UNIT-VI

AI-based Power system Protection

Introduction to ANN:

Structure of a biological neuron Model of ANN

Application of Artificial Intelligence to Power System Protection :

• The application of Artificial Intelligence (AI) to power system

protection has been a significant area of research and
development in the energy sector.

• Power systems are critical infrastructures requiring high reliability,

fast fault detection, and efficient protection mechanisms.

• AI offers robust solutions to enhance these aspects.

1. Fault Detection and Classification

• AI Techniques Used:
• Neural Networks, Fuzzy Logic, Support Vector Machines (SVM),
and Decision Trees.

• Application:
• AI algorithms analyze data from current and voltage signals to detect and
classify faults in power systems (e.g., line-to-ground faults, three-phase
faults).
• Machine learning models learn the characteristics of various faults from
historical data and predict fault types with high accuracy.
2. Adaptive Protection Systems

• AI Techniques Used:
• Reinforcement Learning and Adaptive Neural Networks.

• Application:
• Traditional protection systems rely on fixed settings that may not adapt to
changes in system topology or load conditions.
• AI-based systems dynamically adjust protection settings based on real-
time conditions, ensuring optimal protection even under varying
operational scenarios.
3.Prediction and Prevention of Faults

• AI Techniques Used:
• Predictive Analytics, Anomaly Detection, and Deep Learning.

• Application:
• AI models predict potential faults by analyzing patterns in operational
data.
• Preventive measures can be taken before faults occur, reducing
downtime and damage to equipment.
4. Relay Coordination:

• AI Techniques Used:
• Genetic Algorithms and Optimization Algorithms.
• Application:
• AI is used to optimize the settings of protective relays to ensure proper
coordination, minimizing the impact of faults on the network.
5. Power Quality Monitoring

• AI Techniques Used:
• Expert Systems and Data-Driven Models.
• Application:
• AI analyses disturbances such as voltage sags, harmonics, and flickers.
• Automated systems ensure that power quality remains within acceptable
limits.
6. Protection of Renewable Energy Systems

• AI Techniques Used:
• Hybrid AI Models and Transfer Learning.
• Application:
• Integration of renewable energy sources like solar and wind creates
variability and uncertainty.
• AI ensures effective protection by addressing challenges such as
distributed generation and inverter-based systems.
7. Cybersecurity in Power Systems Protection

• AI Techniques Used:
• Intrusion Detection Systems (IDS), Machine Learning for Threat
Detection.
• Application:
• AI protects power systems from cyberattacks by identifying and
mitigating vulnerabilities in real-time.
• Anomaly detection algorithms monitor communication networks within
the power system.
8. Post-Fault Analysis

• AI Techniques Used:
• Case-Based Reasoning and Knowledge-Based Systems.
• Application:
• AI assists in identifying root causes of faults by analyzing post-fault data.
• It accelerates decision-making for repair and restoration processes.
• Benefits of AI in Power System Protection

1.Increased Reliability: Faster and more accurate fault detection and

resolution.
2.Improved Efficiency: Reduction in unnecessary tripping and system
downtime.
3.Scalability: Adaptability to complex and large-scale power networks.

4.Proactive Maintenance: Predictive capabilities reduce the likelihood

of catastrophic failures.
5.Integration with Smart Grids: Seamless handling of modern grid
complexities, including microgrids and distributed energy resources.
• Challenges in Implementation

1.Data Dependency: AI models require large amounts of quality

data for training.
2.Integration Complexity: Combining AI with existing protection
infrastructure can be challenging.
3.Real-Time Processing: Ensuring AI models operate fast enough
for real-time protection needs.
4.Security Concerns: AI systems must be protected against cyber
threats.
• Conclusion :

• AI applications in power system protection enhance the reliability,

efficiency, and adaptability of electrical grids.

• As power systems evolve with smart grid technologies and

renewable energy integration, AI's role will continue to grow,
driving innovation and improving system robustness.
APPLICATION OF ANN TO OVERCURRENT PROTECTION :
• Overcurrent protection is used for the protection of generators, power
transformers, motors , transmission and distribution lines.

• Therefore, overcurrent relays are amongst the most widely used relays in
a power system.

• The overcurrent relay senses the fault current and operates only when
the current in the circuit exceeds the pick- up value.

• The various time–current characteristics for the overcurrent relay can be

realized using artificial intelligence approaches to power system
protection.
• For realization of the overcurrent relay, the magnitude of the
fundamental frequency component of the current signal can be
estimated using appropriate digital filtering algorithm.

