I-athikili ichaza indlela yokuqalisa I-WMS-uhlelo, besibhekene nesidingo sokuxazulula inkinga yokuhlanganisa engajwayelekile nokuthi yiziphi izindlela esizisebenzisayo ukuyixazulula. Sizokutshela ukuthi sisebenzise kanjani indlela ehlelekile, yesayensi yokuxazulula inkinga, ibuphi ubunzima esihlangabezane nabo nokuthi yiziphi izifundo esizifundile.
Lokhu kushicilelwa kuqala uchungechunge lwama-athikili lapho sabelana khona ngolwazi lwethu oluyimpumelelo ekusebenziseni ama-algorithms okuthuthukisa ezinqubweni ze-warehouse. Inhloso yochungechunge lwama-athikili ukujwayela izethameli ngezinhlobo zezinkinga zokusebenza kwe-warehouse eziphakama cishe kunoma iyiphi indawo yokugcina impahla ephakathi nendawo enkulu, kanye nokusho ngolwazi lwethu ekuxazululeni izinkinga ezinjalo kanye nezingibe okuhlangatshezwane nazo endleleni. . Izihloko zizoba wusizo kulabo abasebenza embonini ye-warehouse logistics, sebenzisa I-WMS-amasistimu, kanye nabahleli abanentshisekelo ekusetshenzisweni kwezibalo ebhizinisini kanye nokwenza kahle kwezinqubo ebhizinisini.
I-Bottleneck ezinqubweni
Ngo-2018, saphothula iphrojekthi okufanele siyisebenzise I-WMS-Izinhlelo endaweni yokugcina impahla yenkampani "I-Trading House "LD" e-Chelyabinsk. Sisebenzise umkhiqizo othi β1C-Logistics: Warehouse Management 3β ezindaweni zokusebenza ezingama-20: opharetha I-WMS, abanikazi bezitolo, abashayeli be-forklift. I-warehouse evamile imayelana nezinkulungwane ezi-4 m2, inani lamaseli yi-5000 kanye nenombolo ye-SKUs yi-4500. Indawo yokugcina impahla igcina ama-valve ebhola okukhiqizwa kwethu kobukhulu obuhlukene kusuka ku-1 kg kuya ku-400 kg. Impahla egciniwe igcinwa ngamaqoqo, njengoba kunesidingo sokukhetha izimpahla ngokwe-FIFO.
Lapho siklama izikimu zenqubo ye-warehouse automation schemes, besibhekene nenkinga ekhona yokugcinwa kwempahla okungafanelekile. Imininingwane yokugcina kanye nama-cranes okufaka izinto ngendlela yokuthi iseli elilodwa lokugcina iyunithi lingaqukatha kuphela izinto ezisuka kuqeqebana elilodwa. Imikhiqizo ifika endaweni yokugcina izimpahla nsuku zonke futhi ukufika ngakunye kuyiqoqo elihlukile. Sekukonke, ngenxa yokusebenza kwenyanga eyi-1, amaqoqo ahlukene angama-30 adaliwe, naphezu kweqiniso lokuthi ngalinye kufanele ligcinwe esitokisini esihlukile. Imikhiqizo ivame ukukhethwa hhayi kuma-pallet aphelele, kodwa ibe yizicucu, futhi ngenxa yalokho, endaweni yokukhetha izingcezu kumaseli amaningi isithombe esilandelayo siyabonwa: esitokisini esinomthamo ongaphezu kwe-1 m3 kunezicucu eziningana zama-cranes. thatha ngaphansi kuka-5-10% wevolumu yeseli.
Umfanekiso 1. Isithombe sezingcezu ezimbalwa zezimpahla esitokisini
Kuyacaca ukuthi umthamo wokugcina awusetshenziswa kahle. Ukuze ucabange ubukhulu benhlekelele, nginganikeza izibalo: ngokwesilinganiso, kunamaseli avela ku-1 kuya ku-3 amaseli anjalo anevolumu engaphezu kwe-100 m300 enezilinganiso "zeminuscule" ngezikhathi ezahlukene zokusebenza kwendawo yokugcina impahla. Njengoba indawo yokugcina impahla incane kakhulu, phakathi nezinkathi ezimatasa endaweni yokugcina izinto, lesi sici siba βibhodlelaβ futhi sibambezela kakhulu izinqubo ze-warehouse.
