1) TOAN is a new diagnostic tool that analyzes DGA monitor data to detect actual transformer faults, eliminating many false alarms.
2) It uses advanced computational techniques like artificial neural networks and harmonic regression to analyze large amounts of DGA data and automatically detect four types of faults.
3) TOAN has demonstrated 96% accuracy in identifying fault conditions, equaling or surpassing other diagnostic techniques, and provides notifications and recommendations without the need for daily analysis by experts.
A Critique of the Proposed National Education Policy Reform
Toan 2019 introducing toan-diagnostics
1. TOAN - Transformer Oil Analysis and No fica on - is the first en rely new DGA
diagnos c tool to emerge in recent years. It allows the user to move away from
alarming on DGA gas levels or rate of change and towards alarming only when an
actual fault is developing.
™
Introducing TOAN
TOAN - Transformer Oil Analysis and No fica on
$28M
The concepts that underline TOAN have been developed around
advanced computa onal techniques to solve the “big data” issues
associated with wide scale deployment of online DGA monitors.
®
Developed by an expert team at an electrical u lity in the USA, Serveron
now offer TOAN as an op onal plug-in to the groundbreaking TM View™
so ware suite. TOAN has been specifically designed for u li es where
availability of me and or DGA exper se is inadequate. It provides a
pla orm to move away from day to day analysis of DGA data and towards
automa c alarming on real faults. Virtually elimina ng false alarms, TOAN
simplifies the task of supervision of DGA monitors. TOAN can analyze
data from large or small popula ons of online DGA monitors,
automa cally detec ng faults and providing accurate diagnosis while
minimizing the false alarms o en associated with Rate of Change and
PPM alarm se ngs.
TOAN is available as a plug-in applica on in Serveron’s DGA monitoring
pla orm TM View. TM View provides a broad range of diagnos c and
trending capabili es as standard and free of charge with all Serveron DGA
monitors. TOAN may be ac vated within TM View on purchase of a
license key.
Arizona Public Services had a drama c
transformer failure in 2004. Repair costs
were calculated at $28M!
TOAN was born out of this catastrophic
event. Today TOAN is employed by electrical
u li es and transformer operators globally to
aid them in protec ng their cri cal assets.
TOAN provides mely and accurate fault
condi on alarms from online DGA monitors.
Advanced fault alarming and diagnos cs for your
cri cal assets
Ÿ Automa cally creates ACTIONABLE INFORMATION
from large volumes of data
Ÿ No fies the users only when a fault is present, thereby
filtering out false alarms
Ÿ Uses DATA MINING techniques to let the data tell you
the trends
TOAN - Transformer Oil Analysis and No fica on
TOAN:
1
Automate
No fy
Prevent
TOAN DGA
diagnos c tool allows
asset managers to
2
Automate the monitoring of
DGA data.
Receive no fica on of
abnormali es in near-real me
Take ac ons necessary to
prevent outages or more
transformer failure
2. TOAN - Transformer Oil Analysis and No fica on - is the first en rely new DGA
diagnos c tool to emerge in recent years. It allows the user to move away from
alarming on DGA gas levels or rate of change and towards alarming only when an
actual fault is developing.
™
Introducing TOAN
TOAN - Transformer Oil Analysis and No fica on
$28M
The concepts that underline TOAN have been developed around
advanced computa onal techniques to solve the “big data” issues
associated with wide scale deployment of online DGA monitors.
®
Developed by an expert team at an electrical u lity in the USA, Serveron
now offer TOAN as an op onal plug-in to the groundbreaking TM View™
so ware suite. TOAN has been specifically designed for u li es where
availability of me and or DGA exper se is inadequate. It provides a
pla orm to move away from day to day analysis of DGA data and towards
automa c alarming on real faults. Virtually elimina ng false alarms, TOAN
simplifies the task of supervision of DGA monitors. TOAN can analyze
data from large or small popula ons of online DGA monitors,
automa cally detec ng faults and providing accurate diagnosis while
minimizing the false alarms o en associated with Rate of Change and
PPM alarm se ngs.
TOAN is available as a plug-in applica on in Serveron’s DGA monitoring
pla orm TM View. TM View provides a broad range of diagnos c and
trending capabili es as standard and free of charge with all Serveron DGA
monitors. TOAN may be ac vated within TM View on purchase of a
license key.
Arizona Public Services had a drama c
transformer failure in 2004. Repair costs
were calculated at $28M!
TOAN was born out of this catastrophic
event. Today TOAN is employed by electrical
u li es and transformer operators globally to
aid them in protec ng their cri cal assets.
TOAN provides mely and accurate fault
condi on alarms from online DGA monitors.
