Interactive Power BI dashboard analysing generation data from 2 Indian solar power plants to identify performance patterns, equipment efficiency and maintenance requirements using inverter-level data analysis.
🔗 [View Dashboard Screenshots and Analysis]
(Based on original Kaggle dataset objectives)
- Can we predict power generation patterns for better grid management?
- Can we identify the need for panel cleaning or maintenance?
- Can we identify faulty or suboptimally performing equipment?
The 24% Generation Gap: Plant 1 generates 21M AC power units versus Plant 2's 16M — despite both receiving identical average solar irradiation of 0.23 kW/m². This 24% gap cannot be explained by weather or environmental differences — it points directly to operational and equipment factors at Plant 2.
Inverter Performance Analysis: Plant 1 shows tight linear DC to AC conversion clustering — most inverters operating consistently at 300-320 AC output with only 2-3 outliers. Plant 2 shows significantly higher conversion variability — multiple inverters operating at 160-200 AC output against a cluster average of 240-280. This pattern strongly indicates underperforming or faulty inverters requiring immediate maintenance inspection.
Predictable Generation Pattern: Both plants follow a Gaussian (bell curve) distribution of power output throughout the day — peaking between 10AM and 1PM — confirming solar irradiation as a reliable generation predictor for grid management planning. Plant 1 consistently peaks higher than Plant 2 at every hour of the day.
Maintenance ROI Calculation: If Plant 2 inverter maintenance restored performance to Plant 1 levels, theoretical generation improvement would be approximately 31% — from 16M to 21M units. For solar plant operators, this analysis provides a data-driven justification for maintenance investment before the next generation season.
Seasonal Generation Pattern: Both plants show peak generation in May-June months — consistent with Indian summer solar irradiation patterns. Plant 1 outperforms Plant 2 across every month in the dataset — the performance gap is not seasonal but structural.
Equipment Fault Identification: DC to AC scatter plot analysis reveals Plant 2 has multiple inverters operating 30-40% below the cluster average output level. This pattern is consistent with dust accumulation, inverter degradation or connection losses — all addressable through targeted maintenance without full equipment replacement.
Dataset does not include installed capacity information for each plant. Definitive diagnosis of operational fault versus capacity difference would require cross-referencing maintenance logs and installed capacity specifications not available in this dataset. However identical irradiation with systematic inverter underperformance strongly suggests operational factors rather than capacity differences.
- Microsoft Power BI Desktop
- DAX (Data Analysis Expressions)
- Power Query for data transformation
- Data Modelling and Relationship Management
- Many-to-One relationship management across 4 tables
Solar Power Generation Data — Kaggle
- 2 Indian solar power plants
- Plant 1 ID: 4135001 | Plant 2 ID: 4136001
- 4 files — Generation data and Weather Sensor data per plant
- Date range: May to September 2020
- 15-minute interval readings
Page 1 — Generation Overview:
- 5 KPI cards — Total AC Power Plant 1 and 2, Avg Daily Yield Plant 1 and 2, Avg Solar Irradiation
- Monthly AC Power Generation comparison — Plant 1 vs Plant 2
- Hourly bell curve — Average AC Power Output by hour and Plant ID
- PLANT_ID slicer synced across both pages
Page 2 — Equipment Performance Analysis:
- DC vs AC Power scatter plot — Plant 1 inverter performance
- DC vs AC Power scatter plot — Plant 2 inverter performance
- Side by side comparison revealing equipment efficiency gap
- Solar orange theme with professional formatting
- Cross-table relationship modelling across 4 datasets
- DAX measure creation for combined plant metrics
- Many-to-One cardinality with bidirectional cross filtering
- Hour extraction from DateTime for time-series analysis
- Append Queries for combined generation analysis
- Interactive PLANT_ID slicer synced across multiple pages
🔗 LinkedIn: www.linkedin.com/in/ankitasahawork96/