Labarin ya bayyana yadda ake aiwatarwa WMS-tsarin, mun fuskanci buƙatar magance matsalar tari mara daidaituwa da abin da algorithms muka yi amfani da su don magance shi. Za mu gaya muku yadda muka yi amfani da tsari, tsarin kimiyya don magance matsalar, waɗanne matsalolin da muka fuskanta da kuma irin darussan da muka koya.
Wannan ɗaba'ar ta fara jerin labaran da a cikinta muke raba nasarorin ƙwarewarmu wajen aiwatar da ingantaccen algorithms a cikin matakan sito. Manufar jerin kasidu ita ce sanar da masu sauraro nau'ikan matsalolin inganta ayyukan ajiyar kayayyaki da ke tasowa a kusan kowane matsakaici da babban rumbun ajiya, da kuma ba da labarin kwarewarmu wajen magance irin wadannan matsalolin da kuma ramukan da aka fuskanta a hanya. . Labaran za su kasance da amfani ga waɗanda ke aiki a cikin masana'antar kayan aikin sito, aiwatarwa WMS-tsarin, kazalika da masu shirye-shirye waɗanda ke da sha'awar aikace-aikacen lissafi a cikin kasuwanci da haɓakar matakai a cikin kamfani.
Bottleneck a cikin matakai
A cikin 2018, mun kammala aikin aiwatarwa WMS- tsarin a cikin sito na kamfanin "Trading House" LD a Chelyabinsk. Mun aiwatar da samfurin "1C-Logistics: Warehouse Management 3" don wuraren aiki 20: masu aiki WMS, ma'ajiyar ajiya, direbobin forklift. Matsakaicin sito yana da kusan 4 dubu m2, adadin sel shine 5000 kuma adadin SKU shine 4500. Gidan ajiya yana adana bawul ɗin ball na samar da namu masu girma dabam daga 1 kg zuwa 400 kg. Ana adana kaya a cikin ɗakunan ajiya a cikin batches, tun da akwai buƙatar zaɓar kaya bisa ga FIFO.
Lokacin zayyana tsare-tsaren sarrafa kayan sito, mun fuskanci matsalar da ba ta dace ba ta ajiyar kaya. Ƙayyadaddun ƙayyadaddun kayan ajiya da kirƙira sun kasance kamar yadda tantanin ma'auni ɗaya zai iya ƙunsar abubuwa daga rukuni ɗaya kawai. Kayayyakin suna zuwa wurin ajiyar kullun kuma kowane isowa keɓantacce ne. Gabaɗaya, sakamakon aikin da aka yi na tsawon wata 1, an ƙirƙiri batches 30 daban-daban, duk da cewa kowane ɗayan ya kamata a adana shi a cikin tantanin halitta daban. Sau da yawa ana zaɓar samfuran ba a cikin pallets gaba ɗaya ba, amma a cikin guda ɗaya, kuma a sakamakon haka, a cikin yanki na zaɓin yanki a cikin sel da yawa ana lura da hoton da ke gaba: a cikin tantanin halitta tare da ƙarar fiye da 1 m3 akwai nau'ikan cranes da yawa waɗanda ya mamaye ƙasa da 5-10% na ƙarar tantanin halitta.
Hoto na 1. Hoton kayayyaki da yawa a cikin tantanin halitta
A bayyane yake cewa ba a amfani da ƙarfin ajiya da kyau. Don yin la'akari da ma'auni na bala'i, zan iya ba da adadi: a matsakaita, akwai daga 1 zuwa 3 kwayoyin irin waɗannan kwayoyin halitta tare da ƙarar fiye da 100 m300 tare da ma'auni na "minuscule" a lokacin lokuta daban-daban na aikin sito. Tunda ma'ajiyar tana da ƙanƙanta, a cikin lokutan shagulgulan shagulgulan shagunan wannan lamarin ya zama "kwalba" kuma yana rage tafiyar matakai na sito sosai.
Ra'ayin warware matsalar
Wani ra'ayi ya taso: a rage ragowar ragowar kwanakin da suka fi kusa da shi zuwa juzu'i guda, kuma irin wannan ragowar tare da dunƙule guda ɗaya yakamata a sanya su gaba ɗaya a cikin tantanin halitta guda ɗaya, ko kuma da yawa, idan babu isasshen sarari a cikin ɗaya don ɗaukar ɗakin. duka adadin ragowar.
