🤖 AI-Powered IoT Platform
🛡️ AssetGuard™

Master Plan Bundling IoT + AI

Strategi Bertahap: Administrasi → Energi → AssetGuard™ (Predictive Maintenance Semua Aset)

📄 Berdasarkan Makalah Riset Pandawa Techno 2026
🎯 Executive Summary

Konsep bundling IoT bertahap dengan integrasi AI di setiap stage dan AssetGuard™ sebagai Stage 3 — platform predictive maintenance terintegrasi untuk semua aset kritis: dari server admin hingga mesin produksi.

Total Revenue 3 Tahun
Rp 4.99 M
Dengan AI + AssetGuard
AI Revenue Contribution
32%
Dari total revenue
Target Klien
50+
Semua segmen pabrik
Gross Profit 3 Tahun
Rp 1.90 M
Margin ~38%
🤖 Filosofi AI + AssetGuard di Setiap Stage
🎯 Prinsip Pandawa Techno: AI bukan fitur tambahan, melainkan core value layer di setiap tahapan. Setiap stage memiliki kemampuan AI yang meningkat — dari descriptive (Stage 1) → diagnostic (Stage 2) → predictive maintenance (Stage 3).
📊

Stage 1: Descriptive AI

AI yang menjelaskan apa yang terjadi di setiap divisi — anomali, tren, insight otomatis.

Contoh: "Sales pipeline menurun 12% minggu ini karena konversi di segmen otomotif turun."
🔍

Stage 2: Diagnostic AI

AI yang menjelaskan mengapa hal itu terjadi — root cause analysis, korelasi energi-produksi.

Contoh: "Kenaikan konsumsi energi 8% disebabkan oleh power factor rendah di Panel A."
🛡️

Stage 3: Predictive AI (AssetGuard™)

AI yang memprediksi kapan aset akan rusak — untuk SEMUA aset: admin + produksi.

Contoh: "Motor B: Prediksi breakdown dalam 48 jam — lakukan maintenance preventif sekarang."
🛡️ Mengapa AssetGuard™, Bukan OEE?

📚 Alasan Berdasarkan Makalah Riset

Makalah Anda menyebutkan empat pilar aplikasi IIoT, dengan predictive maintenance sebagai pilar #1 — sebelum OEE. Ini adalah entry point paling natural dalam adopsi IIoT.

  • Pilar #1: Condition monitoring & predictive maintenance
  • Pilar #2: OEE monitoring
  • Pilar #3: Manajemen energi & ESG
  • Pilar #4: Traceability

🎯 Alasan Strategis

  • Pain point lebih jelas — downtime = masalah nyata
  • Sales cycle lebih pendek — resistensi lebih rendah
  • Cakupan lebih luas — admin + produksi
  • Cross-selling lebih kuat — sinergi dengan Stage 1 & 2
  • Retention lebih tinggi — maintenance = kebutuhan berkelanjutan
  • ROI lebih terukur — 20% downtime + 30% biaya perawatan
📊 Perbandingan Hasil: OEE Stage 3 menghasilkan 13 klien dengan revenue Rp 1.65 M. AssetGuard™ menghasilkan 18 klien dengan revenue Rp 1.53 M — jumlah klien +38%, retention lebih tinggi, dan cross-selling lebih kuat.
🏭 Target Pasar (dari Makalah Riset)
Kawasan Industri Lokasi Karakteristik AssetGuard Potential
Jababeka Cikarang, Bekasi 2.000+ tenant, ICTel IoT end-to-end, Smart Manufacturing Hub 🛡️ Sangat Tinggi
MM2100 Cibitung, Bekasi Klaster Jepang: NSK, Toyotetsu, Panasonic, Riken 🛡️ Sangat Tinggi
KIIC & Suryacipta Karawang Principal otomotif Jepang + ratusan supplier tier-2/3 🛡️ Sangat Tinggi
EJIP & Delta Silicon Cikarang Elektronik & komponen otomotif 🛡️ Tinggi
KNIC Karawang HLI Green Power (Hyundai-LG), pabrik sel baterai EV 🛡️ Sangat Tinggi
🏢 Stage 1 — Q4 2026 - Q2 2027
📦 Pandawa AdminGateway™ + AI Assistant

Gateway IoT yang mengumpulkan data dari komputer administrasi, dilengkapi AI Assistant dan Asset Health Basic untuk aset admin dasar.

