A method and device for optimizing quality of cultivated land resources under multi-source data analysis

Abstract

The application discloses a kind of cultivated land resource quality optimization method and device under multi-source data analysis, it is related to cultivated land resource optimization technical field, the method includes: obtaining target cultivated land attribute, constructs the quality relationship matrix of the first matrix containing cultivated land quality main control factor and the second matrix of quality decline driving mechanism, generates optimization strategy;Based on the cyclic structure strategy generator of built-in conditional probability matrix, in combination with the quality matrix of real-time information acquisition, dynamically generates and adjusts optimization strategy;According to quality optimization strategy, implement phased resource management and optimization acceptance to target cultivated land.The application solves the technical problems that the prior art lacks system quantization and dynamic adjustment mechanism for complex correlation of multiple factors, and it is difficult to accurately adapt to the complex changes of actual cultivated land resources, achieves the technical effects of precision and dynamic adaptability of cultivated land quality optimization through multi-source data analysis and feedback optimization mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of arable land resource optimization, in particular to a method and device for optimizing arable land resource quality based on multi-source data analysis. BACKGROUND

[0002] Arable land resources are the foundation of agricultural production, and their quality directly affects food security, ecological environment protection, and economic development. However, with global climate change, soil degradation, over-exploitation, and widespread use of unreasonable agricultural cultivation methods, arable land quality is facing serious threats of decline.

[0003] Traditional methods for optimizing arable land quality are usually based on a single data source, such as soil monitoring data or climate data, lacking comprehensive support from multi-source data, resulting in insufficient adaptability and scientificity of optimization strategies. In addition, existing methods mostly use static analysis and empirical decision-making methods, which are difficult to reflect the dynamic changes and complexity of arable land quality in real time, especially in the face of climate change and soil property variability, lacking effective adjustment mechanisms, affecting the optimization effect. SUMMARY

[0004] The present application provides a method and device for optimizing arable land resource quality based on multi-source data analysis, to solve the technical problem that existing technologies lack systematic quantification and dynamic adjustment mechanisms for complex correlations of multiple factors, making it difficult to accurately adapt to the complex changes of actual arable land resources.

[0005] In a first aspect of the present application, a method for optimizing arable land resource quality based on multi-source data analysis is provided, which includes: obtaining target arable land attributes and mining a quality relationship matrix, wherein the quality relationship matrix includes a first matrix based on arable land quality main control factors and a second matrix based on arable land quality decline driving mechanisms; based on the quality relationship matrix, constructing a strategy generator, wherein the strategy generator has a conditional probability matrix built-in, which is used for strategy generation constraints, and the strategy generator is a loop structure; collecting information on the target arable land, determining a real-time quality matrix, and transmitting it to the strategy generator to generate an optimization strategy under the constraints of the conditional probability matrix, and performing feedback optimization based on grey correlation projection analysis to determine a quality optimization strategy; and based on the quality optimization strategy, managing the arable land resources of the target arable land.

[0006] In a second aspect of the present application, a cultivated land resource quality optimization device under multi-source data analysis is provided, the device comprising: a quality relationship matrix mining module, the quality relationship matrix mining module being configured to obtain a target cultivated land attribute and mine a quality relationship matrix, wherein the quality relationship matrix comprises a first matrix based on a cultivated land quality main control factor and a second matrix based on a cultivated land quality decline driving mechanism; a strategy generator construction module, the strategy generator construction module being configured to construct a strategy generator based on the quality relationship matrix, wherein the strategy generator is internally provided with a conditional probability matrix, the probability matrix being configured to perform strategy generation constraints, and the strategy generator is a cyclic structure; an optimization strategy generation module, the optimization strategy generation module being configured to collect information of a target cultivated land, determine a real-time quality matrix, transmit the real-time quality matrix to the strategy generator, and perform optimization strategy generation under conditional probability matrix constraints and feedback optimization based on grey correlation projection analysis to determine a quality optimization strategy; and a cultivated land resource management module, the cultivated land resource management module being configured to manage cultivated land resources of the target cultivated land based on the quality optimization strategy.

