Regional scale saline-alkali soil humic acid improvement project planning system

Abstract

The invention discloses a regional scale saline-alkali soil humic acid improvement project planning system, and relates to the technical field of saline-alkali soil improvement, and the system comprises a data obtaining and processing module which is used for obtaining an electronic map of a target region, spatial soil salinization characteristic data, environmental factor data and soil improvement project constraint conditions. According to the regional scale saline-alkali soil humic acid improvement project planning system, the defects of an existing general method in special material recommendation are overcome, and humic acid types can be accurately matched for sub-regions with different saline-alkali obstacle characteristics, and the use amount and mode can be determined. Furthermore, a planning engine integrated with multi-objective optimization considers technical parameters and complex constraints such as economy, resources and time and space cooperatively, a detailed engineering scheme which is controllable in cost, reasonable in path and feasible in time sequence is automatically generated, and the problems of extensive planning, poor resource allocation and difficulty in cost control caused by traditional decision-making depending on experience are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of saline-alkali land improvement technology, specifically a regional-scale planning system for saline-alkali land humic acid improvement projects. Background Technology

[0002] Existing saline-alkali land improvement technologies are gradually developing towards intelligence and precision. The patent publication number "CN116703181B" describes "a method for improving saline-alkali land." This method generates a saline-alkali condition coefficient by comprehensively acquiring data such as electronic maps of the area to be treated, groundwater depth, and average annual precipitation, and determines whether to initiate treatment based on preset thresholds. Furthermore, the area is divided into grids, and parameters such as soil conductivity and alkalinity are collected through monitoring points to generate a saline-alkali state coefficient, classifying alkali and non-alkali areas. Combined with vegetation coverage to correct the assessment results, the area is divided into severely, moderately, and mildly saline-alkali areas, and differentiated treatment plans are matched accordingly. For moderately saline-alkali areas, a saline-alkali feature database and a treatment plan database are established, and a latent semantic model is used for plan matching. A BP neural network model is used to predict and optimize the matched plans. This invention effectively improves the targeting and systematic nature of saline-alkali land treatment, avoiding the problem of treatment plans being out of touch with actual conditions.

[0003] However, the above methods still have certain limitations: First, these methods focus on the intelligent matching and prediction of general remediation schemes, without engineering planning for the application of specific amendments such as humic acid. They lack systematic consideration of key engineering parameters such as the source, properties, application rate, application method, and coupling relationship with soil salinity and alkalinity characteristics of humic acid. Second, when implemented at a regional scale, they do not fully consider engineering planning factors such as soil spatial heterogeneity, transportation costs of amendments, construction sequence, long-term monitoring, and benefit assessment, making it difficult to directly guide large-scale, replicable, and sustainable engineering practices for humic acid amendments. Third, while the predictive models for remediation schemes have general applicability, they do not deeply integrate specific mechanisms and long-term effects data of humic acid amendments, potentially resulting in insufficient predictive accuracy and guidance value in humic acid-specific projects.

[0004] Therefore, in practical engineering applications, especially in scenarios requiring large-scale application of humic acid for regional improvement, there is still a lack of a specialized planning system that integrates soil diagnosis, material matching, engineering planning, benefit simulation, and dynamic control. This leads to humic acid improvement projects often relying on experience-based decision-making, resulting in problems such as unscientific planning, difficulty in cost control, unstable long-term effects, and difficulty in regional optimization scheduling, thus hindering the large-scale promotion of humic acid improvement technology and the full realization of its comprehensive benefits. Summary of the Invention

[0005] The purpose of this invention is to provide a regional-scale planning system for humic acid improvement projects in saline-alkali land, in order to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a regional-scale planning system for humic acid improvement projects in saline-alkali land, comprising: The data acquisition and processing module is used to acquire electronic maps of the target area, spatialized soil salinization characteristic data, environmental factor data, and soil improvement engineering constraints. The humic acid material adaptation module is connected to the data acquisition and processing module. It is used to match suitable humic acid improvement materials from the pre-established humic acid material feature library and determine their engineering application parameters based on the spatialized soil salinization characteristic data. The regional engineering optimization planning engine connects the humic acid material adaptation module and the data acquisition and processing module. It is used to generate a preliminary engineering planning scheme that includes material delivery, construction sequence and cost budget based on the engineering application parameters, the electronic map and the engineering constraints. The engineering simulation verification sandbox connects the regional engineering optimization planning engine and the data acquisition and processing module. It is used to dynamically simulate the improvement process and effects of the preliminary engineering planning scheme, and to verify and optimize the preliminary engineering planning scheme based on the simulation results, and output the optimized engineering planning scheme. The engineering implementation and dynamic monitoring module is used to continuously acquire soil condition monitoring data of the implementation area after the improvement project is implemented according to the optimized engineering plan, and to feed the monitoring data back to the engineering simulation verification sandbox for effect evaluation and control.

[0007] Furthermore, the humic acid material adapter module includes: The Humic Acid Material Characteristic Library stores physicochemical property data of humic acid materials from different sources and types. A soil salinity barrier feature database stores the spatialized soil salinization feature data. An adaptation rule set, constructed based on the humic acid material feature library and the soil salinity barrier feature library, is used to output recommended humic acid material types, application rates per unit area, and application methods according to the input soil salinity barrier features, thus forming the engineering application parameters.

[0008] Furthermore, the regional engineering optimization planning engine includes: A multi-objective optimization model whose objective function simultaneously considers the total project cost, the expected overall improvement effect, and the total construction period; The project constraints include: the demand for humic acid materials in each planning sub-region, the accessibility of the road network, the location of material supply points, the operational capacity of construction machinery, the upper limit of the budget, and the overall construction period requirements. The regional engineering optimization planning engine runs the multi-objective optimization model and outputs the preliminary engineering planning scheme, which includes at least a material delivery route diagram, a phased construction sequence diagram, and a budget detail table.

[0009] Furthermore, the engineering simulation verification sandbox includes: Mechanistic process simulation model, based on the construction of soil chemical reaction and physical process mechanism of humic acid improvement of saline-alkali land; The data-driven prediction model was trained based on historical saline-alkali land improvement project case data. The workflow of the engineering simulation verification sandbox includes: First, the preliminary engineering planning scheme and the historical environmental data of the target area are input into the mechanism process simulation model to conduct the first round of long-term dynamic simulation and obtain the first simulation result; Secondly, the first simulation result, together with the preliminary engineering planning scheme and the spatialized soil salinization characteristic data, is input into the data-driven prediction model for cross-validation prediction to obtain the second prediction result; Then, the consistency between the first simulation result and the second prediction result is compared and analyzed; if they are inconsistent or the simulation result does not meet the preset improvement target, a scheme adjustment instruction is generated. Finally, the scheme adjustment instruction is fed back to the regional engineering optimization planning engine, triggering the adjustment of the preliminary engineering planning scheme to form an adjusted scheme. The adjusted scheme is then re-input into the engineering simulation verification sandbox for a new round of simulation verification until the optimized engineering planning scheme that meets the consistency requirements and improvement goals is obtained.

