Anda di halaman 1dari 26

EXPERT SYSTEM

APPLICATIONS

M. Sarayu
Assistant Professor
Department of EEE
CIT, Coimbatore
EXPERT SYSTEM SRUCTURE

3/30/2019 MS/AP/EEE/CIT 2
Expert System for V–Q Control in Power
Systems
To maintain voltage at desired levels and prevent phenomena of
instability and voltage collapse

• planning of compensation means

• V–Q regulation in power systems

• perfecting the wide area monitoring and control techniques

Heuristic techniques are based upon the experience and knowledge


accumulated by specialists from the power engineering field

3/30/2019 MS/AP/EEE/CIT 3
Expert System for V–Q Control in Power
Systems
Purpose of an ES for voltage stability evaluation and V–Q control
• To assist the area operators or dispatcher
– To evaluate the risk of instability phenomenon appearance or voltage collapse
– To decide the appropriate V–Q control settings – most efficient measure

ES suggests solutions for control of the voltage level in the network as a


consequence of events
– Increases in consumption
– Network Topology Changes
• Unavailability of generators
• Unavailability power distribution network branches

3/30/2019 MS/AP/EEE/CIT 4
EXPERTVQ SYSTEM STRUCTURE

3/30/2019 MS/AP/EEE/CIT 5
ROLE OF EXPERTVQ SYSTEM
• Assess the voltage stability of the electrical power system and identify the
most vulnerable nodes in terms of V–Q variation

• Identify, based on the concept of weak connections, the V–Q control areas
and establish the reactive power balance and reserve of each zone

• Detect the nodes with voltage problems and the areas in which they are
located

• Appreciate the voltage fault degree of complexity, and accordingly, launch a


specific procedure to solve the problem

• Through a user friendly interface, provide the operator with timely


information on the state of the power system, and the necessary control
actions needed in order to steer the voltage level to an admissible band
3/30/2019 MS/AP/EEE/CIT 6
ALGORITHMIC CALCULATION MODULE
• Implemented in C++ programming language
• Designed to complete the facts in the knowledge database from the heuristic
module with system state information at the time of analysis
• The steady-state load flow computations - Newton–Raphson method on the
primary database that contains information entered by the user, nodal data,
and network topology obtained directly from the system
Determines
 The local stability indicators SVQ and SQgQc and the nodal voltage
sensitivity with the variation of the control values
 The factors of participation for the PQ nodes to the system V–Q variation
modes, by analyzing the reduced Jacobian matrix

3/30/2019 MS/AP/EEE/CIT 7
ALGORITHMIC CALCULATION MODULE
V–Q Control Areas and Weak Connections
Static methods were used to establish the V–Q correlations of a specific power system, based
on a mathematic model consisting of the linearized load flow equations

where [J′] is the complete Jacobian matrix.

The standard model from was modified in order to include the influences of loads
characteristics, the transformers with control under load, and the limitations imposed on the
generator stator and rotor currents in triggering phenomena of instability and voltage collapse

3/30/2019 MS/AP/EEE/CIT 8
ALGORITHMIC CALCULATION MODULE
Sensitivity Coefficients
For voltage stability evaluation and selection of the most effective means of control
necessary to solve a voltage problem, the following were used

• Sensitivity SVQ of voltage at a consumption node i as the variation of reactive power

consumed in the same node is defined by the relationship

• Sensitivity SQgQc of the generators output reactive power as the variation of the

reactive power consumed in a consumer node i is defined by the relationship

3/30/2019 MS/AP/EEE/CIT 9
ALGORITHMIC CALCULATION MODULE
Sensitivity Coefficients
The sensitivity matrix [H], which gives the relationship between the voltage variations
[ΔVc] in the P–Q nodes and generators output reactive power variations [ΔQg],
respectively, and the variation of the quantities associated with the control variables
(voltage variations in generator nodes [ΔVg], reactive power output variation at capacitor
banks or variation in the reactive power absorbed by compensation reactors [ΔQinj], and
the transformer ratio variation [ΔNij]) as

3/30/2019 MS/AP/EEE/CIT 10
ALGORITHMIC CALCULATION MODULE
Participation factors
Given the strong bound between voltage and reactive power in a power system, the V–Q
dependency is described by [ΔQ] = [JR ][ΔV], where [JR ] is the reduced Jacobian matrix
obtained by eliminating the angle variations [Δθ] & considering [ΔP] = 0

By applying modal analysis to the [JR ] matrix, it can be shown that

where Rki and Lik are right and left elements of the eigen vectors Rk , respectively, left
Lk corresponding to the eigenvalue λk ,

Pki is the factor of participation of node i to the variation mode k and constitutes an
indicator of the node’s vulnerability

3/30/2019 MS/AP/EEE/CIT 11
ALGORITHMIC CALCULATION MODULE
V–Q Controllability and Areas of Control
The phenomena of instability and voltage collapse are determined by the inability of the network to transmit
an additional load to a demand center as a result of reaching the maximum power transfer capacity.
Accordingly, V–Q control areas are defined, consisting of PQ and PV nodes that are similar in terms of
voltage behavior and the links between them are identified as vulnerable links. The zone separating
algorithm relies on the information provided by the submatrix [JQV ] extracted from the complete Jacobian
matrix that has the elements

For a vulnerable electric line, the value Lik has a negative, small value because it either has a high
impedance (Yik is close to zero), or it operates near the limit of maximum power transfer capacity, that is
sin(θi −θ k −γik) ≅ 0.