Umbono wesixazululo senkinga
Kwavela umqondo: amaqoqo okusalile anezinsuku eziseduze kufanele ancishiswe abe iqoqo elilodwa, futhi lokho okusalayo neqoqo elihlangene kufanele kubekwe kuhlanganiswe ndawonye esitokisini esisodwa, noma eziningana, uma singekho isikhala esanele kweyodwa ukuze kuhlaliswe inani eliphelele lezinsalela.
Fig.2. Uhlelo lokucindezela izinsalela kumaseli
Lokhu kukuvumela ukuthi unciphise kakhulu indawo yokugcina impahla ezosetshenziselwa izimpahla ezintsha ezibekwayo. Esimeni lapho umthamo we-warehouse ugcwele ngokweqile, isilinganiso esinjalo sidingeka kakhulu, ngaphandle kwalokho kungase kungabi khona indawo yamahhala eyanele yokufaka izimpahla ezintsha, okuzoholela ekumisweni kwezinqubo zokubekwa kwempahla kanye nokugcwalisa kabusha. Ngaphambili ngaphambi kokuqaliswa I-WMS-amasistimu awenza lo msebenzi mathupha, owawungasebenzi, njengoba inqubo yokufuna izinsalela ezifanele kumaseli yayinde impela. Manje, ngokwethula uhlelo lwe-WMS, sinqume ukwenza inqubo ngokuzenzakalelayo, siyisheshise futhi siyenze ibe ehlakaniphile.
Inqubo yokuxazulula inkinga enjalo ihlukaniswe izigaba ezi-2:
- esigabeni sokuqala sithola amaqembu amaqoqo asondelene ngosuku lokucindezelwa;
- esigabeni sesibili, eqenjini ngalinye lamaqoqo sibala ukubekwa okuhlangene kakhulu kwezimpahla ezisele kumaseli.
Esihlokweni samanje sizogxila esigabeni sokuqala se-algorithm, futhi sishiye ukumbozwa kwesigaba sesibili esihlokweni esilandelayo.
Sesha imodeli yezibalo yenkinga
Ngaphambi kokuthi sihlale phansi ukuze sibhale ikhodi futhi sisungule kabusha isondo lethu, sinqume ukubhekana nale nkinga ngokwesayensi, okungukuthi: siyenze ngokwezibalo, siyinciphise ibe yinkinga eyaziwayo yokwenza kahle ebanzi futhi sisebenzise ama-algorithm akhona ukuyixazulula, noma sithathe lawa ma-algorithms akhona. njengesisekelo futhi uziguqule zibe nemininingwane yenkinga esebenzayo exazululwayo.
Njengoba ilandela ngokucacile kusukela ekwakhiweni kwebhizinisi kwenkinga esibhekene nayo namasethi, sizokwakha inkinga enjalo ngokwetiyori emisiwe.
Vumela β isethi yawo wonke amaqoqo ensalela yomkhiqizo othile endaweni yokugcina izimpahla. Vumela - kunikezwe ukungaguquguquki kwezinsuku. Vumela - iqoqo elincane lamaqoqo, lapho umehluko wezinsuku zawo wonke amapheya amaqoqo kusethi encane ungeqi isilinganiso esingaguquki . Sidinga ukuthola inani elincane lamasethi angaphansi ahlukene , ukuze wonke amasethi angaphansi kuthathwe ndawonye kunganikeza abaningi .
