Power transmission and transformation project surface soil layered stripping system and stripping method based on quality feedback and landform adaptation

Abstract

The invention belongs to the technical field of ecological restoration, and particularly discloses a power transmission and transformation project topsoil layered stripping system and stripping method based on quality feedback and landform adaptation, and the system comprises a stripping decision module, an equipment execution cluster, a quality monitoring module and an intelligent regulation and control module. The stripping method comprises the following steps: generating an accurate stripping plan by means of multi-source data fusion; different function extension modules are configured according to the plain landform, the hill landform and the mountain landform so as to execute layered stripping; the quality of the stripped surface soil is monitored in real time by using a near infrared spectrum technology; and dynamically adjusting the stripping depth based on the quality feedback. According to the method, the technical obstacles of extensive planning, process lag, quality out-of-control, poor equipment adaptability and the like existing in a traditional surface soil stripping technology are broken through, efficient and accurate protection of surface soil resources, especially a soil seed bank, is achieved, and the ecological benefits of a power transmission and transformation project are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of ecological restoration and soil and water conservation engineering technology, and in particular to a topsoil layering stripping system and stripping method for power transmission and transformation projects based on quality feedback and topographic adaptation. Background Technology

[0002] Topsoil, a valuable resource rich in organic matter, microorganisms, and a soil seed bank, undergoes an extremely long formation process under natural conditions. During the construction of power transmission and transformation projects, the excavation of tower foundations and the construction of roads inevitably cause strong disturbance to the surface soil, leading to the waste of topsoil resources and damage to the ecological environment.

[0003] Currently, topsoil stripping work in power transmission and transformation projects mainly relies on general standards such as the "Code for Design of Soil and Water Conservation Engineering" (GB 51018-2014), which has the following obvious shortcomings: 1) Lack of precision in stripping planning: The reference range for stripping thickness (20-80cm) given in the specification is too broad and fails to accurately match the soil profile quality and geomorphological characteristics of the specific site. This results in a large degree of blindness in determining the stripping range and thickness, and the phenomenon of high-quality topsoil being missed or low-quality topsoil being stripped ineffectively is quite prominent.

[0004] 2) Relatively outdated stripping technology: General-purpose engineering machinery such as bulldozers and excavators are commonly used for overall mixed stripping. This stripping method has poor control over the stripping depth and seriously damages the natural profile structure of the soil, causing the topsoil, which is rich in nutrients and has a dense seed bank, to mix with the lower soil layers, greatly reducing the ecological restoration potential of the topsoil.

[0005] 3) Quality control is lagging: Existing technologies mainly rely on laboratory testing in the later stages to control the quality of stripped topsoil. It is impossible to obtain topsoil quality information in real time during the stripping operation. As a result, the operation process cannot be dynamically adjusted according to the actual soil quality, which is an "open-loop" control and makes quality traceability difficult.

[0006] 4) Insufficient equipment adaptability: There is a lack of specialized stripping equipment for different landform units such as plains, hills and mountains, especially for the construction difficulties in mountainous areas with high gravel content. This results in low efficiency and poor results in complex terrain conditions, and even the abandonment of stripping operations due to technical bottlenecks.

[0007] The shortcomings of existing technologies lie not only in their extensive planning and outdated processes, but also in the isolation of each link and the fragmentation of information flow. Planning (decision-making) and execution (equipment) rely on static information transmitted through blueprints, while execution and quality control depend on delayed manual sampling. This results in the entire stripping process being an "open-loop" system, unable to respond promptly to unexpected soil quality variations revealed during construction. For example, when actual organic matter is far higher than expected, the system cannot automatically increase the stripping thickness to collect more high-quality topsoil; conversely, when quality is poor, it cannot promptly reduce ineffective work, leading to wasted energy and time.

[0008] Although researchers have attempted to improve planning or equipment, as of the application date, no publicly available technical solution has been found that deeply integrates digital twins, modular equipment, in-situ quality monitoring, and real-time dynamic control to form a closed-loop control system. Therefore, existing technologies cannot fundamentally solve the problem of precise and intelligent topsoil stripping across geological terrain in power transmission and transformation projects. There is an urgent need in this field for a systematic topsoil stripping technology and equipment capable of achieving precise planning, intelligent execution, real-time monitoring, and dynamic control. Summary of the Invention

[0009] The primary objective of this invention is to overcome the shortcomings of existing technologies and provide a topsoil layering stripping system for power transmission and transformation projects based on quality feedback and terrain adaptation. This system is capable of realizing a closed-loop intelligent control system for the entire process of "perception-decision-execution-optimization" in topsoil stripping, fundamentally solving the problems of blindness and lag caused by information flow interruption in existing technologies.

[0010] Another objective of this invention is to provide a topsoil layering and stripping method for power transmission and transformation projects based on quality feedback and terrain adaptation. This method can dynamically adjust stripping operation parameters according to real-time monitored topsoil quality information, ensuring maximum protection of topsoil resources, especially the soil seed bank. Simultaneously, it possesses a dynamic control method with online learning capabilities, enabling the system to continuously self-optimize as the project progresses, becoming increasingly "intelligent" with use. To achieve the above objectives, this invention employs the following interconnected technical solutions: A topsoil layering stripping system for power transmission and transformation projects based on quality feedback and terrain adaptation includes: 1) Stripping Decision Module: A stripping suitability evaluation algorithm is used to construct a digital twin model of the stripping plan. This module is based on a Geographic Information System (GIS) platform, integrating spatial interpolation data of soil characteristics within the engineering route buffer zone, high-precision digital elevation model (DEM) data, remote sensing vegetation index data, and engineering design parameters. Through a preset stripping suitability evaluation algorithm (specifically a calculation model based on the Soil Quality Index (SQI), it outputs a three-dimensional stripping plan map with accurate stripping thickness and quality expectation attributes (mainly referring to organic matter content).

