A permafrost region embankment along-line ecological corridor intelligent planning and biodiversity dynamic adaptive management system and method considering temperature disturbance

Abstract

The application discloses a permafrost region embankment along-line ecological corridor intelligent planning and biological diversity dynamic adaptive management system and method considering temperature disturbance, and the system comprises a multi-source remote sensing and field sensing collection module, an AI identification and species diversity space-time database module, a seasonal dynamic parameter configuration module, an animal migration path modeling and ecological corridor planning optimization module, a real-time animal activity monitoring and ecological connectivity evaluation module, a dynamic regulation and decision feedback module and a model online learning and parameter calibration unit. The application takes multi-depth geothermal measurement data as the basis for permafrost thermal state discrimination, and uses surface temperature measurement data to represent the instant moving environment of animals. Dynamic resistance surfaces of different ecological seasons are generated through seasonal dynamic parameters, and the permafrost thermal stability index is used as the engineering hard constraint of the candidate corridor scheme. In the operation period, the system performs dynamic evaluation and graded early warning according to the same period activity layer melting depth deviation, animal crossing monitoring, snow accumulation obstacles and settlement displacement state, and continuously calibrates the local resistance parameters by using the successful crossing events, so as to form a closed-loop management system of sensing-evaluation-regulation-learning, and realize the collaborative optimization of ecological connectivity and engineering thermal stability.

Description

Technical Field

[0001] This invention relates to an intelligent planning and biodiversity dynamic adaptive management system and method for ecological corridors along roadbeds in permafrost regions, taking into account temperature disturbances. Background Technology

[0002] In permafrost regions, the construction and operation of linear infrastructure such as highways and railways have created significant ecological barriers. Long-distance, continuous roadbeds disrupt the original migration routes of wild animals, damage the continuity of vegetation communities, and consequently lead to a series of negative ecological consequences, including population isolation, interruption of gene flow, habitat fragmentation, and decline in regional biodiversity. Simultaneously, because permafrost is highly sensitive to thermal disturbance, engineering activities alter the surface heat absorption and cold air convection conditions, disrupting the thermal equilibrium of the underlying permafrost. This triggers localized thaw subsidence, thermokarst development, and even thickening of the active layer. These thermal instabilities not only damage the engineering structures themselves but also further create new thermokarst lakes or muddy areas, exacerbating habitat fragmentation. Therefore, ecological protection and engineering stability maintenance in permafrost regions are not independent issues but rather deeply interconnected and coupled.

[0003] In existing research and engineering practice, technical breakthroughs in roadbed engineering in permafrost regions have long focused on thermal stability control. Currently, a technical system centered on "cooling conduction—heat insulation—temperature regulation" has been formed, including measures such as heat pipes, ventilation pipe roadbeds, rubble roadbeds, and XPS insulation layers, achieving significant results in many projects. Regarding ecological protection, the main approach is to rely on environmental impact assessments supplemented by mitigation measures, such as setting up fixed-width animal passages or culverts along the roadbed based on engineering experience or general standards. In recent years, some projects have also begun to emphasize preliminary biodiversity baseline surveys, attempting to investigate the distribution and migration routes of wild animals along the route through satellite remote sensing interpretation, infrared camera deployment, and on-site transect surveys, and introducing the concept of "avoidance priority" into passage design. However, existing ecological corridor designs along roadbeds in permafrost regions still have the following limitations:

[0004] First, current ecological corridors are essentially static and one-off. The location, width, and number of corridors are usually determined during the planning stage based on species distribution surveys and engineering feasibility studies, and they are permanently operational after construction. However, the active layer of permafrost regions undergoes cyclical freeze-thaw cycles every year, resulting in drastic and non-linear seasonal changes in surface bearing capacity, access conditions, and landscape patterns. Consequently, animal utilization patterns of corridors exhibit fundamental seasonal differences. In other words, the migration resistance surface for animals is not a static constant but dynamically evolves with the freeze-thaw process. Existing design methods cannot adjust corridor management strategies based on real-time temperature monitoring data, nor do they possess the ability to perceive and adaptively respond to temperature disturbances in real time. This leads to a significant deviation between the actual effectiveness of corridors and their design expectations in certain seasons.

[0005] Second, existing ecological corridor designs separate permafrost thermal stability analysis and ecological corridor design into two independent technical paths. The former serves the safety of the engineering structure, while the latter serves the mitigation of environmental impacts, and there is a lack of a quantitative coupling model between the two. The selection of corridor location, determination of width, and design of internal habitat structure fail to quantitatively assess the temperature field distribution and thermal disturbance tolerance threshold of the underlying permafrost as prerequisite constraints. This results in ecological corridors being placed in areas with poor thermal stability or in locations with low ecological connectivity value in order to ensure engineering stability. Consequently, after the roadbed is operational, the corridor structure fails due to thaw settlement and other reasons, weakening its intended ecological functions.

[0006] Third, currently, the assessment of the ecological connectivity restoration effect of ecological corridors mainly relies on manual transect surveys or periodic spot checks using infrared cameras. The data obtained are discontinuous in time series, making it difficult to capture the dynamic characteristics of animal behavioral responses and to form a continuous diagnosis of the ecological connectivity status. More importantly, the existing monitoring system lacks a closed-loop feedback mechanism between the monitoring system and corridor management and control decisions. The system cannot correct management measures based on animal activity monitoring results, nor can it perform correlation analysis between temperature monitoring data and animal behavior data to assess the impact of thermal changes on ecological connectivity.

[0007] Therefore, how to use the thermal stability of permafrost as a hard constraint for engineering throughout the entire life cycle of roadbed engineering in permafrost areas, achieve quantitative optimization of ecological corridor site selection and design, continuously monitor and dynamically evaluate the ecological connectivity of corridors, and further form adaptive control decisions that can be updated based on monitoring results, remains a technical problem that needs to be solved in this field. Summary of the Invention

[0008] To address the problems in existing ecological corridor planning along roadbeds in permafrost regions, such as the disconnect between thermal stability analysis and ecological corridor design, the inability of static corridor schemes to respond to seasonal dynamics during freeze-thaw cycles, and the lack of a closed-loop feedback mechanism between operational monitoring, evaluation, and control decisions, this invention provides an intelligent planning and biodiversity dynamic adaptive management system and method for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances. By differentiating the technical applications of various data such as multi-depth ground temperature, surface temperature, permafrost properties, and settlement displacement, dynamic resistance surfaces corresponding to multiple ecological seasons are constructed, and the permafrost thermal stability index is used as a hard engineering constraint for corridor scheme optimization. During the operational phase, continuous evaluation and adaptive control are conducted using information on the deviation of the active layer thaw depth, snow cover obstacles, animal crossings, and structural deformation, thereby achieving synergistic optimization of ecological connectivity and engineering thermal stability. The specific technical solution is as follows:

[0009] First, this invention proposes an intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, comprising:

[0010] The multi-source remote sensing and field perception acquisition module is configured to acquire multi-source perception data covering the area along the proposed roadbed. The field data includes at least satellite remote sensing images, UAV remote sensing images, acoustic audio data, infrared image data, multi-depth ground temperature measurement data, surface temperature measurement data, settlement displacement monitoring data, and permafrost property data.

[0011] The AI ​​recognition and species diversity spatiotemporal database module is communicatively connected to the multi-source remote sensing and field perception acquisition module. It is used to process the multi-source perception data to obtain species distribution information, diversity indicators, land cover type and environmental spatial data, and store the processing results in the spatiotemporal database.

[0012] The seasonal dynamic parameter configuration module is connected to the multi-depth geothermal measurement data and is used to identify the current ecological season based on the multi-depth geothermal measurement data, preset temperature threshold, and temperature change trend, and output the corresponding seasonal label and the seasonal dynamic parameters corresponding to the seasonal label. The seasonal dynamic parameters include at least the set of resistance value parameters corresponding to the current ecological season.

[0013] The animal migration path modeling and ecological corridor planning optimization module is used to determine the core habitat patches of the target species based on the species distribution information, construct dynamic resistance surfaces corresponding to different ecological seasons based on land cover type, environmental spatial data and resistance value parameter set, calculate the permafrost thermal stability index based on multi-depth ground temperature measurement data and permafrost property data, use the permafrost thermal stability index as the engineering hard constraint of the candidate corridor scheme, and generate ecological corridor design schemes through path search and multi-objective optimization.

[0014] The real-time animal activity monitoring and ecological connectivity assessment module is configured to receive real-time data from on-site sensors after the corridor is put into operation, obtain the current melting depth of the active layer based on real-time multi-depth geothermal measurement data, calculate the deviation of the melting depth of the active layer from the baseline state of the same period, and generate a dynamic ecological connectivity score by combining animal crossing monitoring data.

[0015] The dynamic regulation and decision feedback module is used to determine whether the preset early warning conditions are met based on the dynamic score of ecological connectivity, the deviation of the melting depth of the synchronous active layer, the snow cover obstacle status and / or the monitoring results of settlement displacement, generate management measure suggestions, and record the effect data after the implementation of the measures through the execution feedback interface.

[0016] Additionally, a model online learning and parameter calibration unit is used to periodically calibrate the resistance parameters of the dynamic resistance surface based on the effect data and successful crossing events in the animal crossing monitoring data, so as to update the basis for generating subsequent ecological corridor assessment and management measures recommendations.

[0017] In the aforementioned system, the temperature sensors include a soil temperature and humidity probe, a surface temperature sensor, and a freeze-thaw depth measuring instrument. The freeze-thaw depth measuring instrument uses distributed temperature-measuring optical fibers or a one-dimensional temperature chain, with a vertical temperature measurement point spacing of no more than 10 cm. The maximum permissible error of the soil temperature and humidity probe and the surface temperature sensor does not exceed ±, and the response time is less than 5 seconds. Each temperature measurement channel is timestamped using a unified time reference. The temperature-sensing element buried in the soil adopts a low self-heating design, and the data acquisition terminal, communication equipment, or power supply unit can be configured with low-temperature start-up or temperature-controlled heating measures. Settlement displacement monitoring data is obtained from GNSS monitoring devices, hydrostatic levels, displacement gauges, and / or remote sensing deformation monitoring data.

[0018] The aforementioned intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, identifies ecological seasons using the following threshold criteria in its seasonal dynamic parameter configuration module:

[0019] Complete Freeze Season: A predetermined number of temperature measurement points within the activity layer remain below a preset first temperature threshold for a predetermined duration;

[0020] Spring thawing season: shallow temperature measurement points change from below the first temperature threshold to above the preset second temperature threshold, and a freeze-thaw transition layer exists within the active layer;

[0021] Complete melting season: A predetermined number of consecutive temperature measurement points within the active layer are above the second temperature threshold for a predetermined duration;

[0022] During the autumn freezing transition period: the shallow temperature measurement points change from being above the second temperature threshold to being below the first temperature threshold;

[0023] The first and second temperature thresholds are engineering management thresholds used to identify the freezing, thawing, and transitional states of the active layer. The first temperature threshold is lower than the second temperature threshold and is calibrated based on the soil type, water content, and long-term geothermal observation data of the study area. In a preferred embodiment, a hysteresis threshold can be set around the measured freezing / thawing transition temperature of the study area, for example, the first temperature threshold is -0.5℃ and the second temperature threshold is +0.5℃. The actual thresholds are based on the field calibration results, and the management thresholds are not equivalent to a uniformly fixed actual soil phase transition temperature.

