How to Perform Online Monitoring of Winding Partial Discharge?

—UHF Sensor Solution Design

 

Transformers are the core equipment of power grids, and the insulation health of their windings directly determines grid operation safety. Statistics from the International Council on Large Electric Systems (CIGRE) show that insulation degradation is the leading cause of transformer failures, accounting for over 70% of cases. Partial discharge (PD) is the most sensitive early indicator of insulation degradation. Traditional offline detection methods (such as the IEC 60270 standard pulse current method) cannot capture the true discharge characteristics during operation.

Menu

1.Limitations of Traditional Online Monitoring Methods and the Rise of UHF Technology

● Challenges of the Pulse Current Method in Online Applications:

This method, based on the IEC 60270 standard, measures transient voltages across a detection impedance to quantify discharge magnitude (in picocoulombs, pC). However, in substations, transformers and connected equipment form a vast electrical network, generating wideband electromagnetic interference (e.g., corona discharge, switching operations, power electronics noise) with frequencies ranging from tens of kHz to hundreds of MHz. These interference signals often overwhelm genuine PD signals, resulting in poor signal-to-noise ratios and significantly reduced detection sensitivity and reliability.

 

● Field Challenges of the Acoustic Method:

The acoustic method detects ultrasonic waves (20kHz–300kHz) generated by PD to locate discharge sources. However, inside an operating transformer, complex structures like insulation oil, paperboard, and windings cause sound waves to attenuate severely, especially at higher frequencies. Additionally, operational noise (e.g., core magnetostriction, cooling fans) often overlaps with PD signals, making extraction and accurate localization difficult.

 

● Breakthrough Advantages of the UHF Method (Physical Layer):

The UHF method detects electromagnetic waves (300MHz–3GHz) emitted by PD. Its key advantages include:

(1)Natural Noise Immunity:Substation interference (e.g., corona, switching) is concentrated below 100MHz, while UHF signals experience minimal interference due to the transformer's Faraday cage effect.
(2)Low Propagation Loss:UHF waves attenuate less in oil-paper insulation compared to ultrasonic waves, enabling detection over several meters.

 

2. Core Design Considerations for UHF Sensors

● Frequency Band Selection and Optimization (Key Trade-offs):

The choice of UHF frequency band impacts sensitivity, noise immunity, and sensor feasibility. Key factors include:

(1)Sensitivity vs. Noise:PD energy peaks between 300MHz–1.5GHz. Higher frequencies (>1.5GHz) suffer greater attenuation.

(2)Taille de l'antenne:Antenna dimensions must match the wavelength (λ = c / (f × √εr)). For 800MHz in oil-paper (εr ≈ 2.3), λ ≈ 0.22m, allowing compact designs (5–10cm).

Facteur	Low Band (300–500MHz)	High Band (700–1500MHz)	Gamme recommandée
Atténuation du signal	Lower, longer range	Higher (∝ f²), shorter range	300 à 800 MHz
Immunité au bruit	Susceptible to switching noise	Lower background noise	500 à 1500 MHz
Taille d'antenne	Larger (λ/4 ≈ 0.58m for 300MHz)	Compact (λ/4 ≈ 0.055m for 1.5GHz)	700 à 1500 MHz
Résolution d'impulsion	Lower time resolution	Higher, better for waveform analysis	> 500 MHz
Portée optimale			500 MHz à 1 GHz

 

● Sensor Types and Installation Locations:

(1)Internal Sensors (Optimal but Requires Planning):
Embedded in insulation oil or near windings for minimal signal loss. Ideal for new transformers or retrofits. Locations: winding pressure plates, riser flanges.

(2)External Sensors (Practical Solutions):

- Oil Valve Sensors: Installed on sampling valves, leveraging them as waveguides.

- Bushing Tap or GIS Sensors: Use capacitive coupling at bushing ground leads.

- Tank-Mounted Sensors:Non-invasive but less sensitive due to metal shielding.

● Sensitivity and Directionality Optimization:

(1)Antenna Gain and Matching:Maximize gain/bandwidth with designs like fractal or patch antennas (VSWR < 2:1).

