Utafiti huu unalenga kubaini hali zisizo za kawaida katika hatua za mwanzo kwenye actuators za electro-pneumatic zinazoweza kutumika katika matumizi ya nishati mbadala kama vile wind turbines, mifumo ya kufuatilia jua na vituo vya kuhifadhi nishati. Watafiti wamemodeli kwa njia ya kihisabati uhusiano kati ya pressure, volume, piston position, air flow rate na force katika double-acting pneumatic cylinder; kisha wakachunguza data za sensor zinazodaiwa kupatikana kutoka FluidSIM simulation na FESTO training bench kwa kutumia support vector machine classifier.

Modeli ya SVM iliyowasilishwa ilifikia usahihi wa asilimia 90,2 katika balanced test set yenye hali za kawaida na zisizo za kawaida za uendeshaji. Kwa anomaly class, precision iliripotiwa kuwa asilimia 87, recall takribani asilimia 94,5 na F1 score asilimia 90,6. Katika confusion matrix, takribani anomalies 120 zilibainishwa kwa usahihi, anomalies 7 zikaainishwa kama kawaida; samples 18 za kawaida zikatoa false alarms. Values hizi zinaonyesha kwamba modeli inaweza kutenganisha classes mbili katika laboratory au simulation data, lakini hazithibitishi ni kwa kiwango gani inaweza kutabiri kwa mapema na kwa kuaminika failures zitakazotokea baadaye katika actual renewable-energy facility.

Umuhimu kwa mtazamo wa Uturuki: Mbinu hii inatoa initial model inayoweza kuchunguzwa kwa lengo la kupunguza unplanned downtime katika wind turbines, solar-tracking systems, energy-storage facilities na viwanda vinavyotumia pneumatic automation nchini Uturuki. Kabla ya matumizi halisi, long-term pressure, position, flow, temperature na maintenance records zinapaswa kukusanywa kutoka actuators halisi zinazotumika Uturuki; failures kama leakage, valve sticking, seal wear, increased friction na sensor drift ziwekwe labels tofauti; na modeli ifanyiwe external validation katika seasons, loads na facilities tofauti. Kwa kuwa utafiti hautoi Turkey-specific field data, cost calculation au emission reduction, haupaswi kutumiwa moja kwa moja kama investment au maintenance-policy recommendation.

Dai la “sustainability” la utafiti linategemea assumption kwamba early maintenance inaweza kupunguza failures, part replacement na unplanned downtime. Utafiti haujapima actual maintenance time, energy-production loss, compressed-air consumption, part life, maintenance cost au carbon emissions. Kwa hiyo, main success iliyoonyeshwa si sustainability performance, bali normal-anomaly classification performance katika limited dataset.

Tatizo kuu la utafiti ni nini?

Katika renewable-energy facilities, si turbine blades, solar panels au batteries pekee zinazohusika; kuna pia mechanical na electro-pneumatic subsystems nyingi zinazozisogeza, kuziweka katika nafasi na kuzidumisha katika hali salama. Failure ya valve, cylinder, seal au air-supply line inaweza kusababisha system kusimama hata kama main energy-generation equipment bado inaweza kufanya kazi.

Katika traditional maintenance approaches, equipment hubadilishwa baada ya operating hours fulani au intervention hufanywa baada ya failure kutokea. Predictive maintenance, kwa upande mwingine, inalenga kubaini failure kabla haijafikia critical level kwa kufuatilia small changes katika pressure, position, vibration, flow, temperature na electrical control signals.

Swali linaloshughulikiwa na utafiti ni hili: Je, normal na early-stage abnormal operating states zinaweza kutenganishwa pale mathematical model inayoeleza physical behavior ya double-acting electro-pneumatic cylinder inapounganishwa na SVM classifier inayojifunza kutoka sensor data?

Electro-pneumatic actuator ni nini?

