Power System Voltage Stability Assessment through ... - ScienceDirect

Available online at www.sciencedirect.com

Procedia Engineering 00 (2011) 000–000

Procedia Engineering 30 (2012) 53 – 60

Procedia Engineering www.elsevier.com/locate/procedia

International Conference on Communication Technology and System Design 2011

Power System Voltage Stability Assessment through Artificial Neural Network O.P. Rahib, Amit Kr Yadavb, Hasmat Malika, Abdul Azeemb, Bhupesh Krb, a* a,b

Electrical Engineering Department,National Institute of Technology, Hamirpur 177005,India

Abstract Voltage stability is concerned with the ability of a power system to maintain acceptable voltages at all buses in the system under normal conditions and after being subjected to a disturbance. This paper presents a new technique to determine the static voltage stability of load buses in a power system for a certain operating condition and hence identifies load buses which are close to voltage collapse. A voltage stability index with respect to a load bus is formulated from the voltage equation derived from a two bus network and it is computed using Thevenin equivalent circuit of the power system referred to a load bus. . Buses with values of voltage stability factors close to 1 .0 are identified as the critical buses. And after that, an Artificial Neural Network is developed for voltage stability monitoring. In ANN applications, selection of input variables is an important aspect

© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of ICCTSD 2011

Open access under CC BY-NC-ND license.

"Keywords: Voltage stability; Voltage Stability Index; Artificial Neural Network (ANN); Security Assessment;critical bus; Stability Index Prediction"

1. Introduction Voltage stability is a major concern in planning and operations of power systems. It is well known that voltage instability and collapse have led to major system failures; with the development of power markets, more and more electric utilities are facing voltage stability-imposed limits .A system enters a state of voltage instability when a disturbance increases in load demand, or change in system condition causes a progressive and uncontrollable decline in voltage and the process may result in voltage collapse. The main factor causing instability is the inability of the power system to meet the demand for reactive power. Knowledge of the voltage stability margin is of vital importance to utilities in order to operate their system with maximum security and reliability. The system operator must be provided with an accurate and fast method to predict the voltage stability margin so as to initiate the necessary control actions. Several works have been conducted previously for the prediction of voltage stability and proximity to collapse conditions based on conventional techniques, using sensitivity indices and based on continuation methods [1-11]. Continuation method based voltage stability analysis techniques are fairly accurate but hampered by the fact of taking longer computational time being based on repetitive power flow. For online voltage stability, index suits well, if it is accurate indicator of voltage stability margin and can be calculated fast. Some of the works focus on static voltage stability analysis using such instability measuring indicators as that of the popular L-index [6-7]. This index

* Hasmat Malik. Tel.: +91-9736788386; E-mail address: [email protected]

1877-7058 © 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license. doi:10.1016/j.proeng.2012.01.833

54

O.P.Malik/ Rahi etProcedia al. / Procedia Engineering 30 (2012) 53 – 60 Hasmat Engineering 00 (2011) 000–000

gives sufficiently accurate as well as practical means of the assessment, and can express the stability analysis in simple and operator friendly way. Application of Artificial Neural Network (ANN) to the above-mentioned problem has attained increasing importance mainly due to the efficiency of present day computers. Moreover real-time use of conventional methods in an energy management center can be difficult due to their significant large computational times. One of the main features, which can be attributed to ANN, is its ability to learn nonlinear problem offline with selective training, which can lead to sufficiently accurate online response. ANN approach to voltage stability assessment and improvement has been proposed and various neural network combinations have been used for solving the problem in [12-18]. The ability of ANN to understand and properly classify such a problem of highly non-linear relationship has been established in most of them and the significant consideration is that once trained effectively ANN can classify new data much faster than it would be possible with analytical model. 2. DERIVATION OF VOLTAGE STABILITY INDEX The mathematical formulation for the voltage stability index is derived from voltage equation used in the Improved Distribution load flow technique [19]. The derivation for the voltage stability index is as follows, Consider again a line connecting bus i to i+1,

P i, Qi, V i,

θi Pi+1, Q i+1, V i+1,

θi+1

ri +jxi PLi, QLi

PLi+1, QLi+1

The voltage equation would be:Vi+14+Vi+12[2(Pi+1ri+Qi+1xi)-Vi2] + Pi+12+Qi+12ri2+xi2=0 Where,