📊 Data yang Dikumpulkan (Divisi)

DivisiMetrik untuk CEO
SalesPipeline value, conversion rate
FinanceDSO, cash position
PurchasingProcurement cycle, cost savings
HRLabor cost ratio, absenteeism
PPICOrder fulfillment, WIP value
ITSystem uptime, incidents

🖥️ Asset Health Basic (Admin)

AsetMetrik
Server RoomSuhu, kelembaban, power
NetworkUptime, latency
UPSBattery status, voltage
Komputer AdminDisk health, temperature

⚙️ Spesifikasi Gateway

  • CPU: ARM Cortex-A72 quad-core 1.5GHz
  • RAM: 4GB LPDDR4
  • Storage: 64GB eMMC + microSD
  • AI Accelerator: NPU 2 TOPS (opsional)
  • Ethernet: 2x Gigabit RJ45
  • USB: 2x USB 3.0, 2x USB 2.0
  • Wi-Fi: Dual-band 802.11ac
  • Power: 12V DC, 2A
  • BOM Cost: Rp 1.5jt - Rp 2.2jt
🤖 AI Layer Stage 1 — Descriptive AI
📝

Auto-Generated Executive Summary

AI menghasilkan ringkasan eksekutif harian/mingguan dalam bahasa natural.

Output: "Minggu ini: Sales naik 8.2%, Finance DSO turun 8.7%, HR overtime naik 6.5%."
🚨

Anomaly Detection

AI mendeteksi anomali pada KPI divisi dan aset admin secara otomatis.

Contoh: "Anomali: Suhu server room 28°C (di atas normal 22°C) — cek AC."
💬

Natural Language Query

CEO dapat bertanya dalam bahasa natural dan AI menjawab dengan data.

Teknologi: LLM fine-tuned + RAG pada data dashboard
📈

Trend Forecasting (Basic)

AI memprediksi trend KPI dan aset admin 7-30 hari ke depan.

Algoritma: Prophet / ARIMA
💰 Model Bisnis Stage 1
Paket Hardware + AI Setup Langganan/Bulan AI Features
Starter Rp 3.500.000 Rp 2.500.000 Rp 1.000.000 Auto-summary, Anomaly detection
Business Rp 7.000.000 Rp 5.000.000 Rp 2.000.000 + NL Query, Trend forecasting
Enterprise Rp 14.000.000 Rp 10.000.000 Rp 4.000.000 + Custom AI model, API access
Target Klien Y1
15
Kawasan Bekasi-Karawang
Revenue Y1
Rp 420 jt
+26% vs tanpa AI
AI Premium Revenue
Rp 87 jt
~21% dari total
⚡ Stage 2 — Q3 2027 - Q2 2028
🔋 Pandawa EnergyWatch™ + AI Analytics

Memantau konsumsi energi listrik dengan AI Analytics — diagnostic insight, prediksi konsumsi, dan korelasi energi dengan kesehatan aset.

🎯 Nilai Jual

  • Cost savings: Identifikasi pemborosan energi 10-15%
  • ESG compliance: Laporan emisi otomatis
  • Diagnostic AI: Root cause analysis kenaikan energi
  • Predictive AI: Prediksi konsumsi & emisi
  • Energy-Asset Correlation: Deteksi kerusakan aset dari pola energi

⚙️ Spesifikasi Hardware

  • Energy Meter: PZEM-004T V4.0 (100A)
  • CT Clamp: 100A, non-invasive
  • Modbus Converter: USB-to-RS485
  • Edge AI: NPU 4 TOPS untuk inference lokal
  • BOM Cost: Rp 2jt - Rp 3.5jt per panel
🤖 AI Layer Stage 2 — Diagnostic AI
🔍

Root Cause Analysis

AI menganalisis penyebab kenaikan/penurunan konsumsi energi dengan korelasi multi-variabel.

Contoh: "Kenaikan energi 8% di Panel A disebabkan power factor 0.89 — pasang capacitor bank."
🔮

Energy Demand Forecasting

AI memprediksi konsumsi energi 7-30 hari ke depan berdasarkan pola historis dan jadwal produksi.

Algoritma: LSTM / Gradient Boosting
💡

Optimization Recommendation

AI memberikan rekomendasi konkret untuk penghematan energi berdasarkan benchmark industri.

Contoh: "Geser beban produksi ke off-peak hours — potensi hemat Rp 4.2jt/bulan."
🔗

Energy-Asset Health Correlation

AI mendeteksi kerusakan aset dari anomali konsumsi energi — early warning sebelum breakdown.

Contoh: "Konsumsi motor B naik 12% — kemungkinan bearing wear, cek dalam 7 hari."
💰 Model Bisnis Stage 2
Paket Hardware + AI Langganan/Bulan AI Features
Energy Basic Rp 6.500.000 Rp 1.500.000 Root cause, Basic forecasting
Energy Pro Rp 15.000.000 Rp 3.500.000 + Optimization, Carbon prediction
Energy Enterprise Custom Rp 7.000.000+ + Energy-Asset correlation, API access
Target Klien Y2
18
7 konversi + 11 baru
Revenue Y2
Rp 720 jt
+31% vs tanpa AI
AI Premium Revenue
Rp 216 jt
~30% dari total
🛡️ Stage 3 — Q3 2028 - Q4 2029
🛡️ Pandawa AssetGuard™ — Predictive Maintenance Unified

Platform predictive maintenance terintegrasi untuk SEMUA aset kritis — dari server admin hingga mesin produksi. AI memprediksi kapan aset akan rusak sebelum kerusakan terjadi.