Electro-pneumatic actuator ni mfumo unaobadilisha electrical control signal kuwa mechanical motion kwa kutumia compressed air. Katika setup iliyochunguzwa, double-acting cylinder inatumika. Compressed air inapopelekwa upande mmoja wa piston, piston husogea katika mwelekeo mmoja; inapopelekwa upande mwingine, husogea katika mwelekeo wa kinyume.

Main components za mfumo ni:

• Double-acting pneumatic cylinder,
• 5/3-way directional control valve,
• Solenoid valve na electrical control circuit,
• Compressor inayozalisha compressed air,
• Air-preparation au service unit,
• Pressure, position na flow measurement elements,
• Relay, switch na 24 V supply connections.

5/3-way valve ina connection ports tano na operating positions tatu. Valve huamua compressed air itaelekezwa kwenye chamber ipi ya cylinder na chamber nyingine itaondolewa hewa kwa namna gani. Degradation katika valve, cylinder, air line au sensor yoyote inaweza kusababisha mabadiliko katika movement time, pressure difference, piston position au air consumption.

Volume model ya double-acting cylinder

Piston inaposogea, volumes za chambers mbili za cylinder hubadilika katika directions zinazopingana. Main volume relationships zilizotumiwa katika utafiti ni:

\[ V_1 = V_{1,0} + A_1x \]

\[ V_2 = V_{2,0} - A_2x \]

Hapa V₁ na V₂ zinawakilisha instantaneous volumes za chambers mbili, V_1,0 na V_2,0 initial volumes, A₁ na A₂ effective areas za pande mbili za piston, na x linear displacement ya piston.

Piston inaposogea kuelekea chamber ya kwanza, chamber moja hupanuka huku nyingine ikipungua. Volume change huathiri pressure, na pressure difference huathiri piston force.

Ideal-gas relation na pressure change

Utafiti unamodeli hewa katika chambers kwa ideal-gas approach:

\[ P_1V_1 = n_1RT \]

\[ P_2V_2 = n_2RT \]

P₁ na P₂ zinaonyesha chamber pressures, n₁ na n₂ molar amounts, R universal gas constant na T absolute temperature.

Katika model, temperature imechukuliwa kuwa constant. Volume ya chamber ikipungua huku air mass ikibaki ileile, pressure huongezeka; volume ikiongezeka, pressure hupungua. Kwa kuwa katika real rapid pneumatic motion compression na expansion zinaweza kubadilisha temperature, constant-temperature assumption ni moja ya simplifications muhimu za model.

Force balance ya piston

Forces zinazoundwa na pressures katika chambers mbili juu ya piston areas ndizo chanzo kikuu cha piston motion. Dynamic equation iliyotumiwa na utafiti ni:

\[ M\ddot{x} + b\dot{x} + kx = P_1A_1 - P_2A_2 - F_{\mathrm{load}} \]

Hapa M ni moving mass, ẍ piston acceleration, b viscous damping coefficient, ẋ piston velocity, k equivalent spring coefficient na F_load external load force.

Pressure difference inapopungua, friction inapoongezeka au external load inapoongezeka, piston speed na position response hubadilika. Kwa hiyo, pressure, position na speed signals zinaweza kubeba dalili za possible failure.

Friction model

Utafiti unagawanya friction katika static, Coulomb na viscous components:

\[ F_{\mathrm{friction}} = F_{\mathrm{static}} + F_{\mathrm{Coulomb}} + \mu_v\dot{x} \]

Static friction inawakilisha force inayopaswa kushindwa kabla piston haijaanza kusogea, Coulomb friction takribani constant friction wakati wa motion, na viscous component resistance inayoongezeka pamoja na velocity.

Seal wear, loss of lubrication, surface degradation au mechanical misalignment vinaweza kubadilisha friction components. Hata hivyo, utafiti haujaeleza failures hizi ziliigwa kwa mabadiliko gani ya namba katika friction.

Air flow ilishughulikiwaje?

Kiasi cha hewa kinachoingia na kutoka katika cylinder chambers kupitia valves huamua jinsi chamber pressures zinavyobadilika kwa muda. Utafiti unawasilisha pressure-dynamics equations zinazounganisha inlet na outlet flow rates na chamber volume kwa kutumia conservation of mass.