[0024] The seasonal dynamic parameter configuration module has a built-in temperature state machine. It takes the daily multi-depth ground temperature measurement sequence as input, combines the spatial layer of the temperature measurement point, temperature range, temperature change trend and predetermined duration to determine the ecological season, and outputs the corresponding seasonal label.

[0025] Furthermore, the temperature state machine is equipped with a hysteresis comparator to prevent temperature measurement noise from causing the seasonal label to oscillate frequently at the boundary.

[0026] The aforementioned intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, includes an animal migration route modeling and ecological corridor planning optimization module comprising:

[0027] The resistance surface construction unit assigns basic resistance values ​​to spatial pixels based on surface cover type, vegetation cover, topographic conditions, distance from roadbed, and other animal movement environmental factors, and loads the resistance value parameter set corresponding to different ecological seasons provided by the seasonal dynamic parameter configuration module to generate multiple seasonal dynamic resistance surfaces.

[0028] The constraint calculation unit calculates the permafrost thermal stability index based on the multi-depth geothermal measurement data and the pre-acquired spatial data of volumetric ice content. The spatial data of volumetric ice content is obtained through borehole sampling, permafrost ice content tests, and / or geophysical inversion, and is then spatially interpolated to form raster data. The formula for calculating the permafrost thermal stability index is:

[0029]

[0030] In the formula: It is the thermal stability index of permafrost, which is dimensionless. The higher the value, the lower the thermal stability and the greater the risk of thermal disturbance and thaw settlement. The annual temperature range of the active layer is relatively large; The annual average ground temperature at a predetermined depth below the upper limit of permafrost; This represents the normalized volumetric ice content. This is the upper limit of the multi-year statistical values ​​for the annual temperature range of the active layer in the study area; This represents the upper limit of the absolute value of the annual average ground temperature of permafrost in the study area. , , The non-negative weighting coefficients are calibrated using historical geothermal and meltwater deposition observation data, and satisfy the following conditions: Each normalization term is limited to Within the range.

[0031] The path search and optimization unit runs the minimum cost path algorithm on multiple seasonal dynamic resistance surfaces to calculate candidate corridor paths, conducts cross-seasonal joint evaluation of candidate corridors, and uses the permafrost thermal stability index to meet the preset allowable conditions as a hard constraint on the feasibility of candidate schemes. It performs multi-objective iterative optimization with ecological connectivity and engineering economy as optimization objectives, and outputs ecological corridor design schemes.

[0032] In the aforementioned system, the drag surface construction unit is further equipped with a temperature disturbance input interface for receiving surface temperature measurement data and obtaining the spatial distribution of surface temperature through surface temperature sensors, UAV thermal infrared data, and / or spatial interpolation; for spatial pixel x, the temperature disturbance correction coefficient is calculated according to the following formula:

[0033]

[0034] In the formula: For pixels Temperature disturbance correction factor, For pixels The current surface temperature measurement value, For the same ecological season pixels The historical average surface temperature for the same period This represents the normal fluctuation range of historical surface temperature during the same ecological season. This is a target species-ecological seasonal temperature sensitivity coefficient, which can be positive or negative, calibrated based on the target species' historical traversal data and / or species temperature response data; when When the deviation exceeds a preset threshold, only the current seasonal base resistance value of the corresponding pixel or affected area is locally corrected, and... And the corrected resistance value is limited to a preset reasonable range.

[0035] The aforementioned intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, taking into account temperature disturbances.

[0036] The specific implementation of the permafrost thermal stability index as a hard constraint is as follows: when the average value of the permafrost thermal stability index of the area traversed by the candidate corridor path exceeds the first allowable threshold, or the maximum value exceeds the second allowable threshold, the candidate scheme is marked as infeasible and eliminated in the optimization iteration.

[0037] The path search and optimization unit also obtains the spatial distribution of current density in the candidate corridor path buffer zone based on the circuit theory model, and determines the functional boundaries on both sides of the corridor based on the position where the current density decreases to a preset ratio relative to the maximum current density in the corridor center area, so as to obtain the corridor functional width.

[0038] In the aforementioned system, the real-time animal activity monitoring and ecological connectivity assessment module characterizes the degree of deviation in permafrost thermal state using the deviation of the melting depth of the active layer during the same period, and calculates it according to the following formula:

[0039]

[0040] In the formula: For a moment Corresponding deviation of melting depth of the concurrent active layer; The melting depth of the active layer is obtained based on the location of the freeze / thaw interface in the current multi-depth geothermal measurement data; The baseline value for the active layer melting depth corresponding to the same date or ecological season during the baseline period; This is a preset positive minimum reference depth used to avoid relative deviation distortion when the reference melt depth is 0 or too small; when When the preset threshold is exceeded, the dynamic score of ecological connectivity is multiplied by a preset weighting coefficient;

[0041] The real-time animal activity monitoring and ecological connectivity assessment module also includes a snow connectivity coefficient calculation unit, used to obtain the normalized snow cover rate of key corridor crossing areas based on winter remote sensing imagery, UAV imagery, and / or snow depth sensor data. ,in , and according to Calculate the snow connectivity coefficient; when the snow cover rate and its duration meet the preset conditions, generate an "snow-blocked obstacle" event label independently.

[0042] The aforementioned intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, includes a dynamic control and decision feedback module with built-in multi-level early warning rules, including:

[0043] Frozen soil thermal state early warning: triggered when the deviation of the thawing depth of the active layer exceeds the limit; Melting settlement early warning: triggered when the cumulative melting settlement calculated from the settlement displacement monitoring data exceeds the limit;

[0044] Snow cover warning: Triggered based on snow cover rate and duration exceeding the limit;

[0045] Connectivity warning: Triggered when the dynamic score of ecological connectivity falls below the corresponding preset score within a preset number of consecutive assessment periods;

[0046] Each warning level corresponds to a different management measure push strategy; the dynamic control and decision feedback module also includes a corridor management strategy knowledge base, which stores adaptive management measures in a structured manner, and each management measure is associated with at least the applicable seasonal tag, triggering conditions and historical implementation effects.

[0047] In the aforementioned system, the online model learning and parameter calibration unit uses a local resistance perturbation method based on an adaptive grid for calibration. Specifically, it extracts a set of verified successful crossing event trajectory points, determines the adaptive grid size based on the target species' activity scale and positioning error, and calculates the ratio of the actual crossing intensity to the model-predicted crossing intensity for each grid. When the ratio is higher than the upper limit of a preset interval, the corresponding grid resistance value is reduced by a preset adjustment step size. When the ratio is lower than the lower limit of a preset interval, the corresponding grid resistance value is increased by a preset adjustment step size. After adjustment, the animal migration simulation is rerun and calibration is repeated until the convergence criterion is met for a preset number of consecutive calibration cycles.

[0048] The aforementioned system also includes a data security and privacy protection module and a system robustness and fault tolerance module; the data security and privacy protection module performs anonymization and access permission classification on the precise distribution points of sensitive species; the system robustness and fault tolerance module is used to maintain data continuity and core monitoring and early warning functions in the event of communication link interruption, sensor anomaly, or core service failure through edge caching and breakpoint resume mechanism, sensor redundancy and voting mechanism, and core service degradation operation mechanism.

[0049] Secondly, this invention provides a method for intelligent planning and dynamic adaptive management of biodiversity along roadbeds in permafrost regions, taking into account temperature disturbances. This method uses the aforementioned system for planning and management, and specifically includes the following steps:

[0050] S1: Deploy multi-source sensing devices along the proposed roadbed area to acquire multi-source sensing data and store it in a spatiotemporal database. The multi-source sensing data includes at least species activity data, land cover data, multi-depth ground temperature measurement data, land surface temperature measurement data, permafrost property data, and settlement displacement monitoring data.

[0051] S2: Perform AI recognition processing on the multi-source sensing data to extract species distribution information, diversity indicators, land cover type and environmental spatial data, and determine the core habitat patches of the target species based on the species distribution information;

[0052] S3: Identify the current ecological season based on multi-depth geothermal measurement data, preset temperature thresholds, and temperature change trends; generate a seasonal label and a set of resistance value parameters corresponding to the seasonal label.

[0053] S4: During the planning stage, multiple seasonal dynamic resistance surfaces are generated based on the land cover type, environmental spatial data, and resistance value parameter sets corresponding to different ecological seasons. The permafrost thermal stability index is calculated based on multi-depth geothermal measurement data and volumetric ice content spatial data. The permafrost thermal stability index is used as a hard constraint for candidate corridor schemes. Ecological corridor design schemes are generated through path search, circuit theory analysis, and multi-objective optimization.

[0054] S5: After the corridor is put into operation, the deviation of the melting depth of the active layer during the same period is calculated based on real-time multi-depth geothermal measurement data. Temperature disturbance correction is made on the current seasonal resistance surface based on the spatial distribution of surface temperature. Ecological connectivity dynamic score is generated by combining animal crossing monitoring data.

[0055] S6: Based on the dynamic score of ecological connectivity, the deviation of melting depth of the active layer during the same period, the status of snow cover obstacles and / or the cumulative melting and settling, trigger the corresponding early warning, output management measures suggestions, and record the effect data after the implementation of the measures;

[0056] S7: Based on the effect data and verified successful crossing events, periodically calibrate the local resistance parameters of the dynamic resistance surface, and use the calibration results for subsequent planning evaluation and management decisions.

[0057] The beneficial effects of this invention are as follows:

[0058] 1) This invention, through a seasonal dynamic parameter configuration module, automatically identifies four ecological seasons—the complete freezing season, the spring thaw season, the complete thaw season, and the autumn freezing transition period—based on real-time ground temperature measurement data. It then drives the resistance surface construction unit to switch the corresponding resistance value parameter set. This allows animal migration simulation to move beyond a single static resistance map, reflecting the unique winter-passage, summer-resistance alternation pattern in permafrost regions: swamps and water bodies harden into temporary passages during the freezing season, and reach peak passage resistance during the thaw season when they become muddy and saturated. The resulting corridor paths are the result of comprehensive testing of resistance surfaces across the four seasons, ensuring feasibility of passage in all seasons. This avoids the problem of traditional methods neglecting seasonal dynamics, which leads to significant deviations between actual corridor performance and design expectations.

[0059] 2) This invention introduces a spatialized temperature perturbation correction coefficient into the drag surface. (x) quantifies the degree to which the current surface temperature at different locations deviates from the normal fluctuations of the same historical period as the local adjustment amount of the corresponding pixel resistance value; in the connectivity assessment, it is expressed as the deviation of the melting depth of the active layer during the same period. (t) Characterizes changes in permafrost thermal state, avoiding direct comparisons between instantaneous thaw depths in different seasons and the maximum active layer thickness in a fixed year. The two mechanisms respectively serve to correct for the immediate movement of animals in the environment and assess the thermal state of permafrost, establishing a clear computational relationship between temperature data and planning and operational processes.

[0060] 3) This invention uses the thermal stability index of frozen soil ( As a hard engineering constraint for corridor site selection and structural design, it is directly embedded into the feasible solution screening conditions of the multi-objective optimization model, without repeatedly weighting it with animal movement resistance factors such as ground cover and slope; when the candidate scheme When the average or maximum value exceeds the corresponding allowable threshold, it is directly determined to be infeasible, thereby realizing the stratified treatment of ecological mobility resistance and permafrost engineering safety constraints in the planning stage, and further optimizing ecological connectivity and engineering economy within the feasible domain that meets the thermal stability conditions.

