Before01
Company overview
- HVAC drives energy use
- Carbon data demanded
- Uneven manual control
OPTIMIZATION
Energy efficiency optimization
Energy-cost optimization realized through data-driven intelligent control
OVERVIEW
Not just measuring what energy is used —controlling it down to the least possible
Beyond watching consumption: the operating conditions of the equipment itself come into the AI’s control range.
Sample scenario
Daily power load (kW)
- Current operation
- AI-optimized(low → high)
- Off-peak
- Mid-peak
- On-peak
KEY FEATURES
Four capabilities that optimize energy
From real-time monitoring through equipment efficiency and demand forecasting to carbon management.
Capability Index04
Energy Capability
01 / 04
01
Real-time energy monitoring
Equipment efficiency optimization
Energy demand forecasting
Integrated carbon emissions management
It collects and visualizes the power consumption of key equipment and lines across the plant in real time. Energy consumption is grasped at a glance so waste is caught immediately, and peak-demand periods are managed to realize cost savings.
Management Scope
- Per-equipment power collection
- Real-time consumption board
- Peak demand control
- Abnormal consumption alerts
- Energy data retention
- Period-over-period analysis
Real-time energy monitoring
Tracks electricity, gas, steam and other energy use across the whole plant and its main equipment in real time
WORKFLOW
The workflow behind each capability
For each capability, the order it runs in and the check it passes before moving on.
Reading the production plan alongside energy demand
- 01
History collection
hourly usage
- 02
Context linking
production · weather · operating condition
- 03
Demand forecast
load · peak risk
- 04
Actual comparison
forecast error · baseline calibration
Recorded alongside
01
Energy/
02
Operating/
03
Forecast/
Operating check
Is the current condition within the training range?
YES
Within range → demand outlook
NO
New condition → uncertainty · re-review
An example workflow · the detailed links and criteria are set to suit your floor.
SYSTEM ARCHITECTURE
What it connects,and the work it leads to
From the data that goes in to the results on the floor.
01 / INPUT
01Energy historyPower · gas · tariff · peak
02Production conditionsVolume · item · run plan
03Operating environmentTemperature · load · set points
CONNECTED OPERATIONS
AI Energy
Forecast load and demand patterns
Find waste and idle running
Recommend operation and load spreading
03 / OUTPUT
✓Demand outlookLoad risk by time of day
✓Operating optionsStart-up order · set points
✓Improvement checkProduction-adjusted · unit energy comparison
01 / INPUT
01Energy historyPower · gas · tariff · peak
02Production conditionsVolume · item · run plan
03Operating environmentTemperature · load · set points
CONNECTED OPERATIONS
AI Energy
Forecast load and demand patterns
Find waste and idle running
Recommend operation and load spreading
CONNECTED OPERATIONS
AI Energy
Forecast load and demand patterns
Find waste and idle running
Recommend operation and load spreading
03 / OUTPUT
✓Demand outlookLoad risk by time of day
✓Operating optionsStart-up order · set points
✓Improvement checkProduction-adjusted · unit energy comparison
Linked by common identifiers, master data and historyResults checked → fed into the next run
Who uses itEnergy managersEquipment operatorsProduction planners
METHOD
How it is built
Floor energy data collected and analyzed, then returned to how the equipment runs.
Collection01 →
Data Ingestion
PLC
Modbus / OPC-UA
SENSOR
4-20mA / RS-485
IoT gateway
MQTT / HTTP
Load patternTemp · humidity / pressurePower consumptionOutput
Signal processing
Outlier removal · gap filling · unit aggregation
Time syncNTP
Aligns process and meter timestamps
Target System
Time-series DBInfluxDB, Timescale
Data lakeRaw waveform
MetadataLOT, work orders, equipment state
Modeling & analysis02 →
AI/ML Modeling
Analysis & prediction
Energy waste detection
XGBoost · LightGBM
Optimal scheduling
Reinforcement learning
Demand forecasting
LSTM AutoEncoder
Carbon emission forecast
Feedback & refinement
Model evaluation
Digital twin
MLOps retraining
Service03
Energy saving KPIs
Dashboard
KPI Report
Carbon neutrality report
Scope 1/2/3 emissions
Automatic ESG report
Alerts & guidance
Instant anomaly alerts / action guidance
Minimize unnecessary energy consumption and cut costs.
Power management
HVAC systems
Smart buildings
Renewable energy
Data centers
Lighting and utilities
Power management
HVAC systems
Smart buildings
Renewable energy
Data centers
Lighting and utilities
Energy is not used sparingly. It is used when it should be.
- Electricity and fuel use
- Equipment operating conditions
- Production plan
- Tariff and peak
- Renewable generation
AI Energy
- Per-machine energy intensity analysis
- Load and peak prediction
- Operating and switching scenario recommendation
- Energy intensity
- Peak demand
- Energy cost
- Renewable utilization
EFFECT
Business impact
What it leaves behind — in energy intensity, demand peaks, energy cost and renewable share.
- 01
Lower energy intensity
Energy per unit produced is broken down by process to find the waste, and optimal set-points per machine cut standby loss.
Intensitystandby lossset-points - 02
Lower demand peaks
Overlapping loads are predicted and spread across the schedule, so the contracted demand limit is not crossed.
Peak demandload spreadingcontract limit - 03
Lower energy cost
Peak charges and standby loss come down together, and tariff and run plan are matched so the same output costs less to power.
Electricitygasrun plan - 04
Higher renewable share
Generation forecast and load plan are matched so self-generated power is used first, and the renewable share is tracked with the certificates that prove it.
Renewable shareself-consumptioncertificates
CASE STUDY
Case study
Model predictive control cut HVAC energy by 15-30% while carbon data was managed alongside it.
Customer industry
High-layer-count PCB (MLB) manufacturing
After02
Implementation
- HVAC and climate data
- Predictive control
- PMV and VAV control
Business impact
Energy optimization where AI runs the HVAC on its own
- HVAC energy consumption
- Before 100 (baseline), After 70–85 (15–30% less)
Relative to pre-deployment energy use set at 100.