fairfolio
Saved Projects

Constraining Asteroid-Mass Primordial Black Holes via Temporal Convolutional Network Analysis of Space Probe Ephemerides

CSEF · 2026 Physics & Astronomy (Senior Division)

Dhruva Gopal

Overview

In the search for primordial black holes (PBHs) as dark matter candidates, the asteroid mass range between 10^17g and 10^23g remains unconstrained by past microlensing and Hawking radiation searches. This study analyzes deep space probe ephemerides for a solar system PBH flyby. Two synthetic training datasets with injected PBH signals of mass ranging from 10^23g-10^24.5g and 10^22.5g-10^23g, respectively, were created. A seven-layer temporal convolutional network (TCN) was trained on the data, achieving up to 85% recall and accuracy >80% on isolated signals. Over 15,000,000 state vectors spanning nearly three decades of Juno, Cassini, and New Horizons mission data was searched, finding zero high-confidence candidates. Poisson statistical analysis determines with 95% confidence that the average flyby rate of a PBH within the probes’ detection volume (10,000 km) must be <3.0. This was then extended to a 39-dimensional multivariate TCN to detect the synchronized gravitational effects of a PBH flyby across thirteen solar system bodies. The new model achieved >76% accuracy on signals with mass >10^21g, covering ⅓ of the asteroid mass range, in a Newtonian simulation with a Gaussian noise distribution of σ = 1e-5 AU for planets and σ = 1.67e-9 AU for probes. This research introduces a novel TCN-based framework for PBH flyby detection, establishing one of the first ML-based constraints for asteroid-mass PBHs. Further research will achieve full deployment of the 39-dimensional model and improve lower-mass signal detection with kinematic feature expansion and curriculum learning.

Competition history

  • CSEF 2026 Physics & Astronomy (Senior Division) · Entry S-17-04

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: California Science & Engineering Fair public projects

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google