[0010] Furthermore, the workflow of the engineering simulation verification sandbox further includes: The first round of long-term dynamic simulation covers multiple crop growth cycles; The cross-validation prediction includes: calculating the difference between the first simulation result and the second prediction result on key soil indicators; if the difference exceeds a preset allowable deviation threshold, it is determined to be inconsistent. When a discrepancy or non-compliance is determined, the generated scheme adjustment instructions include suggestions for correcting the amount of humic acid applied to a specific planning sub-area, suggestions for adding auxiliary materials, or suggestions for changing the construction sequence. The regional engineering optimization planning engine recalculates the optimization under the engineering constraints according to the scheme adjustment instructions, and generates the adjusted scheme.

[0011] Furthermore, the engineering simulation verification sandbox is also equipped with a risk warning unit; When the first simulation result is inconsistent with the second prediction result, and the difference continues to exceed the allowable deviation threshold for a preset number of rounds, the risk warning unit outputs a high-risk indicator and identifies the corresponding planning sub-area and uncertain soil improvement indicators. The proposed adjustment instructions, under the high-risk label, also include recommendations to conduct supplementary on-site investigations or to adopt alternative improved material solutions.

[0012] Furthermore, the material delivery route map is an optimal set of routes that takes into account vehicle load limits and road conditions; The phased construction sequence diagram is arranged according to the suitable period of soil moisture, availability of mechanical resources and material supply rhythm; The detailed budget table dynamically links material market prices, transportation distances, and labor costs.

[0013] Furthermore, the engineering implementation and dynamic monitoring module includes: A sensor network deployed in the implementation area is used to periodically collect data on soil salinity, pH value, and water content as the monitoring data; The data comparison unit is used to compare the real-time acquired monitoring data with the predicted data of the corresponding region and time point from the engineering simulation verification sandbox.

[0014] Furthermore, the engineering implementation and dynamic monitoring module also includes a dynamic control unit; When the data comparison unit detects that the monitoring data deviates continuously from the predicted data, the dynamic control unit generates dynamic control suggestions based on the degree of deviation and the current environmental data. The dynamic control recommendations include instructions to increase the application of humic acid or adjust the irrigation plan, and specify the implementation sub-region and operation time window.

[0015] Furthermore, the optimized engineering plan output by the system is an engineering guidance document that has undergone multiple rounds of simulation verification and has clear construction steps, a bill of materials, a cost structure, and long-term effect predictions.

[0016] This invention provides a regional-scale planning system for humic acid improvement projects in saline-alkali land. It offers the following advantages: This regional-scale planning system for humic acid improvement projects in saline-alkali land overcomes the shortcomings of existing general methods in recommending specific materials by constructing a deep-coupled feature library and matching rules for humic acid materials and soil obstacles. It can accurately match humic acid types and determine dosage and application methods for sub-regions with different saline-alkali obstacle characteristics. Furthermore, the integrated multi-objective optimization planning engine considers technical parameters in conjunction with complex constraints such as economy, resources, and time and space, automatically generating detailed engineering schemes that are cost-controllable, have reasonable paths, and are time-series feasible. This effectively solves the problems of extensive planning, suboptimal resource allocation, and difficulty in cost control caused by traditional experience-based decision-making.

[0017] This regional-scale humic acid improvement project planning system for saline-alkali land constructs a complete closed loop of "planning-simulation-verification-optimization-execution-feedback" through a simulation verification sandbox driven by a "mechanism-data" dual-drive model and a dynamic monitoring and control network after implementation. The sandbox conducts multiple rounds of cross-validation and iterative optimization of the planning scheme, improving its scientific rigor and resilience, and enabling pre-assessment of long-term effects and identification of potential uncertainties. After project implementation, the IoT-based monitoring and dynamic control mechanism continuously tracks and maintains the improvement effects, ensuring their sustainability and stability. This closed-loop system enhances the overall reliability, predictability, and long-term success guarantee of regional-scale humic acid improvement projects. Attached Figure Description

[0018] Figure 1 This is a data flow diagram of a regional-scale saline-alkali land humic acid improvement engineering planning system according to the present invention; Figure 2 This is a soil condition monitoring and control feedback diagram for a regional-scale saline-alkali land humic acid improvement engineering planning system of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 and Figure 2 This invention provides a technical solution: a regional-scale planning system for humic acid improvement projects in saline-alkali land, comprising: The data acquisition and processing module is used to acquire electronic maps of the target area, spatialized soil salinization characteristic data, environmental factor data, and soil improvement engineering constraints. The humic acid material adaptation module is connected to the data acquisition and processing module. It is used to match suitable humic acid improvement materials from a pre-established humic acid material feature library and determine their engineering application parameters based on spatialized soil salinization characteristic data. The regional engineering optimization planning engine connects the humic acid material adaptation module and the data acquisition and processing module. It is used to generate a preliminary engineering planning scheme that includes material delivery, construction sequence and cost budget based on engineering application parameters, electronic map and engineering constraints. The engineering simulation verification sandbox connects the regional engineering optimization planning engine and the data acquisition and processing module. It is used to dynamically simulate the improvement process and effects of the preliminary engineering planning scheme, and to verify and optimize the preliminary engineering planning scheme based on the simulation results, and output the optimized engineering planning scheme. The engineering implementation and dynamic monitoring module is used to continuously acquire soil condition monitoring data of the implementation area after the improvement project is implemented according to the optimized engineering plan, and to feed the monitoring data back to the engineering simulation verification sandbox for effect evaluation and control.

[0021] It should be further explained that the system first collects electronic maps of the target area, spatialized soil salinization characteristic data, environmental factor data, and engineering constraints through the data acquisition and processing module. The spatialized soil salinization characteristic data includes, but is not limited to, soil electrical conductivity, sodium adsorption ratio, alkalinity, texture, and organic matter content. The environmental factor data includes historical data on precipitation and evaporation. The engineering constraints include total budget, construction period, types and quantities of available machinery, road network, and locations of material supply points.

[0022] The humic acid material adaptation module, based on the acquired soil salinization characteristic data, calls upon a pre-established humic acid material feature library and a soil salinity barrier feature library. The humic acid material feature library stores attributes such as molecular weight, functional group composition, pH, and ion exchange capacity of humic acids from different sources. It runs a built-in adaptation rule set, which is constructed based on historical data and mechanistic knowledge, such as through decision trees or rule engines. For each planning sub-unit with different salinity barrier characteristics, it outputs a matching humic acid material model, recommended application rate per unit area, and specific application method, such as tillage and mixing or surface application, thereby generating accurate engineering application parameters.

