Intelligent measurement and evaluation method and system for land resource bearing capacity

Abstract

The invention discloses a land resource bearing capacity intelligent measurement and evaluation method and system, and belongs to the technical field of data processing and intelligent decision support, and the method comprises the steps: obtaining multi-source land resource data, carrying out the preprocessing, and generating a unified multi-modal data set; key indexes are mined by using a big data analysis technology, weights are distributed, and a dynamic evaluation index system is generated; training and adaptively optimizing the intelligent evaluation model by using the unified multi-modal data set and the dynamic evaluation index system, and generating an optimized intelligent evaluation model; carrying out bearing capacity calculation and trend prediction on the newly added land resource data by using the optimized intelligent evaluation model, and generating a bearing capacity evaluation result; and generating land resource utilization optimization strategy suggestions based on the bearing capacity evaluation result. A technical path combining multi-modal data fusion, dynamic index construction and intelligent model adaptive optimization is adopted, and dynamic evaluation, prediction and intelligent decision support of the land resource bearing capacity can be realized.

Description

Technical Field

[0001] This invention relates to the field of data processing and intelligent decision support technology, and in particular to a method and system for intelligent calculation and assessment of land resource carrying capacity. Background Technology

[0002] Land resource carrying capacity refers to the number of people and the scale of economic and social activities that a specific region's land resource ecosystem can sustainably support under certain time, technological, and economic conditions. Scientific and dynamic assessment of this capacity is a crucial step in achieving regional sustainable development, optimizing land spatial layout, and formulating effective management strategies. It is widely applied in urban planning, agricultural production, ecological protection, and other business management fields.

[0003] Existing methods for assessing land resource carrying capacity typically rely on manually defined indicator systems and static mathematical models. The assessment process often depends on historical statistical data and expert experience, calculating carrying capacity indices through weighted summation. For example, some methods select a limited number of indicators such as land use type, water resources, and environmental capacity, assigning them fixed weights, and then using simple linear models for assessment. Data sources are also relatively limited, often focusing on statistical data or remote sensing image analysis from a single aspect.

[0004] However, existing technical solutions have shortcomings. First, fixed indicator systems cannot adapt to the complexity and dynamism of land resource systems, easily overlooking key influencing factors and leading to biased assessment results. Second, static models struggle to capture the complex nonlinear relationships between various indicators and carrying capacity, resulting in limited assessment accuracy. Furthermore, the single source and processing method of data restrict the comprehensiveness of the assessment, while heavy reliance on human experience makes the entire process subjective and inefficient, failing to meet the decision-making needs of modern refined and dynamic land space management. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides an intelligent method and system for assessing and evaluating land resource carrying capacity. It employs a technical approach that combines multimodal data fusion, dynamic index construction, and intelligent model adaptive optimization, enabling dynamic assessment, prediction, and intelligent decision support for land resource carrying capacity.

[0006] The above objectives can be achieved through the following approach: A method for intelligently assessing and evaluating land resource carrying capacity includes: acquiring multi-source land resource data and preprocessing the data to generate a unified multimodal dataset; based on the unified multimodal dataset, using big data analytics to mine key indicators and assign weights to generate a dynamic evaluation indicator system; using the unified multimodal dataset and the dynamic evaluation indicator system to train and adaptively optimize an intelligent evaluation model to generate an optimized intelligent evaluation model; using the optimized intelligent evaluation model to calculate the carrying capacity and predict trends of newly added land resource data to generate carrying capacity evaluation results; and based on the carrying capacity evaluation results, generating suggestions for optimizing land resource utilization strategies.

[0007] Optionally, generating a unified multimodal dataset includes: cleaning the multi-source land resource data to remove noise and outliers, generating cleaned data; converting the cleaned data to a new format and coordinate system to generate standardized data; and filling missing values ​​in the standardized data to generate a unified multimodal dataset.

[0008] Optionally, the generation of the dynamic evaluation index system includes: performing feature engineering based on the unified multimodal dataset to extract candidate indicators related to carrying capacity and generate an initial index set; performing cluster analysis on the initial index set to eliminate redundant information among indicators and select core indicators to generate an optimized index set; assigning weights to the optimized index set in combination with preset multidimensional carrying capacity requirements for land resources to generate final index weights; and constructing the optimized index set into a multi-level, multi-dimensional evaluation system based on the final index weights to generate the dynamic evaluation index system.

[0009] Optionally, generating the final indicator weights includes: objectively assigning weights to the indicators in the dynamic evaluation indicator system using the entropy weight method to generate an initial weight distribution; acquiring expert knowledge and using the analytic hierarchy process (AHP) to correct the initial weight distribution to generate adjusted weights; and iteratively optimizing the adjusted weights to generate the final indicator weights.