[0011] 2) Equipment execution cluster: Composed of a basic stripping unit and multiple functional expansion modules; The basic stripping unit constitutes the power and execution foundation of the equipment execution cluster, and its tracked walking mechanism provides the mobility of the entire unit. The hydraulic control system is integrated on the walking mechanism and is connected to the lifting cylinder of the wear-resistant alloy blade through high-pressure oil pipes and control valve groups, driving the blade to complete the cutting and lifting actions. The depth sensor (such as an ultrasonic sensor or a wire displacement sensor) is fixedly installed on the blade arm to detect the blade's height off the ground in real time and feed this depth signal back to the hydraulic control system, thus forming a local closed loop of measurement-feedback-execution for the stripping depth.

[0012] The functional expansion modules include a flat-type layered conveying module, a hilly-type screening-layered composite module, and a mountain-type crushing-screening-layered three-stage processing module. Each module has a different physical structure and function, and can be quickly connected and configured with the basic stripping unit according to the terrain features and construction requirements of the work area.

[0013] 3) Quality Monitoring Module: Integrated into the equipment execution cluster, this module includes a near-infrared spectral sensor, a GPS positioning module, and an image acquisition module. The GNSS positioning unit, in collaboration with the lidar, acquires precise three-dimensional coordinates of the work surface and simultaneously assigns this coordinate information to the real-time spectral data collected by the near-infrared spectral sensor. This generates soil quality parameters and operational geometric parameters with precise geographic location identifiers, which are then uploaded to the intelligent control module. The near-infrared spectral sensor is used during the stripping process to perform real-time scanning of the freshly stripped, unmixed topsoil in situ. It then uses a built-in partial least squares regression (PLSR) model to invert key quality parameters such as organic matter, moisture content, clay content, and total nitrogen. The data acquisition and processing frequency is no less than once per second.

[0014] 4) Intelligent Control Module: Deployed on a cloud server, this module receives real-time data from the quality monitoring module (i.e., soil quality parameters and operational geometric parameters with precise geographic location identifiers) and preset target data from the stripping decision module (i.e., the dynamic stripping thickness T and expected quality value at the corresponding geographic location on the three-dimensional stripping planning map). The real-time data includes at least the real-time organic matter content (OM). real Real-time peeling depth (D) real The preset target data includes at least the target peel thickness (T) at the corresponding coordinate point and the real-time location coordinates of the equipment. target ) and target organic matter content (OM) targetBy running a rule-based adaptive adjustment algorithm (control algorithm), a peeling depth adjustment command is generated and sent to the hydraulic control system of the equipment execution cluster. The core of this system—the electro-hydraulic proportional valve—precisely adjusts its valve core opening and direction according to the received command signal, thereby changing the flow rate and direction of hydraulic oil to the blade lifting cylinder, driving the cylinder piston rod to extend or retract. The cylinder piston rod is connected to the blade, ultimately translating into the raising or lowering of the blade height. This achieves dynamic and precise adjustment of the peeling depth with a millisecond-level response, realizing dynamic optimization of the peeling process.

[0015] The stripping decision module, equipment execution cluster, quality monitoring module, and intelligent control module are coupled together through data flow and control commands to form a real-time perception-decision-execution closed-loop control system for topsoil stripping.

[0016] Furthermore, in the stripping decision module, the stripping suitability evaluation algorithm uses the Soil Quality Index (SQI) as the core judgment criterion, and its calculation formula is as follows: SQI =α×OM norm +β×SD norm +γ×(1 - GC norm ) + δ×VI norm Among them, OM norm SD is the normalized value of organic matter content. norm GC is a normalized value of the seed bank density estimate. norm VI is the normalized value of gravel content. norm α represents the normalized value of the vegetation index; α, β, γ, and δ are weighting coefficients determined by the Delphi method, and α+β+γ+δ=1.

[0017] The peel thickness T has a dynamic functional relationship with SQI: T = T min + (T max - T min )×[(SQI - SQI min ) / (SQI max - SQI min )] k , where k is a nonlinear adjustment coefficient, with a value range of 0.8 to 1.5.

[0018] Furthermore, the hilly screening-layering composite module sequentially completes gravel screening and topsoil layering in a single workflow. Its screening mechanism employs a double-layer vibrating screen, with an upper screen mesh size of 20mm and a lower screen mesh size of 10mm. The screened clean topsoil directly enters the subsequent chain-plate layering conveyor mechanism, achieving automatic separation of the 0-10cm topsoil from the underlying soil.

[0019] Furthermore, the mountain-type crushing-screening-layering three-stage processing module integrates a hydraulic impact crushing mechanism, a vibrating screening mechanism, and a chain plate layering conveyor mechanism to form a continuous three-stage processing production line. The crushing mechanism has an impact power of no less than 120kJ and a frequency adjustable within the range of 300-500 times / minute, effectively crushing parent rock or boulders larger than 30mm.

