Please use this identifier to cite or link to this item:
|Title:||Intelligent techniques for home electric load forecasting and balancing|
Powerline ampacity -- Forecasting
Hong Kong Polytechnic University -- Dissertations
|Publisher:||The Hong Kong Polytechnic University|
|Abstract:||The work presented in this thesis is related to an intelligent power system in a modern home. Results in the following areas will be reported: a power line based data network infrastructure with hardware modules for an intelligent home, a home electric load forecasting system, and a home electric load balancing system.|
A power line data network based on the spread-spectrum technology is proposed and implemented. It facilitates digital data communications at a rate of 10 Kbps in the noisy and signal-distorting environment of the AC power line. This power line data network serves as a backbone of communication in an intelligent home through which electrical appliances can be controlled via line/mobile phones, personal digital assistants (PDAs), keypads or personal computers anytime and anywhere, inside or outside the home. It can provide a basis for the plug-and-play features of electrical appliances without the need of installing additional cables.
Short-term electric load forecasting (STELF) is essential to improve the reliability of the AC power line data network and provide optimal load scheduling in an intelligent home. Three computational intelligence techniques are developed to realize STELF. The first approach is by using a fuzzy genetic algorithm (GA)-based neural network (NN). It can forecast the electric load accurately with respect to different day types and weather information. The proposed fuzzy GA is modified from a published GA with arithmetic crossover and non-uniform mutation. Fuzzy logic is used to incorporate expert knowledge and experience (in terms of linguistic rules) into the crossover and mutation operations. With fuzzy logic, the number of iteration and the rate of change of fitness value can be fine-tuned. By using some benchmark test functions, it can be shown that the fuzzy GA performs better than the traditional GA. In many applications of NNs, the networks are fully connected. However, the performance of a fully connected NN may not be better than that of a partly connected NN with the same number of hidden nodes. This is because some links in an NN could be redundant. A three-layer NN with a switch introduced to each link is proposed to facilitate the tuning of the network structure. By turning on or off these link-switches during the training process, the optimal neural network structure can be obtained. This implies that the cost of implementing the proposed NN in terms of hardware, processing and simulation time can be reduced.
The second approach for realizing STELF involves a fuzzy GA-based neural fuzzy network (NFN). The optimal NFN structure can be found by the fuzzy GA when switches in the links of the network are introduced. The membership functions and the number of rules of the NFN can be generated automatically. Results for implementing STELF in an intelligent home by using the proposed NEN will be given.
The third approach for realizing STELF is based on a modified fuzzy GA-based neural network, which involves a new neuron model. Under this model, the neuron has two activation transfer functions and exhibits a node-to-node relationship within the hidden layer. The proposed neural network can offer a better performance and a smaller number of hidden nodes than the traditional feed-forward neural network. This network is trained by the fuzzy GA.
The electric load forecasting system is further applied in a home electric load balancing system. After an electric load forecasting system can successfully forecast the power consumption profile of a home, the load balancing system can adjust the amount of energy stored in batteries accordingly and prevent it from reaching some practical limits. A steady consumption pattern can then be obtained which will benefit both the power users and the utility company. An example will be given to illustrate the accuracy of the forecaster, and its ability of achieving load balancing.
|Description:||xx, 118 leaves : ill. (some col.) ; 30 cm.|
PolyU Library Call No.: [THS] LG51 .H577M EIE 2002 Ling
|Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b16677547_link.htm||For PolyU Users||167 B||HTML||View/Open|
|b16677547_ir.pdf||For All Users (Non-printable)||5.53 MB||Adobe PDF||View/Open|
Checked on Jan 31, 2016
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.