[0023] The regional engineering optimization planning engine receives the aforementioned engineering application parameters and all constraints, constructs and solves a multi-objective optimization model that simultaneously minimizes the total engineering cost, maximizes the expected improvement effect, and meets the schedule requirements. It automatically generates a preliminary engineering plan that includes optimal material delivery routes, detailed phased construction schedules, and a precise cost budget. Subsequently, the engineering simulation verification sandbox initiates its core verification and optimization process: this sandbox incorporates a mechanism process simulation model based on the chemical and physical processes of humic acid improvement, and a data-driven prediction model trained on a large amount of historical improvement case data.

[0024] The sandbox first inputs the preliminary engineering plan and multi-year historical climate data of the target area into the mechanistic model to simulate the dynamic changes of key soil indicators over several future growth cycles, obtaining the first simulation result. Next, the sandbox inputs the first simulation result, the preliminary plan, and the original soil data into the data-driven model for cross-validation, obtaining the second prediction result. The system compares and analyzes these two sets of results. If the difference in their key indicators exceeds a preset tolerance threshold, or if the simulation result fails to meet the preset improvement target, the preliminary plan is deemed to have uncertainties or deficiencies, and an adjustment instruction containing specific adjustment suggestions is generated.

[0025] This instruction is fed back to the regional engineering optimization planning engine, which recalculates under the original constraints and produces an adjusted engineering solution. This adjusted solution is then sent back to the simulation verification sandbox for a new round of "mechanism simulation-data verification" cycle. This process can be iterated multiple times until the output engineering solutions demonstrate consistency under dual model verification and achieve the expected improvement goals, ultimately forming a robust and scientifically reliable optimized engineering planning scheme. After implementing the improvement project based on this optimized scheme, the engineering implementation and dynamic monitoring module continuously collects real-time monitoring data on soil salinity, pH, etc., in the implementation area through a pre-deployed sensor network and compares this data with the predicted baseline at the corresponding time point in the simulation verification sandbox. If a persistent unfavorable deviation is detected in the monitoring data, the system can generate dynamic adjustment suggestions, thereby achieving closed-loop feedback and precise maintenance of the engineering effect and ensuring the stability of long-term improvement results.

[0026] The humic acid material adapter module includes: The Humic Acid Material Characteristic Library stores physicochemical property data of humic acid materials from different sources and types. The Soil Salinization Barrier Feature Database stores spatialized soil salinization feature data. The adaptation rule set, built on the humic acid material feature library and the soil salinity barrier feature library, is used to output recommended humic acid material types, application rates per unit area, and application methods based on the input soil salinity barrier features, thus forming engineering application parameters.

[0027] It should be further explained that the humic acid material characteristic library is a structured database. Its core fields include at least the source type of humic acid, production process, average molecular weight range, content of functional groups such as carboxyl and phenolic hydroxyl groups, E4 / E6 ratio, pH value, cation exchange capacity, and reference cost per unit mass. Among them, the source type includes mineral source and biochemical source, and the E4 / E6 ratio reflects the degree of molecular condensation. This library is constructed by aggregating product data from different suppliers and laboratory characterization data to ensure that each record corresponds to a specific commodity or raw material with clear physicochemical properties.

[0028] The soil salinity barrier feature library is built based on spatialized data provided by the data acquisition and processing module. Its records correspond one-to-one with the planned grid units. The feature vector of each unit includes the soil electrical conductivity value, sodium adsorption ratio, exchangeable sodium percentage, residual sodium carbonate content, soil texture classification, organic matter background value, and groundwater level data of that unit. The adaptation rule set is the core processing logic of the module. Its construction is not a simple table lookup and matching, but is formed by integrating mechanistic knowledge and machine learning methods: First, a series of basic rule frameworks are established based on the scientific principles of humic acid improvement of saline-alkali land. The scientific principles include ion exchange reaction kinetics and colloidal flocculation conditions. Second, using a large amount of historical field test data, the contribution weight of each humic acid material attribute to the improvement effect is determined through decision tree or random forest algorithms under different combinations of soil obstacle features. The improvement effect includes desalination rate and pH reduction value. The final rule set can receive a soil obstacle feature vector as input, and through internal calculation, output a recommended humic acid material ID, a calculated precise application range, and a specific application technology suggestion, such as: it is recommended to mix it with gypsum in a specific ratio and then deep plow and compact it. Among them, the humic acid material ID comes from the feature library, and the precise application range includes recommending the application of mineral humic acid with a carboxyl content greater than X mmol / g to Z tons / hectare for medium loam with high sodium adsorption ratio. This process transforms abstract soil problems into quantifiable and actionable engineering material parameters, enabling precise integration from macro-regional diagnosis to micro-material selection.

[0029] The specific logic of the adaptation rule set is constructed based on the matching relationship between soil salinity and alkalinity barrier characteristics and humic acid material properties. For example, when the soil electrical conductivity is 8-12 mS / cm and the sodium adsorption ratio is 15-20, mineral-derived humic acid with a carboxyl content greater than 5 mmol / g and a phenolic hydroxyl content greater than 3 mmol / g is preferred, with an application rate of 2.5-3.5 tons / hectare and the application method is to mix it with soil during tillage, with the tillage depth controlled at 20-30 cm. When the soil electrical conductivity is 4-8 mS / cm and the sodium adsorption ratio is 10-15, biochemical-derived humic acid with a carboxyl content greater than 3 mmol / g and an E4 / E6 ratio between 3 and 5 is recommended, with an application rate of 1.5-2.5 tons / hectare and the application method can be surface application followed by irrigation leaching. The decision tree nodes are divided based on the critical values ​​of key soil indicators. For example, soil electrical conductivity of 8 mS / cm and sodium adsorption ratio of 15 are used as primary nodes, and humic acid functional group content and molecular weight range are used as secondary nodes to ensure the relevance and operability of the rules.

[0030] The regional engineering optimization planning engine includes: The multi-objective optimization model considers the total project cost, the expected overall improvement effect, and the total construction period simultaneously in its objective function. The optimization direction of this multi-objective optimization model is to minimize the total project cost, maximize the expected overall improvement effect, and minimize the total construction period, while satisfying the engineering constraints. The engineering constraints include the upper limit of the budget and the overall construction period requirement. The engineering constraints are the rigid boundaries for solving the model, and the objective function is the optimization objective within the boundary range. The two work together to achieve the comprehensive optimization of the solution.