🎯 Positioning: "Platform Predictive Maintenance Terintegrasi untuk Semua Aset Kritis — dari Server Admin hingga Mesin Produksi."

🖥️ Aset Administrasi

AsetSensor/Metrik
Server & Data CenterSuhu, RH, power, disk health
Network EquipmentUptime, latency, packet loss
UPS & GensetBattery health, voltage, runtime
AC & CoolingSuhu, refrigerant pressure, current
CCTV & SecurityCamera health, storage, network
Komputer AdminDisk health, RAM, temperature
Printer & ScannerPage count, toner, error rate
Forklift AdminBattery, motor current, hours

⚙️ Aset Produksi

AsetSensor/Metrik
Motor ListrikGetaran, suhu, arus, PF
PompaGetaran, tekanan, flow, suhu
KompresorTekanan, suhu, arus, getaran
ConveyorGetaran, arus motor, belt tension
Mesin CNCGetaran, spindle load, tool wear
Chiller & HVACSuhu, tekanan, arus kompresor
Panel ListrikSuhu, arus, power factor
Forklift & AGVBattery, motor, hours
🤖 AI Layer Stage 3 — Predictive Maintenance AI
🔮

Predictive Failure (Admin + Production)

AI memprediksi kapan aset akan rusak berdasarkan pola sensor — untuk semua aset kritis.

Contoh: "Prediksi breakdown motor B dalam 48 jam — lakukan maintenance preventif sekarang untuk hindari downtime 4 jam."

Remaining Useful Life (RUL)

AI mengestimasi sisa umur pakai aset — membantu perencanaan penggantian spare part.

Contoh: "Bearing motor A: RUL 720 jam (±30 hari) — pesan spare part sekarang."
🔍

Root Cause Analysis (Asset)

AI menganalisis akar penyebab degradasi aset dengan SHAP dan correlation analysis.

Contoh: "Getaran abnormal motor B disebabkan misalignment (impact 0.65) + bearing wear (0.35)."
💡

Maintenance Recommendation

AI memberikan rekomendasi maintenance yang spesifik, terprioritas, dan cost-effective.

Contoh: "Prioritas tinggi: Ganti bearing motor B dalam 24 jam — biaya Rp 2jt vs potensi downtime Rp 50jt."
📝

Auto-Generated Maintenance Report

AI menghasilkan laporan maintenance otomatis untuk tim teknis dan manajemen.

Output: Laporan harian/mingguan dengan status aset, prediksi, dan rekomendasi.
📦

Spare Part Prediction

AI memprediksi kebutuhan spare part berdasarkan RUL dan histori maintenance.

Contoh: "Prediksi kebutuhan bearing SKF 6205: 4 unit bulan ini — stok saat ini 2 unit."
💰 Nilai Tambah AssetGuard™: Predictive maintenance mengurangi downtime hingga 20% dan biaya perawatan hingga 30% (sesuai benchmark makalah). Cakupan admin + produksi membuat platform ini menjadi single source of truth untuk kesehatan aset perusahaan.
💰 Model Bisnis Stage 3 — AssetGuard™
Paket Cakupan Hardware Langganan/Bulan Target Segmen
AssetGuard Admin 5-10 aset admin Rp 12.000.000 Rp 2.500.000 UKM, kantor
AssetGuard Production 3-5 mesin produksi Rp 25.000.000 Rp 5.000.000 Pabrik kecil-menengah
AssetGuard Unified Admin + Produksi Rp 45.000.000 Rp 9.000.000 Pabrik menengah-besar
AssetGuard Enterprise 20+ aset, multi-lokasi Custom Rp 18.000.000+ Enterprise, multi-plant
Target Klien Y3
18
10 konversi + 8 baru
Revenue Y3
Rp 1.53 M
Hardware + SaaS
AI Premium Revenue
Rp 580 jt
~38% dari total
📊 Perbandingan dengan OEE: AssetGuard™ menghasilkan 18 klien (vs 13 klien OEE) dengan revenue Rp 1.53 M (vs Rp 1.65 M). Meskipun revenue sedikit lebih rendah, jumlah klien +38% dan retention lebih tinggi — lebih sustainable untuk jangka panjang.
OEE sebagai Fitur Tambahan (Opsional)

OEE tidak hilang — ia menjadi fitur tambahan dalam AssetGuard™ untuk klien yang sudah mature.

