1. Gabatarwa
Makamashin hasken rana yana wakiltar ɗaya daga cikin mafi arha da tsaftataccen tushen makamashi mai dorewa a duniya. Duk da haka, samar da shi yana da wuyar hasashe saboda dogaro ga yanayi, yanayi, da yanayin muhalli. Wannan takarda ta gabatar da mai hasashen makamashin hasken rana na duniya ta amfani da Naive Bayes classifier don hasashen jimillar makamashin da ake samu daga na'urorin hasken rana na yau da kullun.
Binciken ya magance buƙatar mahimmanci na daidaitaccen hasashen makamashin hasken rana don inganta tsarin makamashi da haɓaka inganci. Tare da tsammanin samar da wutar lantarki ya kai triliyan kWh 36.5 nan da shekara ta 2040, kuma samar da makamashin hasken rana yana girma da kashi 8.3% a kowace shekara, hanyoyin hasashe masu dogaro sun zama mafi mahimmanci ga tsarawa da sarrafa makamashi.
2. Binciken Adabi
Binciken da ya gabata ya binciko hanyoyi daban-daban don hasashen makamashin hasken rana. Creayla et al. da Ibrahim et al. sun yi amfani da dazuzzukan bazuwar, hanyoyin sadarwar jijiyoyi na wucin gadi, da hanyoyin da suka dogara da algorithm na firefly don hasashen hasken rana na duniya, suna samun kurakurai na son kai daga kashi 2.86% zuwa 6.99%. Wang et al. sun yi amfani da dabarun koma baya da yawa tare da ƙimar nasara daban-daban.
Hanyoyin gargajiya sau da yawa sun dogara da ilimin ƙwararrun fanni, wanda ya zama mara amfani ga daidaita tsarin ci gaba. Hanyoyin koyon inji suna ba da koyon alaƙa ta atomatik tsakanin yanayin muhalli da samar da makamashi daga bayanan tarihi.
3. Hanyar Aiki
3.1 Tattara Bayanai
Binciken ya yi amfani da bayanan tarihi na shekara guda waɗanda suka haɗa da:
- Matsakaicin zafin jiki na yau da kullun
- Tsawon lokacin hasken rana na yau da kullun
- Jimillar hasken rana na duniya na yau da kullun
- Jimillar samar da makamashin hasken rana na yau da kullun
Waɗannan sigogi suna aiki azaman siffofi masu ƙima na rukuni don ƙirar hasashe.
3.2 Zaɓin Siffofi
Zaɓin siffa yana mai da hankali kan sigogi masu mafi girman alaƙa da samar da makamashi. Hanyar rukuni tana ba da damar sauƙaƙe rarrabuwa yayin riƙe daidaiton hasashe.
3.3 Aiwatar da Naive Bayes
Naive Bayes classifier yana amfani da ka'idar Bayes tare da "naive" zato na 'yancin kai na yanayin tsakanin siffofi. Lissafin yuwuwar yana biye da:
$P(y|X) = \frac{P(X|y)P(y)}{P(X)}$
Inda $y$ ke wakiltar ajin samar da makamashi, kuma $X$ ke wakiltar vector ɗin siffa. Classifier ɗin ya zaɓi ajin da ke da mafi girman yuwuwar baya don hasashe.
4. Sakamakon Gwaji
4.1 Ma'aunin Aiki
Hanyar da aka aiwatar tana nuna ingantacciyar haɓaka a cikin daidaito da hankali idan aka kwatanta da hanyoyin gargajiya. Manyan alamomin aiki sun haɗa da:
Haɓaka Daidaito
Haɓaka mai mahimmanci akan hanyoyin tushe
Nazarin Hankali
Ingantaccen gano tsarin samar da makamashi
Alaƙar Sigogi
Bayyanannen gano sigogin hasken rana masu tasiri
4.2 Nazarin Kwatance
Hanyar Naive Bayes tana nuna aiki mai gasa da ƙarin ƙirar rikitarwa kamar dazuzzukan bazuwar da hanyoyin sadarwar jijiyoyi, musamman a cikin ingantaccen lissafi da fahimta.
Bayanin Chati: Chati na kwatancen aiki yana nuna kashi-kashi na daidaito a cikin hanyoyin hasashe daban-daban. Naive Bayes classifier yana nuna daidaitaccen aiki a duk ma'auni tare da ƙananan buƙatun lissafi.
