Canyon-type pumped storage power station earthwork balance optimization system and method

Abstract

This invention relates to the field of earthwork balancing technology, providing an earthwork balancing optimization system and method for canyon-type pumped storage power stations. The system includes the following steps: acquiring three-dimensional point cloud and image data; calculating excavation reduction zones and filling increment zones based on point cloud differential calculations, classifying and identifying topsoil, strongly weathered rock, weakly weathered rock, and fill bodies according to reflection intensity and color, and generating a material source distribution map; extracting a list of available excavated materials from the material source distribution map and an immediate demand list from the filling plan; performing conflict detection based on quality matching rules, spatial reachability rules, and time window rules, and filtering feasible pairings; setting each loading point as a material source seller and each filling area as a demand buyer, initiating a round of negotiation at set intervals to generate allocation instructions. This invention employs three rules for conflict detection and then uses a virtual market negotiation mechanism to allocate material sources, making it suitable for real-time on-site operation and easy to debug and adjust.

Description

Technical Field

[0001] This invention relates to the field of earthwork balance technology, specifically to an earthwork balance optimization system and method for canyon-type pumped storage power stations. Background Technology

[0002] Canyon-type pumped storage power stations are typically located in deeply incised canyons, characterized by narrow sites, steep slopes, and compact construction layouts. Their dams are mostly concrete-faced rockfill dams, requiring large volumes of fill and demanding high-quality fill materials. In traditional construction, the initial excavation often uses strongly weathered material, while a large amount of weakly weathered material is urgently needed in the early stages of filling, leading to either "high-quality material used in underutilized areas" or "material available but unusable." Furthermore, the excavation of the dam foundation and abutments must be completed before dam filling, but the downstream slope has not yet formed in the early stages of filling, making it difficult to prepare large quantities of waste material for disposal.

[0003] Existing technologies mostly employ static earthwork balance planning, relying on empirical formulas and manual scheduling, which cannot cope with dynamic changes during construction. Furthermore, existing methods lack online classification of material source quality, dynamic obstacle avoidance along transportation routes, and closed-loop feedback adjustment of fill compaction degree. Therefore, there is a need to provide an earthwork balance optimization system and method for canyon-type pumped storage power stations, aiming to solve the above problems. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an earthwork balance optimization system and method for canyon-type pumped storage power stations, so as to solve the problems existing in the background technology.

[0005] This invention is implemented as follows: a method for optimizing earthwork balance in a canyon-type pumped storage power station, the method comprising the following steps: The construction area is scanned to obtain three-dimensional point cloud and image data; based on the point cloud difference, the excavation reduction zone and the filling increase zone are calculated, and the topsoil, strongly weathered rock, weakly weathered rock and filling body are classified and identified according to the reflection intensity and color, and a material source distribution map is generated. Extract the available excavation material list from the material source distribution map and the immediate demand list from the filling plan; perform conflict detection based on quality matching rules, spatial accessibility rules and time window rules, screen feasible pairings, and issue material shortage or excess warnings; Each loading point is set as a material source seller, and each filling area is set as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score and bids based on the quality matching coefficient, urgency coefficient, and distance attenuation factor, and generates allocation instructions. This allows trucks to transport materials to the filling area according to allocation instructions and dynamic planning routes. At the loading point, machine vision is used to detect the gradation and sort oversized stones. Unmanned road rollers receive the compaction task and monitor the compaction degree in real time to complete the filling. After each filling layer is completed, the actual filling volume is obtained and compared with the cumulative transportation volume to calculate the deviation rate. Based on the deviation rate, the quality matching tightness, transportation distance attenuation index and loose laying coefficient are automatically adjusted, and construction sequence optimization suggestions are triggered.

[0006] Another objective of this invention is to provide an earthwork balance optimization system for a canyon-type pumped storage power station, the system comprising: The material source distribution module is used to scan the construction area and obtain three-dimensional point cloud and image data; based on the point cloud difference calculation, the excavation reduction zone and the filling increase zone are calculated, and the topsoil, strongly weathered rock, weakly weathered rock and filling body are classified and identified according to the reflection intensity and color to generate a material source distribution map. The rule matching module is used to extract the list of available excavated materials from the material source distribution map and the list of immediate needs from the filling plan; to perform conflict detection based on quality matching rules, spatial accessibility rules and time window rules, to screen feasible pairs, and to issue warnings of material shortage or excess. The allocation instruction module is used to set each loading point as a material source seller and each filling area as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score based on the quality matching coefficient, urgency coefficient, and distance attenuation factor and submits a bid, generating an allocation instruction. The filling execution module enables trucks to transport material sources to the filling area according to allocation instructions and dynamic planning paths. At the loading point, machine vision is used to detect the gradation and sort oversized stones. The unmanned road roller receives the compaction task and monitors the compaction degree in real time to complete the filling. The adjustment and optimization module is used to obtain the actual filling volume after each filling layer is completed, compare it with the cumulative transportation volume to calculate the deviation rate, automatically adjust the quality matching tightness, transportation distance attenuation index and loose laying coefficient based on the deviation rate, and trigger construction sequence optimization suggestions.

[0007] Compared with the prior art, the beneficial effects of the present invention are: This invention combines point cloud differential and reflection intensity classification to automatically update the material source distribution map, accurately grasping the type, volume, and location of excavated materials, overcoming the shortcomings of traditional manual material collection, such as long cycles and large errors. It employs quality matching rules, spatial reachability rules, and time window rules for conflict detection, and then uses a virtual market negotiation mechanism to allocate material sources. All decisions are based on simple and interpretable rules and arithmetic operations, requiring no linear programming or genetic algorithms, resulting in low computational load, suitability for real-time operation on on-site computers, and ease of debugging and adjustment. Trucks automatically drive according to allocation instructions and dynamic planning paths, utilizing virtual force field methods for real-time obstacle avoidance; machine vision automatically detects gradation and sorts oversized stones at loading points; unmanned road rollers automatically compact and monitor compaction. The entire process achieves mechanization and automation, reducing labor costs and improving construction efficiency. After each fill layer is completed, the quality matching tightness, transport distance attenuation index, and loose paving coefficient are automatically adjusted based on the deviation rate, triggering construction sequence optimization suggestions. It can effectively cope with uncertainties such as changes in geological conditions and fluctuations in blasting effects, continuously optimizing allocation strategies, reducing secondary transportation, and shortening the construction period. Attached Figure Description

[0008] Figure 1 This is a flowchart of the earthwork balance optimization method for a canyon-type pumped storage power station.

