SERVEXO
Field service management platform
One platform for work orders, technician dispatch, scheduling and reporting, in Arabic and English.
Data science and energy forecasting
MATLAB Time-Series & PV Forecasting
ARIMA, neural networks and the LSTM family compared on solar and weather data with one shared evaluation.
Overview
A MATLAB environment that prepares PV and weather data, tests stationarity, and fits ARIMA alongside a family of neural models: MLP, ANN, LSTM, stacked and multivariate BiLSTM and CNN-LSTM. Every model is scored with the same metric functions for a fair comparison.
The challenge
Solar output follows weather that changes hour by hour. Picking a forecasting model needs a fair comparison with the same preparation, the same delays and the same metrics.
Our approach
Key capabilities
- PV calculation and data preparation scripts
- ADF stationarity test, ACF/PACF analysis and ARIMA
- MLP, ANN, LSTM, BiLSTM and CNN-LSTM models
- Shared normalisation, delay selection and metrics
Inside the project
Outcomes
- One benchmark for classical and deep-learning forecasters
- Reusable pipeline for any site's solar and weather data
Technology
Try it
Related projects
Industrial Pulse
Industrial IoT condition monitoring
From sensor to screen: equipment telemetry over MQTT, stored in PostgreSQL and monitored live on a bilingual dashboard.
Full-Stack IoT Monitoring Platform
Full-stack IoT software
Device registry, live readings, analytics and alerts for building sensors, in one web platform.
Not sure where your project fits?
Tell us where you are starting. We reply with the simplest sensible scope.
Start a project