[0010] Optionally, generating the optimized intelligent evaluation model includes: constructing an initial evaluation model, training the initial evaluation model using the unified multimodal dataset and the dynamic evaluation index system to generate a trained model; and validating and adjusting the parameters of the trained model using preset historical evaluation feedback data to generate an optimized intelligent evaluation model.

[0011] Optionally, the step of using preset historical evaluation feedback data to verify and adjust the parameters of the trained model includes: acquiring new land resource data and verification information of historical evaluation results to generate model update input; adjusting the trained model using an incremental learning strategy based on the model update input to generate an updated model; and iteratively optimizing the updated model based on its performance in practical applications to generate an optimized intelligent evaluation model.

[0012] Optionally, generating the carrying capacity assessment result includes: acquiring assessment standards and grading rules for defining carrying capacity levels; using the optimized intelligent assessment model to calculate the newly added land resource data and generate preliminary calculation results; analyzing the preliminary calculation results according to the assessment standards and grading rules to generate assessment levels and carrying capacity status; and using a time series prediction algorithm to predict the assessment levels and carrying capacity status to generate the carrying capacity assessment result.

[0013] Optionally, the generation of land resource utilization optimization strategy suggestions includes: obtaining a regional land spatial planning demand and risk avoidance knowledge base; spatializing the carrying capacity assessment results to generate a carrying capacity visualization thematic map; and generating land resource utilization optimization strategy suggestions through a rule reasoning engine based on the carrying capacity visualization thematic map and in conjunction with the regional land spatial planning demand and the risk avoidance knowledge base.

[0014] Based on the same inventive concept, this invention also provides an intelligent calculation and assessment system for land resource carrying capacity, the system comprising: The data fusion module is used to acquire multi-source land resource data and preprocess the multi-source land resource data to generate a unified multimodal dataset. The indicator construction module is used to mine key indicators and assign weights based on the unified multimodal dataset using big data analysis technology, and generate a dynamic evaluation indicator system. The model optimization module is used to train and adaptively optimize the intelligent evaluation model using the unified multimodal dataset and the dynamic evaluation index system, and generate the optimized intelligent evaluation model. The calculation and prediction module is used to use the optimized intelligent assessment model to calculate the carrying capacity and predict the trend of newly added land resource data, and generate carrying capacity assessment results. The decision support module is used to generate suggestions for optimizing land resource utilization strategies based on the carrying capacity assessment results.

[0015] Compared with the prior art, the present invention has the following advantages: 1. By constructing a fully automated process from data collection and fusion to decision support, the scientific nature and efficiency of land resource management have been improved. Intelligent processing and fusion of multi-source heterogeneous data overcomes the problems of data inconsistency and incomplete information in traditional methods, laying a solid data foundation for subsequent assessments and making the entire management decision-making process more transparent and reliable.

[0016] 2. It enables the dynamic construction of the evaluation index system and the adaptive optimization of the model, allowing the evaluation system to continuously evolve in line with the dynamic changes in the land resource system. This data-driven self-learning capability solves the problem of lagging evaluation results and decreased accuracy caused by the inability of traditional static evaluation models to adapt to environmental changes, ensuring the effectiveness and forward-looking nature of the evaluation conclusions.

[0017] 3. By transforming complex assessment results into intuitive visualizations and concrete strategic recommendations, the technical threshold for decision-making is lowered, and the operability of management solutions is enhanced. By closely integrating scientific analysis with management needs, it provides powerful intelligent support for business decision-making activities such as land spatial planning and ecological environmental protection, thereby effectively promoting the sustainable utilization and coordinated development of regional resources.

[0018] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the intelligent calculation and assessment method for land resource carrying capacity according to an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the training process of the intelligent evaluation model in an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram illustrating the adaptive optimization effect of the intelligent evaluation model in an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the intelligent calculation and evaluation system for land resource carrying capacity according to an embodiment of the present invention.

[0024] Figure 5 This is a schematic diagram of the city's carrying capacity satellite cloud image according to an embodiment of the present invention.

[0025] Figure 6 This is a schematic diagram illustrating the predicted trend of land resource carrying capacity according to an embodiment of the present invention.

[0026] Figure 7 This is a schematic diagram illustrating the multi-dimensional carrying capacity visualization of an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0028] Reference Figure 1 One embodiment of the present invention proposes an intelligent measurement and evaluation method and system for land resource carrying capacity. It adopts a technical approach that combines multimodal data fusion, dynamic index construction and intelligent model adaptive optimization, which can realize dynamic evaluation, prediction and intelligent decision support for land resource carrying capacity.