[0020] Furthermore, the adaptive adjustment algorithm logic executed by the intelligent control module is as follows: 1) If the real-time detected organic matter content OM real Below the target value OM target 85%, that is, OM real <0.85×OM target Then, an instruction is generated to adjust the peeling depth to: D new = D current × (OM real / OM target ).

[0021] 2) If the real-time organic matter content OM real OM is higher than the target value target 115%, that is, OM real >1.15×OM target And the current depth D current Less than the maximum allowable depth D max Then the generation instruction is pressed by D. new = min(D max D current × [1 + η × (OM real - OM target ) / OM target Increase the peeling depth, where η is the gain coefficient, with a value ranging from 0.1 to 0.3.

[0022] 3) If OM real In OM target If the peeling depth is between 85% and 115%, then maintain the current peeling depth D. current .

[0023] Simultaneously, the algorithm couples the adjustment of the travel speed V; when the peeling depth adjustment exceeds ±10%, the travel speed V is adjusted synchronously. new = V current × (D current / D new To maintain an approximately constant processing volume per unit time.

[0024] A topsoil layering stripping method for power transmission and transformation projects based on quality feedback and terrain adaptation, comprising the following steps based on the aforementioned system: S1. Data Fusion and Planning Generation: A digital twin model is constructed based on multi-source data fusion, the Soil Quality Index (SQI) is calculated, and a three-dimensional stripping planning map containing the stripping thickness and expected quality value of each work unit is generated. Specifically, the construction process is carried out in a Geographic Information System (GIS) platform, through spatial interpolation and fusion algorithms, to unify soil attributes, topography, and remote sensing vegetation index data into grid cells under the same spatial coordinate system, and based on a preset Soil Quality Index (SQI) calculation model, each cell is assigned a stripping thickness and expected quality value. S2. Equipment Configuration and Terrain Adaptation: Based on the terrain type of the work area (plains, hills, mountains), configure corresponding functional expansion modules for the basic stripping unit to perform the layered stripping process; the functional expansion modules include a plain-type layered conveying module, a hill-type screening-layered composite module, or a mountain-type crushing-screening-layered three-stage processing module; S3. Layered stripping and real-time monitoring: The equipment cluster is activated to strip the topsoil according to the planned thickness. During the layered stripping process, the topsoil quality and operation parameters are monitored in real time and in situ using near-infrared spectroscopy and lidar. The topsoil quality includes at least organic matter content and moisture content, and the operation parameters include at least the real-time stripping surface elevation calculated based on lidar point cloud data and the equipment travel speed calculated based on GNSS / IMU data. S4. Quality Feedback and Dynamic Control: The intelligent control module compares real-time quality data with the planned target value. If the deviation exceeds the preset threshold (±15%), it generates a depth adjustment command through a multi-parameter coupled adaptive algorithm to optimize and adjust the depth and equipment travel speed of the stripping operation in progress or in the future. S5. Ecological Separation and Protection: Throughout the stripping process, the equipment execution cluster simultaneously completes the physical separation of the 0-10cm top seed bank enrichment layer from the underlying topsoil, and transports them to the storage yard for differentiated protection and marking.

[0025] Furthermore, the method also includes step S6: Model self-learning and optimization: The system collects planning data, real-time operation data, adjustment records and subsequent vegetation restoration monitoring data of each operation point, and iteratively optimizes the weight coefficients (α, β, γ, δ) in the SQI calculation formula and the parameters of the depth adjustment algorithm through machine learning algorithms to improve the accuracy of subsequent planning.

[0026] The core of this invention lies in constructing a real-time data closed loop. The stripping decision module no longer generates a static map, but rather a dynamic digital twin model. During operation, this model receives feedback data from the intelligent control module and fine-tunes its parameters, providing a more accurate basis for subsequent planning of similar terrains.

[0027] The modular design of the equipment cluster is key to its terrain adaptability. The three functional expansion modules are not simply additions or subtractions of functions, but rather specialized, integrated solutions addressing the core challenges of different terrains (plains: efficient transport; hills: gravel screening; mountains: boulder crushing). For example, the hill module seamlessly integrates screening and stratification on a single rigid frame, avoiding secondary soil mixing and structural damage caused by intermediate transfers—a process effect that general-purpose mechanical combinations cannot achieve.

[0028] The innovation of the quality monitoring module lies in its "in-situ" and "real-time" operation. A near-infrared spectral sensor is directly mounted behind the shovel blade to scan the soil profile that has just been stripped and has not yet reached the ground, ensuring the representativeness of the tested samples. Its high-frequency detection of 1 scan per second provides a data foundation for real-time control.

[0029] The innovation of the algorithm in the intelligent control module lies in its "multi-parameter coupling". It does not adjust the depth in isolation, but rather adjusts the depth in relation to the travel speed. While ensuring the quality control target, it also takes into account the stability of the operation efficiency, preventing the equipment from overloading due to a sudden increase in depth or running idle due to a sudden decrease in depth.

[0030] In the topsoil layering and stripping method of this invention, step S1, "constructing a digital twin model," is the key starting point, creating a virtual mapping of the physical entity. Step S3, "real-time, in-situ monitoring," is the sensing link that forms a closed loop. Step S4, "multi-parameter coupled adaptive algorithm," is the brain and nerves of the closed loop. Step S5, "physical separation and labeling," ensures traceability for later use. The newly added "online model optimization step" in claim 9 is the core of this method's scalability; it frees the system from the constraints of fixed parameters, enabling it to continuously approach optimal decisions through data-driven approaches.