💡 Strategi: Tawarkan OEE sebagai add-on module bagi klien AssetGuard™ yang sudah memiliki fondasi data maintenance yang matang. Ini menciptakan upsell pathway yang natural.
Modul Opsional Deskripsi Harga Add-on
OEE Monitoring Perhitungan OEE real-time untuk lini produksi +Rp 2.000.000/bulan per lini
Traceability Module Ketertelusuran produksi untuk tuntutan principal +Rp 1.500.000/bulan per lini
Quality Analytics Analitik kualitas & prediksi reject +Rp 2.500.000/bulan per lini
🤖 Arsitektur AI Terintegrasi — 3 Layer

AI di Pandawa Techno dibangun dalam 3 layer yang saling terintegrasi — dari edge hingga cloud — memberikan kemampuan yang meningkat di setiap stage.

Edge AI Layer

AI yang berjalan langsung di gateway — latensi rendah, privasi terjaga, bandwidth efisien.

Teknologi: TensorFlow Lite / ONNX Runtime pada NPU
Fungsi: Anomaly detection real-time, filter data, inference lokal
Hardware: NPU 2-8 TOPS
☁️

Cloud AI Layer

AI yang berjalan di cloud — model kompleks, training, dan analitik lanjutan.

Teknologi: Python (scikit-learn, TensorFlow), MLflow
Fungsi: Predictive modeling, forecasting, RUL estimation
Infrastruktur: Cloud VPS dengan GPU (opsional)
🧠

LLM Layer

Large Language Model untuk natural language interface dan auto-generated insights.

Teknologi: LLM fine-tuned + RAG
Fungsi: NL Query, auto-summary, chatbot CEO
Model: GPT-4o-mini / Llama 3 / Gemini Flash
📊 Matriks AI Capability per Stage
Kemampuan AI Stage 1 Stage 2 Stage 3 Teknologi
Auto-Generated Summary LLM + Template
Anomaly Detection Isolation Forest
Natural Language Query LLM + RAG
Trend Forecasting Prophet, ARIMA
Root Cause Analysis SHAP, Correlation
Optimization Recommendation Rule-based + ML
Carbon Footprint Prediction Regression Model
🛡️ Predictive Failure XGBoost, LSTM
🛡️ Remaining Useful Life (RUL) Survival Analysis
🛡️ Spare Part Prediction Time-series
🛡️ Maintenance Recommendation Rule-based + ML
🛡️ Auto-Generated Maintenance Report LLM + Template
🚀 Roadmap Pengembangan AI
📅

Q4 2026 — AI Foundation

Membangun infrastruktur AI dasar: data pipeline, model registry, LLM integration.

Deliverable: Auto-summary + Anomaly detection
📅

Q2 2027 — AI Expansion

Menambahkan kemampuan diagnostic AI dan forecasting lanjutan.

Deliverable: Root cause + Forecasting + NL Query
📅

Q4 2027 — AI Diagnostic

Mengembangkan model diagnostic untuk energi dan kesehatan aset admin.

Deliverable: Energy-Asset correlation + Optimization
📅

Q2 2028 — AI Predictive (Admin)

Model predictive maintenance untuk aset administrasi (server, network, AC).

Deliverable: Predictive failure + RUL aset admin
📅

Q4 2028 — AI Predictive (Production)

Model predictive maintenance untuk mesin produksi (motor, pompa, kompresor).

Deliverable: Predictive failure + RUL mesin produksi
📅

Q2 2029 — AI Unified Platform

Platform unified untuk semua aset — admin + produksi dalam satu dashboard.

Deliverable: Unified Asset Health Platform
📅

Q4 2029 — AI Prescriptive

AI prescriptive yang tidak hanya memprediksi, tapi juga merekomendasikan aksi optimal.

Deliverable: Prescriptive Maintenance + AutoML
📊 Executive Dashboard — Live Preview dengan AI + AssetGuard

🏢 STAGE 1 — Monitoring Administrasi + AI

Overall Health
87/100
Critical Alerts
0
Warnings
2
🤖 AI Insights
12
Pipeline Value
Rp 4.85 M
🤖 Forecast Accuracy
94%
Server Room Temp
22.5°C
🤖 Anomalies Detected
1

⚡ STAGE 2 — Monitoring Energi + AI

Total Konsumsi
45,230 kWh
Total Biaya
Rp 67.8 jt
🤖 Prediksi Besok
47,100 kWh
Efisiensi
5.77%
🤖 Optimasi Potensial
Rp 4.2 jt
Emisi CO₂
38.4 t
🔗 Energy-Asset Alert
1
Power Factor
0.89

🛡️ STAGE 3 — AssetGuard™ Predictive Maintenance

Total Aset Dimonitor
142
🛡️ Aset Admin
58
🛡️ Aset Produksi
84
🛡️ Prediksi Breakdown
3
🛡️ RUL < 30 hari
5
🛡️ Maintenance Due
7
🛡️ Spare Part Alert
4
Asset Health Score
91/100
🤖 AI Insight Panel — Live Feed
📊

Executive Summary — Hari Ini

"Sales pipeline naik 8.2% dipimpin segmen otomotif. Finance DSO turun ke 42 hari. HR overtime naik 6.5% — perlu perhatian. Energy efficiency tercapai 5.77% di bawah baseline. Asset health score 91/100."