[0029] The method described in this embodiment specifically includes: Acquire multi-source land resource data and preprocess the multi-source land resource data to generate a unified multimodal dataset; Based on the unified multimodal dataset, big data analytics are used to mine key indicators and assign weights to generate a dynamic evaluation indicator system. Using the unified multimodal dataset and the dynamic evaluation index system, the intelligent evaluation model is trained and adaptively optimized to generate the optimized intelligent evaluation model; Using the optimized intelligent assessment model, the carrying capacity of the newly added land resource data is calculated and the trend is predicted to generate carrying capacity assessment results. Based on the carrying capacity assessment results, suggestions for optimizing land resource utilization strategies are generated.

[0030] Specifically, multimodal data fusion technology integrates scattered and heterogeneous land resource information into a unified data foundation, solving the problem of information silos. Based on this, data mining and artificial intelligence algorithms are used to dynamically construct a scientific evaluation index system, overcoming the shortcomings of fixed indicators and strong subjectivity in traditional methods. Next, advanced machine learning or deep learning models are employed to uncover the complex nonlinear relationship between indicators and carrying capacity. Through an adaptive optimization mechanism, the model can continuously learn and evolve, maintaining the accuracy and timeliness of the assessment. Finally, the quantitative results output by the model are transformed into intuitive visualizations and actionable decision-making recommendations, thus achieving an end-to-end automated process from data perception and intelligent analysis to scientific decision-making. The entire process is tightly coupled through data flow and feedback loops, forming a self-improving and continuously optimizing intelligent evaluation ecosystem. This enhances the comprehensiveness, accuracy, and dynamic adaptability of land resource carrying capacity assessment. By integrating multi-source data, the assessment results can more comprehensively reflect the true state of land resources, avoiding the bias caused by a single data source. The dynamically constructed indicator system and adaptively optimized intelligent model ensure that the assessment process can accurately capture the dynamic changes of the land resource system, improving the reliability and foresight of the assessment results. Furthermore, transforming complex analytical results into intuitive visualization interfaces and intelligent decision-making suggestions enhances the usability and decision-support value of the assessment results.

[0031] Optionally, generating the unified multimodal dataset includes: The multi-source land resource data is cleaned to remove noise and outliers, generating cleaned data. The cleaned data is format-converted and mapped to coordinate system 1 to generate standardized data; Missing values ​​are filled into the standardized data to generate a unified multimodal dataset.

[0032] Specifically, for the acquired multi-source land resource data, noise, outliers, and duplicate records are identified and removed, logical errors are corrected, and cleaned data is generated. For example, cloud and fog removal is performed on satellite remote sensing image data, and outlier detection and correction are performed on socio-economic statistical data. The cleaned data undergoes format conversion to unify different formats, including projecting geospatial data from different coordinate systems to the same geographic coordinate system, resampling time series data with different time resolutions to unify the time base, and converting vector data, raster data, and tabular data into a preset unified data structure to generate standardized data. Missing values ​​are filled into the standardized data. Due to equipment failures or environmental limitations during data acquisition, missing items often exist in the data. Interpolation algorithms or prediction models are used to fill these missing values, generating a complete unified multi-dimensional dataset. For example, missing values ​​in time series meteorological data can be filled using linear interpolation. The calculation method can be expressed as: ; in, It is the moment to be filled. The target value, and These are the two most recent valid observations before and after the missing value. and This corresponds to the observation time. The resulting unified multimodal dataset is characterized by data completeness, uniform format, and consistent spatiotemporal benchmarks, providing a reliable data foundation for subsequent indicator construction and model training.

[0033] Optionally, the dynamic evaluation index system includes: Feature engineering is performed based on the unified multimodal dataset to extract candidate indicators related to carrying capacity and generate an initial indicator set. Cluster analysis is performed on the initial indicator set to eliminate redundant information among the indicators and select core indicators to generate an optimized indicator set. Based on the preset multidimensional carrying capacity requirements of land resources, weights are assigned to the optimized index set to generate the final index weights. Based on the final indicator weights, the optimized indicator set is constructed into a multi-level and multi-dimensional evaluation system, generating a dynamic evaluation indicator system.