Advanced fault alarming and diagnos cs for your
cri cal assets
Ÿ Automa cally creates ACTIONABLE INFORMATION
from large volumes of data
Ÿ No fies the users only when a fault is present, thereby
filtering out false alarms
Ÿ Uses DATA MINING techniques to let the data tell you
the trends
TOAN - Transformer Oil Analysis and No fica on
TOAN:
1
Automate
No fy
Prevent
TOAN DGA
diagnos c tool allows
asset managers to
2
Automate the monitoring of
DGA data.
Receive no fica on of
abnormali es in near-real me
Take ac ons necessary to
prevent outages or more
transformer failure
3. Surpasses all other diagnos c techniques
Ÿ The severity of the fault category is assigned and rated
within a 6-level scale, with 1 being the most severe
Ÿ The applica on window shows the final score and
recommenda on for the monitor, then the individual
scores for each fault types
Ÿ The weight values represent the ANN scores for each
fault type. These values are between 0 and 1 and
represent likeliness for that type of fault, as
determined by the neural network analysis
Ÿ No fica ons can be customized by fault category and
severity to enable excep on-based analysis
Ÿ A rule-based step at the end of the analysis yields a
final 'score' for that transformer. If the score is not
within an acceptable range, an alarm is triggered and
emails can be sent to selected users of the system
4 fault condi ons are iden fied by TOAN
96%
Ar ficial Neural Network
Harmonic Regression
Correct fault condi on
iden fica on
-
TOAN has a demonstrated
capability for correct fault
condi on iden fica on in 96%
of cases - equaling or surpassing
all other diagnos c techniques
and TOAN does it all automa cally
and con nuously
Data flow for a TOAN analysis:
3 TOAN - Transformer Oil Analysis and No fica on 4
TOAN - Transformer Oil Analysis and No fica on
HEDA
High
Energy Discharge
LED
Low
Energy Discharge
OHO
Over Heated
Oil
CD
Cellulose
Decomposi on
Harmonic Regression to remove
harmonic components in the data,
clearly revealing the underlying
trends - also Piecewise Linear
Approxima on to accurately
assess gassing rate of change
ANN is trained on large data
sets that reference pre-failure
DGA data with post-failure
inspec on results
TOAN u lizes programming
and an Expert System for
fault condi on iden fica on
Use ASTM
Correc on?
ASTM
Correc on
Harmonic
Regression
CO2?
Harmonic
Regression
for CO, CO2
Determine
Gassing Rates
Run
Neural Nets for
Each Fault Type
Run
Rule Based
Expert System
Fuzzy Logic
Combine
All Decisions
Send Email
to All Recipients
Is there
an Alarm?
Monitors
®
SERVERON
Poller
END
®
SERVERON
Database
YES
YES
YES
NO
NO
NO
4. Surpasses all other diagnos c techniques
Ÿ The severity of the fault category is assigned and rated
within a 6-level scale, with 1 being the most severe
Ÿ The applica on window shows the final score and
recommenda on for the monitor, then the individual
scores for each fault types
Ÿ The weight values represent the ANN scores for each
fault type. These values are between 0 and 1 and
represent likeliness for that type of fault, as
determined by the neural network analysis
Ÿ No fica ons can be customized by fault category and
severity to enable excep on-based analysis
Ÿ A rule-based step at the end of the analysis yields a
final 'score' for that transformer. If the score is not
within an acceptable range, an alarm is triggered and
emails can be sent to selected users of the system
4 fault condi ons are iden fied by TOAN
96%
Ar ficial Neural Network
Harmonic Regression
Correct fault condi on
iden fica on
-
TOAN has a demonstrated
capability for correct fault
condi on iden fica on in 96%
of cases - equaling or surpassing
all other diagnos c techniques
and TOAN does it all automa cally
and con nuously
Data flow for a TOAN analysis:
3 TOAN - Transformer Oil Analysis and No fica on 4
TOAN - Transformer Oil Analysis and No fica on
HEDA
High
Energy Discharge
LED
Low
Energy Discharge
OHO
Over Heated
Oil
CD
Cellulose
Decomposi on
Harmonic Regression to remove
harmonic components in the data,
clearly revealing the underlying
trends - also Piecewise Linear
Approxima on to accurately
assess gassing rate of change
ANN is trained on large data
sets that reference pre-failure
DGA data with post-failure
inspec on results
TOAN u lizes programming
and an Expert System for
fault condi on iden fica on
Use ASTM
Correc on?
ASTM
Correc on
Harmonic
Regression
CO2?
Harmonic
Regression
for CO, CO2
Determine
Gassing Rates
Run
Neural Nets for
Each Fault Type
Run
Rule Based
Expert System
Fuzzy Logic
Combine
All Decisions
Send Email
to All Recipients
Is there
an Alarm?
Monitors
®
SERVERON
Poller
END
®
SERVERON
Database
YES
YES
YES
NO
NO
NO