🤖 AI Generated · 08:30
🔮

🛡️ Predictive Maintenance Alert — Motor B

"Prediksi breakdown motor B dalam 48 jam berdasarkan pola getaran abnormal (RMS 4.2 mm/s, di atas threshold 3.5 mm/s) dan suhu 78.5°C. RUL estimasi: 720 jam. Rekomendasi: lakukan maintenance preventif dalam 24 jam untuk hindari downtime 4 jam."

🛡️ AssetGuard AI · 08:25
🖥️

🛡️ Asset Health Alert — Server Room

"Suhu server room 28°C (di atas normal 22°C). Prediksi risiko overheating dalam 6 jam jika AC tidak diperbaiki. Rekomendasi: cek AC server room sekarang — potensi downtime ERP 4-8 jam."

🛡️ AssetGuard AI · 08:20
💡

Optimization Recommendation — Energi

"Power factor Panel Produksi A 0.89 (di bawah target 0.90). Rekomendasi: pasang capacitor bank 50 kVAR — potensi hemat Rp 2.8jt/bulan dan menghindari penalti PF."

🤖 Diagnostic AI · 08:20
💰 Proyeksi Finansial 3 Tahun (dengan AI + AssetGuard)
Tahun Stage 1 Stage 2 Stage 3 (AssetGuard) Total Revenue AI Revenue Gross Profit
Y1 (2026-2027) Rp 420 jt Rp 420 jt Rp 87 jt Rp 145 jt
Y2 (2027-2028) Rp 620 jt Rp 720 jt Rp 1.34 M Rp 402 jt Rp 480 jt
Y3 (2028-2029) Rp 750 jt Rp 950 jt Rp 1.53 M Rp 3.23 M Rp 1.03 M Rp 1.28 M
TOTAL 3 TAHUN Rp 1.79 M Rp 1.67 M Rp 1.53 M Rp 4.99 M Rp 1.52 M Rp 1.90 M
💰 Dampak AssetGuard™: Dengan AssetGuard™ menggantikan OEE, revenue total sedikit lebih rendah (Rp 4.99 M vs Rp 5.2 M) tetapi jumlah klien lebih tinggi (50+ vs 45+), retention lebih tinggi, dan cross-selling lebih kuat. Lebih sustainable untuk jangka panjang.
🤖 Detail AI Revenue Contribution
AI Revenue Y1
Rp 87 jt
21% dari total
AI Revenue Y2
Rp 402 jt
30% dari total
AI Revenue Y3
Rp 1.03 M
32% dari total
Total AI Revenue
Rp 1.52 M
30% dari total 3 tahun
📈 Detail Biaya AI
Cloud AI Infrastructure
Rp 30 jt/thn
GPU cloud + storage
LLM API Cost
Rp 22 jt/thn
GPT-4o-mini / Llama
AI Engineer Salary
Rp 200 jt/thn
1 senior AI/ML engineer
Edge AI Hardware
Rp 800rb/unit
NPU tambahan
Model Training
Rp 18 jt/thn
Compute + experiment
AI Gross Margin
~65%
Setelah biaya AI
🗓️ Roadmap Eksekusi Bertahap dengan AI + AssetGuard

🏢 Stage 1 + AI

Q4 2026 - Q2 2027
  • Launch AdminGateway™
  • 15 klien pilot
  • AI: Auto-summary + Anomaly detection
  • AI: NL Query basic
  • Asset Health Basic (admin)
  • Profesionalisasi website
  • Pendaftaran TKDN
  • Revenue: Rp 420 jt

⚡ Stage 2 + AI

Q3 2027 - Q2 2028
  • Launch EnergyWatch™
  • 18 klien (7 konversi + 11 baru)
  • AI: Root cause + Forecasting
  • AI: Optimization + Energy-Asset correlation
  • Integrasi dengan Stage 1
  • Sertifikasi ISO 27001
  • Revenue: Rp 720 jt

🛡️ Stage 3 + AI (AssetGuard™)

Q3 2028 - Q4 2029
  • Launch AssetGuard™
  • 18 klien (10 konversi + 8 baru)
  • AI: Predictive failure (Admin + Production)
  • AI: RUL + Spare part prediction
  • AI: Maintenance recommendation
  • Integrasi penuh 3 stage
  • Sertifikasi IEC 62443
  • Revenue: Rp 1.53 M
🤖 Roadmap AI Development (Revisi dengan AssetGuard)
Periode Fokus AI Deliverable Teknologi
Q4 2026 AI Foundation Auto-summary, Anomaly detection LLM + Isolation Forest
Q2 2027 AI Expansion NL Query, Trend forecasting LLM + RAG, Prophet
Q4 2027 AI Diagnostic Root cause, Optimization, Energy-Asset correlation SHAP, Correlation Analysis
Q2 2028 🛡️ AI Predictive (Admin) Predictive maintenance aset admin XGBoost, RUL
Q4 2028 🛡️ AI Predictive (Production) Predictive maintenance mesin produksi LSTM, Survival Analysis
Q2 2029 🛡️ AI Unified Platform Unified asset health platform Multi-modal AI
Q4 2029 🛡️ AI Prescriptive Prescriptive maintenance + AutoML Reinforcement Learning
🎯 Faktor Kunci Keberhasilan