[0061] 4) The system of this invention adopts industrial-grade wide-temperature devices and low self-heating temperature sensing design at the sensor level. The buried temperature sensing element does not maintain operation through continuous active heating. Instead, necessary low-temperature start-up or temperature-controlled heating measures are set in the data acquisition terminal, communication equipment or power supply unit to reduce the interference of the equipment's own heat generation on the ground temperature measurement. Each temperature measurement channel uses a unified time reference for synchronous recording and combines distributed temperature measurement fiber optic high-density acquisition to provide a continuous temperature time series for the calculation of the thermal state of permafrost. The system also sets up edge caching and breakpoint resume, sensor redundancy and voting, core service degradation operation, data desensitization and hierarchical access control to improve the engineering deployability in cold environments.

[0062] 5) During the operation phase, this invention continuously receives real-time data from on-site sensors. The ecological connectivity score is calculated on a rolling basis according to a preset cycle and cross-validated with the deviation of the active layer melting depth, snow connectivity coefficient, and settlement displacement status during the same period. When connectivity or physical passage conditions are abnormal, the early warning rules trigger tiered alarms based on the corresponding indicators; the corridor management strategy knowledge base pushes specific measures applicable to the current season accordingly; the execution feedback interface records the effects after the measures are implemented; the model online learning and parameter calibration unit periodically extracts verified successful crossing events and adaptively corrects the local resistance parameters in the dynamic resistance surface, forming a closed loop from perception to evaluation, regulation, and model update.

[0063] In general, this invention processes multi-depth ground temperature, surface temperature, permafrost properties, and animal activity data into layers according to different technical applications. It generates multi-temporal animal movement resistance surfaces using seasonal dynamic parameters, uses the permafrost thermal stability index as the engineering hard constraint for candidate corridor schemes, and establishes an operational period dynamic assessment and adaptive control mechanism driven by the concurrent permafrost thermal state, snow cover obstacles, structural deformation, and successful animal crossing events, forming a complete closed-loop management system of perception-assessment-control-learning. Attached Figure Description

[0064] Figure 1 This is a schematic diagram of the overall architecture and information flow of the intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, taking into account temperature disturbances, as described in this invention.

[0065] Figure 2 This is a schematic diagram of the sensor layout for the multi-source remote sensing and field perception acquisition module of the present invention.

[0066] Figure 3This is a data processing flowchart for the AI ​​recognition and species diversity spatiotemporal database module of the system of the present invention;

[0067] Figure 4 This is a flowchart of the algorithm processing for the animal migration path modeling and ecological corridor planning optimization module of the present invention.

[0068] Figure 5 This is a flowchart illustrating the closed-loop management process of the real-time monitoring and dynamic control modules of the system of the present invention.

[0069] Figure 6 Cross-sectional structural diagram of a cold-region ecological corridor in an embodiment of the present invention;

[0070] Figure 7 A longitudinal section layout diagram of the cold region ecological corridor in this embodiment of the invention.

[0071] In the picture:

[0072] 101-Multi-source remote sensing and on-site perception acquisition module; 102-AI recognition and species diversity spatiotemporal database module; 103-Animal migration route modeling and ecological corridor planning and optimization module; 104-Real-time animal activity monitoring and ecological connectivity assessment module; 105-Dynamic regulation and decision feedback module; 106-Model online learning and parameter calibration unit; 107-Seasonal dynamic parameter configuration module; 108-Data security and privacy protection module; 109-System robustness and fault tolerance module; 110-Standardized data interface / spatiotemporal data bus;

[0073] 201-High-resolution multispectral satellite; 202-Synthetic aperture radar satellite; 203-UAV-borne lidar and multispectral / thermal infrared camera; 204-Acoustic sensor array; 205-Infrared triggered camera network; 206-Soil temperature and humidity probe and surface temperature sensor; 207-Freeze-thaw depth measuring instrument; 208-Roadbed; 209-Corridor planning / monitoring area; 210-Insulation layer / shallow buried helical pile;

[0074] 301-Multispectral / SAR remote sensing image input; 302-Remote sensing image semantic segmentation unit; 303-Acoustic audio stream input; 304-Speakerprint recognition unit; 305-Image / video stream input; 306-Image target detection unit; 307-Species abundance and multiplicity statistics unit; 308-Spatiotemporal database;

[0075] 401 - Species Distribution Model Unit; 402 - Resistance Surface Construction Unit; 403 - Minimum Cost Path Fast Sweep Unit; 404 - Circuit Theory Corridor Width Quantification Unit; 405 - Roadbed Barrier Effect Quantification Unit; 406 - Hard Constraint Sub-Unit for Permafrost Thermal Stability of Cold Region Corridors; 407 - Cooling Facility Thermal Performance Simulation Sub-Unit; 408 - Eco-Engineering Collaborative Design Sub-Unit; 409 - Multi-Objective Optimization Framework;

[0076] 501 - Animal activity event stream processing engine; 502 - Approach and crossing behavior analysis unit; 503 - Ecological connectivity index calculation unit; 504 - Thermal stability deviation (same-time activity layer melt depth deviation) and snow connectivity coefficient auxiliary index; 505 - Early warning push unit; 506 - Corridor management strategy knowledge base; 507 - Execution feedback interface; 508 - Management action instructions;

[0077] 601 - Active layer; 602 - Permafrost; 603 - Active layer / permafrost interface; 604 - Gravel slope protection; 605 - Heat pipe; 606 - Elevated removable grid walkway; 607 - Support column; 608 - Height-adjustable spiral base; 609 - Thermal insulation backfill material; 610 - Adjustable snow fence; 611 - Sunny side snow accumulation micro-topography; 612 - Animal passage direction;

[0078] 701 - Melting and settling reserved adjustment section. Detailed Implementation

[0079] To address the problems in existing ecological corridor planning along roadbeds in permafrost regions, such as the disconnect between permafrost thermal stability analysis and ecological corridor design, the inability of corridors to respond to seasonal changes in traffic conditions caused by freeze-thaw cycles after construction, and the lack of closed-loop feedback between continuous monitoring and evaluation and control decisions during operation, this invention provides an intelligent planning and biodiversity dynamic adaptive management system and method for ecological corridors along roadbeds in permafrost regions that considers temperature disturbances. Figure 1 As shown, the system includes a multi-source remote sensing and on-site perception acquisition module, an AI recognition and species diversity spatiotemporal database module, an animal migration route modeling and ecological corridor planning and optimization module, a real-time animal activity monitoring and ecological connectivity assessment module, a dynamic regulation and decision feedback module, a model online learning and parameter calibration unit, and a seasonal dynamic parameter configuration module. These modules are interconnected through standardized data interfaces and a unified spatiotemporal data bus, forming a complete information loop from data acquisition, intelligent analysis, planning optimization to monitoring and regulation and adaptive model updates. The specific modules of this system are as follows:

[0080] 1. Multi-source remote sensing and field perception acquisition module. This module is the system's data entry layer, configured to acquire multimodal, multi-scale field data covering the area along the proposed roadbed. For example... Figure 2 As shown, this module specifically includes:

[0081] (1) A high-resolution multispectral satellite remote sensing data interface is used to periodically acquire large-scale multispectral images of the study area with a spatial resolution better than 10 meters to support regional-scale vegetation index calculation, land use classification, and frozen soil surface feature extraction. In particular, during the winter snow season, this interface simultaneously acquires synthetic aperture radar (SAR) satellite images, utilizing its all-weather observation capabilities and response characteristics to underlying surface information under partial dry snow conditions to assist in identifying snow cover, frozen water bodies, and surface freeze-thaw status; snow depth can be jointly estimated and verified by combining SAR features, UAV images, and / or snow depth sensor observations.

[0082] (2) The UAV-borne sensor system is equipped with a lidar and a high-resolution multispectral camera, and optionally a thermal infrared camera, for low-altitude aerial photography operations in key sections of corridor planning or key monitoring areas. The lidar is used to acquire high-precision digital elevation models and vegetation canopy height models, the multispectral camera is used to acquire centimeter-resolution images to finely interpret vegetation types and habitat structures, and the thermal infrared camera is used to identify surface thermal anomalies and generate surface temperature spatial distribution data verified by on-site temperature measurement data, providing data for temperature disturbance correction.

[0083] (3) An acoustic sensor array deployed along the roadbed consists of multiple all-weather, broadband outdoor acoustic recording units, arranged at intervals of 200m to 500m along the roadbed to continuously collect animal acoustic events and locate the sound source azimuth or spatial location when the signal quality meets the requirements. Each acoustic recording unit can be equipped with multiple synchronous microphones to obtain the spatial information of the target sound source through single-station azimuth estimation and multi-station joint positioning. The array mainly records identifiable acoustic events generated by animal calls and close-range activities. Micro weather stations can be configured at the acoustic monitoring points to synchronously record wind speed, wind direction, and snow depth, providing background data for subsequent wind noise suppression and propagation environment correction.

[0084] (4) The infrared-triggered camera network consists of passive infrared-triggered cameras deployed at key monitoring points to acquire animal images or video streams, supporting species identification and activity behavior recording; when the image quality and appearance characteristics of the target species meet the requirements, it can also be used for individual re-identification. The foundation of the camera deployment points adopts a thermal insulation design, which reduces the impact of thermal bridging of the metal pole and heat transfer to the permafrost layer by setting a heat insulation pad at the bottom of the mounting pole or using shallow buried helical piles.

[0085] (5) Temperature and humidity and permafrost environment sensors, consisting of soil temperature and humidity probes, surface temperature sensors, and freeze-thaw depth measuring instruments deployed along the roadbed and corridors, are used to synchronously collect microenvironmental parameters. The freeze-thaw depth measuring instrument uses distributed temperature-measuring optical fibers or a one-dimensional temperature chain, with temperature measurement points deployed at vertical intervals of no more than 10 cm to capture the annual freeze-thaw process of the active layer. The maximum permissible error of the soil temperature and humidity probes and surface temperature sensors does not exceed ±0.2℃, and the response time is less than 5 seconds. Each temperature measurement channel is timestamped using a unified time reference. Among them, multi-depth ground temperature is used to distinguish the freeze-thaw state and the melting depth of the active layer, and surface temperature is used to correct for the immediate movement environment of animals and the temperature disturbance of the resistance surface. The system also deploys one or more of GNSS monitoring devices, hydrostatic levels, or displacement gauges to obtain the cumulative melt settlement.

[0086] The aforementioned sensors employ industrial-grade wide-temperature devices that meet the requirements of cold-region environments. The temperature-sensing elements buried in the soil adopt a low self-heating design and do not have continuous active heating. When extreme low temperatures may affect the startup and communication of electronic equipment, only the data acquisition terminal, communication module, or power supply unit uses low-power temperature-controlled heating and thermal insulation packaging to minimize the disturbance of the ground temperature field caused by the heat generated by the equipment itself.

[0087] 2. AI Recognition and Spatiotemporal Database Module for Species Diversity. This module is the core of the system's intelligent analysis, responsible for transforming raw, perceived data into structured biodiversity information. For example... Figure 3 As shown, this module specifically includes:

[0088] (1) The remote sensing image semantic segmentation unit uses a deep convolutional neural network to perform pixel-level classification of satellite and UAV remote sensing images, automatically interpreting categories such as vegetation type, water body, bare land, and internal physical structure of corridors, and generating vegetation cover map and corridor structure map. This model specifically introduces two typical land cover types in cold regions, namely "ice and snow" and "thermal melting lakes and ponds", to enhance the ability to capture dynamic changes in landscape during the freeze-thaw cycle.