Hoto.2. Tsari don damfara ragowar a cikin sel
Wannan yana ba ku damar rage yawan wuraren da aka mamaye da za a yi amfani da su don sanya sabbin kayayyaki. A halin da ake ciki inda karfin sito ya yi yawa, irin wannan ma'auni yana da matukar mahimmanci, in ba haka ba za a iya samun isasshen sarari kyauta don ɗaukar sabbin kayayyaki, wanda zai haifar da tsayawa a cikin jeri na sito da matakan sake cikawa. A baya kafin aiwatarwa WMS-Tsarin sun yi wannan aiki da hannu, wanda ba shi da tasiri, tun da tsarin neman saura masu dacewa a cikin sel ya daɗe sosai. Yanzu, tare da ƙaddamar da tsarin WMS, mun yanke shawarar sarrafa tsarin, hanzarta shi kuma mu sanya shi mai hankali.
Hanyar magance irin wannan matsala ta kasu kashi biyu:
- a matakin farko mun sami ƙungiyoyin batches kusa da kwanan wata don matsawa;
- a mataki na biyu, ga kowane rukuni na batches muna ƙididdige mafi ƙaƙƙarfan wuri na sauran kayan da ke cikin sel.
A cikin labarin na yanzu za mu mayar da hankali kan matakin farko na algorithm, kuma mu bar ɗaukar hoto na mataki na biyu don labarin na gaba.
Nemo samfurin lissafi na matsalar
Kafin mu zauna don rubuta code da sake farfado da dabararmu, mun yanke shawarar tunkarar wannan matsala ta hanyar kimiyya, wato: tsara ta ta hanyar lissafi, rage ta zuwa sanannen matsalar ingantawa da kuma amfani da ingantaccen algorithms da ake da su don magance ta, ko ɗaukar waɗannan algorithms da ake da su. a matsayin tushe da kuma gyara su zuwa takamaiman matsalar aiki da ake warwarewa.
Tun da yake a fili ya biyo baya daga tsarin kasuwanci na matsalar da muke fama da shi, za mu tsara irin wannan matsala ta hanyar ka'idar saiti.
Bari – saitin duk batches na ragowar wani samfur a cikin sito. Bari – ba akai-akai na kwanaki. Bari - wani yanki na batches, inda bambancin kwanakin ga duk nau'i-nau'i na batches a cikin rukunin baya wuce akai-akai. . Muna buƙatar nemo mafi ƙarancin adadin ɓangarorin rarrabuwa , kamar yadda duk subsets tare zai ba da yawa .
A takaice dai, muna buƙatar nemo ƙungiyoyi ko gungu na jam'iyyu iri ɗaya, inda aka ƙayyade ma'aunin kamance ta hanyar dindindin. . Wannan aikin yana tunatar da mu sanannen matsalar tari. Yana da mahimmanci a ce matsalar da aka yi la'akari da ita ta bambanta da matsalar tari domin matsalarmu tana da ƙayyadaddun sharuddan ma'auni na kamancen abubuwan cluster, wanda aka ƙaddara ta akai-akai. , amma a cikin clustering matsalar babu irin wannan yanayin. Ana iya samun bayanin matsalar tari da bayanai kan wannan matsalar
Don haka, mun sami nasarar tsara matsalar kuma mun sami matsala ta gargajiya tare da tsari iri ɗaya. Yanzu ya zama dole a yi la'akari da sanannun algorithms don warware shi, don kada a sake farfado da dabaran, amma don ɗaukar mafi kyawun ayyuka da amfani da su. Don magance matsalar tari, mun yi la'akari da shahararrun algorithms, wato: -ma'ana -ma'ana, Algorithm don gano abubuwan haɗin gwiwa, mafi ƙarancin bishiyar algorithm. Ana iya samun bayanin da bincike na irin waɗannan algorithms
Don magance matsalar mu, tari algorithms -ma'ana kuma -ma'anar ba a aiwatar da su kwata-kwata, tunda ba a taɓa sanin adadin gungu a gaba ba kuma irin waɗannan algorithms ba sa la'akari da ƙayyadaddun kwanakin kullun. Irin waɗannan algorithms an fara jefar da su daga la'akari.
Don magance matsalarmu, algorithm don gano abubuwan haɗin da aka haɗa da mafi ƙarancin bishiyar bishiyar algorithm sun fi dacewa, amma, kamar yadda ya fito, ba za a iya amfani da su "kai-kai" ga matsalar da ake warwarewa da samun mafita mai kyau. Don bayyana wannan, bari mu yi la'akari da dabarun aiki na irin waɗannan algorithms dangane da matsalarmu.