[0034] Specifically, based on a unified multimodal dataset, feature engineering methods are used to extract key indicator features. Feature engineering is the process of creating more representative features from raw data using professional knowledge and technical means. Specific operations include calculating the correlation between each potential indicator and the proxy variable for land resource carrying capacity, or analyzing the information entropy and variance of each indicator to identify the variables most sensitive to changes in carrying capacity. By setting correlation or information thresholds, a batch of features highly correlated with carrying capacity is selected, thus generating an initial indicator set. To address the potential information overlap or high correlation among indicators in the initial indicator set, cluster analysis is performed. Cluster analysis is an unsupervised learning algorithm that automatically groups similar indicators into one class. In this step, each indicator is treated as a feature vector, and their similarity is measured by calculating the distance between vectors. For example, Euclidean distance can be used to measure the similarity between two indicator vectors, and its calculation method is as follows: ; in, Indicators and The Euclidean distance between them and These represent the i-th dimension components of two indicators after normalization. The smaller the distance, the more similar the two indicators. After grouping similar indicators into the same cluster using a clustering algorithm, the most representative indicator from each cluster is selected as the core indicator, for example, the indicator closest to the cluster center. These selected core indicators together constitute the optimized indicator set. Finally, based on the optimized indicator set and combined with the multi-dimensional carrying capacity assessment needs of land resources (ecological, production, and living aspects), a structured multi-level, multi-dimensional assessment indicator system is constructed. The indicators in the optimized indicator set are logically classified and layered according to their attributes, forming a tree-like or network-like hierarchical structure, ultimately generating a dynamic assessment indicator system.

[0035] Optionally, the weights for generating the final metric include: The entropy weight method is used to objectively assign weights to the indicators in the dynamic evaluation indicator system to generate an initial weight distribution. The initial weight distribution is corrected by acquiring expert knowledge and using the analytic hierarchy process (AHP) to generate adjusted weights. The adjusted weights are iteratively optimized to generate the final index weights.

[0036] Specifically, the initial weights of each indicator are calculated using the entropy weight method. The entropy weight method is an objective weighting method that determines the weight based on the degree of variation in the indicator data. The greater the degree of variation, the more information the indicator provides, and therefore the greater its weight should be. Specifically, for the j-th indicator in the dynamic evaluation indicator system, its information entropy is first calculated. : ; Where k is the adjustment coefficient, usually taken as 1 / ln(m), and m is the number of evaluation samples; The normalized value of the i-th sample under the j-th indicator is obtained from the original data in the unified multimodal dataset after dimensionless processing. The lower the information entropy, the lower the information redundancy of the indicator and the more information it provides. Based on this, the initial weights of each indicator are calculated, forming a weight distribution. The analytic hierarchy process (AHP) is used to adjust the weight distribution. AHP constructs a judgment matrix, transforming the qualitative judgments of experts on the relative importance of indicators into quantitative representations. The subjective weights calculated by AHP are weighted and combined with the objective weight distribution generated in the previous step to generate the adjusted weights. This step compensates for the shortcomings of relying solely on data, which may overlook certain key domain knowledge. Intelligent optimization algorithms are used to optimize the adjusted weights. Using the prediction accuracy or error of the evaluation model as the objective function and the adjusted weights as the initial solution, iterative search algorithms such as genetic algorithms or particle swarm optimization are used to continuously adjust the weight combination until the optimal weight scheme for the objective function is found; this scheme is the final indicator weight.

[0037] Optionally, the generation of the optimized intelligent evaluation model includes: Construct an initial evaluation model, train the initial evaluation model using the unified multimodal dataset and the dynamic evaluation index system, and generate a trained model; The trained model is validated and its parameters are adjusted using preset historical evaluation feedback data to generate an optimized intelligent evaluation model.

[0038] Specifically, given the complexity of the land resource carrying capacity problem, a deep neural network architecture capable of handling high-dimensional nonlinear data is used for initial setup to generate an initial assessment model. This initial model is trained using a unified multimodal dataset and a dynamic assessment index system. During this process, the feature data corresponding to the dynamic assessment index system in the unified multimodal dataset are used as the model input, and the known land resource carrying capacity status or values ​​in the dataset are used as labels for supervised learning. The model iteratively calculates and adjusts its internal parameters to minimize the difference between its predicted output and the true label. This difference is typically quantified by a loss function; for example, the mean squared error loss function can be expressed as: ; Where Loss represents the model's prediction loss, and N is the total number of training samples. It is the model's prediction for the i-th sample. This represents the true carrying capacity label value of the i-th sample. Through continuous optimization using algorithms such as backpropagation, the loss function converges to a low level. At this point, the model has initially learned the patterns in the data, generating the trained model. To ensure the model's generalization ability and achieve optimal performance, the trained model needs to be validated and its parameters tuned. A portion of the dataset is reserved as a validation set to evaluate the performance of the trained model on unseen data. Based on the validation results, the model's hyperparameters, such as the learning rate and the number of network layers, are systematically adjusted, and the training and validation process is repeated until the optimal combination of hyperparameters is found. The model refined in this way is the final intelligent evaluation model.

