Intelligent generation method of emergency plan for hydraulic engineering based on knowledge graph – take the South-to-North Water Diversion Project as an example

Figures & data

Figure 1. The three routes of the South-to-North Water Transfer Project, with the blue line representing the western route of the South-to-North Water Transfer Project, the red line representing the middle route of the South-to-North Water Transfer Project and the pink line representing the eastern routeof the South-to-North Water Transfer Project.

Figure 2. The process of construction of an emergency plan knowledge map for the middle route of the South-to-North Water Transfer Project.

Figure 3. Knowledge graph schema of the emergency plan of the middle route of the South-to-North Water Transfer Project.

Table 1. Entity examples from the Risk Prevention and Control Manual.

Entity category	Entity instance
Engineering	The Yi County high fill channel, the Fushan tunnel, the Jingkeshan bleeder, the Donglou Mountain Xigou drainage inverted siphon
Risk event	The erosion instability of the pipe body, the reduction of the flow capacity, the flood-soaked channel slope, and the upper channel foundation damage
Address	Dengzhou city, Taocha county, Nanyang city, Hui county
Management office	Daning Management Office, West Fourth Ring Management Office, Weihui Management Office, Zhenping Management Office

Table 2. Entity examples of the overall scheme of the emergency rescue system.

Entity category (keyword)	Entity instance (short text)
Equipment	Backhoe, manual auxiliary tools; auger, grouting machine, manual auxiliary operation tools.
Materials	The plastic waterproof film, reinforcement, cement, natural sand and stone.
Storage location	The plastic waterproof film is stored in the management office emergency materials warehouse, and cement, natural sand, and stone materials are purchased locally.
Risk consequence	Serious deformation forms hidden danger of landslides; the small-scale landslide forms the outburst of hidden danger; massive landslides block channels.
Engineering risk part	Top of channel slope; middle channel slope; near Grade 1 bridleway; the outer slope of the building.

Figure 4. The pipeline of intelligent emergency plan generation for the middle route of the South-to-North Water Transfer Project.

Figure 5. The BERT+BiLSTM+CRF model framework.

Figure 6. The emergency plan template.

Table 3. Some parameters of the bidirectional encoder representation from transformers (BERT) and bidirectional long-short-term memory with conditional random felds (BiLSTM+CRF) model.

Experimental parameter	Meaning	Training parameter value
Epoch	Number of training wheels	50
Max_len	The maximum length of the sentence to be entered	128
Batch_size	Number of sentences per training session	16
Lstm_units	Number of neurons in the BiLSTM monolayer	128
Drop_rate	Random loss rate	0.1
Learning_rate	Training initial learning rate	5e-5
Optimiser	Optimiser	Adam

Table 4. Statistics on entity extraction of the middle route of the South-to-North Water Transfer Project.

Entity type	Number of entities	Entity type	Number of entities
Engineering	3353	Management office	47
Risk event	411	Material	25
Risk factor	222	Facility	35
Risk value	95	Storage location	6
Prevention measure	640	Risk consequence	101
Control measure	680	Engineering risk part	102
Address	47	Material point	188
Risk level	95

Table 5. Statistics on extraction of relationships for the middle route of the South-to-North Water Transfer Project.

Entity type	Relationship type	Entity type	Number of relationships
Engineering	Contain risk event	Risk event	13,162
Engineering	Value	Risk value	3567
Engineering	Grade	Risk level	3567
Engineering	Be located in	Address	3599
Address	Contain material point	Material point	188
Engineering	Subordinate to	Engineering	3600
Engineering	Belong to	Management office	3599
Risk event	Contain risk factor	Risk factor	8718
Risk event	Lead to	Risk consequence	411
Risk event	Relative event	Risk event	564
Risk event	Happen to	Engineering risk part	102
Engineering	Engineering part	Engineering risk part	102
Risk event	Be controlled	Control measure	2918
Control measure	Required equipment	Facility	92
Control measure	Required material	Material	92
Facility	Be stored in	Storage location	92
Material	Be stored in	Storage location	92
Risk factor	Be prevented	Prevention measure	8487

Figure 7. Part entities and relationships of the knowledge map.

Table 6. Model training results on bidirectional encoder representation from transformers (BERT) and bidirectional long-short-term memory with conditional random felds (BiLSTM+CRF).

Label	Precision	Recall	F1 score
Engineering	0.9768	0.9756	0.9762
Risk event	0.9946	0.9956	0.9951
Risk factor	0.9353	0.9298	0.9321
Engineering risk part	0.9278	0.9296	0.9367
Stake number	0.9384	0.9325	0.9354
Management office	0.9763	0.9788	0.9775
Address	0.9449	0.9190	0.9318
Risk value	1.000	1.000	1.000
Average value	0.9626	0.9615	0.9621

Table 7. Three groups of algorithm evaluation results (%).

Algorithm	Precision	Recall	F1
Word2Vecc	86.67	69.53	77.16
Jaccard	88.667	73.43	80.33
Word2Vec+Jaccard	90.33	77.34	83.32

Table 8. The matching situation of some risk entities on the knowledge graph.

Risk_1	Risk_2	Returned risk	Returned risk	Returned risk	Returned risk	Returned risk
Terrorist attack	Terrorist attack	Terrorist attack	Drone attacks	Terrorist hijacking	Control system failure	The drainage aqueduct overflowed
Debris flow silted up the channel	The excavation channel is blocked	Channel congestion	Canal deposition	The pipeline siltation	The underground pipeline is silted up	Silt-clogged valve well
Massive landslide blockage	The excavation channel is blocked	The excavation channel is blocked	The channel slope collapsed and slid	Channel slope cracking	The slope cracks and collapses	The dike slope collapsed and slid
The seepage control system of the canal slope is destroyed	Leakage from underpass pipe	Leakage from underpass pipe	Leakage of the first slope of the dike	Leakage of the wrapped head slope	Steel pipe leakage	Erosion of channel slope
Slope protection is damaged	Failure of slope protection cracks	Failure of slope protection cracks	Improper operation, maintenance and construction	Erosion of channel slope	Seepage failure of the dike	Damage of auxiliary facilities

Table 9. List of risk candidate entities.

Candidate entities
Culvert body leakage	Structure leakage	Control gate leakage	Surge tank leakage
Box culvert leakage	Concentrated site leakage	Surge tank leakage	Seepage and instability of the foundation
Seam leakage	Box culvert leakage	Tunnel lining leakage	Leakage of pipeline bridge
Waterproof leakage	Structural seam leakage	Check the water leakage	Concentrated leakage, soil flow
Canal seepage	Open trough leakage	Water conservation weir leakage	Leakage of lock chamber

Figure 8. Emergency plan page.

Data availability statement

The data that support the findings of this study are available from the corresponding author, H. K. Lu, upon reasonable request https://github.com/luhankang/emergency_plan