Therefore, increasing the reactive power transmitted through this electric line leads to Lik becoming zero or
even positive, which may lead to a singularity of the [J′] matrix and system instability because of the voltage
and/or angle bifurcations

3/30/2019 MS/AP/EEE/CIT 12
HEURISTIC MODULE
Developed in the ‘C’ Language Integrated Production System (CLIPS)
environment
– Supports building expert system applications

– It has a high performance IE with forward chaining

– Allows a relatively easy interface with algorithmic calculation modules


written in C++

Knowledge engineer can customize many different strategies within CLIPS

It is intended for facts-based inference from the knowledge database by using a


system of production rules of the form

IF condition_1 & condition_2 &….& condition_n THEN conclusion


(action)

3/30/2019 MS/AP/EEE/CIT 13
HEURISTIC MODULE - TASKS

3/30/2019 MS/AP/EEE/CIT 14
HEURISTIC MODULE - TASKS

3/30/2019 MS/AP/EEE/CIT 15
HEURISTIC MODULE - TASKS

3/30/2019 MS/AP/EEE/CIT 16
HEURISTIC MODULE - TASKS

• Voltage-related issues are quantified in the EXPERTVQ system by a


heuristic function,
• Control mean selected is that for which the function takes the maximum
value
• After selecting the most effective means of control, based on sensitivity
coefficients, the heuristic module estimates the effect of its action
• if necessary, it requests the recalculation of sensitivity coefficients by the
algorithmic module

3/30/2019 MS/AP/EEE/CIT 17
CASE STUDY
In order to test the ability of the EXPERTVQ system, to assess the stability of
the voltage and to solve a voltage problem, the standard IEEE 30-bus network
was used

3/30/2019 MS/AP/EEE/CIT 18
CASE STUDY - IEEE 30-bus test network

• Note that the values of the vulnerability factors of the branches and nodes
are specific to each network, but the admissible values of the voltage have a
general character
• The specific scenario of interest is a voltage problem due to a load increase
at bus 29 correlated with a disconnecting of the line 1–2 and the generator
at node 13

3/30/2019 MS/AP/EEE/CIT 19
CASE STUDY - IEEE 30-bus test network
In the first stage, based on the initial facts and on the tasks that it has to
complete, the EXPERTVQ program offers the following summary of the state of
the system
Task T1: The system is stable and V–Q controllable, but the risk of voltage
instability and voltage collapse phenomena is high
Task T2: The system is divided into two zones: area 1 is exporting and
controllable, and area 2 is importing and controllable
Task T3: The mode of variation of V–Q is dispersed, the most vulnerable
nodes being node 9 in area 1 and node 29 in area 2
Task T4: There are 11 nodes in the network with voltage problems (14, 15,
18, 19, 23, and 24 in area 1, respectively, 25, 26, 27, 29, and 30 in area 2).
The severity degree of the voltage problem is 1, and the most affected node
is 29 from area 2
• At this stage, because the severity degree is 1, the heuristic procedure is
triggered
• For the purpose of solving this problem, three iterations are performed

3/30/2019 MS/AP/EEE/CIT 20
Synthesis of actions proposed by the expert system
The suggested actions and the estimated effects are summarized in Table

• In order to validate the actions suggested by the expert system, they have been
input to the primary database and then system continues in algorithmic
calculation mode
• Obtained results are presented in the histogram alongside with those existing in
the initial database

3/30/2019 MS/AP/EEE/CIT 21
OTHER APPLICATIONS OF
EXPERT SYSTEM
Expert System for Distribution Networks
Reconfiguration
• Distribution electrical networks are designed in simple loop or meshed configuration,
but they are operated radially or in arborescent configuration

• Depending on the distribution of loads in different time periods of a day or during the
year season, as well as other characteristics, the network configuration can be
modified by changing the operating state of some branches with the purpose of
achieving technical or economic benefits, such as minimization of the total active
power losses or improvement of the voltage level

• Various optimization techniques aiming various objectives may be employed to


determine the optimal configuration

• While the heuristic methods are the most used, the tabu search is an efficient solution
used today

3/30/2019 MS/AP/EEE/CIT 23
Expert System for Distribution Networks
Reconfiguration
• ES assists the network operator in a real-time activity to perform the network
reconfiguration

• Reconfiguration process consists in a series of steps, proposed and verified by calculations,


which include connection/ disconnection of network branches

• In an on-line system, ES uses real-time data available within the SCADA system from the
dispatching center, among which the state quantities, the state of branches operation, etc.

• For the action proposed at each step, ES calls an algorithmic module to calculate the
network load flow and checks if the security constraints are satisfied

• If no violation is identified, the decision proposed in the current step is sent to the
maneuvers module to assist the human operator in the equipment switching

• Once the current step is validated by the system operator, a new action is proposed in a new
step, and the reconfiguration process continues until the final objective is achieved
3/30/2019 MS/AP/EEE/CIT 24
Expert System Power System Restoration After
Blackouts
• Power systems around the world have been subjected to a large number blackouts, of
various severity. In order to minimize the effects of a blackout, immediate restoration
actions must be taken in order to supply the loads.

• Specific grid codes regarding the procedures to be followed by the system operators in
case of a blackout are available in any power system.

• However, automatic tools may help the human operator to rapidly perform
calculations and equipment energization.

• ES can be a component of the energy management system (EMS) of a dispatch center,


and it can communicate with other tools and systems.

3/30/2019 MS/AP/EEE/CIT 25
Expert System Power System Restoration After
Blackouts
• Several modules - database module, reasoning module, data update module,
algorithmic computation module, on-line assisting equipment maneuverings
module

• Determines the spread of the blackout, identifies the possible sources to energize
the affected areas, and proposes the steps to be followed by the human operator

• Steps can be the result of previous procedures predefined by the human expert
into the ES modules

• Searching strategy is based on two cores

– one core aims to synchronize the power plants,

– one core aims to reconnect the loads

• Human operator may also decide if alternative steps should be considered


3/30/2019 MS/AP/EEE/CIT 26

Anda mungkin juga menyukai