Mathematical arrangement ya relation iliyotolewa kwa valve flow haiko wazi katika maandishi. Inaonekana kwamba square-root sign inayotarajiwa katika standard orifice-flow models inaweza kuwa imekosekana au kuhamishwa vibaya. Pia haijaelezwa jinsi choked na unchoked flow regions zilivyotenganishwa katika compressible flow. Kwa hiyo, valve-flow model haiwezi kujengwa upya independently.

Mathematical model inategemea assumptions zipi?

Main assumptions zinazotumika wazi au kwa njia isiyo ya moja kwa moja katika utafiti ni:

• Pneumatic system kutenda takribani linearly,
• Supply pressure kuwa specified au stable,
• Gas temperature kuchukuliwa kuwa constant,
• Kutokuwepo kwa air leakage katika valves na connections,
• Sensor data kuwa sufficient na accurate,
• Selected features kuweza kuwakilisha failures,
• Training data na usage data kutoka distributions zinazofanana.

No-leakage assumption ni muhimu sana. Air leakage ni mojawapo ya important failure types katika pneumatic systems. Kuacha leakage nje ya initial model kunaonyesha kwamba model haiwakilishi moja kwa moja sehemu ya real maintenance problems.

Experimental na simulation setup

Kielelezo 2 katika ukurasa wa tisa wa utafiti kinaonyesha pneumatic na electrical connections za double-acting cylinder pamoja na 5/3 directional control valve. Kielelezo 3 katika ukurasa huohuo kinaonyesha circuit ikiwa imeendeshwa katika FluidSIM environment na active air-electric paths.

Katika picha ya ukurasa wa kumi, 5/3 solenoid valve, double-acting cylinder, air-service unit na electrical connections zinaonekana kwenye FESTO training bench. Watafiti wanaeleza kwamba setup hii inafanana na industrial components na inaruhusu simultaneous sensor-data collection.

Hata hivyo, haiko wazi experimental na simulation data ziliunganishwa kwa namna gani. Katika sehemu moja kazi inaitwa “controlled simulation-based experiment”, katika sehemu nyingine “experimental data”, na katika comparison graph “real data model”. Idadi ya samples zilizopatikana kutoka actual bench, idadi iliyozalishwa na FluidSIM na data source ambayo results table inategemea havijaelezwa kwa kina.

Ni sensors na features zipi zilitumika?

Utafiti unataja variables zifuatazo katika data-collection process:

• Pressures katika cylinder chambers mbili,
• Piston position au displacement,
• Piston velocity,
• Air flow au flow-rate value,
• Pneumatic force,
• Temperature,
• Valve na control signals.

Kwa kila variable haijaelezwa tofauti kama ilipimwa directly na sensor, ilikokotolewa kutoka mathematical model au ilitolewa kutoka raw signals. Sensor model, measurement range, accuracy, sampling frequency na calibration information pia hazijawasilishwa.

Machine-learning process iliundwaje?

Proposed data-processing chain katika utafiti inafuata mpangilio huu:

1. Kukusanya sensor data,
2. Kusafisha na kufanya preprocessing ya data,
3. Kuunda normal na anomaly labels,
4. Kutoa pressure, position, speed, force, flow na temperature features,
5. Kufanya standardization ya features,
6. Kugawa data katika training na test sets,
7. Kufundisha SVM model,
8. Kutathmini kwa accuracy, precision, recall na F1 score,
9. Real-time normal-anomaly decision.

Kielelezo 4 kinaonyesha blocks za data collection, preprocessing, feature selection, hyperparameter tuning, model selection, training, evaluation na deployment. Vielelezo 5 na 6 vinafupisha decision flow inayoanza na data collection na kusonga hadi real-time anomaly detection ikiwa performance inatosha.

SVM hufanyaje uamuzi?