[0031] Engineering constraints include: the demand for humic acid materials in each planned sub-area, the accessibility of the road network, the location of material supply points, the operational capacity of construction machinery, the upper limit of the budget, and the overall construction period requirements; The regional engineering optimization planning engine runs a multi-objective optimization model and outputs a preliminary engineering planning scheme. The preliminary engineering planning scheme includes at least a material delivery route map, a phased construction sequence map, and a budget detail table.

[0032] It should be further explained that the engine receives precise humic acid demand data from each planning sub-region of the humic acid material adaptation module, and combines it with electronic maps, vectorized road network data, known material supply warehouse coordinates, types of available construction machinery and their daily operating efficiency, the upper limit of the total project budget, and the total construction period requirements stipulated in the contract provided by the data acquisition and processing module.

[0033] The core multi-objective optimization model of the engine is solved using an intelligent algorithm, which can employ a non-dominated sorting genetic algorithm with an elitist strategy. Mathematically, the model constructs three objective functions that need to be optimized simultaneously: the first objective function aims to minimize the total project cost, which consists of the cost of humic acid material procurement, transportation fuel and labor costs from the supply point to each sub-region, machinery rental and operation costs, and management expenses; the second objective function aims to maximize the expected overall improvement effect, which is characterized by summing the improvement potential of each sub-region, where the improvement potential is weighted by indicators such as the expected reduction in salinity calculated by previous modules; and the third objective function aims to minimize the total construction period.

[0034] The model solution process must strictly adhere to a series of hard and soft constraints, including but not limited to: the demand of each sub-region must be met, the load limit of a single transport vehicle, traffic restrictions on specific roads, the total number of available machine shifts per day, the total amount of funds that cannot be exceeded, and the final deadline that must be met. Through iterative calculations of this optimization model, the engine ultimately outputs a complete and executable preliminary engineering plan.

[0035] It should be further explained that in the specific implementation of the multi-objective optimization model, the quantification of the total project cost needs to cover the procurement cost of humic acid materials, transportation cost, machinery rental cost, labor cost, and management cost. The procurement cost is calculated by summing the market unit price corresponding to the appropriate type of humic acid for each sub-region and the actual demand. The transportation cost is calculated based on the transportation distance from each supply point to the sub-region, the vehicle's fuel consumption per unit mile, and the load factor. The machinery rental cost is calculated based on the unit price per shift of construction machinery and the actual number of shifts worked. The labor cost is determined by combining the number of construction workers, average daily wages, and number of working days. The management cost is calculated as a fixed percentage of the sum of the first four items. The weighted calculation of the expected overall improvement effect is based on the weight of the soil salinization degree of each sub-region. The weight of severely salinized areas is higher than that of moderately and lightly salinized areas. The weight values ​​are determined by analyzing the production capacity improvement potential after improvement in areas with different salinization degrees. The final overall effect is the sum of the products of the predicted improvement effect of each sub-region and its corresponding weight. The multi-objective coordination strategy adopts the Pareto optimal solution logic. It selects non-dominated solutions that simultaneously satisfy the following conditions: cost not exceeding the budget limit, improvement effect reaching the preset benchmark value, and construction period within the required range. Then, it combines the actual project priorities, such as cost priority or effect priority, to determine the final solution. The solution that is in the better range in all three objective dimensions is given priority as the planning basis.

[0036] The material delivery route map in this solution is generated based on vehicle routing problem optimization, clearly defining the loading list, driving route, and service sub-area sequence for each vehicle. The phased construction sequence diagram is presented in Gantt chart form, clearly indicating the start and end times of different machines' operations in different sub-areas, and taking into account agronomic factors such as suitable soil moisture conditions to avoid construction under unfavorable conditions. The budget details list all cost items, including material costs, transportation costs, machinery costs, and labor costs, along with their calculation basis, and are dynamically linked to route distance and operation time. This engine achieves a leap from technical parameters to physical engineering plans, integrating scattered improvement needs, limited resources, and complex spatiotemporal constraints into a holistic optimal action blueprint.

[0037] The engineering simulation and verification sandbox includes: Mechanistic process simulation model, based on the construction of soil chemical reaction and physical process mechanism of humic acid improvement of saline-alkali land; The data-driven prediction model was trained based on historical saline-alkali land improvement project case data. The workflow of the engineering simulation verification sandbox includes: First, the preliminary engineering planning scheme and historical environmental data of the target area are input into the mechanism process simulation model to conduct the first round of long-term dynamic simulation and obtain the first simulation results; Secondly, the first simulation results, together with the preliminary engineering planning scheme and spatialized soil salinization characteristic data, are input into the data-driven prediction model for cross-validation prediction to obtain the second prediction results. Then, the consistency between the first simulation results and the second prediction results is compared and analyzed. The core object of the consistency comparison is the key soil indicators that play a decisive role in the improvement effect of saline-alkali land, specifically including soil electrical conductivity, sodium adsorption ratio, and pH value. Subsequent cross-validation prediction and difference calculation are all carried out around these three key indicators. If there is no consistency or the simulation results do not reach the preset improvement target, the scheme adjustment instruction is generated. Finally, the scheme adjustment instruction is fed back to the regional engineering optimization planning engine, triggering the adjustment of the preliminary engineering planning scheme to form the adjusted scheme. The adjusted scheme is then re-input into the engineering simulation verification sandbox for a new round of simulation verification until an optimized engineering planning scheme that meets the consistency requirements and improvement goals is obtained.

[0038] It should be further explained that this sandbox contains two independently constructed and complementary prediction models: a mechanistic process simulation model and a data-driven prediction model. The mechanistic process simulation model is a numerical model built based on the physicochemical process mechanism of humic acid in improving saline-alkali land. Its core is to describe the migration and exchange reactions of salt ions, such as Na⁺, Ca²⁺, and Mg²⁺, between the soil solution and the exchange phase, the dissolution and neutralization reactions of carbonates and acidic functional groups of humic acid, and the change in the stability of soil aggregates after improvement over time through a set of coupled differential equations. This model requires initial soil parameters, humic acid characteristic parameters, application scheme, and daily meteorological data to drive the calculation. The daily meteorological data includes precipitation, evaporation, and temperature.

[0039] It is important to further clarify that the implementation of the mechanism simulation model needs to focus on soil salt migration, ion exchange, and the interaction between humic acid and soil components. The salt migration process requires consideration of soil texture (sandy, loamy, clayey), porosity, and water content to simulate the dynamic process of water infiltration driving salt leaching. Sandy soils have higher porosity, resulting in a higher salt migration rate than clayey soils. The ion exchange process mainly focuses on the exchange between functional groups such as carboxyl and phenolic hydroxyl groups in humic acid and sodium and calcium ions in the soil. The degree of exchange reaction is simulated based on the content of functional groups in humic acid and the content of exchangeable sodium in the soil; higher functional group content results in stronger ion exchange capacity. During the simulation, initial soil parameters of the target area, such as initial salt content, ion composition, humic acid application parameters (dosage, application method), and daily meteorological data (precipitation, evaporation, temperature), need to be considered. The changes in key soil indicators at different time periods are continuously and dynamically calculated. Temperature affects the reaction rate, precipitation affects salt leaching efficiency, and evaporation leads to salt accumulation in the surface soil.