Ngamanye amazwi, sidinga ukuthola amaqembu noma amaqoqo afanayo, lapho umbandela wokufana unqunywa ngokuqhubekayo . Lo msebenzi usikhumbuza inkinga eyaziwa kakhulu yokuhlanganisa. Kubalulekile ukusho ukuthi inkinga ecatshangelwayo ihlukile enkingeni yokuhlanganisa ngoba inkinga yethu inesimo esichazwe ngokuqinile sombandela wokufana kwezakhi zeqoqo, ezinqunywa ukungaguquki. , kodwa enkingeni yokuhlanganisa asikho isimo esinjalo. Isitatimende senkinga yokuhlanganisa nolwazi ngale nkinga singatholakala
Ngakho-ke, sikwazile ukwenza inkinga futhi sithole inkinga yakudala ngendlela efanayo. Manje kuyadingeka ukucabangela ama-algorithms aziwayo wokuyixazulula, ukuze ungaqali kabusha isondo, kodwa uthathe imikhuba engcono kakhulu futhi uyisebenzise. Ukuxazulula inkinga yokuhlanganisa, sicabangele ama-algorithms adume kakhulu, okuyilawa: - izindlela -izindlela, i-algorithm yokuhlonza izingxenye ezixhunyiwe, ubuncane be-algorithm yomuthi ohlukanisayo. Incazelo kanye nokuhlaziywa kwama-algorithms anjalo kungatholakala
Ukuxazulula inkinga yethu, sihlanganisa ama-algorithms - kusho futhi -izindlela azisebenzi nhlobo, njengoba inani lamaqoqo alaziwa kusenesikhathi futhi ama-algorithms anjalo awabheki umkhawulo wezinsuku ongashintshi. Ama-algorithm anjalo ekuqaleni alahlwa ekucatshangelweni.
Ukuze sixazulule inkinga yethu, i-algorithm yokuhlonza izingxenye ezixhunyiwe kanye ne-algorithm encane yomuthi we-spanning ifaneleka kakhulu, kodwa, njengoba kwavela, ayikwazi ukusetshenziswa "ngokubheka phambili" enkingeni exazululwa futhi ithole isisombululo esihle. Ukuze sichaze lokhu, ake sicabangele indlela yokusebenza kwama-algorithms anjalo ngokuphathelene nenkinga yethu.
Cabangela igrafu , lapho ama-vertices ayiqoqo lamaqembu , kanye nomkhawulo phakathi kwama-vertices ΠΈ inesisindo esilingana nomehluko wezinsuku phakathi kwamaqoqo ΠΈ . Ku-algorithm yokuhlonza izingxenye ezixhunyiwe, ipharamitha yokufaka icacisiwe kuphi , nakugrafu yonke imiphetho isisindo sayo sikhulu iyasuswa . Amapheya aseduze kuphela ezinto ahlala exhumekile. Iphuzu le-algorithm ukukhetha inani elinjalo , lapho igrafu "ihlukana" ibe izingxenye ezimbalwa ezixhunyiwe, lapho izinhlangothi eziyingxenye yalezi zingxenye zizokwanelisa umbandela wethu wokufana, onqunywa ukungaguquki. . Izingxenye eziwumphumela zingamaqoqo.
Ubuncane be-algorithm yesihlahla esinabile kuqala yakhela kugrafu ubuncane besihlahla esinwebekayo, bese isusa imiphetho ngokulandelana kwayo enesisindo esiphezulu kuze kube igrafu βihlukanaβ ibe izingxenye ezimbalwa ezixhumene, lapho izingxenye zalezi zingxenye zizokwanelisa umbandela wethu wokufana. Izingxenye eziwumphumela zizoba amaqoqo.
Uma usebenzisa ama-algorithms anjalo ukuxazulula inkinga ecatshangelwayo, isimo singase siphakame njengakuMdwebo 3.
Umdwebo 3. Ukusetshenziswa kokuhlanganisa ama-algorithms enkingeni exazululekayo
Ake sithi ukungaguquki kwethu komehluko phakathi kwezinsuku zeqoqo yizinsuku ezingama-20. Igrafu iboniswe ngesimo sendawo ukuze kube lula ukubona izinto ezibonakalayo. Womabili ama-algorithms akhiqize isisombululo se-3-cluster, esingathuthukiswa kalula ngokuhlanganisa amaqoqo abekwe kumaqoqo ahlukene nomunye nomunye! Kusobala ukuthi ama-algorithms anjalo adinga ukuguqulwa ukuze avumelane nemininingwane yenkinga exazululwa, futhi ukusetshenziswa kwawo ngendlela emsulwa ekuxazululeni inkinga yethu kuzonikeza imiphumela engemihle.