Katika utafiti, SVM decision function imetolewa kama:

\[ f(\mathbf{x}) = \mathbf{w}^{T}\mathbf{x}+b \]

x inawakilisha feature vector, w weights za decision plane na b bias term. Katika linear case, sign ya function huamua sample itawekwa katika class ipi.

Hata hivyo, results table inaonyesha kwamba RBF kernel ilitumika. RBF kernel hutenganisha samples kwa njia isiyo ya moja kwa moja katika higher-dimensional feature space, na decision function haiishii tu kwenye linear expression iliyo juu. Utafiti hautoi basic hyperparameters za RBF kernel kama gamma na regularization parameter.

Kutolingana kuhusu uchaguzi wa kernel

Katika method section, polynomial kernel inaelezwa kutoa more complex decision boundaries kuliko linear kernel na kuchaguliwa. Katika sehemu inayofuata imeandikwa kwamba linear na RBF kernels zilitunishwa, huku results table ikionyesha wazi kernel iliyotumika kuwa RBF.

Kauli hizi tatu hazitoi model definition ileile. Ingawa table inaonyesha kwamba results ni za RBF model, haijaelezwa kama polynomial model ilijaribiwa, ililinganishwa vipi na model ya mwisho ilichaguliwa kwa criterion gani.

Confusion matrix inaonyesha nini?

Confusion matrix katika ukurasa wa 13 wa utafiti ina classes mbili: “Normal” na “Anomaly”. Matrix inaonyesha takribani matokeo yafuatayo:

• Takribani normal samples 110 ziliwekwa kwa usahihi katika normal class.
• Normal samples 18 ziliwekwa kimakosa kama anomaly.
• Anomalies 7 ziliwekwa kimakosa kama normal.
• Takribani anomalies 120 zilibainishwa kwa usahihi.

Idadi ya false alarms ni 18, na missed anomalies ni 7. Kwa predictive maintenance, false negatives ni muhimu hasa; kwa sababu model ikitoa normal decision wakati system kwa kweli inaharibika, maintenance intervention inaweza kuchelewa. False positives, kwa upande mwingine, zinaweza kusababisha unnecessary inspection au maintenance activity.

Performance metrics zinamaanisha nini?

Accuracy inaonyesha ni sehemu gani ya samples zote imeainishwa kwa usahihi. Precision inapima ni sehemu gani ya samples ambazo model imesema ni anomaly zilikuwa anomaly kweli. Recall inaonyesha ni sehemu gani ya actual anomalies iliyogunduliwa. F1 score ni harmonic mean ya precision na recall.

Values zilizoripotiwa kwa anomaly class ni:

• Precision: 0,87
• Recall: 0,95 katika table; asilimia 94,55 katika comparison table
• F1 score: 0,91 katika table; asilimia 90,6 katika comparison table
• Overall accuracy: asilimia 90,2

Recall kuwa juu kuliko precision inaonyesha kwamba model inakamata sehemu kubwa ya anomalies lakini pia inazalisha false alarms zaidi.

Ulinganisho wa experimental na mathematical data

Utafiti unalinganisha matokeo mawili. Model inayoitwa “experimental simulation” ilitoa accuracy ya asilimia 90,2; model ya mathematical simulation pekee ikatoa accuracy ya asilimia 90,3. Overall accuracies ni karibu sawa.

Kwa upande mwingine, anomaly class ikiangaliwa, data inayoitwa experimental ilitoa values za juu zaidi:

• Anomaly precision: asilimia 87,0 dhidi ya asilimia 74,6
• Anomaly recall: asilimia 94,55 dhidi ya asilimia 81,5
• Anomaly F1 score: asilimia 90,6 dhidi ya asilimia 77,9

Comparison hii inaashiria kwamba physical au more physically representative data inaweza kuwa na manufaa katika kutambua anomaly class. Hata hivyo, kwa kuwa haijaelezwa mathematical data na experimental data zilizalishwa vipi, kama datasets zilikuwa size sawa, na kama hyperparameters zilezile zilitumika kwa models zote mbili, sababu ya difference haiwezi kubainishwa kwa uhakika.