[0039] Optionally, the step of using preset historical evaluation feedback data to validate and adjust the parameters of the trained model includes: Acquire new land resource data and verification information of historical assessment results to generate model update inputs; Based on the model update input, the trained model is adjusted using an incremental learning strategy to generate an updated model; Based on the performance of the updated model in practical applications, iterative optimization is performed to generate an optimized intelligent evaluation model.

[0040] Specifically, new land resource data or historical assessment feedback is periodically input. New land resource data refers to the latest acquired satellite remote sensing imagery, socio-economic statistics, etc., while historical assessment feedback is verification information of the accuracy of previous assessment results or data on the actual effectiveness of adopted decision-making recommendations. This new information is integrated to generate the model update input. Based on this model update input, the model initiates an automatic adjustment process, encompassing both parameter adjustment and structural adjustment. Parameter adjustment is the core process, typically employing online learning or incremental learning strategies to fine-tune the internal weights of the existing model, rather than completely retraining. This process aims to allow the model to absorb new information without forgetting previously learned knowledge. The parameter update process can be abstractly represented as follows: ; in, For the updated model parameters, Here, α represents the current model parameters, and α is the learning rate, which determines the step size for each update. In adaptive optimization, it is usually set relatively small to achieve smooth fine-tuning. It is the loss function L with respect to the current parameters The gradient, calculated based on the new model update input, guides the direction of parameter optimization. These adjustments generate an updated model. To achieve deeper optimization, the updated model is iteratively optimized based on actual application results. This step compares the model's predictions with subsequent field monitoring data, forming a long-term feedback loop to continuously verify and correct the model, generating an optimized intelligent evaluation model. Figure 2 , Figure 3 The diagram shows the training and adaptive optimization curves of the intelligent evaluation model.

[0041] Optionally, the generated carrying capacity assessment results include: Obtain the assessment criteria and classification rules used to define the carrying capacity level; The optimized intelligent assessment model is used to calculate the newly added land resource data and generate preliminary calculation results. The preliminary calculation results are analyzed based on the evaluation criteria and grading rules to generate an evaluation level and load-bearing status; The carrying capacity assessment result is generated by combining the assessment level and carrying capacity status with a time series prediction algorithm.

[0042] Specifically, the optimized intelligent assessment model is used to calculate the carrying capacity value. New or real-time land resource data for the area to be assessed is formatted according to the structure of the dynamic assessment index system to form a standardized input feature vector. This feature vector is then input into the optimized intelligent assessment model for forward propagation calculation. The model outputs a quantified continuous value, which is the raw score of the area's current land resource carrying capacity, generating a preliminary calculation result. The preliminary calculation result is analyzed based on preset assessment standards and grading rules. The assessment standards and grading rules are threshold ranges predefined according to relevant policies, regulations, or expert knowledge. The preliminary calculation result is compared with these thresholds, mapping it to a discrete level or state, such as carrying capacity surplus, carrying capacity equilibrium, or carrying capacity overload, thereby generating an assessment level and carrying capacity status. A trend prediction algorithm is then used to predict future carrying capacity. Using the preliminary calculation results from historical time series, a time series prediction model is constructed, such as an autoregressive moving average model, whose prediction logic can be simplified as follows: ; in, f is a prediction function that represents the predicted carrying capacity value at the next time point, based on the carrying capacity values ​​at the current time t and the past n time points. Make inferences. This represents the random error term. The model calculates the future trend of carrying capacity, such as increasing, remaining stable, or decreasing. The current assessment level, carrying capacity status, and future predicted trends are integrated to form the carrying capacity assessment result.

[0043] Optionally, the proposed land resource utilization optimization strategy includes: Acquire a knowledge base on regional land spatial planning needs and risk avoidance; The carrying capacity assessment results are spatialized to generate a visual thematic map of carrying capacity. Based on the aforementioned carrying capacity visualization thematic map, and combined with the regional land spatial planning needs and the risk avoidance knowledge base, a rule-based reasoning engine is used to generate suggestions for optimizing land resource utilization strategies.