[0040] The data-driven prediction model is a machine learning model trained on a large dataset of historical saline-alkali land improvement projects. Each sample in the dataset contains soil characteristics before the project, the improvement scheme adopted, soil index monitoring values ​​at different time points after implementation, and the corresponding environmental background. The model learns the complex nonlinear mapping relationship between the "cause" and the "effect" through algorithms. The improvement scheme includes materials and dosage.

[0041] The training data for the data-driven prediction model must contain complete dimensional information. Soil indicators include soil electrical conductivity, sodium adsorption ratio, exchangeable sodium percentage, texture, organic matter content, and pH value. Environmental factors include annual precipitation, annual evaporation, average temperature, and seasonal precipitation distribution characteristics. Engineering parameters include humic acid type, application rate, application method, type and amount of auxiliary materials, and construction sequence. The evaluation metrics for model training are mean squared error and coefficient of determination. The mean squared error must be controlled within 20% of the standard deviation of historical monitoring data, and the coefficient of determination must be no less than 0.85 to ensure the model's prediction accuracy. Model parameters are selected within industry norms. The decision tree depth is controlled at 8-12 layers to avoid overfitting or underfitting. The number of trees in the random forest is set to 100-200. Ensemble predictions from multiple decision trees reduce the error of a single model. During training, 70% of the data is used as the training set, and 30% as the test set to verify the model's generalization ability.

[0042] The workflow of the sandbox is as follows: First, the preliminary engineering planning scheme generated by the regional engineering optimization planning engine and the historical meteorological sequence of the target area over many years are input into the mechanism process simulation model. Dynamic simulation covering multiple complete annual cycles is performed, and the change curves of key indicators such as soil salinity, pH value, and sodium adsorption ratio of each sub-region in the next few years are output as the first simulation result. Among them, the preliminary engineering planning scheme includes the materials, usage, and construction time of each sub-region.

[0043] Secondly, the sandbox uses the predicted values ​​at specific future time points from the first simulation results, the preliminary planning scheme itself, and the original soil spatial data as input feature vectors. This input data drives the prediction model to perform forward inference, obtaining a second prediction result based on data pattern recognition at the same time point. Then, the system compares and analyzes these two sets of results, calculating, for example, the difference between the predicted soil salinity values ​​at the end of the third year. If this difference exceeds a pre-set tolerance threshold based on model uncertainty, or if the prediction result of either model fails to reach the pre-set improvement target threshold, the preliminary scheme is deemed risky or ineffective, and a scheme adjustment instruction containing specific adjustments is generated.

[0044] Finally, the instruction is fed back to the regional engineering optimization planning engine. Based on the adjustment suggestions in the instruction, the engine recalculates the optimization while satisfying all engineering constraints, generating an adjusted engineering scheme. This new scheme is then sent back to the engineering simulation verification sandbox, initiating a new cycle of "mechanism simulation-data verification." This iterative process continues until the output engineering scheme's dual-model prediction results are consistent within the allowable deviation range and both meet the preset improvement targets. At this point, the scheme is determined as the final optimized engineering planning scheme. This "dual-model cross-validation and dynamic iterative optimization" mechanism is the core innovation of this system. Through the mutual verification and calibration of two different paradigms—mechanism and data—it significantly improves the reliability prediction capability and risk resistance of the planning scheme in complex real-world environments.

[0045] The workflow of the engineering simulation verification sandbox further includes: The first round of long-term dynamic simulations covers multiple crop growth cycles; Cross-validation prediction includes: calculating the difference between the first simulation result and the second prediction result on key soil indicators. If the difference exceeds the preset allowable deviation threshold, it is determined to be inconsistent. When a plan is deemed inconsistent or not up to standard, the generated adjustment instructions include suggestions for revising the amount of humic acid applied to a specific planning sub-area, suggestions for adding auxiliary materials, or suggestions for changing the construction sequence. The regional engineering optimization planning engine recalculates the optimization under engineering constraints based on the scheme adjustment instructions, and generates the adjusted scheme.

[0046] The determination of critical thresholds needs to combine historical engineering data and industry experience. The permissible deviation threshold is determined based on the statistical results of historical monitoring data for key soil indicators such as soil electrical conductivity and sodium adsorption ratio. Specifically, it is 30% of the standard deviation of multiple monitoring data for the same area. For example, if the historical standard deviation of soil electrical conductivity is 1.2 mS / cm, the permissible deviation threshold is set to 0.36 mS / cm. The preset number of rounds is set to 3. This means that if the difference between the first simulation result and the second prediction result exceeds the permissible deviation threshold in three consecutive simulation verifications, a risk warning is triggered. The criterion for continuous deviation is that the difference between the monitoring data and the predicted data for the same sub-region exceeds the preset threshold for four consecutive sampling cycles, each sampling cycle lasting 7 days, and the direction of deviation is unfavorable to soil improvement, such as a continuous increase in salinity or pH value.

[0047] It should be further explained that the time range of the first round of long-term dynamic simulation was set to three to five complete annual cycles to fully capture the year-by-year accumulation of the humic acid improvement effect and the impact of different climate year types. In the cross-validation prediction process, the key soil indicators selected were soil electrical conductivity and sodium adsorption ratio, which have the most direct impact on agricultural production. The allowable deviation threshold was a quantitative standard jointly determined by analyzing the error distribution between the predicted values ​​of the mechanism model and the actual observed values ​​in historical cases and combining it with expert experience. For example, when the relative deviation between the predicted values ​​of soil electrical conductivity of the two models for the same sub-region at the end of the third year is greater than a certain percentage, an inconsistency judgment is triggered.

[0048] When the system determines that the simulation results are inconsistent or have not met the preset goals, the generated scheme adjustment instructions are specific and actionable. These instructions may include: for specific planning sub-areas where the simulation results are insufficient, the instructions may suggest increasing the amount of humic acid applied by a specific value based on the initial recommended value; or, for areas with specific alkalization obstacles, the instructions may suggest applying a specific amount of gypsum or other soil conditioners as an adjunct to the application of humic acid; in addition, the instructions may also suggest changing the construction sequence of specific blocks based on the climate risks revealed by the simulation, such as adjusting the operation time from spring to autumn to utilize precipitation for natural leaching.