[0007] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] The cultivated land resource quality optimization method and device under multi-source data analysis provided in the present application relate to the technical field of cultivated land resource optimization, and the quality relationship matrix of cultivated land quality main control factors and decline driving mechanisms is constructed, a strategy generator internally provided with a conditional probability matrix is combined, an optimization strategy is dynamically generated, real-time quality matrix information collection and grey correlation projection analysis are combined, feedback optimization adjustment is realized, the target cultivated land is managed and accepted in stages according to the optimization strategy, efficient resource utilization and optimization target achievement are ensured, the technical problem that the prior art lacks a system quantization and dynamic adjustment mechanism for complex correlation of multiple factors and is difficult to accurately adapt to complex changes of actual cultivated land resources is solved, and the technical effects of realizing accuracy and dynamic adaptability of cultivated land quality optimization through a multi-source data analysis and feedback optimization mechanism are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort.

[0010] Figure 1 A cultivated land resource quality optimization method flowchart under multi-source data analysis provided in the embodiment of the present application is shown in the figure.

[0011] Figure 2This is a schematic diagram of a device for optimizing the quality of cultivated land resources under multi-source data analysis, provided in an embodiment of this application.

[0012] Figure labeling: Quality Relationship Matrix Mining Module 11, Strategy Generator Construction Module 12, Optimization Strategy Generation Module 13, Farmland Resource Management Module 14. Detailed Implementation

[0013] This application provides a method and apparatus for optimizing the quality of arable land resources based on multi-source data analysis, which addresses the technical problem that existing technologies lack a systematic quantification and dynamic adjustment mechanism for complex relationships among multiple factors, making it difficult to accurately adapt to the complex changes in actual arable land resources.

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0015] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0016] Example 1, as Figure 1 As shown, this application provides a method for optimizing arable land resource quality based on multi-source data analysis, the method comprising:

[0017] P10: Obtain the target arable land attributes and mine the quality relationship matrix, wherein the quality relationship matrix includes a first matrix based on the main control factors of arable land quality and a second matrix based on the driving mechanism of arable land quality decline.

[0018] Specifically, the attributes of the target arable land are acquired and analyzed. The attributes of the arable land can be divided into various types according to different geographical regions and agricultural conditions, such as black soil, northern dry land, southern paddy field, southern dry land, facility farmland, saline-alkali arable land, and the like. Each type of arable land has significant differences in quality performance and utilization mode, and therefore, personalized quality analysis and optimization must be performed for different types of arable land.

[0019] In order to effectively describe various elements of the arable land quality and reveal the mutual relationship, a quality relationship matrix is constructed. The matrix consists of two parts: the first part is a first matrix based on the main control factors of the arable land quality, and the second part is a second matrix based on the driving mechanism of the decline of the arable land quality.

[0020] The first matrix, i.e., the arable land quality main control factor matrix, is mainly constructed based on the core elements affecting the arable land quality, including factors such as the texture of the plough layer, pH value, organic matter content, and trace element content. The texture of the plough layer refers to the proportion of different particles in the soil (such as sand, loam, clay, and the like), which determines the water retention capacity, aeration, and nutrient supply capacity of the soil. The pH value affects the acidity and alkalinity of the soil, directly affecting the absorption of nutrients by crops and the activity of microorganisms. The organic matter content is closely related to the fertility of the soil and is the basis of soil health, while the trace element content (such as boron, zinc, copper, and the like) is an important component required for crop growth, affecting the nutritional status and disease resistance of crops. By comprehensively analyzing these main control factors, a basic matrix of the arable land quality can be formed, revealing the basic composition of the arable land quality.

[0021] The second matrix, i.e., the arable land quality decline driving mechanism matrix, is used to describe and quantify various driving factors leading to the decline of the arable land quality. These factors can be caused by natural environmental changes (such as drought, excessive precipitation) or human activities (such as unreasonable cultivation, improper fertilization, excessive cultivation, and the like). For example, excessive cultivation can lead to soil degradation, such as compaction of the plough layer and loss of organic matter; unreasonable fertilization can cause nutrient imbalance in the soil, thereby reducing the productivity of the arable land. Through analysis of these driving mechanisms, the potential causes of the decline of the arable land quality can be revealed, thereby providing a basis for subsequent optimization strategies.

[0022] By combining the two matrices (the first matrix and the second matrix), the current situation and the trend of the arable land quality can be comprehensively analyzed. This quality relationship matrix not only provides a theoretical basis for the quantitative analysis of the arable land quality, but also provides important data support and decision-making basis for the subsequent optimization strategy generation.

[0023] Further, after mining the quality relationship matrix, the step P10 of the embodiment of the present application further includes:

[0024] P11a: Obtain the target farmland attributes, and perform agricultural big data retrieval with the region of the target farmland as a constraint, and call farmland monitoring data of the same attributes; P12a: Traverse the farmland monitoring data to mine the matrix element range of the quality relationship matrix; P13a: Based on the matrix element range, perform quality optimization constraints on the target farmland, including natural optimization that meets the matrix element range and non-natural condition optimization that does not meet the matrix element range.

