Novel Algorithms And Artificial Neural Network Techniques For Optimal Operation And Control Of Electrical Distribution Networks

Mohammad Abul Kashem, (2000) Novel Algorithms And Artificial Neural Network Techniques For Optimal Operation And Control Of Electrical Distribution Networks. PhD thesis, Multimedia University.

Full text not available from this repository.
Official URL:


In electrical distribution network, problems related to loss minimisation, voltage stability and load balancing generally arise. This thesis is to explore novel algorithms and Artificial Neural Network (ANN) based techniques for solving the above-specified problems efficiently for online and real-time implementation. In this thesis, a systematic feeder reconfiguration technique is presented that develops an optimal switching scheme to achieve a maximum reduction of losses in a distribution network. A minimal tree-search is proposed to find the possible switching-options for loss reduction, and a loss change formula is derived and used to determine the switching option that gives the maximum loss reduction in the system. A computer aided graphical visualisation technique for loss minimisation is also presented and a graphical representation of power losses is given. In this technique, a circle for each loop in the network is drawn from the relationship between the change of loss due to the branch-excahange and the power-flows in the branches. All the circles are superimposed and visualised and the largest one is identified as the maximum loss reduction loop. The circle for every possible branch-exchange in the maximum loss-reduction loop is observed and the corresponding branch-exchange of the smallest circle is found to give the maximum loss-reduction in the network. Two different solution approaches: Index Measurement Technique (IMT) and Distance Measurement Technique (DMT) are presented for load balancing. Each of the proposed algorithms is developed based on a two-stage solution methodology. The first stage finds a loop, which gives the maximum improvement in load balancing in the network. In the second stage, a switching -option is determined in the loop to obtain maximum improvement in load balancing. The IMT uses various indices all over the algorithm to achieve the optimal or near optimal confiduration for load balancing, whereas the DMT employs a graphical method in which differentcircles are drawn and the distances of various point from the centre of the loop circle are computed to obtain the same. The relationship amongst voltage stability, load balancing and loss minimisation is explored in which it is found that voltage stability and load balancing are maximised when power losses are minimised in the networks. A Generalised Simultaneous Switching Algorithm ( GSSA) is proposed to reconfigure distribution networks for loss minimisation. A systematic logical approach is used to find the best combination of switches for minimum loss. In the first stage of the algorithm, a limited number of switching combination is formed by trinary logic principle and the minimum loss configuration is found. In the next stage, the search is extended by considering the branches next to the open-branches one after another in the configuration found above and the best switching configuration that gives the minimum loss is identified. The results obtained by GSSA are compared with those of established methods reported earlier and are found to give better solutions. A scheme is proposed to implement the GSSA for real-time on-line control of distribution networks. The GSSA is extended to enhance voltage stability and improve load balancing in distribution network by using the respective objective functions. In this thesis, four-generalised Artificial Neural Network ( ANN ) models are proposed for on-line network reconfiguration under varying load conditions. Actual loads(P,Q) are used as the input vectors for the ANN model-II. The training data are generated from the Daily Load Curves(DLCs0 with the same load trends for P and Q in the case of ANN model-III, and different load trends in the case of ANN model-IV. The developed ANN models can be used to predict the switching status of the dynamic switches to optimise the networks for loss minimisation, voltage stability and /or load balancing. A 16-bus test system is considered to demonstrate the performance of all the developed ANN models. All the four proposed ANNs are trained using Conjugate Gradient Descent Back-propagation Algorithm, and tested by applying arbitrary input data generated from typicalDLCs. The test results of each of the four ANN models are found to be the same as that obtained by simulation using the GSSA. The proposed ANN techniques are compared with two other methods in literature and found to give better performance.

Item Type: Thesis (PhD)
Subjects: Q Science > QA Mathematics > QA75.5-76.95 Electronic computers. Computer science > QA76.75-76.765 Computer software
Divisions: Faculty of Computing and Informatics (FCI)
Depositing User: Mr Shaharom Nizam Mohamed
Date Deposited: 02 Dec 2009 07:40
Last Modified: 03 Dec 2009 04:10

Actions (login required)

View Item View Item