📜 Kepatuhan & Sertifikasi

  • TKDN — keunggulan pengadaan pemerintah
  • ISO 27001 — keamanan informasi
  • IEC 62443 — keamanan sistem kontrol industri
  • ISO/IEC 42001 — AI management system
  • ISO 55001 — asset management (baru, untuk AssetGuard)

🤝 Kemitraan Ekosistem

  • Jababeka ICTel — akses 2.000+ tenant
  • Telkomsel Enterprise — leverage 5G infrastructure
  • MM2100 Industrial Town — klaster manufaktur Jepang
  • AI/ML Partner — kolaborasi dengan penyedia LLM
  • Sensor Manufacturer — partnership untuk hardware
👥 Tim Minimal yang Dibutuhkan (dengan AssetGuard):
1 Hardware Engineer · 1 Cloud/Dashboard Engineer · 1 Data Analyst · 1 AI/ML Engineer · 1 Maintenance/Reliability Engineer · 1 Sales Engineer
🗄️ Struktur Data JSON — Master Schema dengan AssetGuard™

Schema lengkap yang merepresentasikan keseluruhan sistem monitoring dengan AI layer terintegrasi dan AssetGuard™ sebagai Stage 3.

{
  "system_metadata": {
    "version": "3.0.0-ai-assetguard",
    "company": "PT Contoh Manufaktur Indonesia",
    "location": "MM2100 Industrial Town, Cibitung, Bekasi",
    "gateway_id": "PT-ADM-001",
    "gateway_status": "online",
    "ai_enabled": true,
    "assetguard_enabled": true,
    "ai_models_loaded": [
      "anomaly_detector",
      "forecaster",
      "nl_query",
      "predictive_maintenance_admin",
      "predictive_maintenance_production",
      "rul_estimator",
      "spare_part_predictor"
    ],
    "last_heartbeat": "2026-09-10T08:29:45+07:00",
    "uptime_hours": 1247
  },

  "stage_1_admin": {
    "description": "Monitoring Divisi Administrasi + Descriptive AI + Asset Health Basic",
    "ai_layer": {
      "capabilities": [
        {
          "name": "Auto-Generated Summary",
          "type": "llm",
          "model": "gpt-4o-mini-finetuned",
          "frequency": "daily",
          "output": {
            "timestamp": "2026-09-10T08:30:00+07:00",
            "summary": "Sales pipeline naik 8.2% dipimpin segmen otomotif. Finance DSO turun ke 42 hari. HR overtime naik 6.5% — perlu perhatian.",
            "highlights": [
              "Sales: +8.2% (positif)",
              "Finance: DSO 42 hari (di bawah target 45)",
              "HR: Overtime 1,240 jam (di atas threshold 1,200)"
            ],
            "recommendations": [
              "Monitor HR overtime — pertimbangkan rekrutmen tambahan",
              "Sales momentum positif — pertahankan strategi"
            ]
          }
        },
        {
          "name": "Anomaly Detection",
          "type": "statistical",
          "algorithm": "Isolation Forest + Z-Score",
          "detected_anomalies": [
            {
              "division": "finance",
              "metric": "invoice_outstanding",
              "severity": "medium",
              "message": "Invoice outstanding turun 12.3% — di luar pola normal"
            }
          ]
        },
        {
          "name": "Natural Language Query",
          "type": "llm_rag",
          "model": "gpt-4o-mini",
          "sample_queries": [
            {
              "query": "Bagaimana performa sales bulan ini?",
              "response": "Sales pipeline Rp 4.85 M (97% dari target Rp 5 M). Order masuk 142 (94.7% dari target 150)."
            }
          ]
        },
        {
          "name": "Trend Forecasting",
          "type": "time_series",
          "algorithm": "Prophet",
          "forecasts": {
            "sales_pipeline": {
              "current": 4850000000,
              "forecast_7d": 5100000000,
              "forecast_30d": 5400000000,
              "confidence": 0.87
            }
          }
        }
      ]
    },
    "asset_health_basic": {
      "description": "Monitoring aset admin dasar",
      "assets": [
        {
          "id": "server_room",
          "name": "Server Room",
          "metrics": {
            "temperature": 22.5,
            "humidity": 45.0,
            "power_consumption": 12.5,
            "status": "healthy"
          },
          "ai_anomaly_score": 0.05
        },
        {
          "id": "network_main",
          "name": "Main Network Switch",
          "metrics": {
            "uptime_percent": 99.98,
            "latency_ms": 1.2,
            "packet_loss_percent": 0.01,
            "status": "healthy"
          }
        },
        {
          "id": "ups_main",
          "name": "UPS Server Room",
          "metrics": {
            "battery_health_percent": 92,
            "voltage": 230.5,
            "runtime_minutes": 45,
            "status": "healthy"
          }
        }
      ]
    }
  },