[0044] Specifically, generating visualizations and decision-making recommendations is the final step in transforming complex analytical results into intuitive and actionable management tools. First, the carrying capacity assessment results are converted into interactive charts or maps. Specifically, a geographic information system (GIS) rendering engine and data visualization library are used to spatialize and graphically represent the carrying capacity assessment results, which include information such as assessment level, carrying capacity status, and predicted trends. For example, areas with different assessment levels are rendered in different colors on an electronic map, forming a clear thematic map of carrying capacity; the predicted future carrying capacity trends are plotted as a time-series line chart. These charts and maps are integrated into an interactive dashboard, allowing users to explore them through zooming, panning, and clicking to query, thus generating dynamic and multi-dimensional visualizations. Based on the deep data represented by the visualizations—namely, the carrying capacity assessment results—land resource utilization optimization strategies or risk mitigation plans are automatically generated. This step relies on a built-in expert knowledge base or rule-based reasoning engine. This knowledge base pre-stores a series of IF-THEN rules, which associate different carrying capacity assessment results with corresponding management strategies. For example, a rule might be defined as automatically triggering and generating strategies such as restricting new industrial land use and guiding industries towards environmentally friendly transformation if a region's assessment level is overloaded and the main source of pressure is industrial land expansion. The carrying capacity assessment results are matched with the rule base to screen and combine applicable strategies, forming preliminary land resource utilization optimization strategies or risk mitigation plans. Finally, to ensure the feasibility and compliance of the recommendations, the previously generated plan is refined and customized based on regional land spatial planning needs. These needs include policy and planning documents such as regional development goals and the scope of basic farmland protection. The preliminary plan is then checked against these planning needs to resolve potential conflicts, and the strategies are prioritized and adjusted according to planning priorities, ultimately outputting intelligent decision-making recommendations highly aligned with local development strategies, generating land resource utilization optimization strategy recommendations.

[0045] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides an intelligent calculation and assessment system for land resource carrying capacity, the system comprising: The data fusion module is used to acquire multi-source land resource data and preprocess the multi-source land resource data to generate a unified multimodal dataset. The indicator construction module is used to mine key indicators and assign weights based on the unified multimodal dataset using big data analysis technology, and generate a dynamic evaluation indicator system. The model optimization module is used to train and adaptively optimize the intelligent evaluation model using the unified multimodal dataset and the dynamic evaluation index system, and generate the optimized intelligent evaluation model. The calculation and prediction module is used to use the optimized intelligent assessment model to calculate the carrying capacity and predict the trend of newly added land resource data, and generate carrying capacity assessment results. The decision support module is used to generate suggestions for optimizing land resource utilization strategies based on the carrying capacity assessment results.

[0046] Example 1: To verify the feasibility of this invention in practice, it was applied to the land resource management and planning work of a municipal land spatial planning bureau. Currently, the city is experiencing rapid urbanization, facing multiple challenges such as a shortage of construction land, encroachment on ecological space, and water scarcity. Traditional methods for assessing land resource carrying capacity mainly rely on static indicators and expert experience, resulting in long assessment cycles, untimely updates, and an inability to accurately guide dynamically changing urban development. This leads to urban problems such as excessive resource consumption and declining environmental quality in some areas. The land spatial planning bureau also needs a method that can dynamically, accurately, and intelligently assess land resource carrying capacity to provide scientific decision-making support for sustainable urban development planning.

[0047] In this embodiment, the Land and Spatial Planning Bureau used the method proposed in this invention to construct an intelligent calculation and assessment system for land resource carrying capacity covering the entire city. First, the system collected multi-source land resource data from recent years, including: high-resolution satellite remote sensing imagery with a spatial resolution of 1 meter, containing RGB three bands and near-infrared bands; land use status vector data including land use type, plot area, etc., using the CGCS2000 coordinate system; socio-economic statistical yearbooks including population, GDP, and industrial structure; environmental monitoring data including air quality and water quality; historical meteorological station data; and urban infrastructure distribution data.

[0048] In the data preprocessing stage, the multi-source data were first cleaned. For example, cloud and fog removal was performed on satellite remote sensing images, and individual abnormally high values ​​in economic statistics were identified and corrected. Subsequently, format conversion was performed, projecting geospatial data from different coordinate systems onto the CGCS2000 coordinate system. Data with different time resolutions, such as daily meteorological data and annual economic data, were resampled to a unified quarterly time base. To address the issue of missing data, linear interpolation was used to fill in missing temperature data for a certain month due to equipment failure at the meteorological station. Specifically, given a temperature of 28℃ on July 9th and 32℃ on July 25th, the interpolated temperature on July 15th was calculated as: 28 + (32-28) × (15-9) / (25-9) = 28 + 4 × 6 / 16 = 29.5℃. This ensured the integrity of the dataset, ultimately generating a unified multimodal dataset.