[0089] (2) Voiceprint recognition unit, used to perform animal acoustic event recognition and species classification, automatically processes the continuous audio stream collected by the acoustic sensor array, detects and classifies the vocalization events of the target animal species, and outputs species labels, timestamps, and sound source location information when the positioning conditions are met. To solve the problem of limited sample size of target species, this unit can adopt a few-shot learning framework based on model-independent meta-learning (MAML), pre-training or meta-training on a dataset containing acoustic samples of common species, and then fine-tuning using a small number of locally labeled samples of target species. In addition, this unit integrates a cold-region audio preprocessing workflow, and loads wind noise suppression and ice and snow background noise filtering modules according to wind speed, wind direction and background noise characteristics during the completely frozen season, so as to reduce the interference of non-biological background noise such as permafrost cracking and ice expansion and contraction on the recognition results.

[0090] (3) Image target detection unit: Using a target detection and instance segmentation model, animal individuals are detected and counted in images acquired by infrared-triggered cameras. When the image quality, shooting angle, and appearance characteristics of the target species meet the conditions, cross-time period or cross-location individual re-identification can be carried out in conjunction with a feature extraction network. This unit can also be configured with a cold region image enhancement preprocessing module to perform highlight suppression and dynamic range adjustment in snowy and highly reflective scenes, and to perform defogging or local contrast enhancement in low-contrast snowy weather images to improve the stability of target detection.

[0091] (4) Species richness and diversity statistics unit: The results of remote sensing interpretation, voiceprint recognition and image detection are aggregated and statistically analyzed according to a unified spatiotemporal coordinate system. The diversity indicators such as the number of species, population density of each species, and activity frequency of each spatiotemporal unit are calculated, and standardized species distribution location records and corridor utilization event records are generated.

[0092] (5) The spatiotemporal database is a database system that supports the storage, indexing, and querying of massive spatiotemporal data. Its stored content includes multi-temporal remote sensing image interpretation maps, species distribution point vector data, population density raster data, sound event records, image capture records, individual activity trajectories, land cover and environmental spatial data, and environmental parameter time series. The environmental parameter time series fully records multi-depth ground temperature, surface temperature, snow depth, contemporaneous active layer melting depth, and subsidence displacement data; permafrost property data such as volumetric ice content are obtained through borehole sampling, indoor experiments, and / or geophysical inversion and formed into spatial raster, providing a data foundation for the calculation of permafrost thermal stability constraints.

[0093] 3. Animal Migration Route Modeling and Ecological Corridor Planning Optimization Module. This module is the core of the system's spatial decision support, used to quantify roadbed barrier effects and generate optimized ecological corridor schemes. For example... Figure 4 As shown, this module specifically includes:

[0094] (1) Species distribution model unit, which extracts the occurrence sites of the target species and the corresponding environmental variables from the spatiotemporal database, establishes the statistical relationship between species distribution and environmental factors using model algorithms, and predicts the probability distribution map of suitable habitats for the species in the study area. Among them, the occurrence points need to be preprocessed by spatial dilution and temporal dilution before use: the spatial dilution distance is set to 1 / 10 of the species' average daily movement distance, and the temporal dilution is to retain only one event of the same species within 24 hours on the same sensor.

[0095] (2) The resistance surface construction unit is the core of reflecting the seasonal dynamics of cold regions. Based on environmental factors related to animal movement, such as land use type, vegetation cover, terrain slope, distance from the roadbed, and nighttime light intensity, this unit assigns a basic resistance value to each spatial pixel, representing the degree of obstruction to animal movement. Subsequently, dynamic resistance surfaces for different ecological seasons are formed based on the resistance value parameter set output by the seasonal dynamic parameter configuration module. The permafrost thermal stability index is not repeatedly weighted as a common animal movement resistance factor, but is calculated independently by the constraint calculation unit and serves as a hard engineering constraint for candidate corridor schemes entering the optimization feasible region. Its calculation formula is:

[0096]

[0097] In the formula: It is the thermal stability index of permafrost, which is dimensionless. The higher the value, the lower the thermal stability and the greater the risk of thermal disturbance and thaw settlement. The annual temperature range of the active layer is relatively large; The annual average ground temperature at a predetermined depth below the upper limit of permafrost; This represents the volumetric ice content after normalization from 0 to 1. This is the upper limit of the multi-year statistical values ​​for the annual temperature range of the active layer in the study area; This represents the upper limit of the absolute value of the annual average ground temperature of permafrost in the study area. , , The non-negative weights are calibrated using historical geothermal, ice content, and meltwater deposition observation data, and satisfy the following conditions: .use As an annual average ground temperature term, the closer the annual average ground temperature of permafrost is to... The higher the risk term, the lower the ground temperature; all normalization terms are limited to the range of 0 to 1.

[0098] in, and It is calculated from long-term continuous multi-depth geothermal measurement data from distributed temperature-measuring optical fibers or one-dimensional geothermal chains; The data were obtained through borehole sampling, permafrost ice content tests, and / or geophysical inversion, and spatially interpolated to form a volumetric ice content grid. This data can be combined with historical thaw events to... The weights and allowable thresholds are verified and calibrated.

[0099] The resistance surface construction unit is also equipped with a temperature disturbance input interface, which receives surface temperature data in real time from surface temperature sensors and / or UAV thermal infrared data, and forms a spatial distribution of surface temperature through spatial interpolation; and calculates a temperature disturbance correction coefficient for spatial pixel x:

[0100]

[0101] in, For pixels The current surface temperature, For the same ecological season pixels The historical average surface temperature for the same period This represents the normal fluctuation range of historical surface temperature during the same ecological season. Ecological seasonal temperature sensitivity coefficient, which can be positive or negative, is a target species calibrated using animal history traversal data and / or species temperature response data. The sign represents the direction of the effect of abnormal heating or cooling on moving resistance. When When the deviation exceeds a preset threshold, only the current seasonal resistance value of the corresponding pixel or affected area is locally corrected, and... The corrected resistance value is limited to a preset reasonable range to avoid unreasonable scaling of the resistance surface of the entire study area due to a single temperature anomaly.

[0102] This unit possesses a unique seasonal dynamic resistance mechanism specific to cold regions: it receives seasonal tags pushed by the seasonal dynamic parameter configuration module and automatically switches to the corresponding resistance value table. The ecological significance of this change lies in:

[0103] During the full freeze season, the active layer freezes completely, significantly increasing its bearing capacity. Previously muddy, waterlogged marshes and water surfaces that were muddy and prone to flooding during the thaw season freeze into solid ground, resulting in a significant decrease in resistance and creating temporary shortcuts or alternative passageways for animals. Therefore, the landscape resistance surface is flatter and more interconnected in winter than in summer.

[0104] Spring thaw season: As the snow melts on the ground, the active layer gradually thaws from top to bottom. The surface water content becomes saturated, forming a seasonal waterlogged or muddy layer. During this stage, the passage conditions on most lands are extremely poor, and the resistance value reaches its highest point of the year.

[0105] Full Melt Season: The active layer thaws completely, and the area of ​​the thermal thaw lake reaches its largest size of the year. The surrounding areas become muddy due to poor drainage, and the resistance value remains at a high level.

[0106] Autumn freezing transition period: The surface begins to freeze from top to bottom, but the lower part may still contain water, forming an interlayer of freezing on top and thawing below. In some areas, traffic conditions begin to improve, and resistance values ​​gradually decrease.

[0107] Therefore, instead of using a single static resistance map, the system generates a multi-temporal dynamic resistance surface sequence to characterize the seasonal differences in permafrost surface that may occur in different freeze-thaw stages, such as winter accessibility and summer obstruction or easy access in winter and difficult access in summer. This enables corridor route planning to reflect seasonal changes by combining actual observation data of target species.

[0108] (3) The minimum cost path rapid sweep unit reads the resistance surface and suitable habitat distribution map for the specific season, runs the minimum cost path algorithm, and quickly calculates multiple candidate corridor paths with the minimum cumulative resistance between core habitat patches at the whole region scale, forming a set of candidate corridor alternatives. By running the resistance surface for different seasons, the optimal path for the whole year or key migration window can be identified.

[0109] (4) Circuit theory corridor width quantization unit: For each path in the candidate corridor scheme set, a buffer zone of a certain width is extracted around it. A high-resolution circuit theory model is run in the buffer zone to calculate the current density distribution of the habitat patches connecting the two ends. The spatial decay curve of the current density is analyzed, and the minimum functional width of the corridor is automatically defined at the position where the current density value drops to 1 / e of the maximum current density value of the central path.

[0110] (5) The roadbed barrier effect quantification unit incorporates the precise alignment, roadbed height, pavement width, and information on transverse crossing structures such as bridges and culverts along the proposed or existing roadbed as high resistance barriers into the circuit theory model, re-runs the simulation, quantifies the segmentation intensity of the roadbed and ancillary facilities on the original migration flow, generates a barrier intensity distribution map, and locates the specific fractured road section that causes the maximum cutting-off effect on the complete migration flow.

[0111] (6) A unified multi-objective optimization framework for ecological corridors: Based on a constrained multi-objective genetic algorithm, the planar alignment, functional width, and layout of active cooling facilities of ecological corridors are simultaneously optimized. The hard constraint of this optimization is the long-term thermal stability of permafrost, and the optimization objective is Pareto optimality of ecological connectivity and engineering economy. Specifically, it includes the following sub-units:

[0112] Hard Constraint Sub-Element for Permafrost Thermal Stability in Cold Region Corridors: This sub-element uses permafrost thermal stability as a hard engineering constraint for corridor site selection and design algorithms, and continuously receives the permafrost thermal stability index output by the constraint calculation unit. and based on Numerical simulations are used to pre-screen candidate schemes for thermal stability feasibility. Specific design principles may include: a low-thermal-disturbance foundation principle, prioritizing foundation types with minimal disturbance, such as shallow-buried steel helical piles, based on thermal stability calculations, and configuring heat pipes, crushed stone cooling structures, or thermal insulation backfill materials as needed; the surface radiation characteristics of the foundation, the thermal properties of the backfill material, and the layout of cooling facilities are determined by on-site thermal parameters and numerical simulation results. Regarding snow management, snow barriers or micro-topography are set up based on historical meteorological data and snow redistribution simulations, focusing on preventing abnormal snow accumulation at corridor entrances and key passage sections, while preserving natural snow cover areas that meet the utilization characteristics of target species and the thermal stability requirements of permafrost. Regarding melt settlement adaptation, for corridors traversing high-ice-content permafrost sections, adjustable margins are set based on predicted melt settlement and allowable structural deformation, and height-adjustable helical bases are configured at the support nodes of the elevated walkway to allow for local adjustments based on elevation monitoring results during the operational period.

[0113] The thermal performance simulation subunit for cooling facilities: During the planning phase, local long-term meteorological boundary conditions or climate scenario data, initial permafrost geothermal field, and soil thermal property parameters are input. Numerical models such as heat conduction and convection are run to simulate candidate cooling design schemes and evaluate indicators such as the upper limit of permafrost, active layer thickness, and cumulative melt settlement during the design period. When the simulation results exceed the preset allowable thresholds for geothermal temperature, active layer thickness, or melt settlement, the candidate scheme is judged as not meeting the thermal stability constraints and is removed from the optimal feasible solution to select schemes that meet the thermal stability requirements during the design period.