Yi la'akari da jadawali , wanda a cikinsa ne saitin jam'iyyun , da gefen da ke tsakanin madaidaitan и yana da nauyi daidai da bambancin kwanaki tsakanin batches и . A cikin algorithm don gano abubuwan da aka haɗa, an ƙayyade siginar shigarwa inda , kuma a cikin jadawali an cire duk gefuna waɗanda nauyin ya fi girma . Nau'i-nau'i-nau'i na abubuwa kawai sun kasance suna haɗe. Ma'anar algorithm shine don zaɓar irin wannan ƙimar , wanda jadawali "ya rabu" zuwa sassa da yawa da aka haɗa, inda ƙungiyoyin da ke cikin waɗannan sassan za su gamsar da ma'anar kamancen mu, wanda aka ƙaddara ta akai-akai. . Abubuwan da aka haifar sune gungu.
Mafi ƙanƙantar algorithm na itace yana fara ginawa akan jadawali mafi ƙarancin bishiyar, sannan a bi-da-bi yana cire gefuna tare da mafi girman nauyi har sai jadawali "ya rabu" zuwa abubuwan haɗin kai da yawa, inda ƙungiyoyin da ke cikin waɗannan abubuwan za su gamsar da ma'aunin kamanninmu. Abubuwan da aka haifar za su zama gungu.
Lokacin amfani da irin waɗannan algorithms don magance matsalar da ake la'akari, wani yanayi na iya tasowa kamar yadda yake cikin hoto 3.
Hoto 3. Aikace-aikace na clustering algorithms zuwa matsalar da ake warware
Bari mu ce mu akai-akai don bambanci tsakanin kwanakin batch shine kwanaki 20. Graph an nuna shi a cikin sigar sararin samaniya don sauƙin fahimtar gani. Dukansu algorithms sun samar da mafita na 3-cluster, wanda za'a iya ingantawa cikin sauƙi ta hanyar haɗa batches da aka sanya a cikin gungu daban-daban tare da juna! A bayyane yake cewa irin waɗannan algorithms suna buƙatar gyara don dacewa da ƙayyadaddun matsalar da ake warwarewa, kuma aikace-aikacen su a cikin tsari mai tsabta don magance matsalarmu zai ba da sakamako mara kyau.
Don haka, kafin mu fara rubuta lambar don algorithms ɗin da aka gyara don aikinmu da sake ƙirƙira keken namu (a cikin silhouettes waɗanda za mu iya fahimtar fassarori na ƙafafun murabba'in), mu kuma, mun yanke shawarar tunkarar irin wannan matsalar a kimiyyance, wato: yi ƙoƙarin rage shi zuwa wani ingantaccen ingantaccen matsala, a cikin bege cewa za a iya amfani da algorithms na yanzu don warware shi ba tare da gyare-gyare ba.
Wani bincike na irin wannan matsala na gargajiya ya yi nasara! Mun yi nasarar nemo matsalar ingantawa mai hankali, wanda tsarinsa ya zo daidai da 1 cikin 1 tare da tsara matsalarmu. Wannan aikin ya zama saita rufe matsala. Bari mu gabatar da tsari na matsalar dangane da ƙayyadaddun mu.
Akwai iyakataccen saiti da iyali na dukkan rukunoninsa na jam'iyyu daban-daban, kamar bambancin kwanakin kowane nau'i-nau'i na kowane yanki. daga iyali baya wuce akai-akai . Ana kiran suturar iyali na mafi ƙarancin iko, abubuwan da ke cikin su , irin wannan ƙungiyar ƙungiyoyin daga iyali kamata ya ba da sa na dukan jam'iyyun .
Ana iya samun cikakken nazarin wannan matsala и Ana iya samun wasu zaɓuɓɓuka don aikace-aikacen aikace-aikacen matsalar rufewa da gyare-gyarenta
Algorithm don magance matsalar
Mun yanke shawarar tsarin lissafi na matsalar da za a warware. Yanzu bari mu dubi algorithm don warware shi. Rarraba daga iyali ana iya samun sauƙin samu ta hanya mai zuwa.
- Shirya batches daga saiti a saukowa tsarin kwanakinsu.
- Nemo mafi ƙanƙanta da matsakaicin kwanakin tsari.
- Domin kowace rana daga mafi ƙarancin kwanan wata zuwa mafi girma, nemo duk batches waɗanda kwanakinsu ya bambanta da ba fiye da (don haka darajar Yana da kyau a ɗauki madaidaicin lamba).
Hankali na hanya don kafa iyali na saiti a an gabatar da kwanaki a cikin Hoto na 4.
Hoto.4. Samar da sassan jam'iyyu
Wannan hanya ba lallai ba ne ga kowa da kowa shiga cikin duk sauran batches kuma duba bambancin kwanakin su, ko daga darajar yanzu matsa hagu ko dama har sai kun sami batch wanda kwanan wata ya bambanta da ta fiye da rabin ƙimar madaidaicin. Duk abubuwan da suka biyo baya, lokacin da motsi biyu zuwa dama da hagu, ba za su kasance masu ban sha'awa a gare mu ba, tun da yake a gare su bambance-bambance a cikin kwanaki kawai za su karu, tun da farko an ba da umarnin abubuwan da ke cikin tsararru. Wannan hanya za ta adana lokaci mai mahimmanci lokacin da adadin jam'iyyun da yaduwar kwanakinsu ya yi yawa sosai.