[0009] Figure 2 This is a structural block diagram of the earthwork balance optimization system for a canyon-type pumped storage power station. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0011] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0012] like Figure 1 As shown, this embodiment of the invention provides a data storage method, the method comprising the following steps: S100 scans the construction area to obtain three-dimensional point cloud and image data; based on point cloud differential calculation, it calculates the excavation reduction zone and the filling increase zone, and classifies and identifies topsoil, strongly weathered rock, weakly weathered rock and filling body according to reflection intensity and color, and generates a material source distribution map. S200: Extract the list of available excavated materials from the material source distribution map and the list of immediate demand from the filling plan; perform conflict detection according to the quality matching rules, spatial accessibility rules and time window rules, screen feasible pairings, and issue a warning of material shortage or excess. S300 sets each loading point as a material source seller and each filling area as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score and bids based on the quality matching coefficient, urgency coefficient, and distance attenuation factor, and generates allocation instructions. The S400 system enables trucks to transport materials to the filling area according to allocation instructions and dynamic planning routes. At the loading point, machine vision is used to detect the gradation and sort oversized stones. The unmanned roller receives the compaction task and monitors the compaction degree in real time to complete the filling. S500 obtains the actual filling volume after each filling layer is completed, compares it with the cumulative transportation volume to calculate the deviation rate, automatically adjusts the quality matching tightness, transportation distance attenuation index and loose laying coefficient based on the deviation rate, and triggers construction sequence optimization suggestions.

[0013] In this embodiment of the invention, the UAV automatically flies along a preset zigzag route at an altitude of 50-80m, with a forward overlap rate of no less than 70% and a lateral overlap rate of no less than 60%. A lidar emits multi-echo pulses to acquire topographic point clouds of the ground and vegetation, with a point density ≥ 50 points / m². A visible light camera simultaneously captures orthophotos for subsequent color classification. For areas difficult for the UAV to cover, such as steep slopes, cave exits, and narrow construction passages, tracked or wheeled robots are deployed. These robots are equipped with RTK-GPS (real-time dynamic differential positioning, horizontal accuracy ±2cm) and binocular cameras (baseline distance 30cm, capable of acquiring depth information). The robots travel along a preset path, automatically collecting a set of point clouds and images every 1 meter. The UAV point cloud and the ground robot point cloud are registered and fused using a common control point (such as a fixed reflective target) to fill blind spots and form a complete, high-precision global base point cloud. Then, temporal differential mapping is used to identify excavated and filled areas, and the material sources are classified according to reflection intensity and color to form a real-time digital map. Next, a matching analysis of material sources and demand is conducted. The available excavated material list comes from the digital map, and the demand list comes from the filling plan.

[0014] Then, a virtual market mechanism is used for material allocation. A round of negotiation and matching is initiated at set intervals (e.g., 30 minutes). The formula for calculating the matching score is: Quality Matching Coefficient × (1 / Remaining Demand Time Window) × (1 / (Transportation Distance + 1))^α, where α is the transportation distance decay index; the quality matching coefficient is either 0 or 1. If the material type matches the allowed material type in the demand area, M=1; otherwise, M=0, and the score is 0 (no bidding allowed). The remaining demand time window is defined as the duration between the current time and the planned deadline for the filling area. The urgency coefficient is the reciprocal; the shorter the time, the larger the coefficient, prioritizing urgent demands. (1 / (Transportation Distance + 1)) is the distance decay factor; the shorter the distance, the higher the score. Each demand buyer publishes a tender: including the material type, required volume (e.g., 500 cubic meters), expected arrival time, and maximum acceptable transportation distance. Each material seller (loading point) calculates a score for all tenders and selects the highest-scoring tender for bidding. The tender includes the seller ID and the available volume (not exceeding the current inventory and excavation rate of the material). All bids are aggregated. For each demand bid, all bids are sorted by score from highest to lowest, and the volume is allocated sequentially until the demand is met or all bids are exhausted. If the total volume of bids received for a bid exceeds the demand, the highest-scoring bids are given priority; if not, the demand is marked as "unmet" and enters the next round of negotiation 30 minutes later. All winning bids generate a dispatch instruction table, including: material source point → demand area, transport volume, latest departure time, route number, etc.

[0015] In this embodiment of the invention, trucks transport material to the filling area according to allocation instructions and dynamically planned paths. At the loading point, machine vision is used to detect the gradation and sort oversized stones. An unmanned road roller receives the compaction task and monitors the compaction degree in real time to complete the filling. The dynamically planned path generates multiple candidate paths based on the latest digital terrain and road closure information of the construction area. Each truck obtains the position of other vehicles within 50 meters through V2V communication and uses a virtual force field method to adjust its speed and steering in real time, where the target point generates attraction and surrounding vehicles generate repulsion. After each filling layer is completed, the actual filling volume is obtained and compared with the cumulative transport volume to calculate the deviation rate. Deviation rate = (cumulative transport volume - actual filling volume) ÷ cumulative transport volume. Based on the deviation rate, the quality matching tightness, transport distance attenuation index, and loose paving coefficient are automatically adjusted, and construction sequence optimization suggestions are triggered. Construction sequence optimization suggestion trigger: Real-time monitoring of the following conditions - action pair: Condition A - The filling progress of the upstream rockfill area of ​​the upper reservoir is more than 5% behind the plan and the deep weakly weathered rock (buried depth greater than 30m) of the lower reservoir quarry is expected to be exposed within the next 48 hours; Action A - Generate a suggestion to "exploit the lower reservoir quarry in advance and transport it to the upstream rockfill area of ​​the upper reservoir", and at the same time estimate that it can reduce the amount of secondary transfer.

