Need Expert in Machine Learning (Random Forest + SVM) for Electrical Engineering Thesis (Predictive Maintenance + Simulation)
Budget: $250 – $750 AUD
Hi, I’m working on my Master’s thesis in Electrical Engineering (Power Systems) and looking for an experienced freelancer to help complete key technical components.
The project focuses on AI-based predictive maintenance for distribution transformers, combining machine learning models and optimization-based scheduling.
Scope of Work:
1. Machine Learning Models
Implement and finalize:
Random Forest (already partially done)
SVM (RBF kernel)
Decision Tree (optional benchmarking)
Perform:
Hyperparameter tuning (GridSearchCV / RandomizedSearch)
5-fold cross-validation
Performance evaluation (Accuracy, F1, AUC)
2. Data Handling & Feature Engineering
Work with BRAVO dataset (15k+ samples, 16 features)
Handle:
Missing values (<5%)
Class imbalance (SMOTE already used ~87:13 → ~60:40)
Feature engineering (already defined but may refine):
Load Stress, Maintenance Overdue, Failure Risk Score, etc.
3. Synthetic Data Generation
Generate additional fault data using Monte Carlo simulation
Based on DGA gas ratios (IEC standards)
Validate distribution (e.g., KS test)
4. Optimization & Simulation
Develop maintenance scheduling:
Baseline model
Greedy algorithm
Integer Programming (PuLP or similar)
Integrate ML outputs into scheduling decisions
5. Analysis & Results
Compare:
AI-based vs traditional maintenance
Perform:
Cost-benefit analysis (target ~10–15% improvement)
Sensitivity analysis
Tech Stack Required:
Python (must)
scikit-learn, pandas, numpy
Optimization tools (PuLP / OR-Tools)
Jupyter Notebook
Experience with power systems / predictive maintenance (bonus)
Deliverables:
Clean, well-documented Python code
Model outputs + evaluation metrics
Simulation results (graphs, comparisons)
Brief explanation of methodology (for thesis writing)
The project focuses on AI-based predictive maintenance for distribution transformers, combining machine learning models and optimization-based scheduling.
Scope of Work:
1. Machine Learning Models
Implement and finalize:
Random Forest (already partially done)
SVM (RBF kernel)
Decision Tree (optional benchmarking)
Perform:
Hyperparameter tuning (GridSearchCV / RandomizedSearch)
5-fold cross-validation
Performance evaluation (Accuracy, F1, AUC)
2. Data Handling & Feature Engineering
Work with BRAVO dataset (15k+ samples, 16 features)
Handle:
Missing values (<5%)
Class imbalance (SMOTE already used ~87:13 → ~60:40)
Feature engineering (already defined but may refine):
Load Stress, Maintenance Overdue, Failure Risk Score, etc.
3. Synthetic Data Generation
Generate additional fault data using Monte Carlo simulation
Based on DGA gas ratios (IEC standards)
Validate distribution (e.g., KS test)
4. Optimization & Simulation
Develop maintenance scheduling:
Baseline model
Greedy algorithm
Integer Programming (PuLP or similar)
Integrate ML outputs into scheduling decisions
5. Analysis & Results
Compare:
AI-based vs traditional maintenance
Perform:
Cost-benefit analysis (target ~10–15% improvement)
Sensitivity analysis
Tech Stack Required:
Python (must)
scikit-learn, pandas, numpy
Optimization tools (PuLP / OR-Tools)
Jupyter Notebook
Experience with power systems / predictive maintenance (bonus)
Deliverables:
Clean, well-documented Python code
Model outputs + evaluation metrics
Simulation results (graphs, comparisons)
Brief explanation of methodology (for thesis writing)