Anda di halaman 1dari 7

Determination of Locational Marginal Price using Artificial Neural Networks

M.Suman , Dr.M.V.Rao Vignans Lara Institute Technology and Science,Vadlamudi-522213,India

Abstract: The methodology used in the pricing of electrical energy is a fundamental characteristic of electricity market design. In deregulated power systems the utilization of Locational Marginal Pricing is the dominant approach to pricing electrical energy.The conventional optimal power flow techniques such as Newtons method and Decoupled method are used for the determination of Locational Marginal Price. The conventional methods for computing LMP is no more suitable for spot electricity market as these are too slow and case sensitive methods.In this paper we proposed a soft computing based approach i.e Back Propagation Neural Network (BPNN) for projecting Locational Marginal Prices in a spot power market.This method provides fast and accurate results when compared with the conventional methods. Keywords: Electricity market, Locational marginal price, optimal power flow Artificial neural networks.

1. INTRODUCTION Locational Marginal Price (LMP) is defined as the marginal cost of supplying the next increment of electric energy at a specific bus while considering the generation marginal cost and the physical aspects of the transmission system. The ISO accepts energy and price bids submitted by market participants (e.g., sellers and buyers) and has the authority and responsibility to sort and select them based on a given objective while maintaining the system security operation. The ISOs objectives may range from a least cost formulation to the one that results in minimum possible adjustments to initial schedules. In addition, to maximize the social surplus may also become the ISOs objective. LMP reflects the market clearing price (MCP) at each location. In other words, buyers pay ISO based on their LMP for dispatched energy. In the deregulated electricity market[1], the market dispatch (unconstrained) stage is almost same for different market structures. These auction-based dispatches have no consideration for transmission constraints and the resulting price is the market clearing price (MCP). On the other hand LMP reflects the security constrained prices. At the system nodes on which no transmission constraints are violated, the LMPs are same and are equal to the marginal cost to serve load on those nodes. Locational marginal prices are highly volatile under constrained transmission condition. Sensitivities of LMPs are explored with respect to power demands in. Recently soft computing based artificial intelligent (Al) techniques are becoming popular for projecting locational prices in spot power market. Conventionally these prices are calculated by employing an optimal power flow program (OPF) . Conventional OPF[2] methods of computing LMPs are too slow, as every time a computationally intensive OPF program has to be run for any change in supply/ demand/ topology of the power network. Moreover, any noisy and any missing data would greatly affect the accuracy of the results. On the other hand once an appropriate ANN model has been developed, accurate and instant LMP values can be projected, irrespective of noisy and partially missing data for spot electricity pricing. The ANN based techniques are particularly well suited to price projections because of their generalization ability, fault tolerance and capability of mapping highly non-linear data patterns. In this project Locational Marginal Price is projected by using Newton, decoupled methods which are too slow and case sensitive methods and by using a Back Propagation Neural Network (BPNN). This ANN is particularly suitable for function approximation hence employed in this project for

LMP projection in spot power market. In this project LMPs of 3-bus, 6-bus & 14-bus systems are determined. 2. OPTIMAL POWER FLOW METHODS The optimal power flow is a power flow problem in which certain controllable variables are adjusted to minimize an objective function such as the cost of active power generation or the losses, while satisfying physical and operating limits on various controls, dependent variables. The types of controls that an optimal power flow must be able to accommodate are active and reactive power injections, generator voltages and phase-shift angles transformer tap ratios and phase-shift angles. There are many optimal power flow methods are available, but we have considered the following two methods. 2.1 Newton Method Newtons method is well-known in the area of power systems. It has been the standard solution algorithm for the power flow problem for decades. Newtons method is a very powerful solution algorithm because of its rapid convergence near the solution. This property is especially useful for power system applications because an initial guess near the solution is easily attained. System voltages will be near rated system values, generator outputs can be estimated from historical data, and transformer tap ratios will be near 1.0 p.u. Algorithm: 1 .Read cost function coefficients ai ,bi ,ci ,number of nodes n and number of control variables n1. 2. Form Ybus by using Ybus Algorithm 3. Calculate the initial values of pgi (i=1,2,.Ng) and by assuming PL=0
Nl Ng