[0114] Ecological-Engineering Collaborative Design Subunit: In high-current-density sections crossing the main axis identified by the circuit theory model, elevated, detachable walkways or other low-thermal-disturbance structures that balance cold air convection and animal passage can be adopted. The elevated height, effective passage width, and surface opening or grid dimensions are jointly determined based on the target species' body size, group passage scale, behavioral preferences, site topography, and the need for lower-level cold air convection. At corridor boundaries and non-primary crossing areas, gravel slope protection, heat pipes, or other cooling facilities can be retained based on thermal calculations. Necessary safety distances or protective structures should be set between structural components and animal activity areas to prevent cooling facilities from obstructing animal passage.

[0115] The hierarchical multi-fidelity optimization strategy is as follows: The first layer uses a coarse-resolution grid and a heuristic algorithm based on cumulative cost paths for coarse screening; the second layer performs high-resolution accurate circuit simulation on non-dominated individuals; every few generations, all individuals on the Pareto front are accurately evaluated; the termination condition is that the hypervolume change of the front is less than a preset threshold or the preset number of generations is reached for several consecutive generations.

[0116] The quantitative indicators for the economic efficiency of the project include two main categories: construction costs and operation and maintenance costs. Construction costs include land acquisition and compensation for the corridor, the cost of constructing the main structure, the cost of constructing cooling facilities, and the cost of constructing auxiliary facilities. Operation and maintenance costs include routine maintenance costs, repair costs, snow removal operation costs, and the operating costs of the monitoring system. The comprehensive economic efficiency target is the total life-cycle cost, which is obtained by summing the construction costs and the annual operation and maintenance costs during the operation period, discounting them to the same time base, and then adjusting for residual value recovery. Specifically, the total operating period is defined as the design lifespan. The operation and maintenance costs for each future year are discounted to their present value at the beginning of the construction period using a set social discount rate. The present values ​​of the operation and maintenance costs for all years are accumulated and added to the construction costs. Then, the residual value of the corridor facilities at the end of the design lifespan is subtracted, resulting in a life-cycle economic evaluation indicator that can be compared across different schemes.

[0117] 4. Real-time Animal Activity Monitoring and Ecological Connectivity Assessment Module. This module is the core operational module of the system after the corridor is built and put into use, and is used to continuously evaluate the actual effectiveness of the corridor. For example... Figure 5 As shown, this module, together with the dynamic control and decision feedback module, forms a closed-loop management process. This module specifically includes:

[0118] (1) Animal activity event stream processing engine, which maintains data stream connection with the acoustic sensor array and infrared trigger camera network on site, performs noise reduction, pre-screening and trigger buffering on the real-time incoming audio and image data, and extracts candidate event fragments containing animal activity characteristics.

[0119] (2) The approach and crossing behavior analysis unit further classifies the candidate events output by the event stream processing engine to determine whether the animal is merely approaching the edge of the roadbed, lingering at the entrance of the corridor, crossing the interior of the corridor, or forcibly crossing the roadbed from other non-corridor areas. This unit comprehensively utilizes the animal's movement trajectory, speed changes, and acoustic features obtained by cross-sensor collaborative localization of the acoustic array to make a behavior judgment.

[0120] (3) Ecological connectivity index calculation unit: The dynamic reference system and uncertainty quantification method are used to calculate the dynamic score of ecological connectivity. The system can set two reference modes and select them according to the data conditions: Mode 1: When there are no less than 3 reference sample areas with similar ecological and geographical conditions and weak road interference, the equivalent resistance value of each reference sample area is calculated, and a reference baseline for the same period is established after similarity matching based on environmental covariates such as topography, vegetation, water body, and snow cover; Mode 2: When there are insufficient suitable spatial reference sample areas, the historical benchmark landscape is reconstructed using historical remote sensing images and environmental data before the corridor construction and the equivalent resistance value before construction is calculated; if the historical data is still insufficient, it is downgraded to the trend analysis mode, and only the direction, rate and confidence of connectivity change are output. Each time the score is output, the system marks the reference mode used, data integrity and confidence interval. To reduce the interference of regional background changes on the baseline, the system uses similar habitat patches that are not significantly affected by the corridor at a set period to extract the changing trends of vegetation cover, snow cover, water bodies and other ecological environment factors during the same period, and performs background correction on the resistance or conductivity distribution of the baseline landscape; multi-depth geothermal data is still used to evaluate the deviation between the thermal state of permafrost and the melting depth of the active layer during the same period, and does not directly replace the animal movement resistance factor.

[0121] The objective function for ecological connectivity is constructed based on a circuit theory model, defined as the ratio of the effective current intensity of the corridor scheme in the circuit theory model to the baseline current intensity under natural conditions without roadbed obstruction. In the circuit theory model, spatial pixels in the study area are considered as conductive media, and the conductivity of a pixel is inversely related to the animal migration resistance value; that is, the lower the resistance value, the higher the conductivity, corresponding to areas where animals can pass more easily. The core habitat patches connecting the two ends of the corridor are set as the source and sink ends of the circuit, respectively. Under given boundary conditions, the nodal potential distribution of the entire region is solved, and then the total current intensity flowing from the source end to the sink end is calculated as the effective current intensity. The baseline current intensity is calculated using the same method under the assumption of removing the roadbed obstruction effect. The closer this ratio is to 1, the closer the corridor scheme is to the unobstructed natural connectivity state, and the better the ecological connectivity; the closer it is to 0, the more severe the roadbed obstruction effect, and the worse the recovery effect of the corridor scheme on migration flow.

[0122] The multi-objective optimization approach employs two parallel objectives: maximizing the ecological connectivity index and minimizing the total life-cycle cost. It also includes a hard constraint that the permafrost thermal stability index does not exceed a preset maximum allowable value, supplemented by inequality constraints such as upper and lower limits for corridor width, minimum density of cooling facilities, and minimum requirements for overhead height, as well as integer constraints such as the number of design segments and the selection of structural components. The constrained multi-objective optimization algorithm outputs a Pareto optimal solution set, allowing decision-makers to select the final implementation plan based on their actual preferences.

[0123] (4) Auxiliary index and weighting mechanism for physical connectivity in cold regions: This is a key element missing from traditional ecological connectivity assessments in permafrost regions. The system recognizes that in cold regions, physical accessibility is a prerequisite for biological accessibility. Therefore, based on the aforementioned biological connectivity score, two hard cross-validation indicators based on physical sensors are introduced:

[0124] Thermal stability deviation as a function of the melting depth deviation of the active layer (t) Characterization. This index is based on multi-point, multi-depth geothermal monitoring data within the corridor, comparing the active layer melting depth on the current date or in the current ecological season with the corresponding melting depth on the same date or in the same ecological season during the baseline period. The specific calculation formula is as follows:

[0125]

[0126] in, To determine the active layer melting depth after using current measured data from distributed temperature sensing optical fibers or geothermal chains, The melting depth of the active layer corresponds to the same date or the same ecological season during the baseline period, and both are in meters; This is a preset positive minimum reference depth used to avoid relative deviation distortion when the reference melt depth is 0 or too small. When the threshold is exceeded, the system applies a preset weighting factor to the comprehensive score of ecological connectivity. For long-term annual evaluations, the maximum active layer thickness of the current year is compared with the maximum active layer thickness of the baseline year, thus avoiding a direct comparison between the instantaneous melting depth in winter or at the beginning of thawing and the maximum active layer thickness of a fixed year.

[0127] Snow connectivity coefficient Normalized snow cover of key crossing areas was obtained based on winter drone aerial photography, satellite SAR imagery, and / or snow depth sensor data. ,in , and according to calculate; The smaller the value, the more significant the physical obstacle caused by snow cover. Snow depth and duration serve as independent auxiliary criteria. When the snow cover, snow depth, and / or duration of a key crossing path meet preset conditions, the system generates a "snow-blocked obstacle" event label and sends intervention suggestions to the dynamic control module.

[0128] 5. Dynamic Control and Decision Feedback Module. This module is the adaptive management closed-loop execution end of the system, used to translate real-time assessment results into specific protection and management actions. This module specifically includes:

[0129] (1) Early warning push unit, with built-in yellow, orange, and red three-level early warning rules. The ecological connectivity score first determines the basic score based on the effective current intensity of the circuit theory model and the natural reference current intensity; when the melting depth of the concurrent active layer deviates from the degree (t) When the set threshold is exceeded, the base score is multiplied by the corresponding permafrost thermal state weighting coefficient; when the snow cover rate and duration meet the snow-sealing conditions, the system generates a "snow-sealing obstacle" event label, and can make additional corrections to the score according to management rules. Connectivity warnings are triggered when the ecological connectivity score falls below the corresponding value within a consecutive preset number of assessment periods; permafrost thermal state warnings are triggered according to... (t) Exceeding limits trigger: Melt subsidence warning is triggered when the cumulative melt subsidence amount obtained from settlement displacement monitoring exceeds the limit; Snow cover warning is triggered when the snow cover rate, snow depth, and duration exceed the limit. Each type of warning is executed separately to avoid using different physical causes as the same triggering condition.

[0130] (2) The corridor management strategy knowledge base is a structured information repository for adaptive management measures. Each management measure is accompanied by an applicable season tag, which includes the applicable scenario, operation content, expected effect and implementation precautions. Examples of measures specific to cold regions include: pre-dredging the corridor drainage ditches before the spring thaw to prevent the accumulation of freeze-thaw water; using temporary fences to guide animals to use frozen swamps and other alternative paths in winter; and using cooling fillers for rapid cold repair when thermal thawing subsidence pits are found.

[0131] (3) An execution feedback interface is used to record the issuance time, execution scope, and operation details of each management measure, and to continuously track the changes in the ecological connectivity score in the next period after the implementation of the measure, and to evaluate the effectiveness of the measure. For successful strategies that significantly improve connectivity, their recommendation weight in the knowledge base is increased; for inappropriate strategies that have no significant effect or have a negative impact, their recommendation priority is reduced or restrictive conditions are added.

[0132] 6. Online Model Learning and Parameter Calibration Unit. This unit employs an adaptive grid-based local resistance perturbation method based on species activity scales. The specific steps are as follows: periodically extracting verified successful crossing event trajectory point sets from the spatiotemporal database; performing adaptive grid division, with grid size determined based on the target species' average daily movement distance and positioning error; for each grid, calculating the ratio of actual crossing intensity to the current model's predicted intensity; if the ratio exceeds a preset range, adjusting the resistance value of that grid accordingly; rerunning the migration simulation after adjustment; stopping iteration and saving the latest resistance surface when multiple consecutive calibration cycles meet the convergence criterion. Through this process, the system can adaptively learn, for example, a cold habitat corridor formed by microclimate or multi-year ice accumulation, invisible in remote sensing images but serving as a crucial animal passage in winter, and lower its resistance value, achieving seamless approximation between the model and ecological reality.

[0133] 7. Seasonal Dynamic Parameter Configuration Module. This module provides the system's environmental background service. It primarily identifies the current ecological season based on real-time multi-depth ground temperature data and provides corresponding seasonal labels and dynamic parameters to other modules. Information such as snow cover is used as environmental background parameters for the current ecological season in resistance surface and connectivity evaluation. The seasonal division and its criteria are as follows:

[0134] (1) Complete freezing season: If a predetermined number of temperature measurement points in the active layer are below the first temperature threshold for a predetermined duration, it indicates that the active layer is under freezing control and the surface bearing conditions are relatively good.

[0135] (2) Spring thawing season: shallow temperature measurement points change from below the first temperature threshold to above the second temperature threshold, and there is a clear freeze-thaw transition layer in the active layer; at this time, the surface water content and mud level usually increase rapidly, which is a relatively fragile seasonal window for ecological corridors.

