CN-122022478-A - Intelligent diagnosis system for full-period carbon sink loss risk of highway greening engineering

CN122022478ACN 122022478 ACN122022478 ACN 122022478ACN-122022478-A

Abstract

The invention relates to a full-period carbon sink loss risk intelligent diagnosis system for road greening engineering, which comprises a full-period multi-source perception module, a space-time data fusion and processing module, an intelligent diagnosis and prediction module, a visual interaction and decision support module and a control module, wherein the full-period multi-source perception module is used for acquiring heterogeneous data of the whole period from planning design to operation maintenance of the road greening engineering, the space-time data fusion and processing module is used for constructing a space-time knowledge graph based on the heterogeneous data, the intelligent diagnosis and prediction module is used for executing carbon sink dynamic metering, multi-factor risk diagnosis and maintenance decision optimization based on the space-time knowledge graph to acquire carbon sink data, risk probability and maintenance scheme, and the visual interaction and decision support module is used for generating digital twin visual billboard, risk early warning thermodynamic diagram and decision deduction information based on the carbon sink data, and receiving maintenance feedback data to optimize the space-time knowledge graph. The invention realizes the active, accurate and intelligent management of the greening carbon sink capacity of the highway.

Inventors

HE XINPING
ZHU JIAXUAN
XIAO YIQIAN
CUI PENGFEI
WU XIAOGANG
YAN XIANGYANG
You Tianzhou
ZHOU JIAN
LIU JIN
ZHANG YUE

Assignees

中国公路工程咨询集团有限公司
中咨华科交通建设技术有限公司
中咨数据有限公司
山西农业大学

Dates

Publication Date: 20260512
Application Date: 20260131

Claims (9)

