Tien Shan Mountain Station · 3340m · Kazakhstan
Cosmic Ray Data Processing System
Automated filtering and cleaning of particle detector events from the Hadron-55 installation. Separates physical cosmic-ray signals from noise, calibration events, and empty scintillator readings.
Scroll
01About the Installation
Hadron-55 is a scientific installation at the Tien Shan High-Mountain Cosmic Ray Station operated by the Lebedev Physical Institute. It continuously registers the flux of cosmic particles arriving from deep space, capturing extensive air shower (EAS) events created by high-energy particles in the atmosphere.
25 channels across back, front, left, middle and right sectors — each recording particle flux at high and low sensitivity. Raw data arrives as .dat files with timestamped event blocks.
Each event block starts with |EVENT: and ends with #. Inside: arrays of integer counts per detector channel. Physical events show non-zero scintillator readings — noise events show near-zero values across all channels.
Raw data contains outlier spikes, calibration blocks, isolated single-sample noise, and empty "phantom" events. This system identifies and removes them, preserving only physically meaningful cosmic-ray detections.
Filtered .dat files with identical naming — one output file per input file. Cleaned events retain their original structure; outlier values are replaced by context-aware interpolation.
02Data Pipeline
Four-stage processing pipeline. Each stage runs sequentially on every event block in every .dat file found in the input directory.
Read all 6-digit .dat files. Split content into event blocks by |EVENT: / # delimiters.
Drop calibration and test blocks. Check scintillator channels — discard events where ≥70% of values are zero.
Replace outlier spikes with neighbour average. Suppress isolated single-sample noise surrounded by zeros.
Write cleaned event blocks to output directory. Filename preserved. One output file per input file.
SIMULATED DETECTOR SIGNAL — BEFORE / AFTER FILTERING
03Filter Demo
Upload a .dat file — the server will filter it and return a clean file for download.
Drop a .dat file here or click to browse
Files must follow 6-digit naming convention (e.g. 210516.dat)
04Project
Alisher Baitas
Interface, Integration & QA
GUI, Flask API, Telegram bot, deployment and repository architecture. Full QA cycle: vulnerability detection, fault tolerance, test coverage. Student at Tomorrow School, Astana Hub.
Hanni
Signal Processing & Algorithms
Core filtering algorithms: outlier detection, scintillator screening, event classification. Python implementation based on Hadron-55 detector physics.