Pdi
i 1 Ng i 1

pi

i 1

bi 2a i

bi 2a i

Pgi

bi 2 * ai
MB 1

Qdi

Qgi

Pgi *

i 1 NB i 1

Pdi

Initialize
pi = qi =0 Vi =1 i =0

4.By using initial values calculate the Jacobean and Hessian matrix element
Pg
L Pg L L p L V L q

H
p V q

=-

By using Guass elimination determine Pgi, Lagrange Equation is given by L ( Pg , V , ) F ( Pgi )


NB i 1 NB i 1

i,

pi, Vi,

qi.

pi qi

( Pi (V , ) ( Qi (V , )
NB

Pgi Qgi

Pdi ) Qdi )

Active Power at all the nodes is given by


P (V , ) B ij * sin( Vi *
i 1 i j

V j * ( G ij * cos( ))
NB

Reactive power at all the nodes is given by


Q (V , ) B ij * cos( Vi *
i 1 i j

V j ( G ij * sin( ))

7. Check the limits, if any limit of a variable is violated, then impose or remove power flow equation and go to step4 to update the solution. 8. Stop 2.2 Decoupled Method An important characteristic of any practical power transmission system operating in steady state is that the change real power from the specified value at a bus is more dependent on the changes in voltage angles at various buses than the change in voltage magnitudes, and the change in reactive power from the specified value at a bus is more dependent on the changes in voltage magnitudes, and the change in reactive power from the specified value at a bus is more dependent on the changes in voltage magnitudes various buses than the changes in voltage angles. This is due to the fact that transmission lines are mostly reactive, the conductance Gs are very small compared to the susceptance, Bs . Also , under the normal operating conditions the angle ( i- j) is small (typically less than 100 ). In view of this P/ V and Q/ being small can be ignored. Algorithm: 1 .Read cost function coefficients ai ,bi ,ci ,number of nodes n and number of control variables n1. 2. Form Ybus by using Ybus Algorithm 3. Calculate the initial values of pgi (i=1,2,.Ng) and by assuming PL=0 Ng Nl bi Pdi 2a pi i 1 Ng i 1 i bi i 1 2ai

Pgi

bi 2 * ai
MB 1

Qdi

Qgi

Pgi *

i 1 NB i 1

Pdi

Initialize pi = qi =0 Vi =1 i =0 4.By using initial values calculate the Jacobean and Hessian matrix element H1
Pg p

=-

L Pg L L p

H2

V q

=-

L V L q

Guass elimination method is used to solve these simultaneous equations separately to find Pgi, i, pi, Vi, qi. 5. Check convergence ( ( Pgi)2+ ( i)2+ ( pi)2 & ( V)2+ ( qi)2) If condition is not satisfied then go to step6 else go to step8 6.Modify the the variables as below Pgi= Pgi+ Pgi i= i+ i pi= pi+ pi V=V+ V qi= qi+ qi 7. Check the limits, if any limit of a variable is violated, then impose or remove power flow equation and go to step4 to update the solution. 8. stop 3. BACK PROPAGATION ALGORITHM Back Propagation is a systematic method for training multilayer artificial networks. It is a multilayer forward network using extend gradient-descent based delta-learning rule, commonly known as back propagation rule. Back propagation provides a computationally efficient method for changing the weights in a feed forward network, with differential activation function units, to learn a training set of input-output examples. Being a gradient descent method it minimizes the total squared error of the output computed by net. The network is trained by supervised learning method. The aim of this network is to train the net to achieve a balance between the ability to respond correctly to the input patterns that are used for training and the ability to provide good responses to the input that are similar. The architecture of the proposed Back Propagation Neural Network has been shown in fig.1.