[0031] Beneficial effects Compared with the prior art, the present invention has the following significant advantages: 1) A qualitative change from "experience-based decision-making" to "model-driven decision-making": Through digital twin models and the SQI index, the stripping plan is upgraded from "approximately how much" to "why this number", making the decision-making process transparent, quantifiable, and traceable.

[0032] 2) A technological revolution from “mixed stripping” to “original structure protection”: The modular equipment’s mandatory layering process not only separates the soil layers, but more importantly, it preserves the original granular structure of the soil and the microenvironment for the seed bank to the greatest extent possible. This is the fundamental reason for the multiplied effect of subsequent ecological restoration.

[0033] 3) Innovation of the model from "open-loop construction" to "closed-loop intelligent control": The first real-time control closed loop of "perception-analysis-decision-execution" in this field has been established, which transforms the topsoil stripping equipment from a mechanized tool into an intelligent agent with "eye-brain-hand coordination".

[0034] 4) Evolution of capabilities from “fixed algorithm” to “growth-oriented AI”: The online optimization mechanism enables the system to learn continuously and evolve on its own without relying on the perfection of the initial settings. Its control precision and planning accuracy will continue to improve with the accumulation of engineering data, forming a technical barrier that is difficult to replicate.

[0035] 5) By protecting the seed bank and combining water-retaining agents with other measures, the self-recovery capacity of the ecosystem is enhanced from the source, the vegetation recovery cycle is significantly shortened, the later maintenance cost is reduced, and the ecological and engineering benefits are unified. Attached Figure Description

[0036] Figure 1 This is the overall system architecture and workflow diagram of the present invention.

[0037] Figure 2 This is a flowchart illustrating the process of separating the decision-making module in this invention.

[0038] Figure 3 This is a comparison diagram of the working principles of the three functional expansion modules in the equipment execution cluster of this invention.

[0039] Figure 4 This is a closed-loop principle diagram of the quality feedback and dynamic control of this invention. Detailed Implementation

[0040] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides detailed implementation methods and operating procedures; however, the scope of protection of the present invention is not limited to the following embodiments.

[0041] refer to Figure 1-4 This invention provides a topsoil layering stripping system and method for power transmission and transformation projects based on quality feedback and terrain adaptation. The following specific embodiments illustrate this: Example: A 500kV power transmission and transformation project has a total line length of 120 kilometers, which crosses the Huanghuai Plain, the Liaoxi Hills and the Changbai Mountain range in sequence.

[0042] S1. Data Fusion and Digital Twin Model Construction S1.1 Data Acquisition and Preprocessing Soil data: 1:50,000 soil maps along the project route and corresponding physicochemical property data (including organic matter, mechanical composition, and pH value) were obtained from the provincial soil and fertilizer station. All data underwent consistency checks and smoothing, and obvious outliers (such as extreme values ​​with organic matter content >100g / kg or <1g / kg) were removed.

[0043] Topographic data: Purchase ALOS PALSAR DEM data with a resolution of 12.5 meters. In GIS software (such as ArcGIS 10.8), use the "Slope", "Aspect" and "Flow Accumulation" tools in the "Spatial Analyst" toolkit to generate slope maps (unit: degrees), aspect maps (unit: degrees) and topographic humidity index maps (dimensionless), respectively.

[0044] Remote sensing data: Sentinel-2 L2A level data (atmospherically corrected) from the previous growing season prior to construction were downloaded from the ESA Copernicus Open Access Hub. The Normalized Difference Vegetation Index (NDVI) map was calculated using Band 4 (red) and Band 8 (near-infrared) with the formula (B8 - B4) / (B8 + B4). The maximum NDVI value for the entire growing season was calculated as representative of the vegetation condition at that point.

[0045] Engineering design data: Obtain the coordinates of the center piles of the entire tower foundation (CGCS2000 coordinate system), the excavation range of the tower foundation (usually a circle with a radius of 10-15 meters) and the vector file of the center line of the construction road from the design institute.

[0046] S1.2 Spatial Interpolation and Feature Inference Organic matter spatial interpolation: Organic matter data and corresponding geographic locations from 102 soil profile points were imported into GIS. Ordinary kriging was used for interpolation in the Geostatistics Wizard. A spherical model was selected as the semi-variogram model. After cross-validation, the final model's prediction error (RMSE) was 2.8 g / kg, and the coefficient of determination (R²) was 0.76. A 0.5-meter resolution raster map of organic matter spatial distribution, OM_map, was generated.

[0047] Seed bank density estimation: Due to the high cost of directly measuring the seed bank, this invention uses a surrogate model for estimation. A random forest regression model is established: Feature variables: Extract the following from the above data for each sampling point: organic matter content, slope, aspect, topographic humidity index, and maximum NDVI value.

[0048] Target variable: In the early stage of the project, 30 representative locations were selected to determine the seed bank density (unit: seeds / m²) using the seedling germination method, which served as training samples.

[0049] Model training: Using Python's Scikit-learn library, the random forest parameter was set to n. estimators =100,max depth =10. Using 30 samples for training, the model achieved an R² of 0.71 with 10-fold cross-validation.

[0050] Spatial inference: Input the feature variables of each grid cell of the entire line into the trained model to predict the spatial distribution raster map (SD) of the seed bank density of the entire region. map .