[0136] (3) Complete melting season: When a predetermined number of temperature measurement points in the active layer are above the second temperature threshold for a predetermined duration, the active layer enters a stage dominated by melting.

[0137] (4) Autumn freezing transition period: The shallow temperature measurement points change from above the second temperature threshold to below the first temperature threshold, and the freezing front gradually advances downward from the shallow part.

[0138] The seasonal dynamic parameter configuration module incorporates a temperature state machine. This state machine takes daily multi-depth ground temperature measurement sequences as input, performs a moving average filter on the temperature data over time, and then determines the ecological season layer by layer based on the spatial layer of the measurement points, temperature range, and temperature change trend. When a predetermined number of consecutive temperature measurement points covering a representative depth range of the active layer meet the same ecological season criterion and continue for a predetermined time, the state machine outputs a corresponding ecological season label. The state machine is equipped with a hysteresis comparator; in a preferred embodiment, the hysteresis temperature difference can be set to approximately... The system is calibrated based on sensor errors and temperature fluctuation characteristics of the study area to suppress frequent switching of the seasonal label near the threshold.

[0139] This sub-module pushes seasonal labels to each module through the system data bus and triggers linkage behaviors: providing the resistance value parameter set corresponding to the current season to the resistance surface construction unit; the resistance factors such as land cover, vegetation, slope, snow cover, and surface water state and their weights in each seasonal parameter set are mainly calibrated based on the measured movement trajectories and environmental data of the target species in the corresponding season, multi-depth ground temperature is used for discriminating ecological seasons and permafrost thermal states, and surface temperature is used for local temperature disturbance correction. Preprocessing parameters for the current season can be pushed to the AI identification unit; the historical same-period scoring baseline and confidence interval are transmitted to the ecological connectivity assessment unit, and the cold region environmental background values such as current seasonal snow depth, ice-sealed range and active layer thermal state are synchronously provided; management strategies applicable to the current season are screened for the dynamic regulation module and the corresponding monitoring plans are activated.

[0140] 8. Data security and privacy protection module. This module is configured to perform desensitization and access permission grading on the precise distribution points of sensitive species, and only provide public network service data that has undergone spatial blurring processing to the outside. At the same time, a separate storage architecture for metadata and original data is adopted to optimize the response efficiency of high-frequency queries on the basis of ensuring the physical security of core data.

[0141] 9. System robustness and fault tolerance module. This module is configured to maintain data continuity and core monitoring and early warning functions through edge caching and resume transmission from breakpoints mechanisms, sensor redundancy and voting mechanisms, core service degraded operation mechanisms, and health self-check and reporting functions when communication link interruption, sensor abnormality or core service failure occurs.

[0142] The connection relationship and information flow between the above modules are as follows:

[0143] The multi-source remote sensing and on-site perception acquisition module transmits acquired satellite remote sensing images, UAV remote sensing images, acoustic audio data streams, infrared image data, and environmental parameter time-series data to the AI ​​recognition and species diversity spatiotemporal database module via corresponding data links. The AI ​​recognition unit within this module analyzes and processes the various data types, generating structured species information and diversity indicators which are then written into the spatiotemporal database. The animal migration path modeling and ecological corridor planning optimization module retrieves target species distribution data and environmental layers from the spatiotemporal database and receives roadbed engineering design data to complete migration simulation and corridor scheme generation. Once the corridor is completed and enters the operational phase, the real-time animal activity monitoring and ecological connectivity assessment module accesses the data streams generated by on-site sensors and compares them with historical baselines in the spatiotemporal database to generate an ecological connectivity score for the current period. This score is then transmitted to the dynamic regulation and decision feedback module, triggering corresponding early warnings and management recommendations. The effects of the implemented management measures are recorded through the execution feedback interface and transmitted back to the spatiotemporal database and the corridor management strategy knowledge base, updating the records of species utilization corridors and the results of strategy effectiveness assessments. Simultaneously, the online learning and parameter calibration unit periodically extracts successful crossing events from the spatiotemporal database to adaptively calibrate the resistance surface. The seasonal dynamic parameter configuration module identifies the ecological season based on multi-depth geothermal measurement data and continuously broadcasts seasonal labels and freeze-thaw status parameters, driving each module to call the corresponding seasonal parameters. The data security and privacy protection module implements access control for sensitive data, and the system robustness and fault tolerance module monitors the overall system operation status and provides a degradation recovery mechanism.

[0144] Thus, the system consists of a closed-loop data flow from sensing and data collection → AI identification and database construction → migration modeling and corridor planning → real-time monitoring and connectivity assessment → dynamic regulation and effect feedback → resistance surface adaptive calibration → continuous sensing, forming an intelligent management system that can synchronously respond to changes in permafrost thermal state, structural deformation and ecological connectivity, and realize intelligent planning of ecological corridors along roadbeds in permafrost areas and dynamic adaptive management of biodiversity.

[0145] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are merely exemplary embodiments of the present invention, and not all embodiments, nor do they constitute a limitation on the scope of protection of the present invention. Those skilled in the art can make changes, equivalent substitutions, or modifications to the embodiments without departing from the essence of the technical solution of the present invention; other embodiments obtained thereby can still fall within the scope of the technical solution of the present invention.

[0146] Example 1

[0147] This embodiment uses the application of the system of the present invention to a newly built secondary highway project on the Qinghai-Tibet Plateau as an example to describe in detail the technical solution and system operation mode of the present invention. The measured annual average ground temperature in this area is -1.8℃, the active layer (601) thickness is 1.5-2.8m, and permafrost (602) lies beneath the active layer. The interface (603) between the active layer and the permafrost changes dynamically with the seasonal freeze-thaw cycle. The main protected targets are Tibetan antelope and Tibetan wild ass. It should be understood that the following descriptions of various scenarios that the system may deal with in future operation are intended to illustrate the functional logic and execution capabilities of the system, rather than a statement of facts that have already occurred. The modules in the system are interconnected through standardized data interfaces and a spatiotemporal data bus (110).

[0148] I. Planning and Construction Phase: Multi-source temperature data acquisition, thermal stability calculation, and corridor scheme generation

[0149] In the planning and construction phase of this embodiment, the multi-source remote sensing and field perception acquisition module (101) is configured and put into operation to acquire multimodal background data. Among them, the acquisition and fusion calculation of temperature measurement data is the core basic work throughout the entire phase. The corridor planning and monitoring area (209) covers a 10-kilometer buffer zone on both sides of the roadbed. Specifically, images of the buffer zones on both sides of the roadbed (208) are acquired quarterly through the remote sensing data interface of the high-resolution multispectral satellite (201); synthetic aperture radar (SAR) images are acquired simultaneously in winter to help identify snow cover, frozen water bodies and surface freeze-thaw status, and to jointly estimate or verify snow depth with UAV images and snow depth sensor data. During the snowless or low-snow window, the UAV (203) equipped with lidar, multispectral camera and thermal infrared camera is used to conduct strip aerial photography of the planned corridor candidate area to acquire high-density point clouds, high-resolution multispectral orthophotos and the spatial distribution of surface temperature verified by field temperature data. Simultaneously, an acoustic sensor array (204), an infrared trigger camera network (205), and permafrost environment sensors, including soil temperature and humidity probes, surface temperature sensors (206), and freeze-thaw depth measuring instruments (207), are deployed along the roadbed in the selected area. The freeze-thaw depth measuring instrument can use distributed temperature-measuring optical fibers, with a vertical temperature measurement point spacing of no more than 10 cm. The infrared camera mounting base adopts a low-thermal-disturbance form such as a heat-insulating pad and shallow-buried helical piles (210) to reduce the thermal bridge effect. All the above temperature-measuring elements have been calibrated, and the measurement data are timestamped using a unified time base. Simultaneously, spatial data on volumetric ice content and settlement displacement monitoring data formed by borehole drilling, testing, and / or geophysical surveys are acquired to provide a data basis for subsequent calculations of permafrost thermal stability index and thaw settlement early warning.

[0150] Next, the AI ​​recognition and species diversity spatiotemporal database module (102) analyzes and processes the multimodal data collected above. Multispectral and SAR remote sensing images (301) are input to the remote sensing image semantic segmentation unit (302) to extract land cover categories such as arbor forests, shrub forests, herbaceous wetlands, bare land, water bodies, snow and ice, and thermal meltwater lakes and ponds. The acoustic audio stream (303) is input to the voiceprint recognition unit (304), which uses a few-shot learning framework to detect animal acoustic events and classify species in the continuous audio stream, and loads wind noise and snow and ice background noise suppression modules according to the winter background noise characteristics. The image and video stream (305) is input to the image target detection unit (306), and corresponding image enhancement preprocessing is loaded in scenes with strong reflection of snow or low contrast. All identification results are aggregated by the species richness and abundance statistics unit (307), and associated with coordinates and timestamps. For the occurrence record of the target species, the spatial dilution distance can be determined as 1 / 10 of the average daily movement distance of the target species, and the temporal dilution can be achieved by retaining only one representative event of the same species within 24 hours of the same sensor, and then stored in the spatiotemporal database (308).

[0151] Based on the constructed database, the animal migration path modeling and ecological corridor planning optimization module (103) performs quantitative optimization of the corridor scheme. Taking the target species as an example, the species distribution model unit (401) uses the diluted occurrence points combined with environmental factors such as vegetation, altitude, and slope to predict the distribution of suitable habitats and determine the core habitat patches. The resistance surface construction unit (402) generates basic resistance surfaces based on animal movement environmental factors such as land use, vegetation coverage, slope, and distance from the roadbed, and receives four sets of resistance value parameter sets pushed by the seasonal dynamic parameter configuration module (107) to generate dynamic resistance surfaces for the complete freezing season, spring thaw season, complete melting season, and autumn freezing transition period, respectively. The constraint condition calculation unit calculates the spatial data of multi-depth ground temperature and volumetric ice content according to the above. The formula calculates the permafrost thermal stability index for each spatial pixel. , , Calibrated based on historical geothermal, ice content, and meltwater data for this region. The resistance surface construction unit utilizes both surface temperature sensors and UAV thermal infrared data to visualize the spatial distribution of surface temperature, as described above. The formula performs local corrections for temperature anomaly areas, where the sign and amplitude of the temperature sensitivity coefficient are calibrated according to the target species and ecological season, and the correction coefficient and resistance value are truncated with upper and lower limits. The minimum cost path rapid sweep unit (403) runs the algorithm on the resistance surface in different seasons to perform cross-seasonal joint evaluation of candidate paths; the circuit theory corridor width quantification unit (404) further determines the functional width, and the roadbed barrier effect quantification unit (405) locates road sections that have a significant effect on cutting off migration flows.

[0152] Finally, a unified multi-objective optimization framework (409) for the corridor was initiated to generate the final scheme. The hard-constrained sub-element (406) for the permafrost thermal stability of the cold region corridor continuously receives the output of the constraint calculation unit. Numerical values; in this embodiment, the candidate corridor path area can be... The average and maximum values ​​are compared with the first and second allowable thresholds, respectively. If either index exceeds the limit, the scheme is marked as infeasible and eliminated in the optimization iteration. For schemes that meet the following conditions... The scheme with hard constraints is then optimized by combining the thermal performance simulation of cooling facilities, ecological connectivity, and engineering economy. The structural foundation can adopt shallow buried helical piles and be equipped with heat-insulating backfill material (609). Adjustable snow fences (610) are set on both sides of the corridor according to the simulation of snow redistribution. The thaw settlement adjustment amount is preset in the high ice content permafrost section and an adjustable height helical base is configured (608). Under the coordination of the ecological-engineering collaborative design subunit (408), the final output is a Pareto scheme that meets the thermal stability constraints of permafrost and the ecological connectivity objectives. Figure 6 and Figure 7 The above-ground detachable grid walkway (606), heat pipe (605), gravel slope protection (604), snow fence and settlement pre-reserved adjustment section (701) shown are one of the specific structural forms.