  "stage_2_energy": {
    "description": "Monitoring Energi + Diagnostic AI + Energy-Asset Correlation",
    "ai_layer": {
      "capabilities": [
        {
          "name": "Root Cause Analysis",
          "type": "explainable_ai",
          "algorithm": "SHAP + Correlation Analysis",
          "analyses": [
            {
              "id": "rca_001",
              "trigger": "Panel Produksi A power factor 0.89",
              "root_cause": "Beban induktif tinggi tanpa kompensasi capacitor bank",
              "recommendation": "Pasang capacitor bank 50 kVAR — potensi hemat Rp 2.8jt/bulan",
              "confidence": 0.91
            }
          ]
        },
        {
          "name": "Energy Demand Forecasting",
          "type": "time_series",
          "algorithm": "LSTM",
          "forecasts": {
            "total_kwh": {
              "current": 45230.5,
              "forecast_7d": 47100.0,
              "confidence": 0.89
            }
          }
        },
        {
          "name": "Optimization Recommendation",
          "type": "rule_based_ml",
          "recommendations": [
            {
              "id": "opt_001",
              "category": "load_shifting",
              "description": "Geser beban produksi ke off-peak hours",
              "potential_savings_idr": 4200000,
              "priority": "high"
            }
          ]
        },
        {
          "name": "Energy-Asset Health Correlation",
          "type": "correlation_analysis",
          "detections": [
            {
              "asset_id": "motor_b",
              "energy_anomaly": "+12% konsumsi",
              "predicted_issue": "bearing_wear",
              "confidence": 0.78,
              "recommendation": "Cek bearing motor B dalam 7 hari"
            }
          ]
        }
      ]
    },
    "total_consumption_kwh": 45230.5,
    "total_cost_idr": 67845750,
    "total_emission_kgco2": 38445.9,
    "efficiency_percent": 5.77,
    "panels": [
      {
        "id": "panel_main",
        "name": "Panel Utama",
        "status": "online",
        "voltage": 398.5,
        "current": 245.3,
        "power_kw": 98.2,
        "power_factor": 0.92,
        "energy_kwh": 18500.5,
        "ai_anomaly_score": 0.12
      }
    ]
  },