[0051] S1.3 SQI Calculation and Peel Thickness Determination Data normalization: Use the max-min normalization method to normalize the OM. map SD map GC inferred gravel content map map (derived from soil texture conversion), and NDVI map The values ​​are uniformly converted to the [0,1] range. norm =(Value - Value min ) / (Value max - Value min ) Calculate SQI: In the GIS raster calculator, enter the following formula: SQI map = 0.35×OM norm + 0.30×SD norm + 0.20×(1-GC norm ) + 0.15×VI norm Determine the peel thickness: Set T min = 20 cm, T max = 60 cm. Based on previous experiments, the SQI was set... min = 0.2, SQI max = 0.9, nonlinear coefficient k = 1.2.

[0052] Grid calculator formula: T map = 20 + (60 - 20) ×Power( (Con(SQI map <0.2, 0.2, Con(SQI map >0.9, 0.9, SQI map)) - 0.2) / (0.9 - 0.2), 1.2 ) This formula ensures that regions with an SQI below 0.2 are stripped at 20cm, regions with an SQI above 0.9 are stripped at 60cm, and intermediate regions are calculated using a non-linear function.

[0053] Generate job file: (T) map and OM targetmap (by OM) map The data is directly exported as GeoJSON format and sliced ​​into vector tiles suitable for loading on mobile devices, then deployed on a local server. Each tile contains the ID, center point coordinates, peel thickness, and target organic matter content of each 0.5m grid cell.

[0054] The weighting coefficients, adjustment coefficients, and thresholds described above are only one preferred implementation method. They can be adjusted within a reasonable range according to the topography, soil, and vegetation characteristics of the specific engineering area, without departing from the core principles and protection scope of this invention.

[0055] S2. Equipment Configuration and Terrain Adaptation S2.1 Preparation of the basic stripping unit A Shantui SD22 tracked bulldozer was selected as the base platform.

[0056] Modifications: Blade: Replaced with a straight tilt blade made of domestic NM450 wear-resistant alloy steel, 2.8 meters wide, and equipped with an ultrasonic ranging sensor (model: SICK UM30-213111) on one side to monitor the blade's height off the ground in real time.

[0057] Hydraulic system: An electro-hydraulic proportional valve (model: Bosch REXROTH 4WRLE) is installed to receive electrical signals and precisely control the blade lifting cylinder. Essentially, it controls the flow rate and direction of hydraulic oil entering the cylinder by continuously adjusting the valve core opening and direction, thereby achieving precise and stepless control of the blade lifting speed and position.

[0058] Control system: An onboard industrial computer (IPC, model: Advantech ARK-3500) and a CAN bus interface card are added. The IPC has a built-in 4G module for downloading planning data from the server and uploading real-time data.

[0059] Positioning and attitude determination system: A GNSS / IMU integrated navigation system (model: Novatek SPAN-CPT) is installed on the top of the bulldozer, providing centimeter-level positioning accuracy and 0.1-degree pitch / roll angle accuracy to determine the precise three-dimensional coordinates of the blade end.

[0060] S2.2 Functional Expansion Module Connection and Debugging Flat-type layered conveyor module: Structure: This is a separate trailer articulated behind a bulldozer. At its front is a receiving hopper, and behind the hopper is a dual-channel chain conveyor. A movable hydraulic diverter baffle directs soil of 0-10cm to the left channel and soil of less than 10cm to the right channel.

[0061] Connection: Connects to the bulldozer rear axle traction point via a standard pin-type quick connector. Hydraulic lines use quick-change couplings. Upon power-up, the IPC automatically identifies the module type and loads the corresponding control program.

[0062] Hilly screening-layer composite module: Structure: This is a rigidly connected frame. At the front is the receiving hopper, below which is a double-layer eccentric block vibrating screen with a 22kW motor. The screen box inclination angle can be manually adjusted between 15° and 25°. The screened soil falls into the lower chain-plate layered conveyor mechanism.

[0063] Commissioning: Based on the gravel particle size distribution observed on-site, manually replace the screens: install a 20mm aperture manganese steel woven screen on the upper layer and an 8mm aperture polyurethane elastic screen on the lower layer. After startup, run under no-load for 5 minutes to check if the temperature rise of the vibration motor bearings is normal.

[0064] Mountain-type three-stage processing module: crushing, screening, and stratification. Structure: This is a module with its own power unit (diesel engine). It integrates the following components in sequence: Hydraulic impact crushing mechanism: It adopts a mobile hydraulic breaker (model: Sunward Intelligent SWPB170), with a working flow rate of 140-180 L / min, impact energy of 120kJ, and an adjustable impact frequency of 300-500 times / minute.

[0065] Vibrating screening mechanism: similar to the hill type, but the screen box is thicker and heavier to withstand the impact of large stones.

[0066] Chain plate type layered conveyor mechanism.

[0067] Connection: Connect to the base unit via a hydraulic quick-connect coupling and a high-voltage cable plug. After the IPC identification module is activated, the operator sets the impact frequency of the breaker on the touchscreen, for example, to 400 impacts / minute.

[0068] S3. Layering and Real-time Monitoring Operation Initiation: The operator selects the tower base number (e.g., "N158") for the current operation on the IPC touchscreen. The system automatically loads the planned tile data for that area from the server. Once GNSS positioning indicates that the equipment has entered the N158 area, the system issues a "Ready" prompt.