Ngakho-ke, ngaphambi kokuthi siqale ukubhala ikhodi yama-algorithms egrafu alungiselwe umsebenzi wethu futhi siphinde sisungule ibhayisikili lethu (ngemifanekiso esesivele sikwazi ukubona uhlaka lwamasondo ayisikwele), saphinde, sanquma ukubhekana nenkinga enjalo ngokwesayensi, okungukuthi: zama ukuyehlisela kokunye ukwenza kahle kwezinkinga, ngethemba lokuthi ama-algorithms akhona wokuyixazulula angasetshenziswa ngaphandle kokulungiswa.
Okunye ukusesha inkinga efanayo yakudala kube yimpumelelo! Sikwazile ukuthola inkinga yokuthuthukisa ehlukahlukene, ukwakheka kwayo okuhambisana noku-1 koku-1 nokwakhiwa kwenkinga yethu. Lo msebenzi uvele waba setha inkinga yokumboza. Ake sethule ukwakheka kwenkinga ngokuphathelene nemininingwane yethu.
Kukhona isethi elinganiselwe nomndeni kuwo wonke amacembu alo ahlukene amaqembu, ukuze umehluko ngezinsuku zawo wonke amapheya amaqembu esethi ngayinye encane emndenini aweqi ama-constants . Isembozo sibizwa ngokuthi umndeni emandleni amancane, izakhi zawo , kangangokuthi inyunyana yamasethi emndenini kufanele inikeze isethi yawo wonke amaqembu .
Ukuhlaziywa okuningiliziwe kwale nkinga kungatholakala ΠΈ Ezinye izinketho zokusetshenziswa okungokoqobo kwenkinga yokumboza kanye nokuguqulwa kwayo kungatholakala
I-algorithm yokuxazulula inkinga
Sinqume ngemodeli yezibalo yenkinga ezoxazululwa. Manje ake sibheke i-algorithm yokuyixazulula. Amasethi angaphansi emndenini ingatholakala kalula ngenqubo elandelayo.
- Hlela amaqoqo kusuka kusethi ngokulandelana kwezinsuku zabo.
- Thola ubuncane kanye nezinsuku zeqoqo eziphezulu.
- Okwezinsuku zonke kusukela osukwini oluncane ukuya kwephezulu, thola wonke amaqoqo amadethi awo ahluka kuwo akukho ngaphezu (ngakho inani Kungcono ukuthatha inombolo elinganayo).
Umqondo wenqubo yokwenza umndeni wamasethi at izinsuku zivezwe kuMfanekiso 4.
Fig.4. Ukwakhiwa kwezingxenye ezingaphansi zamaqembu
Le nqubo ayidingeki kuwo wonke umuntu dlula kuwo wonke amanye amaqoqo futhi uhlole umehluko ngezinsuku zawo, noma kunani lamanje hambisa kwesokunxele noma kwesokudla uze uthole inqwaba usuku lwalo oluhlukile ngaphezu kwengxenye yenani lokungaguquki. Zonke izakhi ezilandelayo, lapho uhambisa kokubili kwesokudla nakwesobunxele, ngeke kusithakazelise, ngoba kubo umehluko wezinsuku uzokhula kuphela, njengoba izakhi ezikuhlu zaqala uku-oda. Le ndlela izosindisa kakhulu isikhathi lapho inani lamaqembu kanye nokusabalala kwezinsuku zabo kukhulu kakhulu.
Isethi yokumboza inkinga -nzima, okusho ukuthi akukho ukushesha (ngesikhathi sokusebenza esilingana ne-polynomial yedatha yokufaka) kanye ne-algorithm enembile yokuyixazulula. Ngakho-ke, ukuxazulula inkinga yokumboza isethi, i-algorithm yokuhaha esheshayo yakhethwa, okuyinto, yiqiniso, ayinembile, kodwa inezinzuzo ezilandelayo:
- Ezinkingeni zosayizi omncane (futhi lokhu kuyindaba yethu ncamashi), ibala izixazululo ezisondele kakhulu kokuhle. Njengoba ubukhulu benkinga bukhula, ikhwalithi yesisombululo iyawohloka, kodwa namanje ihamba kancane;
- Kulula kakhulu ukusebenzisa;
- Kuyashesha, njengoba isilinganiso saso sesikhathi sokusebenza sinjalo .