5. Nazarin Fasaha
Mahimmin Fahimta
Wannan takarda ta gabatar da hanyar ra'ayin mazan jiya ta asali ga matsala mai rikitarwa. Duk da yake marubutan sun gano daidai buƙatar mahimmanci na hasashen makamashin hasken rana a cikin canjinmu zuwa tushen sabuntawa, zaɓin su na Naive Bayes classifier yana jin kamar amfani da kalkuleta aljihu lokacin da masana'antu suka ƙaura zuwa manyan kwamfutoci. Zaton 'yancin kai na siffa a cikin tsarin makamashin hasken rana yana da matsala musamman—zafin jiki, tsawon lokacin hasken rana, da radiation suna da alaƙa ta asali ta hanyoyin da suka saba wa ainihin jigon Naive Bayes.
Kwararar Hankali
Binciken yana bin madaidaiciyar bututun ruwa: tattara bayanai → zaɓin siffa → aiwatar da ƙira → kimantawa. Duk da haka, wannan hanyar madaidaiciya ta rasa damar don ƙarin fasahohi masu ƙwarewa kamar injiniyan siffa ko hanyoyin haɗakarwa. Kwatancen da adabin da ke akwai yana da zurfi aƙalla—ambaton aikin Creayla da Wang ba tare da shiga cikin ƙayyadaddun hanyoyinsu ba ko kuma bayyana dalilin da yasa ƙirar mafi sauƙi za ta iya fi na rikitarwa a cikin wannan takamaiman mahallin.
Ƙarfi & Kurakurai
Ƙarfi: Mayar da hankali na takarda akan mafita masu amfani yana da yabo. Ƙirar Naive Bayes suna da ingantaccen lissafi kuma suna aiki da kyau tare da ƙayyadaddun bayanai—mahimman abubuwan la'akari don tsarin makamashi na ainihin duniya. Hanyar siffa ta rukuni tana sauƙaƙe aiwatarwa da fassara.
Matsalolin Mahimmanci: Sashen hanyar aiki ya rasa zurfi. Babu tattaunawa game da sarrafa bayanai, sarrafa ƙimar da ba a taɓa, ko magance yanayin yanayi da ke cikin bayanan hasken rana. Da'awar "haɓaka mai ban sha'awa" ba ta da goyan baya na ƙididdiga—waɗanne ma'auni? Idan aka kwatanta da wane tushe? Wannan rashin fayyace yana lalata aminci. Mafi mahimmanci, kamar yadda aka nuna a cikin cikakken bita na Antonanzas et al. a cikin Bita na Makamashi Mai Sabuntawa da Dorewa (2016), hasashen hasken rana na zamani yana ƙara yin amfani da zurfin koyo da ƙirar haɗakarwa waɗanda ke ɗaukar dogaro na lokaci fiye da masu rarraba tsaye.
Fahimta Mai Aiki
Ga masu aiki: Wannan hanyar na iya zama ƙirar tushe mai sauri amma bai kamata ya zama mafita ta ƙarshe ba. Yi la'akari da haɓaka gradient (XGBoost/LightGBM) ko hanyoyin sadarwar LSTM don bayanan jeri. Ga masu bincike: Fannin yana buƙatar ƙarin aiki akan koyon canja wuri tsakanin wurare—mai hasashe na gaske na "duniya". Gasar hasashen hasken rana akan Kaggle da dandamali kamar National Renewable Energy Laboratory's (NREL) Solar Forecast Arbiter sun nuna cewa mafita masu nasara sun haɗa ƙira da yawa da ƙwararrun injiniyan siffa.
Ainihin damar ƙirƙira ba ta cikin zaɓin classifier ba amma a cikin haɗakar bayanai. Haɗa hotunan tauraron dan adam (kamar bayanan NASA's POWER), karatun tashar yanayi, da na'urar lantarki ta hanyar gine-gine iri ɗaya da waɗanda ke cikin hangen nesa na kwamfuta (misali, hanyoyin multimodal a cikin CLIP ko DALL-E) na iya haifar da ci gaba. Marubutan sun taɓa wannan tare da ambaton "aikin aiki na kasuwanci" amma ba su bi shi ba.