[0069] Stripping Execution: The operator presses the "Auto Start" button. Based on the planned thickness at the current location (e.g., 38.5cm) and the real-time blade height calculated by GNSS / IMU, the IPC continuously adjusts the opening of the electro-hydraulic proportional valve using a PID control algorithm, ensuring the blade is precisely maintained at the set depth. An ultrasonic sensor serves as auxiliary feedback for the depth closed-loop control.

[0070] Real-time monitoring process: Spectral data acquisition: In the hilly module, a near-infrared spectral sensor (model: SupNIR-2700 from Juguang Technology) is installed 30cm above the vibrating screen outlet. As soil flows through, the sensor triggers once per second, acquiring diffuse reflectance spectra in the range of 900-1700nm.

[0071] Data Preprocessing and Inversion: Spectral data is transmitted to the IPC via network cable. Dedicated software running on the IPC automatically performs Standard Normal Variable Transform (SNV) to eliminate scattering effects. Then, it calls a pre-stored PLSR model (established by collecting representative soil samples from the engineering area, determining their organic matter, moisture content, and other physicochemical parameters using standard methods, and simultaneously acquiring their near-infrared spectra; after preprocessing the spectral data through SNV, it is imported along with the physicochemical parameter data into chemometrics software (such as Unscrambler) or Python's scikit-learn library, and trained and cross-validated using partial least squares regression to obtain the quantitative analysis model; this model was established by measuring the spectra and organic matter content of 385 local soil samples using a FOS analyzer in the laboratory, with 9 principal factors, a predicted R² = 0.93, and RMSECV = 2.1 g / kg) for real-time inversion to obtain the organic matter content OM. real For example, at a work site in a hilly area, sensors collect a specific set of spectral data. After inputting this data into the model, the organic matter content (OM) is retrieved in real time. real =19.5 g / kg. Meanwhile, the planned target value OM at this point... targetThe value was 22.0 g / kg. The system then initiated the preventative fine-tuning logic based on this deviation (19.5 < 0.85 × 22.0 ≈ 18.7? which is not true, but the deviation is close to the threshold). This involved executing an adaptive adjustment algorithm: when the real-time detected organic matter content OMreal was lower than 85% of the target value OMtarget, an instruction was generated to reduce the peeling depth proportionally by Dnew = Dcurrent × (OMreal / OMtarget); when OMreal was higher than 115% of the target value OMtarget, and the current depth Dcurrent was less than the maximum allowable depth Dmax, an instruction was generated to increase the peeling depth by Dnew = min(Dmax, Dcurrent × [1 + η × (OMreal - OMtarget) / OMtarget]), where η is the gain coefficient, ranging from 0.1 to 0.3; simultaneously, the algorithm coupled with the adjustment of the travel speed V. When the peeling depth adjustment exceeded ±10%, the travel speed Vnew = Vcurrent × (Dcurrent / Dnew) was adjusted synchronously to maintain an approximately constant processing volume per unit time.

[0072] Geometric parameter recording: The GNSS / IMU system records the equipment's latitude, longitude, and elevation coordinates (X, Y, Z) and attitude (Pitch, Roll) at a frequency of 10Hz. Combined with the equipment's mechanical dimension model, the three-dimensional coordinates of the blade tip and the actual working surface elevation are calculated in real time. The actual peeling depth D is calculated by the difference between this elevation and the previous working surface elevation. real .

[0073] Data is packaged and uploaded: IPC uploads the timestamp, device ID, X, Y, Z, and OM. real D real The data is packaged into a JSON data packet and uploaded to the cloud database via a 4G network.

[0074] S4. Quality Feedback and Dynamic Control Decision triggering: The intelligent control module (deployed on a cloud server) continuously monitors the data stream from the device. When a new data packet is received, the control algorithm is triggered.

[0075] Execution of the control algorithm: Data matching: Based on the (X, Y) coordinates in the data packet, find the corresponding planning target value T in the digital twin model. target and OM target .

[0076] Deviation calculation: DeviationOM = (OMreal - OM target ) / OM target DeviationD = (D real - T target ) / T target The logical judgment and instruction generation process is as follows: The intelligent control module will monitor the real-time organic matter content (OM). real With the target value OM target The comparison is performed, and based on the comparison results and the current state of the equipment, one of the following four control strategies is executed: 1) Low-quality control strategy: When OM real Below OM target When the percentage reaches 85%, it is judged as low quality. The system will then proceed according to OM. real With OM target The ratio is used to calculate and set a new peeling depth D proportionally. new Make it smaller than the originally planned thickness T target Meanwhile, to maintain operational efficiency, the equipment's travel speed V was increased. new The depth is increased inversely proportional to the original depth and the new depth. Furthermore, the system sets a safe lower limit for the peeling depth at 15 cm. If the calculated D... new If the value is lower than this, the 15 cm measurement will be enforced.

[0077] 2) High-quality control strategy: When OM real Higher than OM target 115%, and the current actual stripping depth D real Less than the maximum allowable depth T max If the quality is deemed too high, the system will initiate a depth increase procedure, with an increase amount of Delta. D From the target thickness T target The new peel depth D is calculated by combining the gain coefficient η (0.2 in this example) and the relative quality deviation. new For the current depth D real With this increase Delta D The sum. Meanwhile, to avoid excessive processing volume per unit time, the travel speed V... new The depth is reduced inversely proportional to the original depth and the new depth. The system also sets a safe upper limit for the peeling depth as T. max If the calculated D new If this value is exceeded, press T. max implement.