[0049] In the dynamic construction phase of the evaluation indicator system, feature engineering was performed based on the dataset. By calculating the correlation between each potential indicator and proxy variables of land resource carrying capacity, such as GDP per unit area, 112 indicators were initially selected, forming the initial indicator set. To eliminate redundancy, cluster analysis was performed on the initial indicator set, using Euclidean distance to measure the similarity between indicators, grouping highly correlated indicators, such as regional area and resident population, into the same cluster. Subsequently, the most representative indicators were selected from each cluster, ultimately reducing the number of indicators to 45, forming the optimized indicator set. This optimized indicator set was constructed into a multi-level evaluation indicator system encompassing three dimensions: ecology, production, and living conditions.

[0050] In the indicator weight determination stage, the entropy weight method was first used to objectively assign weights to 45 indicators, obtaining initial weights. Then, five experts in urban planning were invited to subjectively revise the initial weights using the analytic hierarchy process (AHP). Finally, a genetic algorithm was used to optimize the adjusted weights, with the prediction accuracy of the evaluation model based on historical data backtesting as the objective function, iteratively optimizing to generate the final indicator weights.

[0051] During the model training and optimization phase, a deep neural network with three hidden layers was constructed as the initial evaluation model. This network consisted of an input layer, three hidden layers, and an output layer. The input layer had 128-dimensional features, and the number of neurons in the hidden layers was 64 / 32 / 16 respectively. The activation function was ReLU+Dropout, the optimizer was Adam, the learning rate was 1e-4, the batch size was 30, and the training epochs were 50. The model was trained using a unified multimodal dataset and a dynamic evaluation index system. It learned the complex relationship between the index and carrying capacity by minimizing the mean squared error loss function. After training, data from a specific year was used as a validation set for parameter tuning. After deployment, the model entered an adaptive optimization phase. When the latest land resource data from the third quarter of this year was input, an incremental learning strategy was used to fine-tune the model parameters, enabling the model to adapt to the latest developments and changes.

[0052] In practical applications, the planning bureau inputs new planning data for a certain area and quickly calculates its carrying capacity. The optimized intelligent assessment model outputs an initial carrying capacity score of 0.68 for the area. Based on preset grading rules, such as a score >0.8 indicating surplus, 0.6-0.8 indicating equilibrium, and <0.6 indicating overload, the carrying capacity of the area is determined to be equilibrium. Simultaneously, combined with a time series forecasting algorithm, it is predicted that if the area continues to develop at the current planning intensity, its carrying capacity will decline over the next five years, with a risk of sliding towards overload.

[0053] Ultimately, the assessment levels, carrying capacity status, and predicted trends of all areas in the city were visualized on a GIS map, forming a dynamic thematic map of carrying capacity. Based on the assessment results of a certain area, the system's built-in rule-based reasoning engine automatically generated decision recommendations: it was recommended to appropriately reduce the development intensity of the area, increase the proportion of public green space, and prioritize the development of water-saving industries to maintain a balanced carrying capacity. After reviewing the visualization results and recommendations, the planners believed that the assessment results for a certain industrial park were slightly inaccurate, and corrected the assessment level of the area to severely overloaded through the interactive interface, adding a note about the high water-consuming projects recently introduced in the area. This feedback was captured by the system, serving as a new training sample for incremental learning of the model, and triggering a re-examination of the indicator system, increasing the weight of the water consumption per unit of GDP indicator.

[0054] Through pilot testing, this invention has demonstrated strong technical effectiveness. Compared with traditional evaluation methods, the dynamic indicator system constructed by this invention is more concise and representative; the accuracy of the intelligent evaluation model is higher than that of traditional expert scoring methods; and the adoption rate and implementation effect of decision-making recommendations have also been highly recognized by planning departments.

[0055] In summary, this invention demonstrates its superiority in indicator system construction, model evaluation accuracy, and decision support effectiveness. Through techniques such as cluster analysis, information redundancy is reduced while retaining key information, making the evaluation system more efficient and focused. The intelligent evaluation model outperforms traditional expert scoring methods in evaluation accuracy across multiple functional areas, and the evaluation time is shorter, achieving timeliness in the evaluation process.

[0056] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0057] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for intelligent calculation and assessment of land resource carrying capacity, characterized in that, The method includes: Acquire multi-source land resource data and preprocess the multi-source land resource data to generate a unified multimodal dataset; Based on the unified multimodal dataset, big data analytics are used to mine key indicators and assign weights to generate a dynamic evaluation indicator system. Using the unified multimodal dataset and the dynamic evaluation index system, the intelligent evaluation model is trained and adaptively optimized to generate the optimized intelligent evaluation model; Using the optimized intelligent assessment model, the carrying capacity of the newly added land resource data is calculated and the trend is predicted to generate carrying capacity assessment results. Based on the carrying capacity assessment results, suggestions for optimizing land resource utilization strategies are generated.