Five-fold cross-validation inatoa nini?

Imeelezwa kwamba five-fold cross-validation ilitumika kwa models zote mbili. Katika method hii, data hugawanywa katika parts tano; model hufundishwa mara tano na kila wakati part tofauti hutengwa kwa validation. Lengo ni kupunguza dependence ya performance kwenye random data split moja.

Utafiti unatoa tu taarifa ya “5-Fold”; individual performances za folds tano, mean na standard deviation, na kama sequential measurements za same operating cycle ziliangukia katika folds tofauti wakati wa data splitting havijaelezwa. Katika time series, samples zilizo karibu sana zikisambazwa kati ya training na test sets zinaweza kuonyesha generalization performance kuwa juu kuliko hali halisi.

Je, utafiti kweli umegundua incipient fault?

Utafiti unadai “incipient fault”, yaani initial au developing fault detection. Hata hivyo, haijaelezwa fault iliundwa kwa physical parameter gani, measurements zilichukuliwa katika degradation levels zipi, na model ilitoa warning muda gani kabla ya critical failure.

Model kutenganisha normal na abnormal samples pekee haimaanishi kwamba imetabiri fault mapema. Kwa dai la incipient fault, experiments zinazoongeza fault severity kadiri muda unavyopita, early-warning time, different degradation levels na remaining time to failure zinahitajika. Utafiti huu hauwasilishi metrics hizi.

Sustainability connection imeundwaje?

Watafiti wanadai kwamba early fault detection inaweza kupunguza unplanned downtime, unnecessary maintenance activities, part replacement na indirect carbon emissions. Uhusiano huu unawezekana kiuhandisi: kutobadilisha equipment yenye afya bila sababu na kurekebisha fault kabla haijapanuka kunaweza kupunguza resource use.

Hata hivyo, hakuna mojawapo ya sustainability indicators zifuatazo iliyopimwa katika utafiti:

• Downtime iliyozuiwa,
• Energy production iliyohifadhiwa,
• Kupungua kwa compressed-air au electricity consumption,
• Kupungua kwa maintenance na spare-part cost,
• Kuongezeka kwa actuator life,
• Greenhouse-gas emissions zilizozuiwa,
• Life-cycle environmental impact.

Kwa hiyo, measured outcome ya utafiti si sustainability gain, bali performance ya SVM classifier. Sustainability impact ni hypothesis inayohitaji kujaribiwa baadaye.

Matokeo yanayoungwa mkono na utafiti

• Normal na abnormal electro-pneumatic operating states zinaweza kuainishwa kwa kutumia features kama pressure, position na flow.
• Physical model na machine learning zinaweza kuunganishwa katika predictive-maintenance framework moja.
• Katika test set iliyowasilishwa, SVM ilitoa overall accuracy ya takribani asilimia 90.
• Model iligundua sehemu kubwa ya anomalies katika test; anomalies 7 pekee ziliainishwa kama normal.
• Data inayoitwa experimental ilitoa anomaly precision, recall na F1 score za juu kuliko mathematical simulation data.
• FESTO training bench inaweza kutoa laboratory environment inayoweza kutumika kwa controlled fault-data generation na algorithm development.

Utafiti hauthibitishi nini?

• Hauthibitishi kwamba model itafanya kazi katika actual wind au solar-energy facility.
• Hauonyeshi kwamba inaweza kutabiri faults muda maalum kabla ya critical degradation.
• Hauthibitishi kwamba inapunguza maintenance cost, downtime au carbon emissions kwa kiwango kinachopimika.
• Hauonyeshi kwamba asilimia 90 accuracy itapatikana kwa aina zote za electro-pneumatic actuators.
• Hauonyeshi kwamba unseen fault types au different facility conditions zinaweza kutambuliwa kwa kuaminika.
• Hauonyeshi kwamba normal-anomaly decision moja inaweza kubainisha fault type, location au remaining useful life.
• Hautathmini real-time processing latency, hardware requirements au communication reliability ya model.