[0078] 3) Depth calibration strategy: If the real-time quality is within the allowable range (i.e., between OM...)target (between 85% and 115%), but real-time stripping depth D real Relative to the planned thickness T target If the absolute value of the deviation exceeds 10%, it is judged as an excessive depth execution deviation. In this case, the system will ignore quality factors and, with the goal of regression planning, directly set the new stripping depth D. new Set as T target And adjust the travel speed V accordingly. new .

[0079] 4) Steady-state operation strategy: If none of the above conditions are met, i.e., the quality is within the allowable range and the depth is executed accurately, then the system maintains the current peeling depth D. current With the speed of travel V current The operation remains unchanged and continues in a steady state.

[0080] Command Issuance and Execution: The cloud server will control the command { "Cmd": "Adjust", "Depth": D new "Velocity": V new The data is transmitted to the vehicle's IPC via the 4G network. The IPC converts the D_new value into the target hydraulic cylinder length via the CAN bus and sends it to the electro-hydraulic proportional valve controller; it also converts the V_new value into the target engine speed and transmission gear position and sends it to the travel controller. The entire response time (from data upload to command execution) is less than 2 seconds.

[0081] S5. Ecological Separation and Protection The topsoil of 0-10cm is diverted to the left conveyor belt and falls into the dump truck numbered "N158-A".

[0082] The lower layer of soil fell into the dump truck with the serial number "N158-B".

[0083] Trucks transported the soil to designated stockpiles. The stockpiles were pre-compacted and covered with HDPE geomembranes.

[0084] When stacking, an excavator is used in conjunction with GPS to form a neat trapezoidal mound, and a "fish scale pit" is dug on top to collect rainwater.

[0085] Immediately cover the 0-10cm "N158-A" soil mound with black permeable non-woven fabric, and use an automatic sprinkler system to spray small amounts of water regularly based on soil moisture sensor data to maintain seed bank activity.

[0086] An RFID electronic tag is inserted next to each mound of soil. After scanning, all traceability information for that mound can be viewed through a mobile app.

[0087] S6. Online Model Optimization Data Collection: Upon project completion, the system automatically generated a dataset containing 15,420 records. Each record includes: point coordinates, planned SQI, and planned thickness T. target Real-time organic matter OM real Real-time Depth D real Whether regulation is triggered, and the vegetation restoration monitoring data (including vegetation cover, aboveground biomass and species richness) of the site one year later (or the corresponding site after backfilling).

[0088] Model training: Objective: To optimize the weights (α,β,γ,δ) in the SQI formula and k in the stripping thickness formula so that the predicted stripping thickness is most positively correlated with the subsequent ecological restoration effect.

[0089] Method: Using Python's XGBoost library. The planned thickness T... target The combined score of recovery effect and subsequent recovery is used as a feature variable. The model learns how to allocate (α, β, γ, δ, k) to ensure T... target It has the highest correlation with recovery results.

[0090] Parameter Update: The new parameters obtained during training (α=0.38, β=0.32, γ=0.18, δ=0.12, k=1.1) are automatically updated in the configuration file of the digital twin model. When proceeding to the next segment or similar project, the system will use these optimized parameters for initial planning, thereby accumulating experience and continuously improving performance.

[0091] Through the implementation of this embodiment, the project achieves the following: 1) Precision and intelligent topsoil stripping; 2) Post-monitoring showed that the time required for vegetation coverage to reach 85% in areas where the topsoil was stripped and backfilled using the present invention was shortened by about 40% compared to areas where the traditional mixed stripping method was used.

[0092] 3) It achieves dynamic optimization based on actual soil quality, reducing the risk of quality control failure in stripping operations by more than 90% from the traditional method of 'post-event discovery'.

[0093] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A topsoil layering and stripping system for power transmission and transformation projects based on quality feedback and terrain adaptation, characterized in that, include: The stripping decision module uses a stripping suitability evaluation algorithm to construct a digital twin model of topsoil stripping based on multi-source spatial data, and generates a three-dimensional stripping planning map containing dynamic stripping thickness and quality expectation value for the corresponding geographical location by calculating the soil quality index (SQI). The equipment execution cluster performs stripping tasks based on the above-mentioned three-dimensional stripping planning map. It includes a basic stripping unit and multiple functional expansion modules. The functional expansion modules include at least a flat-type layered conveying module, a hilly-type screening-layered composite module, and a mountain-type crushing-screening-layered three-stage processing module. The quality monitoring module, integrated into the equipment execution cluster, includes a near-infrared spectral sensor, a lidar, and a GNSS positioning unit. The GNSS positioning unit and the lidar work together to acquire the three-dimensional coordinates of the work surface and simultaneously assign this coordinate location information to the spectral data collected in real time by the near-infrared spectral sensor, thereby generating soil quality parameters with precise geographical location identifiers and uploading them to the intelligent control module. The intelligent control module is used to run a multi-parameter coupled adaptive adjustment algorithm based on the real-time data stream of the quality monitoring module and the preset target of the stripping decision module, and dynamically generate and issue adjustment instructions for stripping depth and travel speed to the equipment execution cluster.