[0025] Optionally, after mining the quality relationship matrix, the accuracy of farmland quality optimization can be further improved. First, for the specific attributes of the target farmland (such as black soil, northern dry land, southern paddy field, etc.), based on its regional characteristics (including climate, soil type, water resources, etc.), agricultural big data retrieval and analysis are performed. By calling historical monitoring data of farmland with the same attributes, the long-term quality change trend and important characteristics of similar farmland can be obtained, such as soil nutrient level, typical crop planting conditions and success rate, etc. These data provide important reference benchmarks for the target farmland, helping to clarify the potential upper limit and limit range that the current farmland can achieve under natural conditions before generating optimization strategies.

[0026] Next, the called farmland monitoring data is comprehensively traversed to mine the matrix element range of the quality relationship matrix, i.e. the universal state range of local farmland under natural conditions. The core goal of this step is to analyze whether the characteristics of the target farmland can meet the optimization requirements. The matrix element range is determined by local climate conditions, soil fertility, water resource conditions, etc. For example, in the northern dry land, the natural conditions are usually not suitable for planting tropical fruits in the south, and additional artificial environmental optimization (such as greenhouse planting) is needed to achieve higher production goals. By traversing the monitoring data, it can be identified which indicators in the target farmland exceed the local universal range, and further determine whether optimization requires additional technical support.

[0027] Finally, based on the mined matrix element range, the quality optimization of the target farmland is constrained. Optimization methods can be divided into two categories: natural optimization and non-natural condition optimization. For indicators that can meet the requirements through natural condition adjustment (such as fertilization, irrigation or soil improvement), i.e. indicators that meet the matrix element range, natural optimization strategies are adopted. For example, by increasing organic matter content or improving soil structure to improve farmland quality. For indicators that do not meet the matrix element range, external technical means need to be introduced, such as greenhouse regulation of temperature and humidity, precise irrigation system control of water supply, artificial light compensation for insufficient light, etc., to break through the limitations of natural conditions. The differentiation of such optimization strategies can ensure the maximum utilization efficiency of resources, while providing practical solutions for specific planting needs.

[0028] Through the above steps, the precision optimization constraints of arable land quality can be systematically completed, both the potential of natural conditions and artificial technical means are flexibly supplemented, and scientific basis is provided for arable land resource management and efficient use.

[0029] P20: constructing a strategy generator based on the quality relationship matrix, wherein the strategy generator is built-in with a conditional probability matrix, the probability matrix is used for strategy generation constraints, and the strategy generator is a loop structure.

[0030] Further, the strategy generator is constructed, and the step P20 of the embodiments of the present application further includes:

[0031] P21: traversing the quality main control factors, calling arable land resource optimization samples for factor-by-factor decomposition, determining N groups of optimization samples, wherein the N groups of optimization samples correspond one-to-one to the quality main control factors; P22: traversing the N groups of optimization samples, performing in-group co-reference fitting to determine a baseline adjustment scheme, wherein co-reference fitting refers to fitting optimization samples of the same mode; and P23: embedding the baseline adjustment scheme in the strategy generator, and based on the arable land resource optimization samples, supervising training of the strategy generator.

[0032] Specifically, the strategy generator is constructed based on the quality relationship matrix to realize the intelligentization and systematization of arable land resource quality optimization. The core of the strategy generator is to build-in a conditional probability matrix, which is used to constrain and guide the strategy generation process, and a loop structure is adopted to continuously optimize the strategy generation, so as to ensure that the output result meets the optimization target.

[0033] In the construction process of the strategy generator, the quality main control factors in the quality relationship matrix need to be analyzed one by one. The quality main control factors include core indicators of arable land quality, such as plough layer texture, pH value, and organic matter content, and each factor is directly related to the optimization direction of arable land quality. For each main control factor, the arable land resource optimization sample library is called to perform associated analysis on the main control factor and sample data. The optimization sample library stores a large number of optimization cases verified in practice, such as the effect of different doses of fertilization on soil organic matter and the effect of different acid-base adjusting agents on soil pH value. This process divides the optimization samples into N groups of optimization samples through factor-by-factor decomposition, and each group of optimization samples corresponds to a main control factor. This grouping method ensures that the optimization strategy of each main control factor is data-driven and can cover its specific variation range, providing sufficient support for the precise design of the strategy generator.

