25 Metro lines planned
>200 kilometers with >130 stations have been built since 2009
Yangzi river: longest river in Asia and the third-longest in the world
Chicago of China
10.6 million people

Wuhan

Data collection

Geotechnical survey
Construction reports
Site photos
...

Semantic enrichment

Ma, L., H.B. Luo, and H.R. Chen, Safety risk analysis based on a geotechnical instrumentation data warehouse in metro tunnel project. Automation in Construction, 2013. 34: p. 75-84.

Data de-noise

Ding, L., L. Ma, H. Luo, et al., Wavelet Analysis for tunneling-induced ground settlement based on a stochastic model. Tunnelling and Underground Space Technology, 2011. 26(5): p. 619-628.

Settlement distribution and development

Ma, L., L. Ding, and H. Luo, Non-linear description of ground settlement over twin tunnels in soil. Tunnelling and Underground Space Technology, 2014. 42(0): p. 144-151.

Guo, J., L. Y. Ding, H. B. Luo, C. Zhou and L. Ma, Wavelet prediction method for ground deformation induced by tunneling. Tunnelling and Underground Space Technology, 2014. 41(0): p. 137-151.

Knowledge discovery from the data

Ma, L., H.B. Luo, and H.R. Chen, Safety risk analysis based on a geotechnical instrumentation data warehouse in metro tunnel project. Automation in Construction, 2013. 34: p. 75-84.

The settlement trough width is correlated with the thickness of soil layers

The settlement trough width is time invariant

Safety classification system

Identification of settlement outlier by Shewhart charts

Area classification zones of ground lose

Dynamic safety classification of ground area and adjacent facilities

Transformation among Data, Information and Knowledge

Data

Information

Knowledge

An integrated data mining system for safety classification of tunneling construction

Data

Information

Knowledge

Semantic enrichment for BIM - SeeBIM

Why Enrichment is Important to BIM

Enrichment Tasks

Previous work

SeeBIM 1.0

Michael Belsky, PhD, “A FRAMEWORK FOR SEMANTIC ENRICHMENT OF IFC MODELS”, Civil and Env. Eng., Technion. 2015

Infravation - SeeBridge

Sacks, R., Kedar, A., Borrmann, A., Ma, L., Singer, D., and Kattel, U., (2016). ‘SeeBridge Information Delivery Manual (IDM) for Next Generation Bridge Inspection’, ISARC 2016 33rd International Symposium on Automation and Robotics in Construction, Sattineni, A. (ed.), Auburn University, Auburn AL, US, July 2016.

Rule engine of SeeBIM 1.0

Rules:

IF a & b -> X
IF b & c -> Y

Advantage:

No computer programing skill is required

Disadvantage:

Lack of rigor

Fact:

IF a & b & c -> ?

SeeBIM 2.0

Features	F1	F2	F3	F4	F5	F6
Type A	y	y	x	y	n	n
Type B	n	n	y	x	y	n
Type C	y	y	n	y	x	n
Type D	n	n	y	y	y	x

‘y’:= Yes
‘n’:= No
‘x’:= Not always

Find the 'unique' string (skip the comparison with ambiguous 'x' )

y	n	y	x	n
y	x	y	y	n

Not unique

SeeBIM 2.0 - pairwise

Physical contact ?

Orthogonality

Length comparison

Width comparison

Elevation comparison

...

Find the 'unique' string (skip the comparison with ambiguous 'x' )

‘y’:= Yes;
‘n’:= No;
‘x’:= Not always;

y	n	y	x	n
y	x	y	y	n

Not unique

SeeBIM 2.0 - an example rule set for bridge object classification

Sacks, R., L. Ma, R. Yosef, A. Borrmann, S. Daum. and U. Kattel, Enhancement of a Semantic Enrichment Tool for Building Information Modeling. Journal of Computing in Civil Engineering, under review.

