Promotes the ad The challengeThe study combines
Unsupervised anom SHAP (Shapley Additive Explanations)
The study under Operators stProposed extensionHDBSCAN for anoma SHAP-based local explanations with a
Highlights the Challenges i Comparison of deeRNN, LSTM, GRU, LIMEC for local interpretability and f
Advances unders Farmers faceIntegration of XAIConvolutional LSTSHAP (global/local explanations), Cou
This work underLack of trustDevelopment of aSVM, KNN, DecisioLIME, SHAP, PDP, ICE.
Highlights the Trust and traSystematic reviewSHAP, LIME, GradLocal methods (SHAP, LIME) dominat
This work demon High failure Development of CTCNN, attention m Temporal-spatial attention weights f
Demonstrates th Accurate andCombines featureXDFM, A-LSTM, GBM SHAP for global and local feature im
Highlights the Addressing th Combines RandomRandom
F Forest, SSHAP for global feature attribution, L
Highlights the Challenges i Utilized FFT-basedSVM, Random Fores LIME for local interpretability, Rand
Highlights the Traditional mPOMDP-based model Deep reinforcemen SHAP for feature importance and dec
This methodoloChallenges i Combines semantic Matrix Profiling, Semantic enrichment and rule-based r
The study demon High costs a Developed a two-lLSTM-AE, AMRules Rule-based explanations (AMRules) w
The study highlTraditional FA physics-informedFederated AveragiFault characteristic frequency (FCF) e
This study highAccurate RULDevelopment of OC LSTM, OCA, ABDAOperational condition attention mech
This study brid Challenges inDeveloped a physic
PIPDN, MS-PIPDN,Physical labelling for feature attribu
This study demo Traditional Developed a stackLightGBM, Random SHAP for feature importance and de
This study highCurrent RUL Developed a hybriSelf-Attention, P Visualization of hidden states and d
The study under Traditional wDeveloped a regreRegression-basedRegression modeling for wear estima
This study brid Traditional Applied SHAP valueXGBRegressor, MLSHAP-based feature importance, depe
This study demo Traditional mDeveloped a hybrid
Grad-CAM, ResNet- Grad-CAM for visualizing critical freq
The study showc ConventionalDeveloped a graph-DIGNN, KDE, SpecCorrelation-based graph analysis, dec
This study provTraditional Implemented a two Bagging, Stacking Feature importance visualization and
This study estaStatic modelsDynamic graph mod Dynamic Graph Neu Graph representation learning and nod
This study highHigh-dimensiDeveloped a three- WrapperRL, SA-CNFeature selection via WrapperRL and
Highlights the Traditional Implemented transf CamemBERT, FlauB Local interpretability through LIME fo
This research bHigh class imDeveloped a compr LightGBM, SHAP, SHAP for global and local interpretabi
This study highTraditional Developed a socio-Digital Intelligent Explainable AI was proposed for mai
 Enhancements Found Limitations Found Insights Mechanism Data description
Type of data
Explainable AI techniques p SHAP incurred high com This study hMechanismThe study uSensor dat
Contextual explanations i Explanations for exte The study eMechanismThe study uSensor dat
LIME reduced features from Transformer models req EmphasizesMechanismThe study uSensor dat
Improved test F1 scores, High computational cos The study de MechanismThe study uSensor dat
XAI techniques provided acHigh computational dem The integraMulti-sensoThe study uSensor dat
XAI methods enhance userComputational costs ofSurvey under MechanismNA NA
Improved fault detection Restricted fault types The CTSAN m MechanismThe study uSensor dat
Feature selection using High computational cos Combining XXDFM withThe S study uSensor dat
Reduced computational com Optimized for APS failuThis study dSequential The study uSensor dat
Achieved 100% accuracy inComputational
f overhead
Demonstrates Multi-sensoThe study uSensor dat
Cost reductions of 53% co Simulated environment Demonstrates Deep reinfoThe study uTabular dat
Reduced false positives anComputational overhea Demonstrate FLAGS integThe study uSensor dat
Improved anomaly detection Domain-specific configCombining aCombining Thisa study Sensor dat
Enhanced generalization an Limited scalability beyCombining FL Dynamic feaThe study uSensor dat
Significantly improved preDependency on high-qua This study Combines op The study uTabular dat
Achieved over 99% diagnost High computational ove Physics-infoPhysics-inf The study uSensor dat
Achieved superior predictiHigh dependency on dat Integrating Stacking enThe study uTabular dat
Achieved higher accuracy High dependency on qua Combining se AttnPINN deThe study uTabular dat