[0034] After completing the sample grouping, it is necessary to conduct in-depth analysis on each group of optimization samples, and extract the internal regularity of sample data by using the in-group co-reference fitting method. In-group co-reference fitting refers to analyzing the relationship between sample data of different optimization amplitudes under the same optimization method, and then determining the optimization baseline and adjustment rule. For example, under the optimization method of adjusting the pH value of soil, different doses of adjusting agents may correspond to different effects. By fitting these sample data, the following information can be extracted: baseline adjustment scheme, i.e. the best basic adjustment strategy under the optimization method, for example, the target range of pH value and the recommended value of the application dose. Linear adjustment relationship, i.e. analyzing the effect of different amplitude optimization methods, establishing a mathematical model (such as linear or nonlinear relationship) between optimization amplitude and its effect. For example, the degree of influence of each unit of adjusting agent on the change of pH value is determined.

[0035] These fitting results not only provide reference values for optimization, but also construct the elastic space of different amplitude optimization. Finally, the baseline adjustment scheme and the linear adjustment relationship jointly serve as the core output of in-group fitting, providing standardized optimization parameters for the strategy generator.

[0036] After completing the extraction of the baseline adjustment scheme, the baseline adjustment scheme and the linear adjustment relationship of each group of optimization samples are embedded into the strategy generator as its core rule module. Subsequently, the strategy generator is supervised trained in combination with all the arable land resource optimization samples. The supervised training iteratively adjusts the conditional probability matrix in the strategy generator by gradually inputting the optimization samples and their corresponding target values. The role of the conditional probability matrix is to constrain the output of the strategy generator, so that the generated strategy meets the logic and effect of actual optimization. At the same time, the strategy generator will use its loop structure to verify and adjust the generated strategy during the training process, and further optimize its internal parameters by comparing and analyzing with the sample target values. This training method ensures that the strategy generator can continuously learn and generate efficient strategies that meet the optimization needs according to the input arable land characteristics, ultimately improving the intelligence and precision of arable land resource management.

[0037] P30: Collect information about the target arable land, determine the real-time quality matrix, transmit it to the strategy generator, and generate an optimization strategy under the constraint of the conditional probability matrix, and feedback optimization based on gray correlation projection analysis to determine the quality optimization strategy.

[0038] Further, the step P30 of the embodiment of the present application further comprises:

[0039] P31: Obtain the real-time quality matrix and the ideal quality matrix of the target farmland, measure the matrix difference, and determine the optimization target; P32: input the optimization target into the strategy generator, determine multiple single-factor strategies based on the conditional probability matrix, and combine to determine a first optimization strategy; P33: take the first optimization strategy as a benchmark to perform feedback optimization until a preset number of iterations is met, and output the quality optimization strategy.

[0040] It should be understood that real-time information collection is performed on the target farmland to form a real-time quality matrix, and an optimization strategy is generated by a strategy generator, while feedback optimization is realized based on gray relational projection analysis to ultimately determine a quality optimization strategy.

[0041] First, real-time quality data of the target farmland is collected to generate a real-time quality matrix. This matrix reflects the current actual quality state of the target farmland, including key factors such as soil structure, nutrient content, pH value, and the like. At the same time, a preset ideal quality matrix is called, which represents the quality standard of the target farmland under optimal conditions. By performing difference analysis on the real-time quality matrix and the ideal quality matrix, the difference degree of each quality main control factor is calculated. These differences are used to determine the optimization target, for example, for some factors (such as insufficient organic matter content), an enhanced optimization strategy needs to be adopted, while for other factors (such as excessive acid-base), a weakened or reverse adjustment may be needed.

[0042] Since there may be mutual influence of optimization strategies between different quality factors, for example, improving soil organic matter may cause slight changes in pH value, a quality coefficient evaluation can be performed to comprehensively consider the interaction between factors to determine an overall optimization guide direction. This guide direction guides the generation of probability to provide a more explicit reference for subsequent strategy generation.

[0043] The optimization target is input into the strategy generator, which decomposes the optimization target by factors based on the conditional probability matrix to generate multiple single-factor strategies. A single-factor strategy is an independent optimization scheme for a single quality main control factor, for example, increasing organic fertilizer to improve soil organic matter or applying an acid regulator to adjust soil pH value. The strategy generator judges the priority and applicability of each single-factor strategy through the conditional probability matrix to ensure that the generated strategy achieves a balance between technical feasibility and reliability of effect.