(2)Modèles directionnels: Focus on high-risk areas (e.g., HV winding ends) using phased arrays or reflectors.

(3)Low-Noise Amplification:Integrated LNAs (NF < 3dB, 20–40dB gain) enhance signal-to-noise ratios.

3. UHF Online Monitoring System Architecture and Key Technologies

● Signal Processing Pipeline (Physical to Information Layer):

(1)UHF Signal Capture:Convert EM waves to electrical signals (µV–mV).

(2)Low-Noise Amplification: Boost signals before noise intrusion.

(3)Bandpass Filtering: Remove out-of-band interference (e.g., radio signals).

(4)High-Speed ADC:Sample at ≥1GSPS to preserve pulse details (Nyquist: 3–5× highest frequency).

(5)Traitement numérique du signal (DSP) :

–Noise reduction via wavelet transforms.

–Feature extraction: amplitude, pulse width, phase, rise time.

(6)Diagnostic IA :Classify PD types (e.g., corona, voids) using SVM, CNN, etc.

 

● Multi-Sensor Localization (TDOA Method):

For large transformers, use 4+ sensors to use 4+ sensors

résoudre :

√[(x−xᵢ)² + (y−yᵢ)² + (z−zᵢ)²] − √[(x−x₁)² + (y−y₁)² +                                                                   (z−z₁)²] = v × Δtᵢ₁

Requires nanosecond time sync (IEEE 1588 PTP) and

known oil propagation speed (~1.5e8 m/s).

● System Integration and Standards Compliance:

(1)matériel:Smart sensors, PTP networks, edge servers.

(2)Logiciel: Real-time DSP, AI diagnostics, IEC 61850/Modbus interfaces.

Métrique	Objectif	Standard	Remarques
Min Detectable PD (pC)	<100 pC (internal), <500 pC (external)	CEI TS 62478	Sensitivity depends on sensor placement
Plage dynamique	> 60 dB	IEEE C57.127	Critical for strong/weak PD detection
Gamme de fréquences	300 MHz à 1.5 GHz	CIGRE WG D1.11	Covers primary PD energy
Time Sync Accuracy	<±1 ns (TDOA)	Norme IEEE 1588 (PTP)	Essential for multi-sensor localization

4. Value and Implementation of UHF Online Monitoring

UHF PD monitoring enables:

(1)Alerte précoce: Detect insulation flaws before catastrophic failures.

(2)Diagnostic précis :Classify PD types via PRPD patterns and localization.

(3)Maintenance prédictive: Reduce unplanned outages and extend asset life.

En résumé

UHF sensors offer a robust solution for online PD monitoring, combining high sensitivity, noise immunity, and AI-driven diagnostics. Global leaders like Siemens and Hitachi Energy deploy UHF systems for smart grid resilience. Advances in edge computing and machine learning will further enhance this technology, ensuring safer, more reliable power networks worldwide.

 

Contactez-Nous

LuShan, HNE.1975, est un fabricant professionnel chinois spécialisé dans les transformateurs de puissance et les réacteurs pour50 années. Les produits phares sont transformateur monophasé, triphasé seul transformateurs, transformateur électrique,transformateur de distribution, transformateur abaisseur et élévateur, transformateur basse tension, transformateur haute tension, transformateur de contrôle, transformateur toroïdal, transformateur à noyau R ;Inductances CC, réacteurs CA, réacteurs filtrants, réacteurs de ligne et de charge, selfs, réacteurs filtrants et produits intermédiaires à haute fréquence.

 

Notre pouvoir Les transformateurs et les réacteurs sont largement utilisés dans 10 domaines d'application : transport rapide, engins de chantier, énergie renouvelable, fabrication intelligente, équipement médical, prévention des explosions dans les mines de charbon, système d'excitation, frittage sous vide (four), climatisation centrale.

 

En savoir plus sur le transformateur de puissance et le réacteur :www.lstransformer.com.

 

Si vous souhaitez obtenir des solutions personnalisées pour les transformateurs ou les réacteurs, veuillez nous contacter.

WhatsApp:+86 17267488565
Courriel : marketing@hnlsdz.com