Improved interpretability Performance limited by Combining iIn-situ sen The study uSensor dat
Improved feature importanRestricted to a single Integrating SHAP-basedThe study uTabular dat
Achieved significant improLimited validation un Integrating The hybrid This study Image data
Achieved earlier fault det Limited to laboratory-sGraph-based DIGNN leverSCADA dataTabular dat
Achieved 95% accuracy in Limited
bi to one turbineA tiered appTwo-tier sySupervisoryTabular dat
Earlier fault detection wit Model performance is Dynamic
d graThe DG-VNFSupervisoryTabular dat
Achieved significant dimenLimited validation to sRL-based feaWrapperRLTime-Freque
i Image data
Enhanced interpretability Applicability restricte NLP techniqu LightGBM wiThe study lTextual dat
Achieved significant perfo Generalizability to dif Integrating SHAP interpThe study uTextual dat
Improved task execution tiConceptual nature of th Integrating DIAs providprovides a Sensor dat
 Predictive Maintenance Tools Interpretability Tools (Indirect)
Isolation Forest (IF); Local Outlier Feature Importance Analysis for Vibratio
Outlier Detection with HDBSCAN; SurrogContextual Explanations (Local and Global
Recurrent Neural Networks (RNNs); LoFeature Importance Ranking; Sensor-Specif
Long Short-Term Memory (LSTM); Extrem SHAP (Shapley Additive Explanations); De
Support Vector Machine (SVM); k-Neare Feature Importance Ranking (Random Fore
SHAP; LIME; CAM; Isolation Forest; Gi Contextual Visualizations (Heatmaps, Sal
Convolutional Temporal-Spatial Atte Temporal and Spatial Attention Mechanis
Extreme Deep Factorization Machine SHAP-Based Feature Selection; RMSE and
Random Forest (RF); Sequential RandoSHAP Decision and Summary Plots; Featur
Multi-Sensor Data Fusion; FFT (Fast F Feature-Level Fusion; Feature Ranking (
Deep Reinforcement Learning (DRL); PaSensitivity Analysis; Kernel SHAP Integrat
Matrix Profiling; Semantic Complex Ev Knowledge Graphs (KGs); Semantic Mappi
LSTM-AE (Long Short-Term Memory Aut Global and Local Explanations (Rule-Based
Federated Learning Framework (FL); FedVisualization of Fault Characteristic Fre
Operational Condition Attention-LST Operational Condition Attention (OCA) S
Physics-Informed Probabilistic Deep N Physical Labeling Module; Epistemic and
Stacking Ensemble Learning (Base LearFeature Impact Analysis with SHAP; Regre
Physics-Informed Neural Networks (PI Hidden State Visualization (Failure Mod
In-Situ MEMS Accelerometers; DecisioGear Dynamics Features (RMS, GMFs, Side
MLPRegressor (Multi-Layer PerceptronTemporal SHAP Analysis (Mean Absolute S
STFT (Short-Time Fourier Transform); RGrad-CAM Heatmaps (Frequency Region Ide
DIGNN (Decision Interpretable Graph NDecision Interpretability Module (Global
Bagging; Voting; Random Forest; CatBoFeature Averaging Based on Tree Models;
Dynamic Graph Representation LearninSelf-Attention Mechanisms (Spatial and T
WrapperRL (Reinforcement Learning ASA-CNN (Attention-Based Convolutional N
CamemBERT; FlauBERT; Random Forest Tokenization
( and Feature Extraction (Cont
LightGBM; XGBoost; Random Forest; CNN Confusion Star Visualization; Global and
Digital Intelligent Assistants (DIAs) Human-in-the-Loop Systems; Enhanced Vi
 Explainability Tools (Direct)
SHAP (Shapley Additive Explanations); Local-DIFFI (Local Depth-based Feature Importance for Isolation Forest)
SHAP (Shapley Additive Explanations)
LIME (Local Interpretable Model-Agnostic Explanations); SHAP (Shapley Additive Explanations)
SHAP Values; Counterfactual Explanations (CFE)
LIME (Local Interpretable Model-Agnostic Explanations); SHAP (Shapley Additive Explanations); PDP (Partial Dependence Plots
SHAP (Shapley Additive Explanations); LIME (Local Interpretable Model-Agnostic Explanations); DeepSHAP; GradCAM
None
SHAP (Shapley Additive Explanations)
SHAP (Shapley Additive Explanations); LIME (Local Interpretable Model Agnostic Explanations)
LIME (Local Interpretable Model Agnostic Explanations); Random Forest Feature Importance
SHAP (Shapley Additive Explanations)
Rule-Based Systems
SHAP (Shapley Additive Explanations)
None
None
None
Shapley Additive Explanations (SHAP)
None
None
SHAP (SHapley Additive exPlanations)
Grad-CAM (Explainable AI Technique)
None
None
None
None
LIME (Local Interpretable Model-Agnostic Explanations)
SHAP (Shapley Additive Explanations); SHAP-based Visualizations
Integrated Explainable AI (XAI) Features; Customizable Explanation Models
artial Dependence Plots); ICE (Individual Conditional Expectation)
HAP; GradCAM