Misalin Tsarin Nazari
Nazarin Shari'a: Kimanta Wurin Gonar Hasken Rana
Yin amfani da tsarin da aka gabatar don kimanta wuraren da za a iya samun gonar hasken rana:
- Lokacin Tattara Bayanai: Tattara bayanan tarihi na shekaru 5 don wuraren da za a iya zaɓa waɗanda suka haɗa da zafin jiki, radiation, da tsarin rufin girgije
- Injiniyan Siffa: Ƙirƙiri siffofi da aka samo kamar matsakaicin yanayi, fihirisar bambance-bambance, da matrices na alaƙa tsakanin sigogi
- Aiwatar da Ƙira: Aiwatar da Naive Bayes classifier don rarraba wurare zuwa babba/matsakaici/ƙananan yuwuwar amfanin ƙasa
- Tabbatarwa: Kwatanta hasashe tare da ainihin amfanin ƙasa daga abubuwan da ake da su a yankunan yanayi iri ɗaya
- Taimakon Yanki: Samar da shawarwarin saka hannun jari bisa ga hasashen fitar da makamashi da ƙirar kuɗi
Wannan tsarin yana nuna yadda koyon inji zai iya ƙara hanyoyin kimanta wuri na gargajiya, ko da yake ya kamata a ƙara shi da ƙirar jiki da shawarwarin ƙwararru.
6. Aikace-aikacen Gaba
Mai hasashen makamashin hasken rana na duniya yana da aikace-aikace masu ban sha'awa da yawa:
- Haɗakar Smart Grid: Hasashen makamashi na ainihin lokaci don daidaita grid da sarrafa amsa buƙatu
- Inganta Zaɓin Wuri: Kimanta wuri mai tushen bayanai na wuraren da za a iya samun sabbin na'urorin hasken rana
- Tsara Kula: Kulawa mai hasashe bisa ga tsammanin vs. ainihin tsarin samar da makamashi
- Cinikin Makamashi: Ingantaccen hasashe don kasuwannin makamashin hasken rana da dandamali na ciniki
- Ƙirar Tsarin Haɗakarwa: Inganta tsarin haɗakar hasken rana-iska-ma'ajiyar ta hanyar daidaitattun hasashe na samarwa
Hanyoyin bincike na gaba yakamata su bincika:
- Haɗakar hotunan tauraron dan adam da hanyoyin sadarwar IoT don ingantaccen ingancin bayanai
- Haɓaka ƙirar koyon canja wuri don daidaita yanki
- Tsarin hasashe na ainihin lokaci tare da iyawar lissafin gefe
- Haɗuwa tare da ingantattun algorithm na ajiyar makamashi
- Aikace-aikace a cikin microgrid da sarrafa albarkatun makamashi masu rarrabawa
7. Nassoshi
- International Energy Agency. (2021). World Energy Outlook 2021. Paris: IEA Publications.
- Antonanzas, J., Osorio, N., Escobar, R., Urraca, R., Martinez-de-Pison, F. J., & Antonanzas-Torres, F. (2016). Review of photovoltaic power forecasting. Solar Energy, 136, 78-111.
- Wang, H., Lei, Z., Zhang, X., Zhou, B., & Peng, J. (2019). A review of deep learning for renewable energy forecasting. Energy Conversion and Management, 198, 111799.
- National Renewable Energy Laboratory. (2020). Solar Forecasting Benchmarking. Golden, CO: NREL Technical Report.
- Creayla, C. M., & Park, S. Y. (2018). Solar radiation prediction using random forest and firefly algorithm. Renewable Energy, 125, 13-22.
- Ibrahim, I. A., Khatib, T., & Mohamed, A. (2017). A novel hybrid model for hourly global solar radiation prediction using random forests technique and firefly algorithm. Energy Conversion and Management, 138, 413-425.
- Wang, Z., & Srinivasan, R. S. (2017). A review of artificial intelligence based building energy use prediction: Contrasting the capabilities of single and ensemble prediction models. Renewable and Sustainable Energy Reviews, 75, 796-808.
- Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. (Don ra'ayoyin koyon inji na tushe)
- NASA Prediction of Worldwide Energy Resources (POWER). (2022). Data Access Guide. Greenbelt, MD: NASA Goddard Space Flight Center.
- European Commission. (2020). Photovoltaic Geographical Information System (PVGIS). JRC Technical Reports.