1. BPNN architecture. Back Propagation Algorithm Initialization of the weights Step1: Initialize weights to small random values Step2: While stopping condition is false do Steps 3-10 Step3: For each training pair do steps 4-9 Feed Forward Step4: Each hidden unit receives the input signal xi and transmits the signals to all units in the layer above i.e. hidden units Step5: Each hidden unit sums its weihted input signals
n

inj

Voj
i 1

( X i * Vij )

applying activation function for to get output Z j f ( Z inj ) Step6: Each output unit sums its weighted input signals
p

ink

Wok
j 1

( Z j * W jk )

and apply activation function to calculate output Yk f (Yink ) Back Propagation of errors Each output unit receives a target pattern corresponding to an input pattern , error information term is calculated as (t k y k ) * f (Y ink ) k Step8: Each hidden unit sums its delta from units in the layer above
m inj k 1

* Wik )

The error information term is calculated as j inj * f ( Z inj ) Updation of the weights Step9: Each unit updates its bias and weights The weight correction term is given by W jk * k * Zi And the bias correction term is given by Wok * k

Therefore W jk ( new) W jk (old )

W jk

Wok ( new) Wok (old ) Wok Each hidden unit updates its bias and weights The weight correction term is given by Vij * j * Xi And the bias correction term is given by Vok * j Therefore Vij (new) Vij (old ) Voj ( new) Voj (old ) Voj Step10: Test stopping condition 4. Test System
5

Vij

Fig. Six bus, seven line power system Cost Functions are given as F1 (Pg1) =120* Pg12 +175* Pg1+75 F2 (Pg2) =110* Pg22 +180* Pg2+50 F3 (Pg3) =160* Pg32 +110* Pg3+80 Bus Data:
From Bus 1 2 2 3 3 1 1 BusNo. 1 2 3 4 5 6 To bus 2 3 5 4 6 5 6 V(p.u) 1.05 0 1 1.02 R 0.01 0.03 0.03 0.01 0.01 0 0 Pg(p.u) 0 0 0 X 0.3 0.2 0.2 0.2 0.2 0.2 0.2 Qg(p.u) 0 0 0 B/2 0.01 0.01 0.01 0.01 0.01 0 0 Pd(p.u) 0.1 0.3 0.1 0.15 0.35 0.3 T 0 0 0 0 0 1.01 1.02 Qd(p.u) 0 0.25 0.3 0.2 0.45 0.5

5. RESULTS AND CONCLUSIONS The methodology used in the pricing of electrical energy is a fundamental characteristic of electricity market design. In deregulated power systems the utilization of locational marginal pricing is thes dominant approach to pricing electrical energy. LMPs are calculated as Lagrangian multipliers (or dual variables) associated with OPF framework and are substantially affected by changes in demands, supply and transmission constraints and hence constitute essential information in an Electricity Market.
Bus No. 1(LMP1) 2(LMP2) 3(LMP3) 4(LMP4) 5(LMP5) 6(LMP6) Method Newtons Decouplede Back Propagation Newton Method 270.6076 269.6519 269.0974 269.9454 271.5628 270.3425 Execution Time 47.14 seconds 32.6880 seconds Immediatly Decoupled Method 270.6076 269.6519 269.0974 269.9454 271.5629 270.3425 BPN Method 270.661 269.728 269.178 270.033 271.621 270.391

In this paper Locational Marginal Price is determined by using Newtons method, Decoupled method and ANN method. The execution time of the Newtons method is 47.14 seconds The execution time for the decoupled method is 32.688 seconds The training for the ANN is 65 minutes. The execution time is very less it will give output immediately From the above conclusions it can be inferred that ANN is superior to conventional methods in view of execution time and accuracy. REFERENCES
[1]. M. Shahidehpour, H. Yamin and Z.Y. Li, Market operations in electric power System, John Wiley

&Sons, Inc., New York, 2002.


[2]. STEVEN STOFT, Power System Economics: Designing Markets for Electricity, [3]. [4].

[5]. [6].

IEEE Press/WileyInterscience, May 2002 D. P. Kothari, J. S. Dhillon, Power System Optimization, Prentice Hall of India Private Limited, New Delhi, 2004. Seema N.pandey, Shasikala Tapaswi, And Laxmisrivastava,,Locational Marginal Price Projection using using a Novel RBFNN Approach in spot power market, Power System Technology and IEEE Power India Conference, 2008. Milano, "Locational Marginal Price Sensitivities" IEEE Transactions on Power Systems, vol.20, no. 4, pp. 20262032, Nov. 2005. S. N. Sivanandam, S. Sumathi, S.N. Deepa, Introduction to Neural Networks using. MATLAB 6.0 , Tata McGraw- Hill, New Delhi, 2006.

Anda mungkin juga menyukai