[0016] Condition B—The remaining capacity of the current spoil heap behind the dam is less than 10%, and the planned excavation volume of unused materials (topsoil and strongly weathered rock that do not meet the filling standards) is greater than 5000 m³ in the next 24 hours; Action B—Generate a suggestion to "suspend excavation of unused materials and prioritize filling the downstream rockfill area above the slope to increase the elevation and capacity of the slope." Condition C—The inventory of a certain transfer yard is higher than 80% of the design capacity for 24 consecutive hours, and the filling rate of the corresponding demand area is consistently lower than the plan; Action C—Generate a suggestion to "increase the number of filling equipment in this demand area or extend the filling operation time"; All suggestions are sent to the construction management platform as structured messages and are executed automatically by the system or after manual confirmation.

[0017] As a preferred embodiment of the present invention, the steps of calculating the excavation reduction zone and the filling increase zone based on point cloud differential calculation, classifying and identifying topsoil, strongly weathered rock, weakly weathered rock, and filling body according to reflection intensity and color, and generating a material source distribution map specifically include: S101, after registering the current point cloud with the previous point cloud, the registration adopts the ICP iterative nearest point algorithm to calculate the elevation change point by point and generate a change raster map; the area with an elevation decrease threshold > 0.3m is marked as the excavation reduction area, and the area with an elevation increase threshold > 0.3m is marked as the filling increase area. For each connected area, the volume change is calculated by integrating the triangular mesh model. S102. Within the excavation reduction zone, the reflection intensity value and RGB color components are read point by point, combined with the local point cloud slope. If the reflection intensity is <80 and the ratio of R component to G component in RGB is between 0.8 and 1.2, and the slope is <30°, it is determined to be topsoil. If the reflection intensity is 80-160 and RGB is yellowish-brown (R≈180, G≈150, B≈100, allowable ±20 deviation) and the point cloud is blocky, it is determined to be strongly weathered rock. If the reflection intensity is >160 and RGB is grayish-white (R, G, and B are all within the range of 200±30), it is determined to be weakly weathered rock. Within the filling increment zone, the presence of a uniformly thick layered structure in the point cloud is checked, with each layer being 40-60cm thick. If it exists, it is determined to be filling material. S103 assigns different labels to the classified topsoil, strongly weathered rock, and weakly weathered rock, and records the spatial location, natural volume, remaining excavation time (estimated based on the current excavation progress rate), and adjacency relationship (whether it is connected to the construction road) of each independent material source block; overlays all material source block information with the filling body information of the filling increment area to generate a material source distribution map, including the material source location, type, volume, status, and a real-time digital map of the demand area location, type, filled volume, and remaining demand, which is automatically updated every 30 minutes.

[0018] As a preferred embodiment of the present invention, the quality matching rule stipulates that weakly weathered rock can only be used in the upstream rockfill area and transition area, strongly weathered rock can be used in the downstream rockfill area or slope protection, and topsoil can be used in temporary storage areas or greening cover soil; the spatial accessibility rule automatically determines whether the transportation route is available based on the real-time vehicle trajectory congestion level. If the congestion index exceeds a preset threshold, it is marked as unavailable, and any transportation route that needs to pass through this section is temporarily blocked until the congestion is relieved; the time window rule requires that the material source excavation time is earlier than the filling deadline of the demand area. If it is violated, it is marked as infeasible; the material shortage warning trigger condition is that the expected material shortage ratio exceeds 10% in the next 4 or 8 hours; the excess warning trigger condition is that a certain material source is expected to be unable to be consumed within 48 hours and the remaining capacity of the stockpile site is lower than the warning line, and it prompts that it is necessary to accelerate consumption or increase the temporary stockpile site.

[0019] As a preferred embodiment of the present invention, the step of performing conflict detection and screening feasible pairings based on quality matching rules, spatial reachability rules, and time window rules specifically includes: Step S201: Traverse all possible combinations of material sources and demand, and use quality matching rules to perform the first layer of hard constraint filtering to generate a preliminary screening pairing set.

[0020] Each material source in the available excavation material list extracted in step S200 is paired one-to-one with each demand area in the immediate demand list. For each pairing, the rock classification of the material source (topsoil, strongly weathered rock, weakly weathered rock) and the permitted material type of the demand area are read. If the rock classification belongs to the permitted material type of the demand area, the pairing passes quality matching, is marked as initially screened as valid, and is included in the initial screening pairing set; otherwise, it is directly rejected. For example, a weakly weathered rock material source paired with a downstream rockfill area that only requires strongly weathered rock will be directly rejected. The output of this step is a set of material source-demand pairings that meet the basic quality requirements.

[0021] Step S202: For each pair in the initial screening pairing set, perform the second-level dynamic constraint filtering using the spatial reachability rule to generate a reachable pairing set.

[0022] In this step, the spatial reachability rule is not a simple binary judgment, but a dynamic quantitative assessment based on the real-time road network status. A pair is selected from the initial screening pairing set to determine the shortest transportation path from the material loading point to the required filling area. The current congestion index of this path is calculated using traffic flow detection equipment deployed on the construction roads or real-time vehicle location data. If the congestion index exceeds a preset unavailability threshold, the path is deemed unavailable, and the pair is temporarily marked as unreachable and removed. If the congestion index does not exceed the threshold, the transportation reliability coefficient of the path is further calculated. This coefficient is a preset function value that decreases as the congestion index increases, representing the reliability of the path in ensuring timely transportation under the current road conditions. Pairs verified by the spatial reachability rule, along with their transportation reliability coefficients, are included in the reachable pairing set. The output of this step is a pairing set that is not only reachable but also includes the current path reliability level.

[0023] Step S203: For each pair in the reachable pair set, perform third-level time-series compliance filtering using time window rules to generate the final list of feasible pairs.

