About The Data
Frame The Problem
In this project, we are trying to build a model that can forecast(predict) the spacecraft telemetry data (That is collected at the Laboratory for Atmospheric and Space Physics).
The way that we will follow:
- What is the telemetry data?
- What the data represent?
- Data descriptions
- What is time-series?
What is the Telemetry Data?
- Link from satellite to the ground station.
- Collection of measurements of sensor or other data at a remote point, their automatic transmission to receiving equipment for monitoring the satellite at the ground control unit, and predicting anomalies in the satellite.
- In this work, we used only the analog telemetry channels that indicate the measurements of interest, related to the battery.
What the data represent?
Our datasets come from NASA satellite in low earth orbit that's using sensors for studying some things. This satellite lifetime around 20 years, most datasets contain between 1.5 million and 2 million records, even there is one with 50 million records, which is a very huge number to think about not even to tackle it. Data are collected from one satellite. The data came from three subsystems as time-series datasets.
Dataset Descriptions
Bus Voltage is a univariate time series with 1,838,088 readings for bus voltage recorded over a period of roughly 18 years. “Univariate” means that it tracks only one variable’s values over time. The cadence of the data (the regular frequency at which measurements are taken) is one measurement every 5 minutes.
Total Spacecraft Bus Current is a univariate time series with 1,518,180 readings for total bus current recorded over roughly 15 years. The cadence of the data (the regular frequency at which measurements are taken) is one measurement every 5 minutes.
Reaction Wheel Temperature is a univariate time series with 1,518,180 readings for wheel temperature recorded over roughly 15 years. The cadence is one measurement every 5 minutes.
Reaction Wheel RPM is a univariate time series with 48,865,494 readings for RPM recorded over roughly 10 years. The cadence is one measurement every second.
Battery Temperature is a univariate time series with 1,518,180 readings for battery temperature recorded over roughly 15 years. The cadence is one measurement every 5 minutes.