[0049] Upon receiving these specific adjustment instructions, the regional engineering optimization planning engine does not simply make local modifications. Instead, it takes the adjustment suggestions as new input conditions and, while fully adhering to all existing constraints such as materials, machinery, funds, and schedule, re-runs its multi-objective optimization model to perform global calculations. This generates a new, overall coordinated adjusted solution, ensuring that the optimized solution meets the new technical requirements while maintaining the economy and feasibility of engineering implementation.

[0050] The engineering simulation and verification sandbox is also equipped with a risk warning unit; When the first simulation result and the second prediction result are inconsistent, and the difference continues to exceed the allowable deviation threshold for a preset number of rounds, the risk warning unit outputs a high-risk flag and identifies the corresponding planning sub-area and uncertain soil improvement indicators. Here, "the difference continues to exceed the allowable deviation threshold for a preset number of rounds" means that the scheme adjustment-simulation verification loop between the sandbox and the regional engineering optimization planning engine reaches a preset number of iterations. For example, after 3 rounds, if the difference between the first simulation result and the second prediction result still exceeds the allowable deviation threshold and the preset improvement target has not been achieved, the risk warning unit will start the high-risk flag output process.

[0051] Under the high-risk label, the adjustment instructions also include recommendations to conduct supplementary on-site investigations or to adopt alternative improved material solutions.

[0052] It should be further explained that when, during continuous simulation and verification iterations in the sandbox, for the same planning sub-region, the difference between the predictions of the mechanistic process simulation model and the data-driven prediction model for a specific soil improvement index consistently exceeds the preset allowable deviation threshold, and this inconsistency is not eliminated within the preset number of iterations, the risk warning unit will be activated. The core logic of this unit lies in identifying this persistent prediction discrepancy that cannot be converged through conventional parameter adjustments, and classifying it as a situation of high uncertainty, rather than simply insufficient effectiveness.

[0053] Once activated, the unit generates an alert report with a high-risk marker, clearly indicating the geographic number of the specific planning sub-area where the discrepancy exists, and detailing one or more soil indicators where the two models' predictions show significant differences in that area. In addition to identifying the risk, the alert report includes automatically generated, more exploratory scheme adjustment instructions. These instructions go beyond routine dosage or timing adjustments, recommending further field actions to obtain more precise input data. For example, the instructions might suggest denser grid soil sampling and testing in the high-risk area to obtain more refined soil spatial variability data for model calibration.

[0054] Simultaneously, based on data from the humic acid material feature library, the instruction will recommend one or more alternative humic acid material solutions with different physicochemical properties than the currently recommended materials, but which may be more suitable for handling such uncertainties, for selection and evaluation. This mechanism elevates the system from purely virtual optimization to an intelligent decision support level that can proactively identify knowledge gaps and guide offline verification and solution selection, enhancing the scientific rigor and robustness of large-scale engineering planning in the face of uncertainties in complex natural systems.

[0055] The material delivery route map is a set of optimal routes that take into account vehicle load limits and road conditions; The phased construction sequence diagram is arranged according to the suitable period of soil moisture, the availability of mechanical resources, and the rhythm of material supply; The budget details are dynamically linked to material market prices, transportation distances, and labor costs.

[0056] Further explanation is needed regarding the implementation details of the material distribution route map, phased construction sequence diagram, and budget breakdown: The material distribution route map is one of the key outputs of the multi-objective optimization model in the regional engineering optimization planning engine. Its generation is based on the classic vehicle routing problem model and has been extended to adapt to the agricultural engineering scenario. During the calculation, the model treats each supply point as a parking lot and denotes the demand points of each planned sub-region. Under the premise of strictly meeting the rated load limit of each transport vehicle, the main optimization objective is to minimize the total transport mileage or the total transport time. At the same time, the actual road conditions are included as weight factors in the calculation. The final route map clearly marks the departure order of each vehicle, the order of passing through each demand point, the unloading volume at each point, and the expected driving route, forming a set of executable optimal or suboptimal distribution solutions.

[0057] The arrangement of phased construction sequence diagrams is a complex resource-constrained project scheduling problem. Phased construction sequence diagrams are usually presented in the form of Gantt charts. When generating the engine, not only are constraints on the quantity and efficiency of mechanical resources considered, but more importantly, the suitable soil moisture period for each sub-region is incorporated into the model as an important time window constraint. For example, tillage operations are not suitable when the soil is too wet or too dry, and the model will avoid these unsuitable periods. At the same time, the timing arrangement is closely coordinated with the material delivery rhythm to ensure that when construction machinery arrives in a certain area, the corresponding improvement materials have also been delivered, thereby avoiding idle work and waiting for materials. The final output sequence diagram clearly shows the operation plans of different types of machinery on different dates and in different sub-regions.

[0058] The detailed budget is not a static report; all costs are dynamically linked to the aforementioned optimization results: material costs are calculated based on the actual type and quantity of humic acid used and a real-time market price database; transportation costs are derived from the optimized model of total mileage and unit mileage cost for each vehicle; machinery usage costs are calculated based on the number of days each machine is occupied and the shift rate in the time sequence diagram; labor costs are linked to the number of construction days and personnel allocation plan; any adjustment to the planning scheme will trigger an automatic recalculation of the detailed budget, thereby ensuring the real-time accuracy of cost estimates and the economic comparability of the schemes. These outputs collectively constitute a complete engineering plan that has been collaboratively optimized in terms of technology, resources, time, and economy.

[0059] The project implementation and dynamic monitoring module includes: A sensor network deployed in the implementation area is used to periodically collect data on soil salinity, pH value, and moisture content as monitoring data; The data comparison unit is used to compare the real-time acquired monitoring data with the predicted data of the corresponding area and time point in the engineering simulation verification sandbox.

[0060] Further explanation is needed regarding the specific implementation details of the engineering implementation and dynamic monitoring module: The sensor network deployed in the implementation area consists of multiple monitoring nodes. Each node integrates a salinity sensor for measuring soil solution conductivity, a pH sensor for measuring soil acidity / alkalinity, and a moisture sensor for measuring soil volumetric water content. These nodes are deployed according to the key monitoring sub-areas defined in the optimized engineering plan, and connect to the system wirelessly. They automatically collect and upload soil salinity, pH, and moisture content data at preset intervals, such as daily or weekly, forming the dynamic monitoring data stream after implementation. Upon receiving the real-time monitoring data, the data comparison unit first indexes the predicted soil index data for the same sub-area, the same time point, or the corresponding growth stage in the engineering simulation verification sandbox database based on the data's timestamp and spatial coordinates. This predicted data originates from the simulation result curve corresponding to the final optimized scheme confirmed in the sandbox.