[0024] For each pair in the reachable pairing set, the estimated excavation completion time of the material source (i.e., the material source availability time) and the planned filling deadline of the demand area are obtained. If the material source availability time is earlier than the planned filling deadline, the pair is ultimately determined to be a feasible pairing; if the material source availability time is later than the deadline, it is determined to be infeasible and is discarded. Pairings verified by the time window rule, together with the transportation reliability coefficient calculated in step S202, constitute the final feasible pairing list. This list serves as input to the virtual market negotiation mechanism in subsequent step S300, for buyers (demand areas) to refer to when issuing tenders, and as the basis data for sellers (loading points) to calculate their bid matching scores. Material shortage or oversupply warnings are triggered based on the comparison between the total supply capacity and total demand of the feasible pairing list.

[0025] In this embodiment of the invention, the static description of "rule listing" is abandoned, and a strict sequential filtering process of "quality → space → time" is constructed. Each layer of filtering has clear inputs, processing logic, and outputs, solving the technical implementation problem of multi-rule comprehensive judgment. The process has a high degree of systematicity and operability. Furthermore, the spatial reachability rule is upgraded from a simple "on / off" condition to a continuous quantitative evaluation of path service quality. This coefficient is not only the basis for filtering in this step, but also serves as a dynamic weight for subsequent virtual market bidding calculations, enabling the final allocation instruction to automatically avoid paths with high congestion risk, realizing an intelligent closed loop from perception to decision-making.

[0026] In a preferred embodiment of the present invention, the step of issuing a material shortage or over-limit warning specifically includes: Step S204: Based on the feasible pairing list, calculate the supply-demand gap rate within the future preset time period and determine the early warning level.

[0027] The data source for this step is the feasible pairing list generated in step S203. The system extracts the total material demand volume for all demand areas and the total available material volume for all material sources from this list. The supply-demand gap ratio is calculated as follows: Supply-demand gap ratio = (Total material demand volume - Total available material volume) / Total material demand volume × 100%. Based on the supply-demand gap ratio, the system classifies the warning level into three levels: When the supply-demand gap ratio is ≤ 5%, it is considered a normal state and no warning is triggered; When the supply-demand gap ratio is between 5% and 10%, it is judged as a mild warning and triggers a yellow warning signal; When the supply-demand gap ratio is between 10% and 20%, it is judged as a moderate warning and triggers an orange warning signal; When the supply-demand gap ratio exceeds 20%, it is considered a severe warning and triggers a red warning signal.

[0028] Simultaneously, the system calculates the projected material shortage ratios for the next 4 hours and 8 hours, as well as the projected consumption time for each material source and the remaining capacity of the storage site, as supplementary judgment criteria. The warning level and supply-demand gap rate data are output to subsequent steps.

[0029] Step S205: Automatically adjust the operating parameters of the virtual market negotiation mechanism according to the warning level.

[0030] Based on the warning level determined in step S204, this step dynamically adjusts three key operating parameters of the virtual market negotiation mechanism in the subsequent step S300: First, adjust the negotiation interval. When the warning level is normal or mild, maintain the default negotiation interval (e.g., 30 minutes); when the warning level is moderate, shorten the negotiation interval to half of the default value (e.g., 15 minutes); when the warning level is severe, shorten the negotiation interval to one-quarter of the default value (e.g., 7.5 minutes) to speed up the supply and demand matching frequency.

[0031] Secondly, adjust the initial value of the transportation distance attenuation index α. When a material shortage warning is detected, it indicates a tight supply of materials. In this case, the restriction on transportation distance should be relaxed to expand the range of available material sources. Specifically: for a mild warning, the α value remains at the default of 1.0; for a moderate warning, the α value is adjusted to 0.8; and for a severe warning, the α value is adjusted to 0.5. The smaller the α value, the smaller the impact of transportation distance on the matching score, and the more the system tends to select material sources that are far away but of acceptable quality.

[0032] Third, when an oversupply warning is detected and a shortage warning is also present, the tender documents for the demand area corresponding to the shortage warning will be processed first. When generating allocation instructions, the system will assign the highest priority to the tender documents for the demand area that triggered the shortage warning, ensuring that limited transportation capacity prioritizes the material needs of key areas; the material sources corresponding to the oversupply warning will be put into the accelerated consumption scheduling queue, and the system will lower the bidding threshold for such material sources in subsequent negotiations.

[0033] Step S206: Based on the warning level and operating parameters, generate a structured warning message, push it to the construction management platform, and record the warning log.

[0034] Based on the processing results of steps S204 and S205, the system generates a structured early warning message. The message includes: early warning level, trigger time, supply-demand gap rate, a list of material shortage areas and their respective expected shortage ratios, a list of excess material sources and their respective expected full-stocking times, and automatically adjusted operating parameters (including the new negotiation interval time and the adjusted transport distance attenuation index α). This early warning message is pushed to the construction management platform in real time in system message format for construction management personnel to view and confirm. Simultaneously, the system writes the complete information of this early warning into an immutable early warning log. The log records include: trigger conditions, decision basis, numerical comparisons before and after parameter adjustments, and adjustment timestamps, for subsequent construction review and strategy optimization.

[0035] In this embodiment of the invention, the invention upgrades the "issuing early warning" from a passive information prompt to an active control trigger node, realizing a complete closed loop of "monitoring → evaluation → early warning → parameter adjustment → re-monitoring". Early warning is no longer an isolated alarm signal, but a direct driving force for the adaptive adjustment of core parameters such as negotiation interval and distance decay index, significantly improving the system's response speed and adjustment capability to supply and demand imbalances. Furthermore, the supply and demand gap rate is quantified into three levels of early warning, with different negotiation interval and decay index adjustment strategies preset for each level. This hierarchical response mechanism avoids the problems of "overreaction" or "underreaction" caused by a single threshold, and makes system control based on evidence and smooth and gradual, making it particularly suitable for the complex and dynamic environment of construction sites. For the contradictory scenario of simultaneous material shortages and excesses, the invention clarifies the principle of "prioritizing the material shortage area" and the accelerated consumption mechanism of excess material sources, resolving the decision-making conflict problem when two types of early warnings coexist, and ensuring the stable operation of the system under complex supply and demand situations.