1. Intelligent diagnosis system of highway greening engineering full cycle carbon sink loss risk, its characterized in that includes: the full-period multisource perception module is used for collecting heterogeneous data of the whole period from planning and design to operation and maintenance of road greening engineering; The space-time data fusion and processing module is used for constructing a space-time knowledge graph based on the heterogeneous data; The intelligent diagnosis and prediction module is used for executing carbon sink dynamic measurement, multi-factor risk diagnosis and maintenance decision optimization based on the space-time knowledge graph to acquire carbon sink data, risk probability and maintenance scheme; and the visual interaction and decision support module is used for generating a digital twin visual billboard, a risk early warning thermodynamic diagram and decision deduction information based on the carbon sink data, the risk probability distribution and the maintenance scheme, and receiving maintenance feedback data to optimize the space-time knowledge map.
2. The intelligent highway greening engineering full-cycle carbon sink loss risk diagnosis system according to claim 1, wherein the full-cycle multi-source perception module comprises: the system comprises a static data sensing unit, a planning and designing unit and a control unit, wherein the static data sensing unit is used for acquiring a BIM model and GIS data in a planning and designing stage, and the BIM model provides a three-dimensional digital model of greening engineering and comprises plant type, specification, initial biomass, planting density and space layout information; the dynamic real-time data sensing unit is used for deploying a dynamic monitoring network integrating space, day and ground in the project operation period, and acquiring real-time dynamic data based on the dynamic monitoring network; the manual recording unit is used for recording maintenance logs, wherein the maintenance logs comprise irrigation, fertilization, pruning and pest control time, frequency and application amount.
3. The intelligent highway greening engineering full cycle carbon sink loss risk diagnosis system according to claim 2, wherein the dynamic monitoring network comprises: The system comprises a ground network monitoring subunit, a monitoring subunit and a monitoring subunit, wherein the ground network monitoring subunit is used for sectionally deploying a large number of low-power-consumption Internet of things sensor nodes along a highway green belt and monitoring soil key parameters in real time, and the soil key parameters comprise volume water content, soil temperature, conductivity, pH value and nutrient content; the air network monitoring subunit is used for carrying hyperspectral cameras, multispectral cameras and laser radars by adopting an unmanned aerial vehicle and collecting inversion vegetation indexes and point cloud data, wherein the point cloud data comprise a canopy structure, a height and a volume; The space network monitoring subunit is used for accessing the public meteorological data interface and the commercial satellite remote sensing data to acquire regional meteorological information and vegetation growth trend.
4. The intelligent highway greening engineering full-cycle carbon sink loss risk diagnosis system according to claim 1, wherein the space-time data fusion and processing module comprises: the preprocessing unit is used for cleaning, normalizing, unifying formats and aligning time and space the heterogeneous data; And the space-time knowledge graph construction unit is used for constructing a space-time knowledge graph based on the preprocessed data.
5. The intelligent highway greening engineering full-cycle carbon sink loss risk diagnosis system according to claim 4, wherein constructing the spatiotemporal knowledge graph comprises: Defining various entities and events in a highway greening system as nodes in a knowledge graph, wherein the entities comprise each plant individual, a sensor, a soil sample point and a highway pile number section, and the events comprise one irrigation and one disease burst; defining the relationship between the nodes as edges, wherein the relationship between the nodes comprises a spatial proximity relationship, a causal relationship and a time sequence relationship; The spatiotemporal knowledge-graph is constructed based on all nodes and edges.
6. The intelligent diagnosis system for the loss risk of carbon sink in the whole period of highway greening engineering according to claim 1, wherein the intelligent diagnosis and prediction module comprises: the carbon sink dynamic metering unit is used for calculating the daily carbon fixed quantity, carbon emission and net carbon sink quantity of each road section by adopting a carbon sink dynamic metering model based on machine learning correction based on full-period heterogeneous data; The multi-factor risk diagnosis and early warning unit is used for acquiring the probability of each road greening management unit for generating a preset risk in a future period by adopting a multi-factor risk diagnosis and early warning model based on a graph convolution-long-short-term memory network based on the space-time knowledge graph, wherein the preset risk comprises drought stress risk, nutrient stress risk, pest and disease damage risk and soil loss risk; and the maintenance optimization unit is used for acquiring an optimal maintenance scheme based on the probability of occurrence of the preset risk and the full-period heterogeneous data.
7. The intelligent diagnosis system for the total-period carbon sink loss risk of the highway greening engineering according to claim 6 is characterized in that a hybrid modeling method is adopted in the carbon sink dynamic metering unit, an ecological process model with a definite mechanism is combined with a data-driven machine learning model, wherein the ecological process model provides a basic framework of carbon circulation, and the machine learning model utilizes collected total-period heterogeneous data to dynamically calibrate key parameters in the ecological process model and correct residual errors.
8. The intelligent diagnosis system for the loss risk of the carbon sink in the whole period of the highway greening engineering according to claim 6, wherein the multi-factor risk diagnosis and early warning model in the multi-factor risk diagnosis and early warning unit comprises a graph convolution network, a long-term and short-term memory network and a full connection layer; The graph convolution network is used for aggregating the information of neighbor nodes along the edges of the space-time knowledge graph, capturing the spatial relevance and transmissibility of risk factors and extracting spatial features; The long-period and short-period memory network is used for capturing the dynamic evolution rule and long-period dependence of data in the time dimension based on the time sequence data of each node and extracting time characteristics, wherein the time sequence data comprises soil humidity and vegetation index change which are continuous for a plurality of days; And the full-connection layer is used for fusing the spatial characteristics and the time characteristics, and finally outputting the probability of occurrence of specific risks of each road greening management unit in a future period of time through a Softmax function.
9. The intelligent diagnosis system for the total cycle carbon sink loss risk of the highway greening engineering according to claim 6 is characterized in that the maintenance optimization unit models a maintenance decision process as a reinforcement learning problem, the state is risk probability output by the multi-factor risk diagnosis and early warning unit and multi-source perception total cycle heterogeneous data, the action is maintenance measures, the reward is a comprehensive utility function, the aim of maximizing the expected carbon sink increment is achieved, the maintenance cost and the negative environmental influence are minimized at the same time, the intelligent body continuously learns in the interaction with the environment through a deep Q network algorithm, and finally an optimal decision strategy is formed, namely, the optimal maintenance action can be recommended under any given system state.

