Hey everyone! 👋

I’m currently working on neutron intensity measurements from the WWR-K research reactor as part of my master’s research in nuclear medicine and radiation safety ⚛️

But I’ve hit a challenge — the detector signals I’m analyzing are often mixed with background noise and fluctuations from the reactor environment.

Has anyone here dealt with noisy neutron flux or radiation detector data before?

What kind of preprocessing, filtering, or denoising techniques worked best for you — e.g. Fourier/wavelet filtering, Kalman filters, or ML-based noise reduction (autoencoders, etc.)?

Would really appreciate any insights or experiences you could share! 🙏💡

10 minutes ago, Zhanerke said:

Hey everyone! 👋

I’m currently working on neutron intensity measurements from the WWR-K research reactor as part of my master’s research in nuclear medicine and radiation safety ⚛️

But I’ve hit a challenge — the detector signals I’m analyzing are often mixed with background noise and fluctuations from the reactor environment.

Has anyone here dealt with noisy neutron flux or radiation detector data before?

What kind of preprocessing, filtering, or denoising techniques worked best for you — e.g. Fourier/wavelet filtering, Kalman filters, or ML-based noise reduction (autoencoders, etc.)?

Would really appreciate any insights or experiences you could share! 🙏💡

Hey! 👋
Yeah, I’ve worked with noisy detector signals before.

👉 First, try background subtraction and simple smoothing (moving average or Savitzky–Golay).
👉 If noise changes over time — wavelet filtering works well.
👉 For tracking slow signal changes — use a Kalman filter.
👉 If you have enough data, you can test ML denoising (autoencoder), but it’s more complex.

Also check shielding, grounding, and detector calibration — sometimes that helps the most ⚛️

49 minutes ago, Zhanerke said:

Hey everyone! 👋

I’m currently working on neutron intensity measurements from the WWR-K research reactor as part of my master’s research in nuclear medicine and radiation safety ⚛️

But I’ve hit a challenge — the detector signals I’m analyzing are often mixed with background noise and fluctuations from the reactor environment.

Has anyone here dealt with noisy neutron flux or radiation detector data before?

What kind of preprocessing, filtering, or denoising techniques worked best for you — e.g. Fourier/wavelet filtering, Kalman filters, or ML-based noise reduction (autoencoders, etc.)?

Would really appreciate any insights or experiences you could share! 🙏💡

Hi there! 👋
Yes, neutron detector data can be quite tricky — especially in reactor environments where background radiation and electronic noise overlap with the signal. I’ve worked with similar flux measurements, and wavelet-based denoising combined with moving average smoothing gave good results for removing short-time fluctuations without losing physical information.

If your data are time-dependent, Kalman filtering works great for tracking real signal variations while suppressing random noise. For more complex patterns, autoencoders or 1D CNNs trained on clean vs. noisy samples can perform surprisingly well — see Zhang et al. (2021, Nuclear Instruments and Methods in Physics Research A) for an example.

Also, don’t forget to characterize the noise spectrum first — that helps you choose the right cutoff frequencies for Fourier or wavelet filters.
Good luck with your analysis — sounds like an exciting project!

1 hour ago, Zhanerke said:

Hey everyone! 👋

I’m currently working on neutron intensity measurements from the WWR-K research reactor as part of my master’s research in nuclear medicine and radiation safety ⚛️

But I’ve hit a challenge — the detector signals I’m analyzing are often mixed with background noise and fluctuations from the reactor environment.

Has anyone here dealt with noisy neutron flux or radiation detector data before?

What kind of preprocessing, filtering, or denoising techniques worked best for you — e.g. Fourier/wavelet filtering, Kalman filters, or ML-based noise reduction (autoencoders, etc.)?

Would really appreciate any insights or experiences you could share! 🙏💡

Hi! Yes, that kind of noise is quite common. You can try a few simple steps:
– remove the background first,
– use wavelet or Fourier filtering,
– if the signal shape is known, apply a Kalman or matched filter.
These usually help clean the signal effectively.💬🙌

Really interesting work! Try using the EEMD + matched filtering method — it works well with non-stationary noise. EEMD helps separate the signal into components and remove the background, while the matched filter enhances weak neutron pulses, improving the signal-to-noise ratio.