[0036] In a preferred embodiment of the present invention, the steps of the seller calculating a matching score based on the quality matching coefficient, the urgency coefficient, and the distance attenuation factor, bidding, and generating an allocation instruction specifically include: Step S301: The seller receives all the demand tenders issued by the buyers in this round of negotiations and extracts the key parameters of each tender. The key parameters include at least the required material type, required material volume, filling deadline, and coordinates of the filling area.

[0037] At set intervals (e.g., 30 minutes), the system initiates a round of negotiation. All filling areas (demand buyers) publish their demand bids, which must include at least: the required material type, the required volume, the filling deadline, and the coordinates of the filling area. Each loading point (material seller) receives all valid demand bids in this round. For each bid, the seller extracts: the required material type, the required volume, the demand deadline, and the coordinates of the demand area. Simultaneously, the seller retrieves from its own status information: the currently available material types, the currently available volume (dynamically calculated based on current inventory and the projected mining rate), and the coordinates of the loading point. This step provides the data foundation for subsequent calculations.

[0038] Step S302: For each tender document, the seller calculates the bid matching score based on the key parameters; this includes three sub-steps: quality matching verification, urgency factor calculation, and distance attenuation factor calculation.

[0039] First, a quality matching verification is performed. The seller compares its current available material sources with the material types required in the tender. If they match, the quality matching coefficient is set to 1, and the calculation continues. If they do not match, the quality matching coefficient is set to 0, the tender's matching score is 0, and the seller does not participate in the bidding for that tender.

[0040] Secondly, for tenders with a quality matching coefficient of 1, an urgency coefficient is calculated. This urgency coefficient is quantified based on the remaining time window of the tender's requirements. The remaining time window is defined as the length of time between the current system time and the tender's deadline for completion. The formula for calculating the urgency coefficient is: Urgency Coefficient = 1 / (Remaining Time Window + T0), where T0 is a preset normal value (e.g., 0.5 hours) to prevent the denominator from becoming too small and causing overflow when the remaining time window approaches zero. This formula ensures that the more urgent the requirement (the shorter the time window), the larger the urgency coefficient of the tender.

[0041] Next, the distance attenuation factor is calculated. Based on the coordinates of the seller's loading point and the required area in the tender, and according to the current road network model of the construction area, the seller calculates the actual transport distance of the shortest feasible path between the two points. The formula for calculating the distance attenuation factor is: (1 / (transport distance + 1))^α, where α is the transport distance attenuation index. The initial value of this index is preset to 1.0, and it can be dynamically adjusted according to the deviation rate in step S500. This formula ensures that the shorter the transport distance, the larger the distance attenuation factor.

[0042] Ultimately, the seller's bid matching score is calculated using the formula: Bid Matching Score = Quality Matching Coefficient × Urgency Coefficient × Distance Attenuation Factor. This score comprehensively reflects three dimensions: quality, time urgency, and spatial economy.

[0043] In step S303, the seller makes a bidding decision based on the bid matching score and its own supply constraints, generates a bidding proposal, and submits it.

[0044] Sellers rank all bids with a quality match coefficient of 1 according to their bid matching scores from highest to lowest. Instead of simply selecting the highest-scoring single bid, sellers use a greedy allocation strategy, taking into account their current available volume, to generate a bidding plan. Specifically: starting with the highest-scoring bid, the seller decides the volume to bid for that bid, with the bid volume not exceeding the smaller of the required volume for that bid and the seller's current available volume. If the seller still has remaining available volume after bidding, they continue bidding for the next highest-scoring bid, until all available volume is allocated or all eligible bids have been submitted. Each bid includes: seller identifier, demand area identifier, bid volume, and bid matching score. Sellers then submit their generated bidding plans (which may include bids for multiple bids) to the system for aggregation.

[0045] Step S304: The system summarizes all the bidding proposals submitted by all sellers, executes the winning bid allocation, and generates allocation instructions.

[0046] The system collects all bids submitted by all sellers in this round of negotiation. For each demand bid, the system sorts all received bids for that bid from highest to lowest according to their matching scores, and accumulates the bid quantities sequentially until the accumulated quantity meets the demand volume of the bid or all bids have been allocated. If the total bid quantity exceeds the demand, the highest-scoring bids are awarded priority, and the last winning bid may only receive a portion of the demand volume. If the total bid quantity is insufficient, the demand bid is marked as "unmet" and proceeds to the next round of negotiation. Based on all the winning bids, the system generates an allocation instruction, which includes: material source loading point, demand filling area, allocated volume, suggested departure time, and route number. This allocation instruction is then sent to the truck dispatch system for execution.

[0047] In this embodiment of the invention, the seller's bidding behavior is upgraded from simply "selecting the highest score" to "greedy allocation of multiple bids based on their own available volume". This allows the seller to meet multiple needs in one round of negotiation, which significantly improves the efficiency of supply and demand matching and resource utilization, and avoids the waste of volume when there is an "oversupply" of high-quality material sources.

[0048] In a preferred embodiment of the present invention, the unmanned road roller receives the compaction task and monitors the compaction degree in real time to complete the filling process, specifically including: Step S401: The unmanned road roller receives the compaction task, parses the task parameters, and obtains the material source characteristics information of the area to be compacted.

[0049] The unmanned road roller receives compaction task information from the allocation instruction generated in step S300 via a wireless communication network. This compaction task information includes at least: the coordinates of the filling area, the filling layer number, the designed pavement thickness, the designed compaction target value, and the type of material source for this batch of fill material (weakly weathered rock or strongly weathered rock). Simultaneously, the unmanned road roller queries the material source distribution module for detailed characteristic parameters of the material source currently used in the filling layer of that filling area, including: rock type, maximum particle size, and uniformity coefficient. These material source characteristic parameters serve as the basis for subsequent adaptive adjustment of compaction parameters. This step provides complete task and material information for the compaction operation.

