Page - 115 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3442 energyforecast isneededfor thepolicy framerswhile takingdecisionfor the future. Thus, thepolicy framersneedto take thisboost inenergyusage inmind. It isalsorecommendedthat theotherenergy forecasting techniquesshall alsobeusedto testify theoutcomeandalsoenergypredictionshallbe recurrentlydoneas thecircumstancesaredynamic. Someof thestateofartwork in thesameresearcharea isdiscussedhere • According to National Energy Map for India: Technology Vision 2030, India’s electricity consumptionwillbecomefourfoldfromabout1.1trillionunits to4trillionunitsby2030.Brookings Institution IndiaCentre, in2013, estimated that the shootup inglobal energyconsumption is attributedmainlydueto IndiaandChina[41]. • Asia-Pacific territory lonelycontributes to79%of thehike in international liquidsuse,whichrises from1281.7Milliontonsofoil equivalent in2010 to1859.3Milliontonsofoil equivalent in2030. Thepercapitaenergyutilization in2030 for India is expected to rise from19.58millionBtu to 29.84millionBtu[42]. • TheformerCoalandpowerministerof India,Mr. PiyushGoyalstatedinMay,2016thatapossible 10%jumpisexpected in theannualelectricitygrowthfor thenext15or16years [43]. • SugandhaChauhan (2017) studied electricitydemandand reported that itwill increase from 1115BU in2015–2016 to 1692BU in2022, 2509BU in2027and3175BU in2030 reflecting the higherendof thedemandforelectricity [44]. • Iniyanetal. 2000. proposedamodel thatallocates therenewableenergydistributionpattern for theyear2020–2021for India [45]. 5.Conclusions Thisworkpresents theanalysisofavailabledataandthepredictedoneregardingwhatwillbe theTotal ElectricityConsumption (TEC) of India for the year 2030usingvarious blackboxbased approaches. The forecastingof totalelectricityconsumptionfor theyear2030–2031 for India is found tobe 1,834,349MWwhiledoing so the forecast for 2017was comparedwith the actual datagiven byEnergystatistics,GOIwhichsitsclose to the forecasteddata.Andtheexpertmodel is forecasted to be the best fit that suits theprediction since theR2 value is 0.997which is comparatively high. Obtainedresults showthat thismodel isofahighprecision. Theadvantagesof themodelare that it canbecomputedeasilywithsimplestatistical softwareandavailable inalmosteveryrecent statistical package.Accessibility isnotanobstacleandtheanalysis shallbeperformedwithadeviceofminimal configuration. The time taken for running themodel isveryminimalwhich isamere00:00:00.06s (processor time). The disadvantage of themodel is that it selects the best suitablemodel on its own.The limitationof thework is thatwecouldnotapply thepopularmethodologiesofblackbox approaches suchasDecisionTrees,ANN,SVM.Thereare several othervariables suchas imports, exports,villageselectrified,pumpsetsenergizedandsoforth,whichhasa futuristic scopefor further extensivestudies. Energyforecastingcanbetakenuptothenext level, forexample, forAsia-Pacific territory. As theneed for energyconsumption is constantly increasing inmanifolds, it is assumed that thefindingsandforecastsgiveninthisarticlewouldbeofuse to thepolicymakersandenergy strategists toevolve futurescenarios for the Indianelectricityconsumptionwhichshouldfocusgreatly in further increasingtheoverall shareof renewableenergyresourcescomparedto theconventional sources of the installed capacity aswell as in the consumptionpattern. The future researchmay bedone consideringmore inputvariables suchas thequantumofCO2 emission,GNPper capita, consumerprice index,powerconsumptionpercapita,wholesaleprice index, imports,grossdomestic savings, exports and so forth. Othermethodologies such as computational intelligence forecasts, beyondpoint forecasts, combinedforecastsmayalsobeapplied inshort termloadforecastingof the electricalenergydemand. 115