[0044] Subsequently, the strategy generator combines multiple single-factor strategies to generate a first optimization strategy. The first optimization strategy is a global optimization scheme that comprehensively considers the actual needs and mutual influence of each quality main control factor. For example, in the conditions of northern dry land, it may be necessary to adjust soil structure and water supply at the same time, and the combination of these strategies needs to be reasonably weighed through the conditional probability matrix to avoid mutual conflict.

[0045] The first optimization strategy is used as the initial benchmark, and feedback optimization is generated through the feedback mechanism of the strategy generator. The feedback optimization is based on real-time evaluation and grey correlation projection analysis to verify and correct the effect of the generated strategy. Grey correlation projection analysis can quantify the difference between the actual effect of the optimization strategy and the target, and gradually narrow the optimization gap through weight adjustment and strategy reconstruction. The strategy generator iterates in each round of feedback optimization, combining the real-time state of the target farmland and the optimization target to adjust the strategy generation direction.

[0046] This process continues until the preset number of iterations is reached or the optimization target is fully met, and the final quality optimization strategy is output. This optimization strategy not only takes into account the real-time conditions of the target farmland, but also combines the refinement adjustments made by the feedback mechanism to the strategy, ensuring the scientificity and feasibility of the generated scheme.

[0047] Further, the step P33 of generating feedback optimization further includes:

[0048] P33-1: Simulate the first optimization strategy and perform grey correlation projection analysis to determine the quality coefficient, wherein the quality coefficient includes the projection angle and the projection value, the projection angle is the difference in direction from the ideal quality matrix, and the projection value is the degree of excellence from the ideal quality matrix; P33-2: Based on the quality coefficient, generate first feedback information; P33-3: Based on the first feedback information and the conditional probability matrix, generate and combine to determine the second optimization strategy.

[0049] Optionally, the specific process of generating feedback optimization can be, on the basis of the first optimization strategy, first performing simulation evaluation, and using grey correlation projection analysis to comprehensively analyze the effect of the strategy. Grey correlation projection analysis is an important tool for evaluating optimization strategies, used to quantify the matching degree between the first optimization strategy and the ideal quality matrix. Specifically, this analysis generates two core indicators: the projection angle and the projection value. The projection angle describes the difference in direction between the optimization strategy and the ideal quality matrix, and the smaller the projection angle, the closer the direction to the ideal state; the projection value reflects the closeness of the optimization strategy to the ideal quality matrix in terms of quality, and the higher the projection value, the closer the optimization strategy to the target effect. Through this process, the effectiveness of the current optimization strategy and its existing deviations can be intuitively identified, providing quantitative basis for subsequent adjustments.

[0050] Then, the first feedback information is generated according to the quality coefficients (including projection angle and projection value) obtained by the grey correlation projection analysis. The feedback information specifically includes two aspects: first, for the indicators with large projection angles, specific direction adjustment suggestions are proposed, such as adjusting the fertilization strategy to better meet the soil nutrient demand; second, for the indicators with low projection values, correction suggestions for the optimization amplitude are provided, such as increasing the application amount of soil conditioner or enhancing the irrigation intensity to improve the effect. These feedback information clearly points out the shortcomings of the current strategy and proposes specific directions for optimization adjustment, laying a foundation for generating new optimization strategies.

[0051] Finally, the first feedback information is combined with the conditional probability matrix built-in the strategy generator to generate a new second optimization strategy by retraining and adjusting the weights. This process includes: first, applying the first feedback information to correct the parameters in the conditional probability matrix, so that the generator can more accurately reflect the latest optimization target; second, under the guidance of the adjusted conditional probability matrix, the single-factor optimization strategy of each quality main control factor is regenerated; finally, the single-factor strategies are combined into the second optimization strategy, and its coordination and applicability in the global range are ensured. The new optimization strategy can better balance the relationship between the actual needs of the target farmland and the ideal state, and can provide an efficient solution for realizing the final quality optimization.

[0052] Through the above steps, the feedback optimization not only realizes the accurate adjustment of the first optimization strategy, but also further improves the scientificity and practicality of the optimization strategy through the combination of dynamic feedback and conditional probability matrix. This iterative optimization mechanism ensures that the output quality optimization strategy can fully adapt to the actual conditions of the target farmland, providing strong support for resource management and farmland improvement.

[0053] Further, after determining the optimization target, the step P31 of the embodiment of the present application further includes:

[0054] P31-1a: initializing the weights of the first matrix to determine the first distribution weight, wherein the first distribution weight is positively correlated with the correlation degree of the optimization target; P31-2a: initializing the weights of the second matrix to determine the second distribution weight, wherein the second distribution weight is negatively correlated with the correlation degree of the optimization target.