[0050] Step S402: Based on the material source characteristics information, automatically calculate and set the initial compaction parameters.

[0051] The unmanned road roller automatically calculates the optimal initial compaction parameters suitable for the current material source by calling a preset compaction parameter matching model based on the acquired material source characteristic parameters. The compaction parameters include: compaction speed, number of compaction passes, vibration frequency, and vibration amplitude. The compaction parameter matching model is a mapping relationship established based on historical filling data and compaction test data, and its specific rules include: When the material source is weakly weathered rock and the maximum particle size is less than the set threshold, a high-frequency low-amplitude vibration mode is adopted, the rolling speed is set to 2-3 km / h, and the number of rolling passes is designed to be 6-8 passes. When the material source is strongly weathered rock, due to its low particle strength and easy breakage, a low-frequency high-amplitude vibration mode is adopted, the rolling speed is set to 1.5-2.5km / h, and the number of rolling passes is designed to be 8-10 passes. When the material source has a large non-uniformity coefficient and good gradation, reduce the number of passes by 1-2 from the initial number of passes; when the non-uniformity coefficient is small and the gradation is poor, increase the number of passes by 1-2.

[0052] The unmanned road roller loads the calculated initial compaction parameters into the onboard control system as the starting control reference for the compaction operation.

[0053] Step S403: Execute the initial compaction parameters and collect compaction monitoring data in real time during the compaction process to calculate the current compaction value.

[0054] During compaction operations, the unmanned road roller continuously collects compaction quality data through its onboard real-time compaction monitoring system. This system includes an acceleration sensor and a temperature sensor integrated into the vibratory drum, as well as an onboard GPS positioning module. The system collects vibration acceleration signals, drum temperature, and compaction trajectory coordinates of the vibratory drum at a preset sampling frequency (e.g., once every 0.5 seconds). Based on the vibration acceleration signals, harmonic analysis is used to calculate the vibration response characteristic parameters of the vibratory drum in real time, including the ratio of the fundamental amplitude to the amplitudes of each harmonic. Based on these vibration response characteristic parameters, a pre-calibrated compaction regression model is used to convert the vibration response characteristic parameters into an estimated compaction value for the current compaction point. This compaction regression model is pre-established through on-site compaction tests, establishing a mapping relationship between the vibration response characteristic parameters and the compaction measured by the water-irrigation method or nuclear density meter.

[0055] Step S404: Compare the real-time compaction degree with the design compaction degree target value, and dynamically adjust the rolling parameters to achieve closed-loop control.

[0056] The unmanned road roller compares the current compaction degree value of the compaction point calculated in step S403 with the design compaction degree target value of the fill layer in real time. Based on the comparison result, a closed-loop feedback control strategy is used to dynamically adjust the compaction parameters. When the real-time compaction degree reaches the set proportion (e.g., 95%) of the design compaction degree target value, reduce the vibration frequency and amplitude, and switch to static pressure mode for finishing and compaction to avoid over-compression and aggregate breakage. When the compaction growth rate of multiple consecutive rolling points (e.g., 5 consecutive points) is lower than the preset threshold and the compaction degree has not yet met the standard, it is determined that there may be an anomaly in the area, and the rolling speed is reduced or the vibration frequency is increased to improve the compaction effect. When the number of compaction passes exceeds the preset proportion of the designed number of compaction passes (e.g., 120%) and the compaction degree still does not meet the standard, the system generates an abnormal warning message, suspends the compaction operation in that area, and prompts that it is necessary to check whether the layer thickness is too thick, whether the gradation is abnormal, or whether the moisture content of the material source deviates from the optimal value.

[0057] The unmanned road roller updates its travel speed and vibration mode in real time based on the adjusted compaction parameters, and records the decision basis for each adjustment and the parameter values ​​before and after the adjustment in the on-board log.

[0058] Step S405: When the real-time compaction degree of all grid points in the compaction area reaches the design compaction degree target value, the compaction of the filling layer is determined to be completed, and a compaction completion report is generated.

[0059] During the compaction process, the unmanned road roller spatially divides the entire filling area into fixed-size grids (e.g., 1m × 1m), recording the number of compaction passes for each grid and the estimated compaction degree of the final pass. When all grids within the area are compacted and the compaction degree of each grid reaches or exceeds the design compaction target value, the compaction of the filling layer is considered complete. The unmanned road roller generates a compaction completion report, which includes: filling area number, filling layer number, compaction start time, compaction end time, actual compaction pass distribution map, compaction degree distribution cloud map, and abnormal event records. The compaction completion report is uploaded to the construction management platform, and the actual filling volume data is simultaneously output to step S500 as one of the input data for calculating the deviation rate.

[0060] In this embodiment of the invention, the traditional method of fixing compaction parameters is abandoned. Instead, material characteristics such as source type, maximum particle size, and gradation features are used as the basis for determining compaction parameters. An optimal initial compaction parameter is automatically set for each batch of different fill material using a preset matching model. This solves the problem of significant differences in compaction characteristics between strongly weathered and weakly weathered rocks, making it impossible to consider uniform parameters. This ensures compaction quality while avoiding aggregate breakage due to over-compaction or quality defects caused by under-compaction. Furthermore, vibration response signals are collected using an accelerometer on the vibrating wheel, and the vibration characteristics are converted into compaction degree estimates in real time through harmonic analysis and a pre-calibrated regression model. Compared to traditional post-construction sampling inspections (such as water injection methods and nuclear density gauges), this invention achieves uninterrupted online monitoring of compaction quality across the entire area and process, eliminating blind spots and lags in sampling inspections, and significantly improving the comprehensiveness and timeliness of quality control.

[0061] Ultimately, a complete closed loop of "real-time monitoring → deviation calculation → parameter adjustment → effect verification" was constructed. Based on the deviation between the real-time compaction degree and the target value, the machine autonomously decides whether to reduce the frequency of static compaction, increase the excitation force, or trigger anomaly warnings. In particular, an over-compression protection strategy (automatic static compaction after reaching the target) and an abnormal stagnation detection strategy were designed, enabling the unmanned road roller to possess intelligent behavior similar to that of a skilled operator, involving "perception-judgment-adjustment".

