Table 1 Summary of Insulator Contamination Assessment Literature.

^8,9,10,11

Acoustic and Ultrasonic Data

Wavelet Transform, SVR Stacking, AE Monitoring, Wavelet Transform, MRA Feature Extraction, Piezoelectric Sensor Acquisition

Reasonable Prediction Accuracy, Non Invasive Monitoring, Immune to electromagnetic Interference

Required High Quality AE sensor, Expensive, Sensitive to Ambient Noise, Sensitive to Signal Quality

^12,13

Electromagnetic & Radio Frequency Data

Wideband electromagnetic radiation signature analysis; statistical analysis (mean and std. deviation), frequency spectrum analysis

Non-invasive; remote monitoring; high sensitivity; effective within 30 MHz to 130 MHz range

Requires specialized equipment; Expensive; sensitive to environmental noise; limited to specific insulator types; sensitive to environmental interference; requires filtering of outliers

^{7,14,15,28,28,29,30,31,32,33,34,35,36,37,38,39,40}

Image Data

DL Models (YOLO, RetinaNet, ResNet); Custom CNNs; Support Machine Learning (SVM, k-NN); NNs (BPNN, RBFNN, ANN); Data Processing (Data Augmentation, Edge Detection, Histogram); Image Analysis Methods (Gabor Filters, Visible/IR Fusion); Feature Extraction (Eigenvalue Extraction)

High accuracy (up to 99.242%); robust under diverse lighting and environmental conditions; non-invasive methods; high discrimination accuracy for specific materials; effective across varying humidity levels.

Requires expensive high quality cameras; computational complexity; sensitive to environmental conditions; reliant on controlled lighting; inefficient in non uniform pollution.

^{16,17,18,19,20,40,41,42}

Image Data (Hyperspectral/UV/IR)

DL Models (CNN, ResNet); Support Machine Learning (MLR, RF); NNs (RBFNN, BPNN, DBN, SAE); Data Processings (MSC, PCA, SPA, LDA); Image Analysis Methods (LatLRR, Image Filtering); Feature Extraction (Infrared Imaging, Fisher Criterion, KPCA, PSO)

Non-Contact Detection & High Accuracy (ESDD, NSDD); High Recognition Accuracy (up to 96.67%); High Accuracy in Varying Humidity Conditions.

Requires Specialized Equipment (Hyperspectral, IR, UV); Sensitive to Lighting and Environmental Conditions; Sensitive to Lightning; Extensive Image Preprocessing; Ineffective for Non Uniform Pollution

^21,22

Partial Discharge

Deep Learning Models (CNN, LSTM); Real-Time Data Analysis (Fuzzy Inference System)

Portable; High Correlation Assessment; Non Invasive

Requires extensive data preprocessing; computationally intensive; limited to specific insulator types; Expensive Equipment; Not suitable for on line monitoring

^{1,2,5,23,24,25,26,27,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59}

Leakage Current Data

Support Machine Learning (RUSBoost, AdaBoost, Bagging, SVM, Random Forest, XGBoost); NNs (ANN, LSTM, GMDH, ANFIS, LS-SVM); Data Processings (PCA, K-means, Rough Set Theory, Statistical Analysis, Harmonic Analysis); Feature Extraction (FFT Analysis, Detrended Fluctuation Analysis, Wavelet Transform, Autocorrelation); Signal Processing Techniques (Short Time Modified Hilbert Transform, Sparse Representation); Innovative Approaches (Fuzzy Logic, D-S Evidence Theory, Particle Swarm Optimization).

High Accuracy (up to 96.82%); Real-Time Monitoring; Environmental Adaptability; Robust Classification; Low Computational Burden; Cheap Equipment; No sensitivity to non uniform pollution.

Invasive, Susceptible to Electromagnetic Noise; Require high performance data acquisition; Require experimental validation to check the accuracies for realistic datasets and multiple contamination levels