[0061] The core function of the comparison unit is to calculate the difference between real-time monitoring values ​​and predicted baseline values. For example, it calculates the difference between the soil electrical conductivity value monitored on a given day and the predicted value in the prediction model for that day. This unit has logical judgment rules set up to identify "persistent deviation" states. For instance, if the difference of a monitoring indicator in the same sub-region changes unfavorably for multiple consecutive sampling periods and exceeds a preset allowable range, it is determined that a persistent deviation requiring attention has occurred. This comparison result is a crucial link connecting virtual planning with the physical world, providing a data-driven decision-making basis for possible subsequent dynamic interventions.

[0062] The engineering implementation and dynamic monitoring module also includes a dynamic control unit; When the data comparison unit detects a continuous deviation between the monitoring data and the predicted data, the dynamic control unit generates dynamic control suggestions based on the degree of deviation and the current environmental data. The dynamic control recommendations include instructions to increase the application of humic acid or adjust the irrigation plan, and specify the implementation sub-regions and operation time windows.

[0063] It should be further explained that the implementation of the dynamic control unit is as follows: When the data comparison unit identifies a sustained unfavorable deviation of the monitoring data for a specific sub-region from the predicted baseline, the dynamic control unit immediately initiates and executes the following decision-making process: First, the unit comprehensively assesses the severity of the deviation, which is achieved by calculating the slope or cumulative amount of the deviation values ​​of key indicators over the most recent monitoring periods. Key indicators include soil electrical conductivity. Simultaneously, the unit accesses real-time or forecast environmental data streams, such as the probability of precipitation, temperature, and evaporation forecasts for the next few days. Based on the degree of deviation and future environmental conditions, the unit calls upon the adaptation rule set and material feature library in the system's humic acid material adaptation module to perform rapid local recalculation.

[0064] Its core logic is to simulate the potential impact of a supplementary intervention on reversing the deviation trend if a project has already been implemented. For example, for areas where desalination is slower than expected, the unit will select a fast-acting liquid humic acid product from the material library that is easy to apply through the irrigation system, based on the current soil conditions and future weather, and calculate the recommended application rate per hectare for the upcoming irrigation using a set of rules.

[0065] The final dynamic control recommendation is a structured instruction that clearly includes the specific measures to be taken, the precise dosage, the specific geographic number of the targeted sub-region, and the recommended time window for implementing the intervention. This recommendation is directly pushed to on-site engineering management personnel through the system interface, thus forming an online closed loop of "monitoring-analysis-control." This enables proactive and precise maintenance of the effects of large-scale improvement projects, representing a fundamental improvement over the traditional "one-time construction, subsequent neglect" model.

[0066] When the dynamic control unit is activated, it first updates the real-time soil monitoring data, including soil salinity, pH value, and water content, to the soil salinity barrier feature library. This ensures that the adaptation rule set of the humic acid material adaptation module is calculated based on the actual soil conditions in the current area, thereby generating dynamic control suggestions that match the real-time soil conditions.

[0067] In the dynamic adjustment recommendations, the degree of deviation is categorized into mild, moderate, and severe based on the proportion of the difference to the predicted data. A difference of 10%-20% indicates mild deviation, 20%-30% indicates moderate deviation, and more than 30% indicates severe deviation. For mild deviation, the proportion of humic acid applied should be increased by 10%-15% of the initial application rate; for moderate deviation, the increase should be 15%-25%, and 50-100 kg / ha of gypsum can be used as an auxiliary material; for severe deviation, the increase should be 25%-35%, and the construction sequence should be adjusted to the next suitable soil moisture period.

[0068] The timing of the operation should be chosen when the soil moisture content is between 15% and 25%, and the weather forecast indicates that there will be no extreme weather such as heavy rain or strong winds in the next 3-5 days. At the same time, it should avoid the critical growth period of crops such as sowing and emergence to ensure that the control measures do not affect crop growth.

[0069] The irrigation plan should be adjusted according to the deviation and soil moisture data. When the deviation is slight, the irrigation frequency should be increased appropriately, while the amount of irrigation per irrigation remains unchanged. When the deviation is moderate or severe, the irrigation frequency should be increased, and the amount of irrigation per irrigation should be increased by 10%-20% to promote salt leaching and uniform distribution of humic acid.

[0070] The final optimized engineering plan output by the system is an engineering guidance document that has undergone multiple rounds of simulation verification and includes clear construction steps, a bill of materials, cost structure, and long-term effect predictions. It should be further explained that the implementation content of the final optimized engineering plan output by the system is as follows: This plan is a comprehensive and executable engineering guidance document. It is not a simple summary of the outputs of various modules in the early stages, but rather a deterministic result formed after multiple rounds of cross-verification and iterative optimization using a "mechanism-data" dual-model approach in the engineering simulation verification sandbox. Its core content includes the following organically integrated parts: First, a detailed construction step instruction manual generated based on the finally verified engineering parameters. This manual, organized by process, clearly specifies the operating procedures and technical points from material inspection upon arrival, specific application methods for each sub-area, to auxiliary measures. Specific application methods include the depth and number of passes of mechanical tillage, and auxiliary measures include supporting irrigation. Second, a complete bill of materials. This bill of materials not only lists the model, total usage, and regional distribution quantity of the main humic acid material, but also includes the precise quantities and corresponding supplier information of all auxiliary materials determined based on the optimization results, such as gypsum and conditioning agents.

[0071] Third is a dynamically linked cost breakdown table. All itemized costs, including material costs, transportation costs, machinery operating costs, and labor costs, are locked to the final optimized path, timing, and usage, forming a budget baseline strictly tied to the construction plan. Fourth is a long-term effect prediction report presented as visual charts. Based on the validation results of the final solution in a simulation sandbox, this report plots predicted changes in soil salinity, pH, and other indicators in key areas over the next few years, and includes sensitivity analyses of effects under different climate scenarios, providing scientific expectations for project acceptance and long-term benefit assessment.

[0072] The system achieves integrated and precise end-to-end processes, from macro-regional diagnosis to micro-material adaptation and physical engineering planning. By constructing a deep-coupled feature library and adaptation rules between humic acid materials and soil barriers, the system overcomes the shortcomings of existing general methods in recommending specific materials. It can accurately match humic acid types and determine dosage and method for sub-regions with different saline-alkali barrier characteristics. Furthermore, the integrated multi-objective optimization planning engine considers technical parameters in conjunction with complex constraints such as economy, resources, and time and space, automatically generating detailed engineering solutions that are cost-controllable, have reasonable paths, and are time-feasible. This effectively solves the problems of extensive planning, suboptimal resource allocation, and difficulty in cost control caused by traditional experience-based decision-making.