Description

Intelligent diagnosis system for full-period carbon sink loss risk of highway greening engineering Technical Field The invention relates to the technical field of intersection of environmental engineering and intelligent traffic systems, in particular to an intelligent diagnosis system for full-period carbon sink loss risk of highway greening engineering. Background At present, in the global context of coping with climate change and achieving the objective of "carbon neutralization", road greening is becoming increasingly important as an important component of traffic infrastructure, and its ecological function, in particular carbon sequestration, is becoming more important. However, there are still significant limitations and challenges in managing and risk controlling the benefits of highway greening engineering carbon sinks, mainly in the following ways: The fragmentation and discontinuity of view angles are managed. Existing highway greening management modes are typically split. In the planning and design stage, the landscape effect, the driving safety and the water and soil conservation function are mainly considered, the survival rate of seedlings and the engineering cost are focused in the construction stage, and the daily watering, fertilization, trimming and the like are mainly used in the operation and maintenance stage. This model lacks a perspective of carbon sink performance assessment and optimization throughout the project's full life cycle. Decision makers cannot systematically understand how design parameters (e.g., plant species, deployment density) affect long-term carbon sink potential, nor can they feed back actual carbon sink performance at the operational stage into future design optimization, resulting in inadequate exploitation of the maximized potential of carbon sink benefit. Limitations of carbon sink monitoring means and data quality problems. Traditional highway greening carbon sink monitoring mainly relies on a sample plot checking method, namely, biomass is estimated by periodically measuring parameters such as breast diameter, tree height and the like of plants in a sample plot and utilizing a different-speed growth equation, so that carbon reserves are estimated. The method has the following defects of high labor intensity, high cost and low updating frequency, and cannot meet the real-time requirement of fine management. Secondly, the road green belt is long and narrow and has high environmental heterogeneity, and the limited sample area is difficult to represent the whole condition, so that the sampling error is obvious. In addition, various error sources exist in the data acquisition process, such as instrument errors, human measurement errors, environmental factor interference (such as wind speed influence measurement) and the like, and the "non-sampling errors" affect the accuracy of the data together. Although the remote sensing technology provides a macroscopic monitoring means, when the remote sensing technology is applied to road greening with small-scale linear distribution, the resolution and the precision are often limited, and carbon information of lower vegetation and soil is difficult to obtain by penetrating through a canopy. The lack of risk diagnosis capability and passive maintenance are that most of current road greening maintenance is 'passive response' management. Maintenance personnel usually take measures after obvious symptoms such as yellow leaves, wilting, lesions and the like of plants are observed. However, when these macroscopic symptoms occur, physiological stress inside plants has persisted for some time, carbon sink capacity has been lost, and even irreversible damage may have been caused. The prior art lacks an effective means to diagnose and pre-warn of potential carbon sink loss risks in advance. For example, a decrease in photosynthetic efficiency of plants caused by soil hardening, nutrient imbalance, or mild water stress, which is a core process of carbon fixation, cannot be recognized at an early stage. This delayed management mode not only results in impaired carbon sequestration benefits, but also increases the cost of later repair. Experience and inefficiency of decision support in the field of highway greening maintenance decisions (such as irrigation amount, fertilization type and time) have greatly relied on the experience of maintenance workers or on intelligent irrigation systems mentioned by fixed work plans, but most are based on simple automation of threshold values, lacking in prediction of future risks. The decision mode based on one cut or subjective judgment cannot be used for accurately matching the actual demands of different road sections and different plant types at a specific time point. This may not only lead to waste of resources such as water and fertilizer, increase maintenance cost and indirect carbon emission, but also may negatively affect plant health and carbon sink function due to improper maintenance (e.