Вы находитесь на странице: 1из 8

Optimal Control Modeling

for Energy Cost Reduction


Development Background and Goals
A Building System has a lot of data points including control points, sensing points and
virtual points.
We want to predict the 24-hour operation schedule of control points by learning the
movement(or correlation) of 15-minute time series values of each point.

• Conventional operating method of the building manager


Typical building managers control plant systems with empirical on / off times and
fixed set-point temperatures independent of forecast weather, base energy rates,
and user satisfaction.

• Improvement Goals
- Optimize control efficiency based on weather and energy cost
- Energy savings without user complaint
- Replace building manager work with automatic schedule

2
Comfort Zone

The internal temperature should be


maintained at the comfort ◆ External Temperature
temperature during the place use ◆ External Humidity
time Comfort
Place

◆ Internal Temperature
◆ On/Off ◆ Set Temperature ◆ Internal Humidity

Facility

Comfort
Temperature
Temperature

Comfort Zone

Time

00:00 24:00
schedule to use place

3
Control Schedule Prediction (simple model)

AI Purpose
Case: Summer - To maintain comfort (during use)
Comfort Temperature ≒ Internal Temperature
External Temperature - To reduce total energy cost (24-hour)

AI Output
- Control Schedule (24-hour)
Temperature

Internal Temperature

Time
00:00 24:00
Schedule to use place

On/Off
Control Points
Set Temperature

Schedule to use place

Internal Temperature

Forecast Weather Data


(External Temperature)

Energy consumption

Energy Unit Price


4
Control Facility & Comfort Place Definition

Type of Control Facility


- AHU
- FCU
- PAC
- VAV
- Chiller
- Cooling Tower
- Heat Exchanger

Comfort Place
Independent user space

Number of each facility and


comfort place is building
dependent.
Each facility is operated by
internal logic based on on/off
and set temperature.

5
AI Input Data (15min time series data)
Chiller On/Off
Cooling Tower On/Off
On/Off Control Point Chiller
Chiller Set Temperature
Set Temperature Control Point
Cooling Tower Set Temperature
AHU Set Humidity

Internal Temperature
Comfort Place
Internal Humidity On/Off Heat
Control Point
Heat Exchanger Set Temperature Exchanger
On/Off
Control Point
FCU Set Temperature

Internal Temperature Comfort Place Power Consumption


Sensing Point Energy
Gas Consumption

On/Off

Set Temperature Control Point

PAC Set Humidity External Temperature


External
Sensing Point
Internal Temperature External Humidity Temp/Hum
Comfort Place
Internal Humidity

Weekdays / Holidays

Set Temperature Control Point Schedule to use comfort place Virtual


VAV Energy Unit Price Point
Internal Temperature Comfort Place
Comfort Temperature

6
AI Output Data (15min time series Control Schedule)

On/Off

AHU Set Temperature Control Point

Set Humidity

Chiller On/Off
On/Off
Cooling Tower On/Off
FCU Control Point
Set Temperature Control Point Chiller
Chiller Set Temperature

Cooling Tower Set Temperature

On/Off On/Off Heat


Control Point
PAC Set Temperature Control Point Heat Exchanger Set Temperature Exchanger
Set Humidity

VAV Set Temperature Control Point

7
Medium-sized building example
Number of Facility
AHU: 15 FCU: 30 PAC: 2
VAV: 25 Chiller: 3 Heat Exchanger: 2

AI Input Data Point


AHU: 75p FCU: 90p PAC: 10p
VAV: 50p Chiller: 12 Heat Exchanger: 4
Other Virtual Points: 9p
Total: 250p

AI Output Data Point (control point)


AHU: 45 FCU: 60 PAC: 6
VAV: 25 Chiller: 12 Heat Exchanger: 4
Total: 152p

Comfort Place: 72

250p Input Data Your Logic 152p Control Schedule


(15min time seriese) (15min time seriese for next day)

DynamoDB DynamoDB

AWS Serverless Platform

We are preparing our database that you can dynamically model for each building.
You should implement logic that predicts energy optimization schedules while maintaining comfort.

Вам также может понравиться