[0062] As a preferred embodiment of the present invention, the steps of automatically adjusting the quality matching tightness, transportation distance attenuation index, and looseness coefficient based on the deviation rate specifically include: S501 classifies the deviation rate into three levels based on the deviation rate value: normal, moderate deviation, and severe deviation. S502: When there is a serious deviation and the deviation direction is that the actual fill volume is significantly lower than expected, and there are ≥3 material shortage warnings in the past 24 hours, the quality matching tightness is adjusted, and the rule that "weakly weathered rock can only be used in the upstream rockfill area" is temporarily relaxed. Up to 15% of strongly weathered rock is allowed to be mixed into the upstream rockfill area. The mixing ratio is dynamically calculated based on the current material shortage rate: Material shortage rate = (estimated material shortage / required material) × 100%. For every 5% increase in the material shortage rate, the allowed mixing ratio increases by 2%, with a maximum of 15%. S503, when the deviation is moderate or greater and the direction of the deviation is that the actual filling volume is lower than expected and the average transportation distance increases by more than 10% compared with the previous cycle, the transportation distance attenuation index is adjusted. S504: When the actual filling volume is consistently lower than expected and the deviation rate is stable between 3% and 8%, it is determined that the loose paving coefficient is too large, resulting in insufficient thickness after compaction. The current loose paving coefficient is increased by multiplying it by (1 + deviation rate / 2). When the actual volume is consistently higher than expected and the deviation rate is negative and the absolute value is greater than 3%, it is determined that the loose paving coefficient is too small. The loose paving coefficient is decreased by multiplying it by (1 - |deviation rate| / 2).

[0063] As a preferred embodiment of the present invention, the method further includes an automatic material borrowing and returning strategy: when a material shortage is detected in the upstream rockfill area while there is weakly weathered rock in the downstream rockfill area that meets the upstream standards, a material borrowing and returning plan is automatically generated; the material borrowing and returning plan includes: recording the amount and source of temporary calls, locking in the deep qualified material source to be mined later as the material returning source, and automatically issuing an equal amount compensation transportation instruction after the qualified material source arrives later; all borrowing and returning operations are recorded in a blockchain-style immutable log to ensure that the quantity is equal and the time is clear.

[0064] like Figure 2 As shown in the figure, this embodiment of the invention also provides an earthwork balance optimization system for a canyon-type pumped storage power station, the system comprising: The material source distribution module 100 is used to scan the construction area and obtain three-dimensional point cloud and image data; based on the point cloud difference calculation, the excavation reduction zone and the filling increase zone are calculated, and the topsoil, strongly weathered rock, weakly weathered rock and filling body are classified and identified according to the reflection intensity and color to generate a material source distribution map. The rule matching module 200 is used to extract the list of available excavated materials from the material source distribution map and the list of immediate needs from the filling plan; to perform conflict detection according to the quality matching rules, spatial accessibility rules and time window rules, to screen feasible pairs, and to issue warnings of material shortage or excess. The allocation instruction module 300 is used to set each loading point as a material source seller and each filling area as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score based on the quality matching coefficient, urgency coefficient, and distance attenuation factor and submits a bid, generating an allocation instruction. The filling execution module 400 enables trucks to transport material sources to the filling area according to allocation instructions and dynamic planning paths. At the loading point, machine vision is used to detect the gradation and sort oversized stones. The unmanned road roller receives the compaction task and monitors the compaction degree in real time to complete the filling. The adjustment and optimization module 500 is used to obtain the actual filling volume after each filling layer is completed, compare it with the cumulative transportation volume to calculate the deviation rate, automatically adjust the quality matching tightness, transportation distance attenuation index and loose laying coefficient according to the deviation rate, and trigger construction sequence optimization suggestions.

[0065] The above description only details the preferred embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0066] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0067] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the disclosure in the specification and embodiments. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. An optimization method for earthwork balance in canyon-type pumped storage power stations, characterized in that, The method includes the following steps: The construction area is scanned to obtain three-dimensional point cloud and image data; based on the point cloud difference, the excavation reduction zone and the filling increase zone are calculated, and the topsoil, strongly weathered rock, weakly weathered rock and filling body are classified and identified according to the reflection intensity and color, and a material source distribution map is generated. Extract the available excavation material list from the material source distribution map and the immediate demand list from the filling plan; perform conflict detection based on quality matching rules, spatial accessibility rules and time window rules, screen feasible pairings, and issue material shortage or excess warnings; Each loading point is set as a material source seller, and each filling area is set as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score and bids based on the quality matching coefficient, urgency coefficient, and distance attenuation factor, and generates allocation instructions. This allows trucks to transport materials to the filling area according to allocation instructions and dynamic planning routes. At the loading point, machine vision is used to detect the gradation and sort oversized stones. Unmanned road rollers receive the compaction task and monitor the compaction degree in real time to complete the filling. After each filling layer is completed, the actual filling volume is obtained and compared with the cumulative transportation volume to calculate the deviation rate. Based on the deviation rate, the quality matching tightness, transportation distance attenuation index and loose laying coefficient are automatically adjusted, and construction sequence optimization suggestions are triggered.

2. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 1, characterized in that, When scanning the construction area, the drone is equipped with a lidar and a visible light camera; the ground robot carries an RTK-GPS and a binocular camera to conduct supplementary scanning along the construction road and slope.

3. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 2, characterized in that, The steps involved in generating a material source distribution map include: calculating the excavation reduction zone and the filling increase zone based on point cloud differential calculations; classifying and identifying topsoil, strongly weathered rock, weakly weathered rock, and fill material according to reflection intensity and color; and generating the material source distribution map. After registering the current point cloud with the previous point cloud, the elevation change is calculated point by point to generate a change raster map; areas with an elevation decrease threshold > 0.3m are marked as excavation reduction zones, and areas with an elevation increase threshold > 0.3m are marked as filling increase zones. Within the excavation reduction zone, the reflection intensity value and RGB color components are read point by point, combined with the local point cloud slope. If the reflection intensity is <80 and the ratio of the R component to the G component in RGB is between 0.8 and 1.2, and the slope is <30°, it is determined to be topsoil. If the reflection intensity is 80-160, the RGB is yellowish-brown, and the point cloud is clustered in blocks, it is determined to be strongly weathered rock. If the reflection intensity is >160 and the RGB is grayish-white, it is determined to be weakly weathered rock. Within the filling increment zone, the presence of a layered structure of uniform thickness in the point cloud is detected. If it exists, it is determined to be filling material. The topsoil, strongly weathered rock, and weakly weathered rock were assigned different labels after being classified, and the spatial location and natural volume of each independent material source block were recorded. All material source block information was overlaid with the filling body information of the filling increment area to generate a material source distribution map.

4. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 3, characterized in that, The quality matching rule stipulates that weakly weathered rock can only be used in the upstream rockfill area and transition area, strongly weathered rock can be used in the downstream rockfill area or slope body, and topsoil can be used for temporary storage sites or greening cover soil; the spatial accessibility rule automatically determines whether the transportation route is available based on the real-time vehicle trajectory congestion level, and if the congestion index exceeds the preset threshold, it is marked as unavailable. The time window rule requires that the material source excavation time be earlier than the filling deadline of the demand area. If this is violated, it will be marked as infeasible. The material shortage warning is triggered when the expected material shortage ratio exceeds 10% in the next 4 or 8 hours. The excess warning is triggered when a material source is expected to be unable to be consumed within 48 hours and the remaining capacity of the stockpiling site is lower than the warning line.

5. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 4, characterized in that, The formula for calculating the matching score is: quality matching coefficient × (1 / remaining time window of demand) × (1 / (transportation distance + 1))^α, where α is the transportation distance decay index; sum up all bids, and if the total number of bids received for a certain demand exceeds the demand, allocate them according to the score from high to low. If the quantity is insufficient, it is marked as unsatisfactory, and the process proceeds to the next round of negotiation; an allocation instruction is generated based on all winning bids.

6. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 5, characterized in that, The dynamic planning path generates multiple candidate paths based on the latest digital terrain and road closure information of the construction area. Each truck obtains the position of other vehicles within 50 meters through V2V communication and uses the virtual force field method to adjust speed and steering in real time, where the target point generates attraction and the surrounding vehicles generate repulsion.

7. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 6, characterized in that, Deviation rate = (cumulative transport volume - actual filling volume) ÷ cumulative transport volume.

8. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 7, characterized in that, The steps for automatically adjusting the quality matching tightness, transportation distance attenuation index, and looseness coefficient based on the deviation rate specifically include: Based on the deviation rate value, the deviation rate is divided into three levels: normal, moderate deviation, and severe deviation; When there is a serious deviation and the deviation direction is that the actual filling volume is significantly lower than expected, and at the same time there are ≥3 material shortage warnings in the past 24 hours, adjust the tightness of the quality matching. When the deviation is moderate or greater and the direction of the deviation is that the actual filling volume is lower than expected and the average transportation distance increases by more than 10% compared with the previous cycle, the transportation distance attenuation index is adjusted. When the actual filling volume is consistently lower than expected and the deviation rate is stable between 3% and 8%, it is determined that the loose paving coefficient is too large, resulting in insufficient thickness after compaction. The current loose paving coefficient is increased by multiplying it by (1 + deviation rate / 2). When the actual filling volume is consistently higher than expected and the deviation rate is negative with an absolute value greater than 3%, it is determined that the loose paving coefficient is too small. The loose paving coefficient is decreased by multiplying it by (1 - |deviation rate| / 2).

9. The earthwork balance optimization method for canyon-type pumped storage power stations according to claim 8, characterized in that, The method also includes an automatic material borrowing and returning strategy: when a material shortage is detected in the upstream rockfill area and there is weakly weathered rock in the downstream rockfill area that meets the upstream standard, a material borrowing and returning plan is automatically generated. The borrowing and returning material plan includes: recording the amount and source of temporary borrowers, locking in qualified deep-seated material sources to be mined later as the source of returning materials, and automatically issuing an equal amount of compensation transportation instructions once the qualified material sources arrive; all borrowing and returning operations are recorded in a blockchain-style immutable log to ensure that the quantity is equal and the time is clear.

10. An earthwork balance optimization system for a canyon-type pumped storage power station, characterized in that, The system applies the earthwork balance optimization method for canyon-type pumped storage power stations as described in any one of claims 1 to 9 above, and the system includes: The material source distribution module is used to scan the construction area and obtain three-dimensional point cloud and image data; based on the point cloud difference calculation, the excavation reduction zone and the filling increase zone are calculated, and the topsoil, strongly weathered rock, weakly weathered rock and filling body are classified and identified according to the reflection intensity and color to generate a material source distribution map. The rule matching module is used to extract the list of available excavated materials from the material source distribution map and the list of immediate needs from the filling plan; to perform conflict detection based on quality matching rules, spatial accessibility rules and time window rules, to screen feasible pairs, and to issue warnings of material shortage or excess. The allocation instruction module is used to set each loading point as a material source seller and each filling area as a demand buyer. A round of negotiation is initiated at set intervals: the buyer publishes a demand tender document containing material type, volume, and time limit; the seller calculates the matching score based on the quality matching coefficient, urgency coefficient, and distance attenuation factor and submits a bid, generating an allocation instruction. The filling execution module enables trucks to transport material sources to the filling area according to allocation instructions and dynamic planning paths. At the loading point, machine vision is used to detect the gradation and sort oversized stones. The unmanned road roller receives the compaction task and monitors the compaction degree in real time to complete the filling. The adjustment and optimization module is used to obtain the actual filling volume after each filling layer is completed, compare it with the cumulative transportation volume to calculate the deviation rate, automatically adjust the quality matching tightness, transportation distance attenuation index and loose laying coefficient based on the deviation rate, and trigger construction sequence optimization suggestions.

Images