• Using data acquisition system 16 samples of the current signal is

acquired and fed as input to the input layer of the ANN.

• The output layer of overcurrent relay has only one neuron, and is used
for estimation of the magnitude of the fundamental component of the
current signal thereafter the operating (tripping) time corresponding to
a particular value of current is computed using the time–current
characteristics
APPLICATION OF ANN TO TRANSMISSION LINE PROTECTION:
• Identification and classification of faults on a transmission line are
essential for relaying decision and auto re-closing requirements.
• The implementation of a pattern recognizer for power system diagnosis
has provided great advances in the protection field.
• An artificial neural network can be used as a pattern classifier for
distance relay operation.
• The magnitudes of three-phase voltage (V a , V b , V c ) and
current (I a , I b , I c ) signals of the transmission line are measured using
current and voltage transformers and surge filters.
• The waveforms are sampled and digitized using analog to digital
converters.
• After the data acquisition, the signals are fed to a pattern recognizer.

• The ANN recognizer module then verifies for the fault, and if it exists, then
issues a trip signal.

• This ANN module can be either trained online or offline and the trip
decision depends on how the module was trained.

• The advantage of ANN based relays is that it has the ability to learn aspects
related to the fault condition and network configuration.

• The ANN approach works as a pattern classifier being able to recognize the
changing power system conditions and consequently improve the performance
of ordinary relays based on digital principle.
Block diagram of the ANN based distance relay
Application of ANN to Power transformer protection :
• A power transformer is an essential and important element of a power
system, and its protection is one of the most challenging problems in the
power system relaying area.

• Differential protection has been recognized as the principal basis for the
protection of power transformer against internal faults.

• The differential protection concepts are based on the assumption that

during an internal fault, the fundamental component of the differential
current becomes substantially higher than the current during normal
operating conditions.
Application of ANN to Power transformer protection :

• A differential relay which is used in the differential protection scheme

converts the primary and secondary currents of the transformer to a
common base and compares them.

• The non-zero value of the differential current indicates an internal fault.

• The direct and differential currents are calculated using the sampled
quantities.

• If there is a significant differential current, the ANN is used to

discriminate the internal fault conditions with other operating conditions.
Flowchart of the power
transformer protection
based on neural
network
• The differential relaying principle in the case of power transformer shows
certain limitations.

• Detection of a differential current does not provide a clear distinction between

internal faults and other conditions.

• Magnetizing inrush currents, over excitation of the transformer core, external

faults combined with current transformers (CTs) saturation and/or CTs and
protected transformer ratio mismatch are the most relevant phenomenon
which may upset the current balance causing the relay to mal-operate.

• The conventional approach to mitigate those problems is to apply percentage

(biased) differential characteristics along with second and fifth harmonic
restraints for inrush and over-excitation conditions, respectively.
• Since power transformer differential protection problem of
distinguishing among magnetizing inrush, over-excitation internal
fault and external fault currents of power transformer can be
considered as a current waveform recognition or pattern
recognition problem, the use of ANN seems to be an excellent
choice.
APPLICATION OF ANN TO GENERATOR PROTECTION:

• Synchronous generators are a major component in a power system and

any faults will hamper the available power to a system.

• Therefore, protecting these vital units against abnormal operating

conditions and faults, while at the same time keeping protection
schemes simple, reliable and fast in operation, has always posed
a challenge to the power system protection engineer.

• Stator winding faults of synchronous generator are considered serious

problems because of the damage associated with high fault currents and
high cost of maintenance.
• A high-speed bias differential relay is normally used to detect three
phase, phase-phase and double phase to ground faults.

• Detection of single line to ground faults depends on the generator

grounding type which can be classified into low and high impedance
grounding. In case of low impedance grounding, a differential relay can
detect and provide protection of only about, 95% of the windings.

• However, for high impedance grounding, ground faults are not normally
detectable by the differential relay because the fault current is, usually,
less than the sensitivity of the relay.
Block diagram of ANN based generator differential protection
• Above Figure shows an ANN based digital differential protection scheme
for generator stator winding protection.

• The waveforms of the generator differential current are sampled and

digitized using analog to digital converters.

• After the data acquisition, the signals are fed to a pattern recognizer.

• Differential current patterns of the generator are used to train the ANN at
different loading conditions and different types of internal and external
faults.

.
• The ANN based fault detector module is used to discriminate between
three generator states, i.e., the normal operation state, external fault state
and internal fault state.

• In the event of an internal fault the trip logic mod ule issues a trip signal
and activates the ANN based fault classifier module, which identifies the
faulted phase.
Thank you