[0073] Secondly, a complete closed loop of "planning-simulation-verification-optimization-execution-feedback" was constructed through a simulation verification sandbox of a "mechanism-data" dual-driven model and a dynamic monitoring and control network after implementation. This sandbox conducts multiple rounds of cross-verification and iterative optimization of the planning scheme, improving its scientific rigor and resilience, and enabling pre-assessment of long-term effects and identification of potential uncertainties. After project implementation, the IoT-based monitoring and dynamic control mechanism continuously tracks and maintains the improvement effects, ensuring their sustainability and stability. This closed-loop system enhances the overall reliability, predictability, and long-term success guarantee of regional-scale humic acid improvement projects.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0075] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A regional-scale planning system for humic acid improvement projects in saline-alkali land, characterized in that, include: The data acquisition and processing module is used to acquire electronic maps of the target area, spatialized soil salinization characteristic data, environmental factor data, and soil improvement engineering constraints. The humic acid material adaptation module is connected to the data acquisition and processing module. It is used to match suitable humic acid improvement materials from the pre-established humic acid material feature library and determine their engineering application parameters based on the spatialized soil salinization characteristic data. The regional engineering optimization planning engine connects the humic acid material adaptation module and the data acquisition and processing module. It is used to generate a preliminary engineering planning scheme that includes material delivery, construction sequence and cost budget based on the engineering application parameters, the electronic map and the engineering constraints. The engineering simulation verification sandbox connects the regional engineering optimization planning engine and the data acquisition and processing module. It is used to dynamically simulate the improvement process and effects of the preliminary engineering planning scheme, and to verify and optimize the preliminary engineering planning scheme based on the simulation results, and output the optimized engineering planning scheme. The engineering implementation and dynamic monitoring module is used to continuously acquire soil condition monitoring data of the implementation area after the improvement project is implemented according to the optimized engineering plan, and to feed the monitoring data back to the engineering simulation verification sandbox for effect evaluation and control.

2. The regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 1, characterized in that: The humic acid material adapter module includes: The Humic Acid Material Characteristic Library stores physicochemical property data of humic acid materials from different sources and types. A soil salinity barrier feature database stores the spatialized soil salinization feature data. An adaptation rule set, constructed based on the humic acid material feature library and the soil salinity barrier feature library, is used to output recommended humic acid material types, application rates per unit area, and application methods according to the input soil salinity barrier features, thus forming the engineering application parameters.

3. The regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 2, characterized in that: The regional engineering optimization planning engine includes: A multi-objective optimization model whose objective function simultaneously considers the total project cost, the expected overall improvement effect, and the total construction period; The project constraints include: the demand for humic acid materials in each planning sub-region, the accessibility of the road network, the location of material supply points, the operational capacity of construction machinery, the upper limit of the budget, and the overall construction period requirements. The regional engineering optimization planning engine runs the multi-objective optimization model and outputs the preliminary engineering planning scheme, which includes at least a material delivery route diagram, a phased construction sequence diagram, and a budget detail table.

4. The regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 3, characterized in that: The engineering simulation and verification sandbox includes: Mechanistic process simulation model, based on the construction of soil chemical reaction and physical process mechanism of humic acid improvement of saline-alkali land; The data-driven prediction model was trained based on historical saline-alkali land improvement project case data. The workflow of the engineering simulation verification sandbox includes: First, the preliminary engineering planning scheme and the historical environmental data of the target area are input into the mechanism process simulation model to conduct the first round of long-term dynamic simulation and obtain the first simulation result; Secondly, the first simulation result, together with the preliminary engineering planning scheme and the spatialized soil salinization characteristic data, is input into the data-driven prediction model for cross-validation prediction to obtain the second prediction result; Then, the consistency between the first simulation result and the second prediction result is compared and analyzed; if they are inconsistent or the simulation result does not meet the preset improvement target, an adjustment instruction for the scheme is generated. Finally, the scheme adjustment instruction is fed back to the regional engineering optimization planning engine, triggering the adjustment of the preliminary engineering planning scheme to form an adjusted scheme. The adjusted scheme is then re-input into the engineering simulation verification sandbox for a new round of simulation verification until the optimized engineering planning scheme that meets the consistency requirements and improvement goals is obtained.

5. The regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 4, characterized in that: The workflow of the engineering simulation verification sandbox further includes: The first round of long-term dynamic simulation covers multiple crop growth cycles; The cross-validation prediction includes: calculating the difference between the first simulation result and the second prediction result on key soil indicators; if the difference exceeds a preset allowable deviation threshold, it is determined to be inconsistent. When a discrepancy or non-compliance is determined, the generated scheme adjustment instructions include suggestions for correcting the amount of humic acid applied to a specific planning sub-area, suggestions for adding auxiliary materials, or suggestions for changing the construction sequence. The regional engineering optimization planning engine recalculates the optimization under the engineering constraints according to the scheme adjustment instructions, and generates the adjusted scheme.

6. The regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 5, characterized in that: The engineering simulation and verification sandbox is also equipped with a risk warning unit; When the first simulation result is inconsistent with the second prediction result, and the difference continues to exceed the allowable deviation threshold for a preset number of rounds, the risk warning unit outputs a high-risk indicator and identifies the corresponding planning sub-area and uncertain soil improvement indicators. The proposed adjustment instructions, under the high-risk label, also include recommendations to conduct supplementary on-site investigations or to adopt alternative improved material solutions.

7. A regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 6, characterized in that: The material delivery route map is the optimal set of routes that takes into account vehicle load limits and road conditions. The phased construction sequence diagram is arranged according to the suitable period of soil moisture, availability of mechanical resources and material supply rhythm; The detailed budget table dynamically links material market prices, transportation distances, and labor costs.

8. A regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 7, characterized in that: The engineering implementation and dynamic monitoring module includes: A sensor network deployed in the implementation area is used to periodically collect data on soil salinity, pH value, and water content as the monitoring data; The data comparison unit is used to compare the real-time acquired monitoring data with the predicted data of the corresponding region and time point from the engineering simulation verification sandbox.

9. A regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 8, characterized in that: The engineering implementation and dynamic monitoring module also includes a dynamic control unit; When the data comparison unit detects that the monitoring data deviates continuously from the predicted data, the dynamic control unit generates dynamic control suggestions based on the degree of deviation and the current environmental data. The dynamic control recommendations include instructions to increase the application of humic acid or adjust the irrigation plan, and specify the implementation sub-region and operation time window.

10. A regional-scale saline-alkali land humic acid improvement engineering planning system according to claim 9, characterized in that: The optimized engineering plan output by the system is an engineering guidance document that has been verified through multiple rounds of simulations and has clear construction steps, a bill of materials, a cost structure, and long-term effect predictions.

Images