2. The system according to claim 1, characterized in that, In the stripping decision module, the stripping suitability evaluation algorithm uses the Soil Quality Index (SQI) as the core criterion, and its calculation formula is as follows: SQI =α×OM norm +β×SD norm +γ×(1 - GC norm ) + δ×VI norm , Among them, OM norm SD norm GC norm VI norm These are the normalized values ​​of organic matter content, seed bank density, vegetation index, and gravel content, respectively; the weighting coefficients α, β, γ, and δ are determined by the analytic hierarchy process and prior engineering data, and satisfy α+β+γ+δ=1. The peel thickness T has a dynamic functional relationship with SQI: T = T min + (T max - T min )×[(SQI - SQI min ) / (SQI max - SQI min )] k , where k is a nonlinear adjustment coefficient, with a value range of 0.8-1.

5.

3. The system according to claim 1, characterized in that, The hilly screening-layer composite module integrates an inclined vibrating screening mechanism and a chain plate layered conveying mechanism on a single rigid frame. The vibrating screening mechanism uses a double-layer screen. The upper screen has a mesh size of 15-25mm to remove medium gravel, and the lower screen has a mesh size of 5-15mm to remove fine gravel. The clean topsoil after screening falls directly into the chain plate layered conveying mechanism to complete the automatic separation and stacking of the 0-10cm topsoil and the lower soil.

4. The system according to claim 1, characterized in that, The mountain-type crushing-screening-layering three-stage processing module integrates a hydraulic impact crushing mechanism, a vibrating screening mechanism, and a chain plate layering conveyor mechanism to form a continuous "crushing-screening-layering" three-stage processing production line; the hydraulic impact crushing mechanism has a single impact energy of not less than 120kJ, and the impact frequency is infinitely adjustable in the range of 300-500 times / minute, crushing parent rock or boulders with a particle size greater than 30cm.

5. The system according to claim 1, characterized in that, The near-infrared spectral sensor in the quality monitoring module is used to collect spectral data of the in-situ topsoil to invert its key quality parameters in real time. Its built-in quantitative analysis model is established using partial least squares regression (PLSR) or support vector machine regression (SVR) algorithms, which can simultaneously invert the organic matter content, moisture content, and clay content of the in-situ topsoil. The data acquisition and processing frequency is no less than once per second.

6. The system according to claim 1, characterized in that, The intelligent control module receives organic matter content data transmitted from the quality monitoring module and executes an adaptive adjustment algorithm. Specifically, When the real-time detected organic matter content OM real Below the target value OM target When it reaches 85%, the generation instruction is proportional to D. new =D current × (OM real / OM target Reduce the peeling depth; When OM real OM is higher than the target value target 115% of, and the current depth D current Less than the maximum allowable depth D max When generating instructions, press D. new = min(D max D current × [1 + η × (OM real - OM target ) / OM target Increase the peeling depth, where η is the gain coefficient, ranging from 0.1 to 0.3; Simultaneously, the algorithm couples the adjustment of the travel speed V; when the peeling depth adjustment exceeds ±10%, the travel speed V is adjusted synchronously. new = V current × (D current / D new To maintain an approximately constant processing volume per unit time.

7. The system according to claim 2, characterized in that, The system also includes a model optimization module, which collects historical operation data and subsequent ecological restoration effect data, and uses machine learning algorithms to iteratively optimize the calculation weights (α, β, γ, δ), nonlinear adjustment coefficient k, and gain coefficient η of the soil quality index SQI, and feeds the optimized parameters back to the stripping decision module and the intelligent control module.

8. A method for topsoil layering and stripping in power transmission and transformation projects based on quality feedback and terrain adaptation, characterized in that, Includes the following steps: S1: Construct a digital twin model based on multi-source data fusion to generate a three-dimensional stripping planning map with quality expectation value and dynamic stripping thickness; specifically, the construction process is to unify soil properties, topography and geomorphology and remote sensing vegetation index data into grid cells under the same spatial coordinate system in the geographic information system (GIS) platform through spatial interpolation and fusion algorithms, and assign stripping thickness and quality expectation value to each cell based on the preset soil quality index (SQI) calculation model; S2: Based on the above three-dimensional stripping planning map, according to the slope and gravel content of the working terrain, the basic stripping unit is configured with corresponding functional expansion modules to perform the layered stripping process; the functional expansion modules include a flat-type layered conveying module, a hilly-type screening-layered composite module, or a mountain-type crushing-screening-layered three-stage processing module. S3: During the layered stripping process, the topsoil quality and operational parameters are monitored in real time and in situ using near-infrared spectroscopy and lidar; wherein, the topsoil quality includes at least organic matter content and moisture content, and the operational parameters include at least the real-time stripping surface elevation calculated based on lidar point cloud data and the equipment travel speed calculated based on GNSS / IMU data. S4: Based on the deviation between the quality monitoring results and the planning target, i.e., the real-time monitored organic matter content OM real Compared with the planned target value OM target The relative deviation between them is dynamically adjusted by a multi-parameter coupled adaptive algorithm to adjust the stripping depth and the equipment travel speed. S5: Physically separate the top 0-10cm seed bank enriched soil from the lower soil layer, and pile, label, and maintain them separately.

9. The method according to claim 8, characterized in that, The dynamic adjustment is triggered by a combination of preset quality deviation threshold (±15%) and rate change threshold (±10%). The adjustment command automatically changes the stroke of the hydraulic cylinder of the blade and the speed of the travel motor in the equipment execution cluster via the CAN bus, thereby achieving precise control of the peeling depth and synchronous adjustment of the equipment travel speed.

Images