I-algorithm ehahayo ikhetha amasethi ngokusekelwe kulo mthetho olandelayo: esigabeni ngasinye, kukhethwa isethi ehlanganisa inani eliphezulu lama-elementi angakahlanganiswa. Incazelo enemininingwane ye-algorithm kanye nekhodi yamanga ingatholakala
Ukuqhathaniswa kokunemba kwe-algorithm ehahayo enjalo kudatha yokuhlola yenkinga exazululwa namanye ama-algorithms aziwayo, afana ne-algorithm yokuhaha enokwenzeka, i-algorithm yekholoni lezintuthwane, njll., ayenziwe. Imiphumela yokuqhathanisa ama-algorithms anjalo kudatha engahleliwe ekhiqizwe ingatholwa
Ukuqaliswa nokusebenza kwe-algorithm
Le algorithm isetshenziswe ngolimi I-1Π‘ futhi yafakwa ekucutshungulweni kwangaphandle okubizwa ngokuthi "Residue Compression" eyayixhunywe kukho I-WMS-uhlelo. Asizange sisebenzise i-algorithm ngolimi C ++ futhi uyisebenzise kusukela engxenyeni yangaphandle yoMdabu, okungaba okulungile kakhulu, njengoba isivinini sekhodi siphansi C ++ izikhathi futhi kwezinye izibonelo ngisho namashumi ezikhathi ngokushesha kunejubane lekhodi efanayo livuliwe I-1Π‘. Olimini I-1Π‘ I-algorithm yenzelwe ukonga isikhathi sokuthuthuka kanye nokulula kokulungisa iphutha endaweni yokukhiqiza yekhasimende. Umphumela we-algorithm uvezwa kuMfanekiso 5.
Fig.5. Icubungula ukuze "cindezela" izinsalela
Umfanekiso wesi-5 ubonisa ukuthi endaweni yokugcina impahla eshiwo, amabhalansi amanje ezimpahla kumaseli okugcina ahlukaniswa abe amaqoqo, lapho amadethi amaqoqo ezimpahla ahluka khona zingekho izinsuku ezingaphezu kwezingu-30. Njengoba ikhasimende likhiqiza futhi ligcina ama-valve ebhola ensimbi endaweni yokugcina impahla, impilo yayo yeshalofu ibalwa ngeminyaka, umehluko onjalo wosuku unganganakwa. Qaphela ukuthi ukucubungula okunjalo okwamanje kusetshenziswa ngokuhlelekile ekukhiqizeni, naku-opharetha I-WMS qinisekisa ikhwalithi enhle yokuhlanganisa amaqembu.
Iziphetho nokuqhubeka
Okuhlangenwe nakho okuyinhloko esikuthole ekuxazululeni inkinga enjalo engokoqobo kuwukuqinisekisa ukusebenza kahle kokusebenzisa ipharadigm: izibalo. isitatimende senkinga umata odumile. imodeli i-algorithm edumile i-algorithm icabangela imininingwane yenkinga. Ukusebenza okuhlukile sekuneminyaka engaphezu kuka-300, futhi ngalesi sikhathi abantu bakwazile ukucabangela izinkinga eziningi futhi baqongelela ulwazi oluningi lokuzixazulula. Okokuqala, kuhle kakhulu ukuphendukela kulokhu okuhlangenwe nakho, bese uqala kabusha isondo lakho.
Esihlokweni esilandelayo sizoqhubeka nendaba emayelana nama-algorithms okwenza kahle futhi sibheke okuthakazelisa kakhulu futhi okuyinkimbinkimbi kakhulu: i-algorithm "yokucindezelwa" okuphelele kwezinsalela zamaseli, esebenzisa idatha etholwe ku-algorithm yeqoqo lokuhlanganisa njengokufaka.
Ilungiselele isihloko
U-Roman Shangin, umhleli womnyango wamaphrojekthi,
Inkampani yokuqala ye-BIT, i-Chelyabinsk
Source: www.habr.com