SeeBIM 2.0 - another thought

	Feature1	Feature2	Feature3	Feature4	Feature5
Type A	n	y	X	n	y
Type B	n	y	y	n	y

‘y’:= Yes;
‘n’:= No;
‘x’:= Not always;

Rules:

Fact:

Obj 1	n	y	y	n	y
Obj 2	n	y	n	n	y

PB>PA

PA>PB

Feature Matching

	Longitudinal	Transverse	Vertical	Convex	...
Primary Girder	1	-1	-1	-1
Transverse Beam	-1	1	-1	1
...

Primary Girder = (1,-1,-1,-1)

	Longitudinal	Transverse	Vertical	Convex
Object 1	-1	1	-1	1
Object 2	1	-1	-1	-1
Object 3	1	-1	-1	1

Object 1 = (-1,1,-1,1)

	Primary Girder	Transverse Beam
Object 1	180 degree	0 degree
Object 2	0 degree	180 degree
Object 3	60 degree	90 degree

Pairwise Feature Matching

	f1	f2	f3
E1,E1	1	1	1
E1,E2	-1	0	0
E1,E3	-1	-1	0
E2,E1	0	0	0
E2,E2	1	1	0
E2,E3	0	0	-1
E3,E1	-1	-1	-1
E3,E2	0	-1	0
E3,E3	0	0	-1

	f1	f2	f3
O1,O2	1	1	1
O1,O3	-1	1	-1
O1,O4	-1	-1	1
O1,O5	1	1	1
O2,O1	1	1	1
O2,O3	-1	-1	1
O2,O4	1	1	1
O2,O5	1	1	1
O3,O1	1	1	1
O3,O2	-1	-1	1
O3,O4	-1	-1	1
O3,O5	1	1	-1
O4,O1	1	1	-1
O4,O2	1	1	-1
O4,O3	-1	-1	-1
O4,O5	-1	-1	1
O5,O1	-1	1	-1
O5,O2	-1	1	-1
O5,O3	-1	1	-1
O5,O4	1	-1	1

3 types of bridge elements (e.g., girder, sheer key etc.)

5 objects in the model

3 types of features

	E1,E1	E1,E2	E1,E3	E2,E1	E2,E2	E2,E3	E3,E1	E3,E2	E3,E3
O1,O2	1	3	-3	1	-5	2	-3	2	1
O1,O3	-5	1	-3	2	-3	-3	2	2	-2
O1,O4	-2	3	-3	5	3	-1	-5	-2	-2
O1,O5	-1	5	1	5	2	1	3	2	1
O2,O1	-3	4	2	0	1	-1	-4	3	-5
O2,O3	-1	3	5	-5	0	0	-3	4	-2
O2,O4	1	3	-4	4	0	4	-1	5	-5
O2,O5	-5	-2	-1	0	1	3	2	0	1
O3,O1	-3	5	0	-2	-3	1	5	-1	5
O3,O2	4	1	-2	3	-2	3	1	0	-1
O3,O4	-2	5	-2	0	-5	4	5	2	-2
O3,O5	-2	-5	-4	3	-2	-3	1	1	-2
O4,O1	-4	1	0	-4	4	-5	-5	1	-4
O4,O2	5	2	5	-5	3	-4	-4	-2	-5
O4,O3	-5	0	3	1	4	-1	2	5	1
O4,O5	-1	-3	3	-5	3	2	-5	-5	-5
O5,O1	1	5	2	5	1	5	0	0	1
O5,O2	-3	-4	0	4	5	0	1	-4	2
O5,O3	-5	0	5	3	-1	4	2	4	2
O5,O4	3	5	-5	5	1	-4	-1	3	2

obj1?=element1

2n(m-1) relations

m : n

SeeBIM 2.0b - example result

All the 331 objects in this bridge model are correctly identified

Ma, L., R. Sacks, U. Kattel and T. Bloch, Building Model Object Classification using Geometric Features and Pairwise Spatial. Computer-aided Civil and Infrastructure Engineering, in preparation.