Mapungufu makuu ya method na editorial

• Scope na namna ya kuunganisha actual experimental data na FluidSIM data havijaelezwa.
• Fault types, severities na creation methods hazijabainishwa.
• Total data size inaeleweka kwa sehemu tu kutoka test table; training dataset haijaelezwa.
• Technical specifications za sensors na sampling frequency hazijatolewa.
• Feature-extraction operations na time windows zilizotumika hazijabainishwa.
• Training-test split ratio haijaelezwa.
• SVM hyperparameters kama C na gamma hazijatolewa.
• Polynomial na RBF kernel statements haziendani.
• Approximate F1 score ya 0,91 katika results table imeandikwa 0,83 katika results section.
• Confusion-matrix values zime-roundiwa katika scientific notation; exact cell counts hazijaandikwa moja kwa moja.
• Hakuna cost, energy au emission measurement inayounga mkono sustainability claim.
• Validation katika actual renewable-energy facility imeachwa kama future work.

Roadmap inayohitajika kwa matumizi nchini Uturuki

Ili mfumo wa aina hii utumike kwa kuaminika nchini Uturuki, katika hatua ya kwanza actual equipment inventory inapaswa kuundwa. Brake, blade-angle au service mechanisms katika wind turbines; tracking na cleaning systems katika solar plants; ventilation, fire-safety na switching actuators katika energy-storage facilities zinapaswa kutathminiwa tofauti.

Katika hatua ya pili, sensors za kufuatilia pressure, flow, position, temperature, vibration na electrical control signals zinapaswa kuwekwa kwenye equipment; data ziunganishwe na maintenance records. Mbali na normal operating data, different fault types ziundwe katika controlled na safe conditions na labeled data zikusanywe.

Katika hatua ya tatu, model haipaswi kuvalidiwa kwa random sample split pekee; inapaswa kujaribiwa kati ya machines tofauti, facilities tofauti na seasons tofauti. Kujaribu model iliyofundishwa katika facility moja kwenye facility nyingine ni muhimu kwa kuonyesha actual generalization performance.

Katika hatua ya mwisho, maintenance economics inapaswa kuongezwa kwenye classification performance. Cost ya false alarm kwa maintenance team, production na safety risk ya missed fault, sensor na data-infrastructure cost, downtime iliyozuiwa na energy-production gain zinapaswa kukokotolewa kwa pamoja.

Mbinu na Matokeo ya Utafiti

Mfumo uliotafitiwa na mtiririko wa data

SVM test results

Kulingana na table, test set ina idadi sawa ya samples katika classes mbili. Hata hivyo, kwa kuwa haijulikani kama sequential samples katika time series ni independent, haiwezi kudhaniwa kwamba rows 254 zinamaanisha physical experiments 254 tofauti.

Tafsiri ya confusion matrix

Kwa kuwa true-normal na true-anomaly cells katika confusion matrix zimewasilishwa kama approximate values katika scientific notation, exact counts haziwezi kusomwa directly kutoka visual. Katika Table 1, support values zimetolewa kama 127 kwa kila class, zikionyesha total test set ina samples 254.

Experimental na mathematical model results

Ingawa overall accuracy ya mathematical model inaonekana kuwa asilimia 0,1 juu zaidi, precision, recall na F1 values katika anomaly class ni za chini. Utafiti haujatathmini statistical significance ya difference hii wala kutoa confidence intervals.

