An artificial neural network (ANN) approach is presented for electric load forecasting. The ANN is used to learn the relationship among past, current and future temperatures and loads. In order to provide the forecasted load, the ANN interpolates among the load and temperature data in a training data set. The average absolute errors of the 1 h and 24 h-ahead forecasts in tests on actual utility data are shown to be 1.40% and 2.06%, respectively. This compares with an average error of 4.22% for 24 h ahead forecasts with a currently used forecasting technique applied to the same data. >

/pdf/electric-load-forecasting-using-an-artificial-neural-network-31c2x1izew.pdf

Electric load forecasting using an artificial neural network

Broad Learning System (BLS) that aims to offer an alternative way of learning in deep structure is proposed in this paper. Deep structure and learning suffer from a time-consuming training process because of a large number of connecting parameters in filters and layers. Moreover, it encounters a complete retraining process if the structure is not sufficient to model the system. The BLS is established in the form of a flat network, where the original inputs are transferred and placed as “mapped features” in feature nodes and the structure is expanded in wide sense in the “enhancement nodes.” The incremental learning algorithms are developed for fast remodeling in broad expansion without a retraining process if the network deems to be expanded. Two incremental learning algorithms are given for both the increment of the feature nodes (or filters in deep structure) and the increment of the enhancement nodes. The designed model and algorithms are very versatile for selecting a model rapidly. In addition, another incremental learning is developed for a system that has been modeled encounters a new incoming input. Specifically, the system can be remodeled in an incremental way without the entire retraining from the beginning. Satisfactory result for model reduction using singular value decomposition is conducted to simplify the final structure. Compared with existing deep neural networks, experimental results on the Modified National Institute of Standards and Technology database and NYU NORB object recognition dataset benchmark data demonstrate the effectiveness of the proposed BLS.

Broad Learning System: An Effective and Efficient Incremental Learning System Without the Need for Deep Architecture

http://elmorigin.weebly.com/uploads/5/2/9/5/52952487/pao_1994.pdf

Learning and generalization characteristics of the random vector functional-link net

Extreme learning machines (ELMs) basically give answers to two fundamental learning problems: (1) Can fundamentals of learning (i.e., feature learning, clus- tering, regression and classification) be made without tuning hidden neurons (including biological neurons) even when the output shapes and function modeling of these neurons are unknown? (2) Does there exist unified frame- work for feedforward neural networks and feature space methods? ELMs that have built some tangible links between machine learning techniques and biological learning mechanisms have recently attracted increasing attention of researchers in widespread research areas. This paper provides an insight into ELMs in three aspects, viz: random neurons, random features and kernels. This paper also shows that in theory ELMs (with the same kernels) tend to outperform support vector machine and its variants in both regression and classification applications with much easier implementation.

An Insight into Extreme Learning Machines: Random Neurons, Random Features and Kernels

A theoretical justification for the random vector version of the functional-link (RVFL) net is presented in this paper, based on a general approach to adaptive function approximation The approach consists of formulating a limit-integral representation of the function to be approximated and subsequently evaluating that integral with the Monte-Carlo method Two main results are: (1) the RVFL is a universal approximator for continuous functions on bounded finite dimensional sets, and (2) the RVFL is an efficient universal approximator with the rate of approximation error convergence to zero of order O(C//spl radic/n), where n is number of basis functions and with C independent of n Similar results are also obtained for neural nets with hidden nodes implemented as products of univariate functions or radial basis functions Some possible ways of enhancing the accuracy of multivariate function approximations are discussed >

Stochastic choice of basis functions in adaptive function approximation and the functional-link net

In the post-fault dynamic analysis of interconnected power systems, the critical fault clearing time (CCT) is one of the parameters of paramount importance. Critical clearing time is a complex function of pre-fault system conditions (operating point, topology, system parameters), fault structure (type and location) and post-fault conditions that are in part dependent on the protective relaying policy. To define analytically such a relationship would be highly desirable but diversity of variable involved makes this task extremely complicated. Our efforts focus on examination of that complex mapping and investigation of the influence of the various parameters on CCT. The evaluation of CCT involves elaborate computations that often include time-consuming solutions of nonlinear on-fault system equations. Existing conventional pattern recognition techniques are incapable of synthesizing such complex and transparent mappings. Thus, when a human operator tells the machine learning unit (that is the pattern recognizer) that system state belongs to a certain class, say "emergency", the pattern recognizer merely records that classification mindlessly and is not able to look at the pattern with insight and discover what underlies the "emergency" nature of pattern. It is, therefore, highly desirable to have a mechanism which when presented with a sequence of class labeled patterns not only learns an internal structure which allows it to generalize and to classify other and to classify other patterns correctly, but also is able to shed some light on what combination of features give rise to the particular class membership.

Artificial neural-net based dynamic security assessment for electric power systems

In this article, we are concerned with neural-nets which can learn to control systems in accordance with a guiding intent, and can also learn how to formulate that control strategy or intent. The overall task of systems control is viewed as being carried out by four components, these being the predictive monitoring net, the control action generator net, the objective function net and the optimization net. This approach and perspective are described and illustrated in this article. In our examples, we show that systems identification can indeed be achieved in the presence of noise and that optimal control can be formulated in a learning mode, by neural nets.

Neural-net computing and the intelligent control of systems

It is highly desirable that the security and stability of electric power systems after exposure to large disturbances be assessable. In this connection, the critical clearing time (CCT) is an attribute which provides significant information about the quality of the post-fault system behavior. It may be regarded as a complex mapping of the prefault, fault-on, and post-fault system conditions in the time domain. Y.-H. Pao and D.J. Solajic (1989) showed that a feedforward neural network can be used to learn this mapping and successfully perform under variable system operating conditions and topologies. In that work the system was described in terms of some conventionally used parameters. In contrast to using those pragmatic features selected on the basis of the engineering understanding of the problem, the possibility of using unsupervised and supervised learning paradigms to discover what combination of raw measurements are significant in determining CCT is considered. Correlation analysis and Euclidean metric are used to specify interfeature dependencies. An example of a 4-machine power system is used to illustrate the suggested approach. >

Combined use of unsupervised and supervised learning for dynamic security assessment

Situation awareness is a key factor in preserving power system security, as it enables effective and timely decision-making and reactions by the operators to an incident. Insufficient situation awareness results in a delayed, incorrect or deficient response, endangering power system stability. This factor was actually identified as one of the main causes of several electrical disturbances in the last decade. This paper identifies numerous factors that govern the formation of situation awareness in a power system control center. A multi-state model based on Markov modeling is proposed for assessing the impact of insufficient situation awareness on the probability of power system blackouts. The proposed model considers the level of situation awareness and the state of the information infrastructure. The workings of this model are illustrated using the IEEE 24-bus Reliability Test System.

Dejan J. Sobajic

Papers

Learning and generalization characteristics of the random vector functional-link net

Artificial neural-net based dynamic security assessment for electric power systems

Neural-net computing and the intelligent control of systems

Combined use of unsupervised and supervised learning for dynamic security assessment

Assessing the Impact of Insufficient Situation Awareness on Power System Operation