Page - IX - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Preface to ”Short-TermLoad Forecasting byArtificial IntelligentTechnologies” In the last fewdecades, short-term load forecasting (STLF)hasbeenoneof themost important research issues forachievinghigherefficiencyandreliability inpowersystemoperation, to facilitate the minimization of its operation cost by providing accurate input to day-ahead scheduling, contingencyanalysis, loadflowanalysis,planning,andmaintenanceofpowersystem.Thereare lots of forecastingmodels proposed for STLF, including traditional statisticalmodels (such asARIMA, SARIMA,ARMAX,multi-variate regression, Kalmanfilter, exponential smoothing, and so on) and artificial-intelligence-based models (such as artificial neural networks (ANNs), knowledge-based expertsystems,fuzzytheoryandfuzzyinferencesystems,evolutionarycomputationmodels,support vectorregression,andsoon).Recently,duetothegreatdevelopmentofevolutionaryalgorithms(EA), meta-heuristicalgorithms(MTA),andnovelcomputingconcepts(e.g.,quantumcomputingconcepts, chaoticmapping functions, andcloudmappingprocess, andsoon),manyadvancedhybridizations with those artificial-intelligence-basedmodels are also proposed to achieve satisfactory forecasting accuracy levels. In addition, combining some superiormechanismswith an existingmodel could empower that model to solve problems it could not deal with before; for example, the seasonal mechanismfromARIMAmodel isagoodcomponent tobecombinedwithanyforecastingmodels to help themtodealwithseasonalproblems. This book contains articles from the Special Issue titled “Short-Term Load Forecasting by Artificial IntelligentTechnologies”,whichaims toattract researcherswithan interest in the research areas described above. As Fan et al. [1] highlighted, the research trends of forecastingmodels in theenergysector inrecentdecadescouldbedivided into threekindsofhybridorcombinedmodels: (1) hybridizingor combining theartificial intelligent approacheswith eachother; (2) hybridizingor combiningwith traditional statistical approaches; and (3) hybridizing or combiningwith the novel evolutionary (ormeta-heuristic) algorithms. Thus, theSpecial Issue, inmethodological applications, was also based on these three categories, i.e., hybridizing or combining any advanced/novel techniques in energy forecasting. The hybrid forecastingmodels should have superior capabilities over the traditional forecastingapproaches, andbeable toovercomesomeinherentdrawbacks, and, eventually, toachievesignificant improvements in forecastingaccuracy. The22articles inthiscompendiumalldisplayabroadrangeofcutting-edgetopicsof thehybrid advanced technologies in STLF fields. The preface authors believe that the applications of hybrid technologieswillplayamoreimportantroleinSTLFaccuracyimprovements,suchashybriddifferent evolutionary algorithms/models to overcome some critical shortcoming of a single evolutionary algorithm/modelor todirectly improvetheshortcomingsbytheoretical innovativearrangements. Based on these collected articles, an interesting (future research area) issue is how to guide researcherstoemployproperhybridtechnologyfordifferentdatasets. Thisisbecauseforanyanalysis models (includingclassificationmodels, forecastingmodels,andsoon), themost importantproblem is how to catch the data pattern, and to apply the learnedpatterns or rules to achieve satisfactory performance, i.e., thekeysuccess factor ishowtosuccessfully look fordatapatterns.However, each model excels in catchingdifferent specific data patterns. For example, exponential smoothing and ARIMAmodels focusonstrict increasing (ordecreasing) timeseriesdata, i.e., linearpattern, though they have a seasonalmodificationmechanism to analyze seasonal (cyclic) change; due to artificial learningfunctiontoadjustthesuitabletrainingrules, theANNmodelexcelsonlyif thehistoricaldata ix