Matsalar rufe saitin shine -wahala, wanda ke nufin babu sauri (tare da lokacin aiki daidai da yawan adadin bayanan shigarwa) da ingantaccen algorithm don warware shi. Don haka, don warware matsalar rufe saiti, an zaɓi algorithm mai saurin haɗama, wanda, ba shakka, ba daidai bane, amma yana da fa'idodi masu zuwa:
- Don ƙananan ƙananan matsalolin (kuma wannan shine ainihin lamarinmu), yana ƙididdige mafita waɗanda ke kusa da mafi kyau. Yayin da girman matsalar ke ƙaruwa, ingancin maganin yana raguwa, amma har yanzu a hankali;
- Mai sauƙin aiwatarwa;
- Mai sauri, tunda kiyasin lokacin tafiyar sa shine .
Algorithm mai haɗama yana zaɓar saiti bisa ga ka'ida mai zuwa: a kowane mataki, ana zaɓar saiti wanda ke rufe matsakaicin adadin abubuwan da ba a rufe su ba tukuna. Ana iya samun cikakken bayanin algorithm da pseudocode ɗin sa
Ba a yi kwatankwacin daidaiton irin wannan algorithm na hadama akan bayanan gwaji na matsalar da ake warwarewa tare da wasu sanannun algorithms, irin su algorithm na kwadayi mai yuwuwa, algorithm na tururuwa, da sauransu, ba a yi ba. Ana iya samun sakamakon kwatanta irin waɗannan algorithms akan bayanan bazuwar da aka samar
Aiwatar da aiwatar da algorithm
An aiwatar da wannan algorithm a cikin harshe 1С kuma an haɗa shi a cikin wani aiki na waje mai suna "Residue Compression" wanda aka haɗa WMS-tsarin. Ba mu aiwatar da algorithm a cikin harshe ba C ++ kuma a yi amfani da shi daga ɓangaren ɗan ƙasa na waje, wanda zai zama mafi daidai, tunda saurin lambar ya yi ƙasa C ++ sau kuma a wasu misalan har sau goma da sauri fiye da gudun irin wannan lambar a kunne 1С. A kan harshe 1С An aiwatar da algorithm don adana lokacin haɓakawa da sauƙi na lalatawa a tushen samar da abokin ciniki. An gabatar da sakamakon algorithm a cikin hoto 5.
Hoto.5. Ana aiwatarwa zuwa "damfara" ragowar
Hoto na 5 ya nuna cewa a cikin ma'auni da aka kayyade, ana rarraba ma'auni na kayayyaki na yanzu a cikin sel masu ajiya zuwa gungu, wanda a cikinsa kwanakin batches na kayan ya bambanta da juna ba fiye da kwanaki 30 ba. Tun lokacin da abokin ciniki ke samarwa da adana bawul ɗin ƙwallon ƙarfe a cikin sito, wanda aka ƙididdige rayuwar rayuwar shi a cikin shekaru, irin wannan bambance-bambancen kwanan wata za a iya sakaci. Lura cewa ana amfani da irin wannan sarrafa a halin yanzu cikin tsari a samarwa, da masu aiki WMS tabbatar da kyakkyawan ingancin tari na jam'iyya.
Ƙarshe da ci gaba
Babban ƙwarewar da muka samu daga magance irin wannan matsala mai amfani ita ce tabbatar da tasiri na amfani da tsarin: lissafi. bayanin matsalar sanannen tabarma. abin koyi sanannun algorithm algorithm yin la'akari da takamaiman matsalar. Haɓakawa mai hankali ya kasance sama da shekaru 300, kuma a wannan lokacin mutane sun sami damar yin la'akari da matsaloli masu yawa tare da tara gogewa mai yawa don magance su. Da farko, yana da kyau a juyo zuwa wannan ƙwarewar, sannan kawai fara sake ƙirƙira dabaran ku.
A cikin labarin na gaba za mu ci gaba da labarin game da ingantawa algorithms kuma mu dubi mafi ban sha'awa kuma mafi rikitarwa: algorithm don mafi kyawun "matsi" na ragowar tantanin halitta, wanda ke amfani da bayanan da aka karɓa daga algorithms clustering batch a matsayin shigarwa.
Wadda ta shirya
Roman Shangin, mai tsara shirye-shirye na sashen ayyuka,
Kamfanin BIT na farko, Chelyabinsk
source: www.habr.com