SeeBIM 3.0

Machine learning:

Classifier: 3 * Feature1 ^ Feature2 + Feature3 / Feature4

	Feature1	Feature2	Feature3	Feature4	Type
Obj 1	1	0	20.5	a	Type A
Obj 2	-1	1	40	b	Type B

	Feature1	Feature2	Feature3	Feature4	Type
Obj 1,001	...	...	...	...	Type ?
Obj 1,002	...	...	...	...	Type ?

SeeBIM 4.0?

Object	Type
Obj 1	Type A
Obj 2	Type B
Obj 3	Type C

Deep learning:

Object	Type
Obj 10,001	Type ?
Obj 10,002	Type ?
Obj 10,003	Type ?

Enrichment of other information

Supplementing information:

Bridge grids
Span
Assembly

Filling in missing objects or fixing their shape caused by occlusion. such as:

Bearings
Girders
Transverse beams

Transformation among Data, Information and Knowledge

Data

Information

Knowledge

Semantic enrichment for BIM - SeeBIM

Data

Information

Knowledge

Automated generating BIM models of earthquake damaged buildings

Modeling damaged reinforced concrete objects

Ma, L., R. Sacks, and R. Zeibak-Shini, Information modeling of earthquake-damaged reinforced concrete structures. Advanced Engineering Informatics, 2015. 29(3): p. 396-407.

Specimen

Four-story Reinforced Concrete and Post-Tensioned Concrete Buildings, December 2010 (from E-defense)

National Building Research Institute @ Technion

Data!!!

By Jiashu Liang

By Ashrant Aryal

Ma, L., R. Sacks, R. Zeibak-Shini, et al., Preparation of Synthetic As-Damaged Models for Post-Earthquake BIM Reconstruction Research. Journal of Computing in Civil Engineering, 2016. 30(3): p. 04015032.

Laser scanner emulator

Ma, L., R. Sacks, R. Zeibak-Shini, et al., A Computational Procedure for Generating Specimens of BIM and Point Cloud Data for Building Change Detection, in Computing in Civil Engineering 2015. 2015, American Society of Civil Engineers. p. 684-691.

Experiment setup

Ma, L., R. Sacks, R. Zeibak-Shini, et al., Preparation of Synthetic As-Damaged Models for Post-Earthquake BIM Reconstruction Research. Journal of Computing in Civil Engineering, 2016. 30(3): p. 04015032.

Structure transformation

Mapping structural cells
Locating the building elements in the scan
Compilation of as-damaged BIM model

Zeibak-Shini, R., R. Sacks, L. Ma, et al., Towards generation of as-damaged BIM models using laser-scanning and as-built BIM: First estimate of as-damaged locations of reinforced concrete frame members in masonry infill structures. Advanced Engineering Informatics, 2016. 30(3): p. 312-326.

Transformation among Data, Information and Knowledge

Synthetic Data

Information

Knowledge

Automated generation BIM model of earthquake damaged building

Synthetic Data

Information

Knowledge

Investigation of construction workflow using Agent-based simulation

Previous work

Lola Ben-Alon, MSc, “Simulating and Visualizing the Flow of Trade Crews using Agents and Building Information Models (BIM)”, Civil and Env. Eng., Technion. 2015

Relational model based simulation design

Ma, L. and Sacks, R. (2016). ‘Agent-based simulation of construction workflows using a relational data model’ Proceedings of the 24th Annual Conference of the International Group for Lean Construction, Pasquire, C., Alves, T., and Reginato, J., (eds.), Boston, MA, US, July 2016, pp. 73-82.

Simulation explaination

Result (no information exchange)

Result (information flow vs work flow)

Transformation among Data, Information and Knowledge

Data

Information

Knowledge

Different problems

Research methods
Data availability
Controlled experiments
Verifiability