P i, Qi =Real & Reactive power injection at bus i V i = Voltage at bus i ri , xi= resistance and reactance of line connecting bus i & i+1

The equation is quadratic inVi+12 and it will have real roots if; Hence from equation (1), 8Pi+1Qi+1rixi - 4Vi2(Pi+1ri + Qi+1xi) + Vi4 – 4(Pi+12xi2 + Qi+12xi2) ≥ 0

(1)

b2-4ac>=0 (2)

This can be simplified to,

4[Vi 2 ( Pi 1ri Qi 1 xi ) ( Pi 1 xi Qi 1ri ) 2 ] 1 Vi 4

(3)

Therefore, the voltage stability index is given by,

L

4[Vi 2 ( Pi 1ri Qi 1 xi ) ( Pi 1 xi Qi 1ri ) 2 ] Vi 4

(4)

55

O.P. RahiMalik/ et al. / Procedia Procedia Engineering Engineering 00 30 (2011) (2012) 000–000 53 – 60 Hasmat

Since,

VV i i 1 cos( VV i i 1 sin(

And,

i i

i 1 i 1

) Vi

2 1

Pi 1ri Qi 1xi

(5)

Pi 1 xi Qi 1ri

)

(6)

Substituting equations (5) and (6) in (4),

L

4[VV i i 1 cos(

i

) Vi Vi 2

i 1

2 1

cos(

i

i 1

)2 ]

(7)

The voltage stability index derived is then applied to the thevenin equivalent circuit looking at the load bus concerned as shown in fig. 2.Vth is the thevenin voltage, which is the open circuit voltage at the bus concerned, ZL φL is the equivalent impedance of the load connected at that bus and Zth is the Thevenin impedance across the bus. The following steps are used to determine the Thevenin’s equivalent circuit with respect to a load bus:a) Run the load flow to obtain the voltage profile of a power system at given loading condition. b) Obtain Vth by running the load flow with the load connected at that bus removed. Applying equation (7) to the Thevenin’s equivalent circuit, gives

L

4[VoVL cos(

o

) VL 2 cos( Vo 2

L

o

L

)2 ]

(8)

L must be kept less than 1.0 to maintain voltage stability. If L exceeds 1.0, according to equation (3) the voltage at the referred bus becomes imaginary which indicates that voltage collapse has occurred in the system. Hence, The Voltage stability index is termed as symbol Land given by equation Voltage Stability Index,  Where L= Stability Index, Vo = No load voltage, and VL= Load Voltage

(9)

In order to maintain a stability of voltage condition in the system network, the value of L any load bus must be kept less than 1.0. If the value of L evaluated at load bus approaches 1.0 the voltage is referred as collapse conditions. This index will be used in the voltage stability analysis caused by the heavy loadings. It will be assigned as the target output of the ANN. 3. Proposed ANN Based Approach A multilayer feed forward Artificial Neural Network with error back propagation learning was used. The developed network consists of three basic elements. 1. Neural network Architecture 2. Suitable back propagation learning algorithm 3. Method of training and testing 3.1 Neural Network Architecture The architecture of a developed network consists of input layer, one hidden layer, and one output layer. Sigmoid activation function is used for each neuron. The architecture of develop network is shown in figure3.

56

O.P.Malik/ Rahi etProcedia al. / Procedia Engineering 30 (2012) 53 – 60 Hasmat Engineering 00 (2011) 000–000

Fig.2. Neural Network Architecture

3.2 Selection of Input Patterns Selection of the training parameters is an important task in designing an ANN. A few important parameters can severely reduce the learn ability of the network. Similarly absence of few important parameters may reduce the accuracy. Total of 300 input/output patterns are generated. Out of 300 patterns, 250 patterns are used for training purpose and 50 patterns are for testing purpose. 28 Inputs are used for IEEE 14-bus system, 25 are bus power injections (active and reactive) and 3 inputs are tap settings of transformers. Numbers of hidden layer neurons are 20 and number of output neurons are 9. 3.3 Training & Testing The training process is performed by using adaptive error back propagation algorithm order to train the developed network with a set of input and targeted output. 250 patterns were randomly selected from the voltage stability results. The network was trained until it achieved a very small MSE typically 0.0022 p.u. The network which was fully trained with the lowest error is capable to be used in the testing process. Performance of Ann to Minimize the RMS Error by showing the training, validation and testing curve is shown in figure5.The generalized capabilities of proposed Neural Network have been investigated by using 50 testing patterns. The reliable network should be able to produce a very good output results. Beside the overall MSE error, the performance of developed neural network can also be justified by using calculating the absolute error for every testing pattern and by using line regression as shown in figure 6.To further identify the accuracy of the developed network, the developed network must be indicating correlation coefficient, R closest to unity (1.0000). Unity value of R indicates zero absolute and RMS error. 4. Results & Discussion Test system presented in this study is the IEEE 14-bus system. Single line diagram of IEEE-14 bus system is shown below:

57

O.P. RahiMalik/ et al. Procedia / Procedia Engineering (2012) 53 – 60 Hasmat Engineering 00 30 (2011) 000–000

1 3

1 4

20

16

19

1 1

1 2 15

17

1 0

18

17

19

6

9

G

G

8

12

10 7

8

11

5

9

4

7 4 5

G

3

6

1 1 2

2

3

G

G

Fig.4. Modified IEEE 14-Bus System

TABLE I VOLTAGE STABILITY INDEX AT DIFFERENT BUSES AT A NORMAL LOADING CONDITION (STABLE CONDITION)

TABLE II. VOLTAGE STABILITY INDEX AT DIFFERENT BUSES WITH 14TH BUS HEAVILY LOADED (UNSTABLE CONDITION)

Bus

Target L

Output L

Error

Bus

4

0.0849

0.0850

0.0001

4

0.2315

0.2314

-0.0001

5

0.0772

0.0774

0.0002

5

0.2115

0.2114

-0.0001

7

0.0944

0.0944

0.0000

7

0.2925

0.2927

0.0002

9

0.1482

0.1481

-0.0001

9

0.4501

0.4508

0.0007

10

0.1465

0.1464

-0.0001

10

0.4032

0.4038

0.0006

11

0.0907

0.0907

0.0000

11

0.2344

0.2345

0.0001

12

0.0654

0.0657

0.0003

12

0.1971

0.1974

0.0003

13 14

0.0985 0.2041

0.0985 0.2035

0.0000 -0.0006

13 14

0.3688 0.9222

0.3691 0.9223

0.0003 0.0001

Target L

Output L

Error

Above table shows the L-index values at each bus for the base case. From the above table it can be observed that Bus-14 is more prone to voltage instability as its L-index is nearer to unity compared to other buses. Stability of bus 5 is more as its L-index is low. Above table shows that with the increase of load on 14th bus, the voltage stability index, L, increases and reached to 0.9222 which is very near to unity value. From the above table it can be observed that Bus-14 is more prone to voltage instability as its L-index is nearer to unity compared to other buses.

58

O.P. Rahi et al. / Procedia Engineering 30 (2012) 53 – 60 Hasmat Malik/ Procedia Engineering 00 (2011) 000–000

Fig. 5.Performance of Ann to Minimize the RMS Error

Figure 5 shows the plotting of training, validation and test error to check the progress of training. The training stopped after 15 iterations because the validation error is increased. The result here is reasonable, because the test set error and the validation set error have similar characteristics. TABLE III RESULT OF VOLTAGE STABILITY INDEX PREDICTION USING ANN ON CRITICAL BUS-14 S.N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Output L 0.2035 0.1864 0.1864 0.1991 0.1832 0.1735 0.1611 0.2080 0.1950 0.1694 0.1553 0.1913 0.1603 0.1656 0.2019 0.1348 0.1906 0.1827 0.1927 0.1778 0.1928 0.1772 0.1887 0.1839

BUS 14 Target L 0.2041 0.1865 0.1861 0.199 0.1844 0.1738 0.1616 0.2079 0.1951 0.1694 0.1554 0.1911 0.1602 0.1661 0.2015 0.1339 0.1906 0.1836 0.1918 0.1779 0.1927 0.178 0.1898 0.1839

Error -0.0006 -0.0001 0.0003 0.0001 -0.0012 -0.0003 -0.0005 0.0001 -0.0001 0.0000 -0.0001 0.0002 0.0001 -0.0005 0.0004 0.0009 0.0000 -0.0009 0.0009 -0.0001 0.0001 -0.0008 -0.0011 0.0000

S.N 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Output L 0.1936 0.1906 0.1919 0.1691 0.1919 0.1825 0.1971 0.1450 0.1833 0.1850 0.1621 0.1730 0.1780 0.1698 0.1772 0.1822 0.1895 0.1938 0.1702 0.1840 0.2043 0.2027 0.2001 0.1814