  "stage_3_assetguard": {
    "description": "🛡️ AssetGuard™ — Predictive Maintenance Unified (Admin + Production)",
    "ai_layer": {
      "capabilities": [
        {
          "name": "Predictive Failure (Admin)",
          "type": "classification",
          "algorithm": "XGBoost + Random Forest",
          "predictions": [
            {
              "asset_id": "server_room",
              "asset_type": "admin",
              "prediction": "overheating_risk",
              "probability": 0.72,
              "estimated_time_to_failure_hours": 6,
              "features_used": [
                { "feature": "temperature", "value": 28.0, "threshold": 25.0, "importance": 0.65 },
                { "feature": "humidity", "value": 55.0, "threshold": 50.0, "importance": 0.35 }
              ],
              "recommendation": "Cek AC server room sekarang — potensi downtime ERP 4-8 jam",
              "potential_downtime_avoided_hours": 8
            }
          ]
        },
        {
          "name": "Predictive Failure (Production)",
          "type": "classification",
          "algorithm": "XGBoost + LSTM",
          "predictions": [
            {
              "asset_id": "motor_b",
              "asset_type": "production",
              "prediction": "breakdown_imminent",
              "probability": 0.87,
              "estimated_time_to_failure_hours": 48,
              "features_used": [
                { "feature": "vibration_rms", "value": 4.2, "threshold": 3.5, "importance": 0.45 },
                { "feature": "temperature", "value": 78.5, "threshold": 75.0, "importance": 0.32 },
                { "feature": "current", "value": 156.8, "threshold": 150.0, "importance": 0.23 }
              ],
              "recommendation": "Lakukan maintenance preventif dalam 24 jam",
              "potential_downtime_avoided_hours": 4
            }
          ]
        },
        {
          "name": "Remaining Useful Life (RUL)",
          "type": "survival_analysis",
          "algorithm": "Weibull + Cox Proportional Hazards",
          "predictions": [
            {
              "asset_id": "motor_b_bearing",
              "rul_hours": 720,
              "rul_days": 30,
              "confidence": 0.82,
              "recommendation": "Pesan spare part bearing SKF 6205 sekarang"
            }
          ]
        },
        {
          "name": "Root Cause Analysis (Asset)",
          "type": "explainable_ai",
          "algorithm": "SHAP",
          "analyses": [
            {
              "asset_id": "motor_b",
              "root_causes": [
                { "cause": "misalignment", "impact": 0.65 },
                { "cause": "bearing_wear", "impact": 0.35 }
              ],
              "recommendation": "Lakukan alignment + ganti bearing"
            }
          ]
        },
        {
          "name": "Spare Part Prediction",
          "type": "time_series",
          "algorithm": "ARIMA + Inventory Optimization",
          "predictions": [
            {
              "part_id": "bearing_skf_6205",
              "predicted_demand_next_month": 4,
              "current_stock": 2,
              "recommendation": "Order 2 unit tambahan"
            }
          ]
        },
        {
          "name": "Maintenance Recommendation",
          "type": "rule_based_ml",
          "recommendations": [
            {
              "priority": "high",
              "asset_id": "motor_b",
              "action": "Ganti bearing dalam 24 jam",
              "cost_idr": 2000000,
              "potential_downtime_cost_idr": 50000000,
              "roi": "25x"
            }
          ]
        },
        {
          "name": "Auto-Generated Maintenance Report",
          "type": "llm",
          "model": "gpt-4o-mini",
          "frequency": "daily",
          "output": {
            "timestamp": "2026-09-10T08:30:00+07:00",
            "summary": "3 aset memerlukan perhatian: Motor B (breakdown 48 jam), Server Room (overheating 6 jam), Panel A (PF rendah).",
            "critical_assets": ["motor_b", "server_room"],
            "recommendations": [
              "Ganti bearing motor B dalam 24 jam",
              "Cek AC server room sekarang",
              "Pasang capacitor bank Panel A"
            ]
          }
        }
      ]
    },
    "asset_summary": {
      "total_assets_monitored": 142,
      "admin_assets": 58,
      "production_assets": 84,
      "assets_healthy": 127,
      "assets_warning": 12,
      "assets_critical": 3,
      "overall_asset_health_score": 91
    },
    "assets": [
      {
        "id": "motor_b",
        "name": "Motor B - Lini Produksi B",
        "category": "production",
        "type": "electric_motor",
        "status": "warning",
        "health_score": 68,
        "metrics": {
          "vibration_rms": 4.2,
          "temperature": 78.5,
          "current": 156.8,
          "power_factor": 0.85,
          "runtime_hours": 12500
        },
        "ai_predictions": {
          "breakdown_probability": 0.87,
          "estimated_time_to_failure_hours": 48,
          "rul_hours": 720,
          "recommended_action": "Ganti bearing dalam 24 jam"
        }
      },
      {
        "id": "server_room",
        "name": "Server Room - Gedung A",
        "category": "admin",
        "type": "data_center",
        "status": "warning",
        "health_score": 72,
        "metrics": {
          "temperature": 28.0,
          "humidity": 55.0,
          "power_consumption": 12.5
        },
        "ai_predictions": {
          "overheating_probability": 0.72,
          "estimated_time_to_failure_hours": 6,
          "recommended_action": "Cek AC server room sekarang"
        }
      },
      {
        "id": "pompa_a",
        "name": "Pompa A - Utility",
        "category": "production",
        "type": "pump",
        "status": "healthy",
        "health_score": 94,
        "metrics": {
          "vibration_rms": 1.8,
          "temperature": 62.0,
          "flow_rate": 45.2,
          "pressure": 4.5
        },
        "ai_predictions": {
          "breakdown_probability": 0.05,
          "estimated_time_to_failure_hours": 2160,
          "rul_hours": 4320
        }
      }
    ],
    "optional_modules": {
      "oee_monitoring": {
        "available": true,
        "price_per_month": 2000000,
        "description": "Add-on module untuk klien mature"
      },
      "traceability": {
        "available": true,
        "price_per_month": 1500000,
        "description": "Add-on untuk tuntutan principal"
      },
      "quality_analytics": {
        "available": true,
        "price_per_month": 2500000,
        "description": "Add-on untuk analitik kualitas"
      }
    }
  },

  "notifications": [
    {
      "id": "notif_001",
      "level": "critical",
      "category": "assetguard",
      "title": "🛡️ Predictive Maintenance Alert",
      "message": "Motor B: Prediksi breakdown dalam 48 jam",
      "ai_generated": true,
      "ai_recommendation": "Ganti bearing dalam 24 jam — hindari downtime 4 jam",
      "timestamp": "2026-09-10T08:25:00+07:00"
    },
    {
      "id": "notif_002",
      "level": "warning",
      "category": "assetguard",
      "title": "🛡️ Asset Health Alert",
      "message": "Server Room: Suhu 28°C — risiko overheating",
      "ai_generated": true,
      "ai_recommendation": "Cek AC server room sekarang",
      "timestamp": "2026-09-10T08:20:00+07:00"
    }
  ]
}