Additional metrics zinazopaswa kufuatiliwa katika field application ya model

Maana ya figures na tables

• Kielelezo 1: Kinaonyesha kwa schematic chambers mbili za double-acting cylinder, piston areas, pressures, valve connections, movement direction na external load.
• Kielelezo 2: Ni connection diagram ya 5/3 directional control valve, double-acting cylinder, air-service unit na electrical control circuit.
• Kielelezo 3: Kinaonyesha active air na electrical paths za circuit ileile katika FluidSIM environment.
• Bench photograph: Kinaonyesha solenoid valve, cylinder, air-preparation unit na electrical connections katika FESTO training setup.
• Kielelezo 4: Kinaonyesha hatua za data collection, preprocessing, feature selection, hyperparameter tuning, training, evaluation na deployment.
• Vielelezo 5 na 6: Vinafupisha normal-anomaly labeling, model training na real-time decision flow.
• Kielelezo 7: Ni confusion matrix yenye takribani 110 true normal, 18 false alarms, 7 missed anomalies na takribani 120 true anomalies.
• Jedwali 1: Linawasilisha precision, recall, F1 score na accuracy values katika balanced test set.
• Jedwali 2: Linalinganisha SVM performance ya data inayoitwa experimental na mathematical simulation data.
• Kielelezo 8: Linalinganisha accuracy, precision, recall na F1 values za models mbili kwa line graph.

Maelezo ya Chanzo na Mbinu

Jina kamili la asili la utafiti: A Sustainability-Oriented Predictive Maintenance Framework for Renewable Energy Actuation Systems

Waandishi na mpangilio sahihi: Samuel Olufemi Amudipe, Adeyinka Adeoye, Aderonke Akinwumi, Ajayi Adebanji na Segun Adebayo.

Equal first author: Hakuna taarifa ya equal contribution au equal first authorship iliyotolewa.

Mwandishi wa mawasiliano: Samuel Olufemi Amudipe. Utafiti unatoa barua pepe ya mwandishi wa mawasiliano kama femioni61@gmail.com.

Uhusiano wa taasisi:

• Mechatronics Engineering Programme, College of Agriculture, Engineering and Science, Bowen University, Iwo, Nigeria
• Department of Mechanical and Mechatronics Engineering, Afe Babalola University, Ado-Ekiti, Nigeria

Samuel Olufemi Amudipe, Ajayi Adebanji na Segun Adebayo wameonyeshwa kuwa affiliated na Bowen University; Adeyinka Adeoye na Aderonke Akinwumi wameonyeshwa kuwa affiliated na Afe Babalola University.

DOI: 10.2139/ssrn.7196243. DOI ni ya SSRN preprint record na haipaswi kutafsiriwa kama DOI ya peer-reviewed journal article.

Jarida: Hakuna peer-reviewed journal name iliyotolewa na published journal version haijathibitishwa.

Jukwaa la uchapishaji: SSRN.

SSRN abstract number: 7196243.

Mwaka wa uchapishaji: 2026.

Aina ya chanzo: Engineering preprint yenye mathematical model, controlled FluidSIM-based simulation, FESTO training bench na SVM classification.

Hali ya mapitio ya kitaalamu: Utafiti haujapitia peer review. Kutolingana katika kernel selection, data source, F1 score na field validity katika method na results kunapaswa kuzingatiwa.

Kiungo rasmi:https://ssrn.com/abstract=7196243

Ufadhili: Hakuna separate funding statement katika maandishi.

Shukrani: Waandishi wameishukuru management ya Bowen University kwa research facilities.

Michango ya waandishi: Utafiti una collective contribution statement; lakini haujatenganisha ni mwandishi gani alifanya task gani.

Mgongano wa maslahi: Hakuna separate conflict-of-interest statement katika maandishi.

Upatikanaji wa data: Open-access link ya dataset, full distribution ya sample count, raw sensor records na model codes hazijawasilishwa.

Makala haya ya Kituruki yameandaliwa kwa kupitia full text ya utafiti, mathematical equations, FluidSIM circuits, FESTO bench image, algorithm flows, confusion matrix na performance tables. Vyanzo vya nje vimetumiwa tu kwa bibliographic verification ya preprint record, DOI na official link; hakuna finding ya nje iliyoongezwa kwenye scientific content.

Mapungufu makuu ya utafiti ni kutofanywa validation katika actual renewable-energy facility, kutokueleza boundaries za experimental na simulation data, kutobainisha fault types na severities, upungufu wa sensor na data-collection details, inconsistency katika SVM kernel, F1-score inconsistency katika results section na sustainability impacts kutopimwa quantitatively.