2. The intelligent calculation and assessment method for land resource carrying capacity according to claim 1, characterized in that, The generation of the unified multimodal dataset includes: The multi-source land resource data is cleaned to remove noise and outliers, generating cleaned data. The cleaned data is format-converted and mapped to coordinate system 1 to generate standardized data; Missing values ​​are filled into the standardized data to generate a unified multimodal dataset.

3. The intelligent calculation and assessment method for land resource carrying capacity according to claim 1, characterized in that, The dynamic evaluation index system includes: Feature engineering is performed based on the unified multimodal dataset to extract candidate indicators related to carrying capacity and generate an initial indicator set. Cluster analysis is performed on the initial indicator set to eliminate redundant information among the indicators and select core indicators to generate an optimized indicator set. Based on the preset multidimensional carrying capacity requirements of land resources, weights are assigned to the optimized index set to generate the final index weights. Based on the final indicator weights, the optimized indicator set is constructed into a multi-level and multi-dimensional evaluation system, generating a dynamic evaluation indicator system.

4. The intelligent calculation and assessment method for land resource carrying capacity according to claim 3, characterized in that, The weights for generating the final metric include: The entropy weight method is used to objectively assign weights to the indicators in the dynamic evaluation indicator system to generate an initial weight distribution. The initial weight distribution is corrected by acquiring expert knowledge and using the analytic hierarchy process (AHP) to generate adjusted weights. The adjusted weights are iteratively optimized to generate the final index weights.

5. The intelligent calculation and assessment method for land resource carrying capacity according to claim 1, characterized in that, The optimized intelligent evaluation model includes: Construct an initial evaluation model, train the initial evaluation model using the unified multimodal dataset and the dynamic evaluation index system, and generate a trained model; The trained model is validated and its parameters are adjusted using preset historical evaluation feedback data to generate an optimized intelligent evaluation model.

6. The intelligent calculation and assessment method for land resource carrying capacity according to claim 5, characterized in that, The step of using preset historical evaluation feedback data to validate and adjust the parameters of the trained model includes: Acquire new land resource data and verification information of historical assessment results to generate model update inputs; Based on the model update input, the trained model is adjusted using an incremental learning strategy to generate an updated model; Based on the performance of the updated model in practical applications, iterative optimization is performed to generate an optimized intelligent evaluation model.

7. The intelligent calculation and assessment method for land resource carrying capacity according to claim 1, characterized in that, The generated carrying capacity assessment results include: Obtain the assessment criteria and classification rules used to define the carrying capacity level; The optimized intelligent assessment model is used to calculate the newly added land resource data and generate preliminary calculation results. The preliminary calculation results are analyzed based on the evaluation criteria and grading rules to generate an evaluation level and load-bearing status; The carrying capacity assessment result is generated by combining the assessment level and carrying capacity status with a time series prediction algorithm.

8. The intelligent calculation and assessment method for land resource carrying capacity according to claim 1, characterized in that, The proposed land resource utilization optimization strategies include: Acquire a knowledge base on regional land spatial planning needs and risk avoidance; The carrying capacity assessment results are spatialized to generate a visual thematic map of carrying capacity. Based on the aforementioned carrying capacity visualization thematic map, and combined with the regional land spatial planning needs and the risk avoidance knowledge base, a rule-based reasoning engine is used to generate suggestions for optimizing land resource utilization strategies.

9. A land resource carrying capacity intelligent calculation and assessment system, applied to the land resource carrying capacity intelligent calculation and assessment method as described in any one of claims 1-8, characterized in that, The system includes: The data fusion module is used to acquire multi-source land resource data and preprocess the multi-source land resource data to generate a unified multimodal dataset. The indicator construction module is used to mine key indicators and assign weights based on the unified multimodal dataset using big data analysis technology, and generate a dynamic evaluation indicator system. The model optimization module is used to train and adaptively optimize the intelligent evaluation model using the unified multimodal dataset and the dynamic evaluation index system, and generate the optimized intelligent evaluation model. The calculation and prediction module is used to use the optimized intelligent assessment model to calculate the carrying capacity and predict the trend of newly added land resource data, and generate carrying capacity assessment results. The decision support module is used to generate suggestions for optimizing land resource utilization strategies based on the carrying capacity assessment results.

Images