Target L 0.1934 0.1904 0.1922 0.1687 0.1923 0.182 0.1971 0.1452 0.1836 0.1843 0.1626 0.1726 0.1784 0.17 0.1776 0.1826 0.1896 0.1939 0.1701 0.1834 0.2036 0.2013 0.1998 0.1813

Error 0.0002 0.0002 -0.0003 0.0004 -0.0004 0.0005 0.0000 -0.0002 -0.0003 0.0007 -0.0005 0.0004 -0.0004 -0.0002 -0.0004 -0.0004 -0.0001 -0.0001 0.0001 0.0006 0.0007 0.0014 0.0003 0.0001

In table 3, Target output which is voltage stability index through load flow studies and calculated output which is voltage stability index calculated through ANN, is compared and the error is very less. The effect of variation of load on voltage stability index “L” of critical bus is shown in table 4.

59

O.P. RahiMalik/ et al. Procedia / ProcediaEngineering Engineering00 30(2011) (2012)000–000 53 – 60 Hasmat TABLE IV RESULT OF VOLTAGE STABILITY INDEX PREDICTION USING ANN ON CRITICAL BUS 14 (WITH HEAVILY LOADED)

S.N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Output L 0.9214 0.8079 0.8289 0.9082 0.8711 0.7185 0.7935 0.8536 0.7545 0.8160 0.9279 0.8332 0.8808 0.8161 0.9024 0.8552 0.9387 0.9307 0.8335 0.7902 0.9429 0.9302

BUS 14 Target L 0.9222 0.8074 0.8291 0.9079 0.8712 0.7186 0.794 0.8553 0.7551 0.816 0.9272 0.8336 0.8812 0.8155 0.9032 0.8552 0.939 0.93 0.8333 0.7908 0.942 0.9297

Error -0.0008 0.0005 -0.0002 0.0003 -0.0001 -0.0001 -0.0005 -0.0017 -0.0006 0.0000 0.0007 -0.0004 -0.0004 0.0006 -0.0008 0.0000 -0.0003 0.0007 0.0002 -0.0006 0.0009 0.0005

S.N 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Output L 0.7817 0.8655 0.8551 0.9192 0.8250 0.9633 0.9526 0.9006 0.8809 0.8072 0.7509 0.7912 0.8326 0.9225 0.9021 0.8945 0.8162 0.8764 0.7128 0.8002 0.7358 0.8794 0.8160

Target L 0.7818 0.8651 0.8558 0.9188 0.8252 0.9649 0.9523 0.9011 0.8814 0.8069 0.7508 0.7916 0.8325 0.9223 0.9019 0.8947 0.8157 0.8762 0.7141 0.8001 0.7361 0.8802 0.8159

Error -0.0001 0.0004 -0.0007 0.0004 -0.0002 -0.0016 0.0003 -0.0005 -0.0005 0.0003 0.0001 -0.0004 0.0001 0.0002 0.0002 -0.0002 0.0005 0.0002 -0.0013 0.0001 -0.0003 -0.0008 0.0001

Fig.7. Accuracy Prediction L- Index on Load Bus 14

5. Conclusion The development of ANN for voltage stability monitoring was presented. The back propagation neural network using MATLAB language has been employed for voltage stability monitoring. From the testing process, it shows that the proposed method is capable in predicting voltage stability index for the load buses in power system network. The proposed method was tested using IEEE-14 bus test system and indicated a good agreement between targeted output and ANN output. Voltage stability index (L-index) has been calculated by conventional method (Newton Raphson;s load flow) and using Artificial Neural Networks. The value of L-index is a measure for voltage stability. The bus with the highest L-index value will be the most vulnerable bus in the system and hence this method helps in identifying the weak areas in the system which need critical reactive power support.