[0153] II. Operational Monitoring Phase: Real-time tracking of thermal stability and closed-loop system control based on temperature monitoring

[0154] After the corridor is built and put into operation according to the above optimization plan, the system will switch to real-time monitoring and control mode. The real-time animal activity monitoring and ecological connectivity assessment module (104) uses the deployed sensor network to continuously carry out real-time monitoring. Among them, the distributed temperature measurement optical fiber and ground temperature chain deployed along the entire corridor continuously collect ground temperature data once per hour to form a dynamic temperature measurement database during the operation period. The following describes the main workflow and logic of this system during the operation period.

[0155] The animal activity event stream processing engine (501) continuously receives on-site audio and image data streams, and the approach and crossing behavior analysis unit (502) determines whether the animal has successfully crossed based on multi-source localization, image sequences, and trajectory information. The ecological connectivity index calculation unit (503) calculates the connectivity dynamic score according to a preset period and cross-validates it with the cold region physical connectivity auxiliary index (504): (1) the deviation of the permafrost thermal state from the melting depth of the active layer during the same period. Characterization, based on the current multi-depth ground temperature, determines the freezing / thawing interface and the melting depth of the active layer, and then compares it with the melting depth of the same date or the same ecological season in the baseline period; when it exceeds the preset threshold, the system applies a weighting coefficient to the comprehensive ecological connectivity score according to the rules. (2) Snow connectivity coefficient is based on the normalized snow cover of the key channels. ,according to Calculate and combine snow depth and duration to determine whether a "snowbound obstacle" event has occurred.

[0156] The dynamic regulation and decision feedback module (105) is responsible for transforming the assessment results into management actions. The early warning push unit (505) is configured with connectivity early warning, permafrost thermal state early warning, thaw settlement early warning, and snow cover early warning: the connectivity early warning is triggered when the ecological connectivity score falls below the corresponding value within a continuous assessment period; the permafrost thermal state early warning is triggered when the ecological connectivity score falls below the corresponding value within a continuous assessment period; the permafrost thermal state early warning is triggered when the ecological connectivity score falls below the corresponding value within a continuous assessment period. (t) Exceeding limits triggering warnings; Melt subsidence warnings are triggered based on cumulative melt subsidence exceeding limits as monitored by GNSS, hydrostatic level, or displacement gauges; Snow cover warnings are triggered based on snow cover rate, snow depth, and duration exceeding limits. For example, if the snow cover rate at the main entrance reaches [a certain threshold] in winter... If the snow depth reaches 50cm and the duration meets the snow closure conditions, the system generates a snow closure warning and pushes a snow removal or backup passage activation suggestion from the corridor management strategy knowledge base (506). During operation, the deformation and thermal status of key structures such as elevated walkways, heat pipes, adjustable bases, and snow fences can be located and evaluated by corresponding sensors and inspection data.

[0157] The execution feedback interface (507) records management measures and their effects. For example, if the success rate of crossing increases for several consecutive evaluation cycles after the implementation of a traffic control measure, the knowledge base can increase the recommendation weight of the strategy and continue to adjust it based on the effect data in subsequent operations. The online model learning and parameter calibration unit (106) can adaptively calibrate the resistance surface using newly added successful crossing events on a quarterly or other preset cycle: when the actual crossing intensity of a grid is continuously higher than the model prediction range, the system lowers the resistance value of the grid by a preset step size; when the actual crossing intensity is continuously lower than the prediction range, the corresponding resistance value is raised. Through continuous calibration, the system gradually improves the consistency between the resistance surface and actual ecological observations, and can identify seasonal passage areas that are not easily found directly in remote sensing images.

[0158] The seasonal dynamic parameter configuration module (107) serves as the system's rhythm controller. It incorporates a temperature state machine, taking daily multi-depth ground temperature measurement sequences as input. After moving average, it combines the spatial layer of the measurement points, temperature range, trend, and duration to determine the ecological season. In this embodiment, a predetermined number of consecutive temperature measurement points at representative depths of the active layer can be used as discrimination points. First and second temperature thresholds calibrated through long-term ground temperature observations in the study area are employed; for example, the first temperature threshold can be -0.5℃ and the second temperature threshold can be +0.5℃. A hysteresis zone of approximately 1℃ is also set, and the actual parameters are calibrated based on soil type, water content, and measurement error. Upon identification of the start of the full freezing season, the module provides the corresponding seasonal parameter set to the resistance surface unit, loads preprocessed parameters adapted to winter background noise and snow accumulation scenarios into the AI ​​unit, transmits historical baselines from the same period to the connectivity assessment unit, and filters out management strategies inapplicable to the current season from the knowledge base. During the spring thawing period, the corresponding thawing and drainage monitoring plans are activated.

[0159] Throughout the system's operation, the data security and privacy protection module (108) encrypts and controls access to all data streams. The system robustness and fault tolerance module (109) monitors the overall system operation status and provides fault tolerance functions such as edge caching and breakpoint resume, sensor redundancy and voting mechanisms, and degraded operation of core services.

[0160] The specific implementation of the sensor redundancy and voting mechanism is as follows: Key sensors deployed along the corridor can be configured with dual or triple redundancy based on risk level; at monitoring points requiring triple redundancy, three independent sensors collect the same physical quantity in parallel. The voting unit can employ algorithms such as majority consensus, weighted fusion, or median method. For continuous measurements, majority consensus is based on a preset consensus tolerance. When the difference between the readings of at least two sensors is within the tolerance range, the fusion result is taken as the valid output, and sensors with larger deviations enter a verification state; if none of the readings meet the consensus condition, an alarm is triggered. Weighted fusion can dynamically allocate weights based on the historical accuracy, health status, and confidence level of each sensor; the median method is used to suppress single large deviations. After voting, the system continuously compares the fusion result with the original data of each sensor. When a sensor exceeds the allowable deviation multiple times consecutively, it is marked as a suspected fault and isolated, while the abnormal pattern is recorded. When the number of available sensors is lower than a preset threshold, the system enters a degraded operation mode and pushes a maintenance alarm to maintain core monitoring and early warning functions.

[0161] III. Emergency Response to Extreme Operating Conditions: Logic for Handling Combined Snow Covering and Melting Events Based on Temperature Measurement Data

[0162] If continuous snowfall accompanied by abnormally high temperatures occurs during the late winter to spring thawing transition period, it could lead to corridor... The snow depth at the main entrance has reached the preset snow closure condition, triggering a snow closure warning and snow removal recommendations. (Assuming the same time period...) Multi-depth geothermal data in the segment showed an abnormal downward shift of the freeze / thaw interface, and the deviation of the melting depth of the active layer during the same period was calculated. If the preset threshold is exceeded, and the settlement displacement monitoring shows that the cumulative thaw settlement has reached the corresponding warning condition, the system will trigger both the frozen soil thermal state warning and the thaw settlement warning, and then... The weighting rules were adjusted to modify the ecological connectivity score, and suggestions for on-site structural, drainage, and abnormal heat source inspections were pushed out.

[0163] Based on the knowledge base suggestions pushed by the system, maintenance personnel can adjust the height of slightly subsided elevated walkways and repair them using backfilling, drainage, or insulation / cooling measures that meet the requirements of permafrost thermal stability. Simultaneously, anomaly tracing can be conducted using the database; if newly formed gullies are found to alter local drainage and cause surface water to accumulate and infiltrate, drainage measures can be implemented to restore a reasonable drainage path, provided that ecological access and engineering safety requirements are met. If abnormal heat exchange efficiency of local heat pipes or other cooling facilities is detected, they should be repaired or replaced, and the thermal recovery process after repair should be continuously verified using multi-depth geothermal data.

[0164] The complete data of such events, including abnormal temperature time series, melt-settlement displacement series, changes in drainage conditions, intervention measures and their effects, can be recorded and fed back to the thermal performance simulation sub-unit of the cooling facility for calibrating boundary conditions and related thermal parameters in subsequent numerical models. Simultaneously, the event handling results are entered into the corridor management strategy knowledge base, adjusting the recommended weights of relevant drainage inspections, cooling facility checks, and other strategies, and will continue to be updated based on newly added effect data during subsequent operation.

[0165] In summary, the system of this invention can be driven by multi-source data such as temperature measurement, structural deformation, snow accumulation and animal activity in cold and extreme event scenarios. It can trigger early warnings, push management suggestions, record execution effects and carry out parameter calibration according to preset logic, forming a closed-loop operation mechanism from temperature perception, state assessment, regulation to learning and updating.

[0166] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered exemplary and not restrictive in all respects. Furthermore, it should be understood that although this specification describes embodiments, it does not encompass only one technical solution. This descriptive method is merely for clarity, and those skilled in the art should consider the specification as a whole. The technical solutions in the embodiments can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A smart planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, characterized in that: include The multi-source remote sensing and field perception acquisition module is configured to acquire multi-source perception data covering the area along the proposed roadbed. The multi-source perception data includes at least satellite remote sensing images, UAV remote sensing images, acoustic audio data, infrared image data, multi-depth ground temperature measurement data and surface temperature measurement data collected by temperature sensors deployed along the proposed roadbed and corridor candidate area, as well as settlement displacement monitoring data and permafrost property data. The AI ​​recognition and species diversity spatiotemporal database module is communicatively connected to the multi-source remote sensing and field perception acquisition module. It is used to process the multi-source perception data to obtain species distribution information, diversity indicators, land cover type and environmental spatial data, and store the processing results in the spatiotemporal database. The seasonal dynamic parameter configuration module is connected to the multi-depth geothermal measurement data. It is used to identify the current ecological season based on the comparison results of the multi-depth geothermal measurement data with the preset temperature threshold and temperature change trend, and output the corresponding seasonal label and the seasonal dynamic parameters corresponding to the seasonal label. The seasonal dynamic parameters include at least the set of resistance value parameters corresponding to the current ecological season. The animal migration path modeling and ecological corridor planning optimization module is used to determine the core habitat patches of the target species based on the species distribution information, construct dynamic resistance surfaces corresponding to different ecological seasons based on the land cover type, environmental spatial data and the resistance value parameter set, calculate the permafrost thermal stability index based on the multi-depth ground temperature measurement data and permafrost property data, use the permafrost thermal stability index as the engineering hard constraint of the candidate corridor scheme, and generate ecological corridor design schemes through path search and multi-objective optimization. The real-time animal activity monitoring and ecological connectivity assessment module is configured to receive real-time data from on-site sensors after the corridor is put into operation, obtain the current melting depth of the active layer based on real-time multi-depth ground temperature measurement data, calculate the deviation of the melting depth of the active layer relative to the same date or the same ecological season of the baseline period, and generate a dynamic ecological connectivity score corrected by the deviation by combining animal crossing monitoring data. The dynamic regulation and decision feedback module is used to determine whether the preset early warning conditions are met based on the dynamic score of ecological connectivity, the deviation of the melting depth of the synchronous active layer, the snow cover obstacle status and / or the monitoring results of settlement displacement, generate management measure suggestions corresponding to the early warning level, and record the effect data after the implementation of management measures through the execution feedback interface. In addition, a model online learning and parameter calibration unit is used to periodically calibrate the resistance parameters of the dynamic resistance surface based on the effect data and successful crossing events in the animal crossing monitoring data, so as to update the basis for generating subsequent ecological corridor assessment and management measures recommendations.

2. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 1, is characterized in that: The temperature sensors in the multi-source remote sensing and field perception acquisition module include a soil temperature and humidity probe, a surface temperature sensor, and a freeze-thaw depth measuring instrument. The maximum permissible error of the soil temperature and humidity probe and the surface temperature sensor shall not exceed ±0.2℃, and the response time shall be less than 5 seconds. The freeze-thaw depth measuring instrument uses distributed temperature measuring optical fiber or one-dimensional temperature chain, with the vertical temperature measuring point spacing not exceeding 10cm. The measurement data from each temperature sensor are timestamped using a unified time base to form a synchronously correlated temperature measurement time series; All temperature sensors operate within a temperature range of at least -45℃ to +50℃; the temperature sensing elements embedded in the soil are designed with low self-heating; the data acquisition terminal, communication equipment, or power supply unit are configured with low-temperature start-up or temperature-controlled heating measures according to the ambient temperature; the settlement displacement monitoring data are obtained from GNSS monitoring devices, hydrostatic levels, displacement gauges, and / or remote sensing deformation monitoring data.

3. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 2, is characterized in that: The seasonal dynamic parameter configuration module identifies the ecological season based on the following threshold criteria: Complete Freeze Season: A predetermined number of temperature measurement points within the activity layer remain below a preset first temperature threshold for a predetermined duration; Spring thawing season: shallow temperature measurement points change from below the first temperature threshold to above the preset second temperature threshold, and a freeze-thaw transition layer exists within the active layer; Complete melting season: A predetermined number of consecutive temperature measurement points within the active layer are above the second temperature threshold for a predetermined duration; During the autumn freezing transition period: the shallow temperature measurement points change from being above the second temperature threshold to being below the first temperature threshold; The first temperature threshold and the second temperature threshold are engineering management discrimination thresholds used to identify the freezing, thawing and transition states of the active layer. The first temperature threshold is lower than the second temperature threshold and is calibrated based on the soil type, water content and long-term geothermal observation data of the study area. The seasonal dynamic parameter configuration module has a built-in temperature state machine and hysteresis comparator. It takes the daily multi-depth ground temperature measurement time series as input, combines the spatial layer of the temperature measurement point, temperature range, temperature change trend and predetermined duration to output the corresponding seasonal label, and suppresses the frequent switching of seasonal labels caused by temperature measurement noise through hysteresis comparison.

4. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 3, is characterized in that: The animal migration route modeling and ecological corridor planning optimization module includes: The resistance surface construction unit assigns basic resistance values ​​to spatial pixels based on surface cover type, vegetation cover, terrain conditions, distance from roadbed, and other animal movement environmental factors, and loads the resistance value parameter set corresponding to different ecological seasons provided by the seasonal dynamic parameter configuration module to generate multiple seasonal dynamic resistance surfaces. The constraint calculation unit calculates the permafrost thermal stability index based on the multi-depth geothermal measurement data and the pre-acquired spatial data of volumetric ice content. The spatial data of volumetric ice content is obtained through borehole sampling, permafrost ice content tests, and / or geophysical inversion, and is then spatially interpolated to form raster data. The formula for calculating the permafrost thermal stability index is: ； In the formula: The thermal stability index of permafrost. The larger the value, the lower the thermal stability of the permafrost and the higher the risk of thermal disturbance and thaw settlement. The annual temperature range of the active layer is relatively large; The annual average ground temperature at a predetermined depth below the upper limit of permafrost; This represents the normalized volumetric ice content. This is the upper limit of the multi-year statistical values ​​for the annual temperature range of the active layer in the study area; This represents the upper limit of the absolute value of the annual average ground temperature of permafrost in the study area. , , The non-negative weighting coefficients satisfy the following conditions: Each normalization term is limited to the range of 0 to 1; The path search and optimization unit runs the minimum cost path algorithm on the multiple seasonal dynamic resistance surfaces to calculate candidate corridor paths, conducts cross-seasonal joint evaluation of the candidate corridors, and uses the permafrost thermal stability index to meet the preset allowable conditions as a hard constraint on the feasibility of the candidate scheme. It performs multi-objective iterative optimization with ecological connectivity and engineering economy as optimization objectives, and outputs ecological corridor design schemes.

5. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 4, is characterized in that: The resistance surface construction unit is also equipped with a temperature disturbance input interface for receiving surface temperature measurement data and obtaining the spatial distribution of surface temperature through surface temperature sensors, UAV thermal infrared data, and / or spatial interpolation; for spatial pixels The temperature disturbance correction factor is calculated using the following formula: ； In the formula: For pixels Temperature disturbance correction factor; For pixels The current surface temperature measurement; For the same ecological season pixels The historical average surface temperature for the same period; This represents the normal fluctuation range of historical surface temperature during the same ecological season. This refers to the target species-ecological seasonal temperature sensitivity coefficient, calibrated based on the target species' historical traversal data and / or species temperature response data. The value can be positive or negative, and its sign is used to characterize the direction of the effect of abnormal temperature rise or fall on animal movement resistance. when When the deviation exceeds a preset threshold of 1, the current seasonal base resistance value of the corresponding pixel is multiplied by [the threshold value]. Perform local corrections and The corrected resistance value is limited to a preset reasonable range to ensure that the correction coefficient is positive and to avoid overcorrection caused by a single temperature anomaly.

6. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 4, is characterized in that: The specific implementation of the permafrost thermal stability index as a hard constraint is as follows: when the average value of the permafrost thermal stability index of the area traversed by the candidate corridor path exceeds the first allowable threshold, or the maximum value exceeds the second allowable threshold, the candidate scheme is marked as infeasible and eliminated in the optimization iteration. The path search and optimization unit also obtains the spatial distribution of current density in the candidate corridor path buffer zone based on the circuit theory model, and determines the functional boundaries on both sides of the corridor based on the position where the current density decreases to a preset ratio relative to the maximum current density in the corridor center area, so as to obtain the corridor functional width.

7. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 6, is characterized in that: In the real-time animal activity monitoring and ecological connectivity assessment module, the deviation of the melting depth of the active layer during the same period is used to characterize the degree of deviation of the permafrost thermal state, and is calculated according to the following formula: ； In the formula: For a moment Corresponding deviation of melting depth of the concurrent active layer; The melting depth of the active layer is obtained based on the location of the freeze / thaw interface in the current multi-depth geothermal measurement data; The baseline value for the active layer melting depth corresponding to the same date or ecological season during the baseline period; A preset positive minimum reference depth is used to avoid The relative deviation is distorted when it is 0 or too small; when When the preset threshold is exceeded, the dynamic score of ecological connectivity is multiplied by a preset weighting coefficient; The real-time animal activity monitoring and ecological connectivity assessment module also includes a snow connectivity coefficient calculation unit, used to obtain the normalized snow cover rate of key corridor crossing areas based on winter remote sensing imagery, UAV imagery, and / or snow depth sensor data. ,in , and according to Calculate the snow connectivity coefficient; when the snow cover rate and its duration meet the preset conditions, generate an "snow-blocked obstacle" event label independently.

8. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 1, is characterized in that: The dynamic control and decision feedback module has built-in multi-level early warning rules, including: Frozen soil thermal state early warning: triggered when the deviation of the thawing depth of the active layer exceeds the limit; Melting settlement early warning: triggered when the cumulative melting settlement calculated from the settlement displacement monitoring data exceeds the limit; Snow cover warning: Triggered based on snow cover rate and duration exceeding limits; Connectivity warning: Triggered when the dynamic score of ecological connectivity falls below the corresponding preset score within a preset number of consecutive assessment periods; Each warning level corresponds to a different management measure push strategy; The dynamic control and decision feedback module also includes a corridor management strategy knowledge base, which is used to store adaptive management measures in a structured manner. Each management measure is associated with at least the applicable seasonal tag, triggering conditions, and historical implementation effects.

9. The intelligent planning and biodiversity dynamic adaptive management system for ecological corridors along roadbeds in permafrost regions, considering temperature disturbances, as described in claim 1, is characterized in that: The model online learning and parameter calibration unit uses a local resistance perturbation method based on adaptive grids for calibration. Specifically, it extracts the verified successful crossing event trajectory point set, determines the adaptive grid size based on the target species activity scale and positioning error, and calculates the ratio of the actual crossing intensity of each grid to the model-predicted crossing intensity. When the ratio is higher than the upper limit of the preset interval, the resistance value of the corresponding grid is reduced by a preset adjustment step size; when the ratio is lower than the lower limit of the preset interval, the resistance value of the corresponding grid is increased by a preset adjustment step size; the animal migration simulation is rerun and the calibration is repeated until the convergence criterion is met for a preset number of consecutive calibration cycles. The system also includes a data security and privacy protection module and a system robustness and fault tolerance module; the data security and privacy protection module is used to desensitize and classify access permissions for the precise distribution points of sensitive species; the system robustness and fault tolerance module is used to maintain data continuity and core monitoring and early warning functions in the event of communication link interruption, sensor anomaly, or core service failure through edge caching and breakpoint resume mechanism, sensor redundancy and voting mechanism, and core service degradation operation mechanism.

10. A method for intelligent planning and dynamic adaptive management of biodiversity along roadbeds in permafrost regions, considering temperature disturbances, characterized in that: Planning and management using the system described in any one of claims 1-9 specifically includes the following steps: S1: Deploy multi-source sensing devices along the proposed roadbed area to acquire multi-source sensing data and store it in a spatiotemporal database. The multi-source sensing data includes at least species activity data, land cover data, multi-depth ground temperature measurement data, land surface temperature measurement data, permafrost property data, and settlement displacement monitoring data. S2: Perform AI recognition processing on the multi-source sensing data to extract species distribution information, diversity indicators, land cover type and environmental spatial data, and determine the core habitat patches of the target species based on the species distribution information; S3: Identify the current ecological season based on multi-depth geothermal measurement data, preset temperature thresholds, and temperature change trends; generate a seasonal label and a set of resistance value parameters corresponding to the seasonal label. S4: During the planning stage, multiple seasonal dynamic resistance surfaces are generated based on the land cover type, environmental spatial data, and resistance value parameter sets corresponding to different ecological seasons. The permafrost thermal stability index is calculated based on multi-depth geothermal measurement data and volumetric ice content spatial data. The permafrost thermal stability index is used as a hard constraint for candidate corridor schemes. Ecological corridor design schemes are generated through path search, circuit theory analysis, and multi-objective optimization. S5: After the corridor is put into operation, the deviation of the melting depth of the active layer during the same period is calculated based on real-time multi-depth geothermal measurement data. Temperature disturbance correction is made on the current seasonal resistance surface based on the spatial distribution of surface temperature. Ecological connectivity dynamic score is generated by combining animal crossing monitoring data. S6: Based on the dynamic score of ecological connectivity, the deviation of melting depth of the active layer during the same period, the status of snow cover obstacles and / or the cumulative melting and settling, trigger the corresponding early warning, output management measures suggestions, and record the effect data after the implementation of the measures; S7: Based on the effect data and verified successful crossing events, periodically calibrate the local resistance parameters of the dynamic resistance surface, and use the calibration results for subsequent planning evaluation and management decisions.

Images