60

O.P.Malik/ Rahi etProcedia al. / Procedia Engineering 30 (2012) 53 – 60 Hasmat Engineering 00 (2011) 000–000

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]

B.Gao, G.K.Morrison and P.Kundur, “Voltage stability evaluation using modal analysis,” IEEE Trans.on Power System, Nov. 1992, vol. 7, no. 4, pp.1529-1542,. V.Balamourougan, T.S. Sidhu and M.S. Sachdev, “Technique for online prediction of voltage collapse,” IEE Proc.-Gener. Transm. Distrb. July 2004, vol.151, pp. 454-460. M. H. Haque. "On-line monitoring of maximum permissible loading of a power system within voltage stability limits," IEE Proc. Gener. Transm. Distrib, Jan. 2003, vol. 150, pp. 107-112. Claudio A.Canizares, Antonio C.Z. de Souza and Victor H.Quintana, “Comparison of performance indices for detection of proximity to voltage collapse,” IEEE Trans. on Power System, Aug. 1996,vol. 11, no. 3, pp. 1441-1450. D. Thukaram et. al., “Voltage stability Improvement: Case Studies of Indian Power networks,” Electric Power Systems Research, 1998, pp. 35-44. T. Esaka, Y. Kataoka, T. Ohtaka, S Iwamoto, “Voltage stability preventive control using a new voltage stability index,” International conference on Power System Technology, POWERCON 2004, Nov 2004, vol. 1, pp. 344-349, 21-24. P. A. Lof, T. Smed, G. Anderson and D. J. Hill: "Fast Calculation of a Voltage Stability Index," Paper # 91 WM 203-0-PWRS. IEEE/PES 1991 Winter Meeting, NY, Feb. 1991. Y.Tamura, H. Mori an S. Iwamoto "Relationship between Voltage Instability and Multiple Load Flow Solutions in Electric Power Systems," IEEE Trans. on PAS, vol. 102, no. 5, May 1983. C. L. Demarco and T. J. Overbye "An Energy Based Security Measure for assessing Vulnerability to Voltage Collapse,” IEEE Trans. on Power Systems, vol. 5, no. 2, May 1990. V. Ajjarapu and C. Christy, "The Continuation Power Flow: A Tool for Steady State Voltage Stability Analysis," IEEE PICA Conference Proceedings, pp. 304-311, May 1991. H. Sun, X. Zhou and R. Li, “Accuracy analysis of static stability indices based on power flow model,” IEEE/PES Transmission and Distribution conference and Exhibition: Asia and pacific, pp. 1-7, 15-18 Aug. 2005. J. A. Momoh, L.G. Dias and R. Adapa, "Voltage Stability Assessment Using Neural Networks," Proceedings of the I994 North American Power Symposium, Kansas, October 1994. A. A. El-Keib and X. Ma, "Application of Artificial Neural Networks in Voltage Stability Assessment," Paper # 95WM 156-0 PWRS, IEEE PES Winter Power Meeting, New York, Jan./Feb. 1995. D Chen and R. R. Mohler, “Neural-network-based load modeling and its use in voltage stability analysis,” IEEE Trans. On Control System Technology, vol. 11, issue 4, pp. 460-470, July 2003. M. Pandit, L. Srivastava, J. Sharma “Fast voltage contingency selection using fuzzy parallel Self-Organizing Hierarchical Neural Network”, IEEE Trans. On Power system, vol. 18, issue 2, pp. 657-664, May 2003. S. Charabarti and B. Jeyasurya, “On-Line voltage stability monitoring using artificial neural network,” Large engineering system conference on power engineering, LESCOPE 2004, pp. 71-75, 28-30 July 2004. V. R. Dinavahi and S. C. Srivastava, “ANN based voltage stability margin prediction,” IEEE PES Summer Meeting, 2001, vol. 2, pp. 12751280, 15-19 July 2001. Bansilal, D. Thukaram and K. H. Kashyap, “Artificial neural network application to power system voltage stability improvement,” Conference on convergent technologies for Asia-Pacific region, TENCON 2003, Vol. 1, pp. 53-57, 15-17 Oct. 2003. T.K.Abdul Rehman and G. B. Jasmon, “A new technique for voltage stability analysis in a power system and improved load flow algorithm for distribution network,” International Conference on EMPD , vol .2, pp. 714-719, Nov 1995 F. Milano, “An Open Source Power System Analysis Toolbox,” IEEE Trans. Power Systems, Vol. 20, No. 3, pp. 1199-1206, August 2005. P. Kundur, Power System Stability and Control. New York: McGraw- Hill, 1994. Hasmat Malik, S.M, R.K. Jarial, YRS, “Application and Implementation of Artificial Intelligence in Electrical System” published in international conference on Advances in Computing & Communication (ICACC-2011) ISBN 978-81-920874-0-5, Sponsored by IEEEMTTS, pp. 499-505.