[0055] In a possible embodiment of the present application, after determining the optimization target, the first matrix (arable land quality main control factor matrix) in the quality relationship matrix is initialized with weights oriented to the optimization target, and the first distribution weight is calculated and generated. Specifically, the setting of the first distribution weight is positively correlated with the relevance of the optimization target, that is, the higher the contribution of a certain main control factor to the realization of the optimization target, the greater the weight value. For example, if the target is to improve the organic matter content of arable land, the main control factors directly related to organic matter in the first matrix (such as soil organic matter level, soil aeration, etc.) will be given higher weights, and the weights of other factors will be relatively reduced.

[0056] The technical core of this process lies in positive correlation evaluation. By analyzing the historical optimization effect data of the optimization target and each main control factor, a correlation model is constructed, and the weights of each factor are distributed according to the model results. Through this weight initialization step, the optimization direction of the first matrix is more clear, providing a scientific basis for subsequent optimization strategy generation.

[0057] Then, the second matrix (quality decline driving mechanism matrix) in the quality relationship matrix is initialized with weights oriented to the optimization target, and the second distribution weight is calculated and generated. Unlike the first matrix, the second distribution weight is negatively correlated with the relevance of the optimization target, that is, the greater the negative impact of a certain driving mechanism on the optimization target, the lower the weight value. For example, if the optimization target is to improve the content of soil trace elements, and some driving mechanisms (such as unreasonable fertilization methods or irrigation caused by element loss) may lead to negative optimization of the target, the weights of these mechanisms will be correspondingly reduced, thereby reducing their impact on strategy generation.

[0058] The key to the weight initialization lies in negative correlation control. By analyzing the historical negative effect data of the driving mechanism on the target optimization effect, the weight value is dynamically reduced using a reverse adjustment algorithm to ensure that the negative guiding effect of the second matrix on the optimization strategy is effectively controlled, while avoiding optimization bias caused by too high weights.

[0059] Through the weight initialization of the above steps, the weight distribution of the first matrix and the second matrix is optimized, forming a dynamic weight configuration system oriented to the optimization target. This configuration method enhances the response capability of the strategy generator to the optimization target, so that the subsequent generated optimization strategy can not only fully play the positive role of the main control factors, but also effectively avoid the negative impact of the quality decline driving mechanism, providing technical support for efficient and precise arable land quality optimization.

[0060] P40: Based on the quality optimization strategy, managing the arable land resources of the target arable land.

[0061] Further, the step P40 of the embodiment of the present application further includes:

[0062] P41: Identify the quality optimization strategy, divide it into M stages, each stage identifies a stage quality matrix; P42: Based on the M strategy stages, conduct the quality optimization acceptance of the target farmland in stages, determine the acceptance quality matrix; P43: Map the acceptance quality matrix and the stage quality matrix, and conduct feedback optimization management.

[0063] It should be understood that based on the generated quality optimization strategy, comprehensive resource management is carried out on the target farmland. By refining the optimization strategy into a phased plan and conducting acceptance and feedback adjustment on the quality target of each stage, it ensures that the optimization effect is gradually achieved, and the efficient management of farmland resources is realized.

[0064] Before executing the optimization strategy, first analyze the quality optimization strategy in depth, and divide the strategy into M stages according to its optimization target and implementation sequence. Each stage takes an independent quality optimization target as the core and generates a corresponding stage quality matrix, which clearly shows the key indicators that the farmland needs to achieve in the stage (such as soil organic matter content, trace element level, pH value range, etc.). Through this phased management method, complex global optimization tasks can be decomposed into several easy-to-execute subtasks, making resource allocation more precise and optimization implementation more organized.

[0065] After each strategy stage is completed, quality optimization acceptance is conducted to evaluate whether the actual optimization effect meets the stage target. Specifically, through real-time data collection and monitoring, an acceptance quality matrix is generated and compared with the stage quality matrix. The acceptance quality matrix covers the current state indicators of the farmland, helping to identify the strengths and weaknesses of optimization implementation.

[0066] The advantage of phased acceptance is that it can dynamically monitor the optimization progress and avoid the risk of global failure. For example, if some soil nutrient indicators do not meet the standards during the acceptance of a certain stage, subsequent measures can be adjusted in time. This phased evaluation mechanism effectively improves the flexibility of management and the accuracy of target achievement.

[0067] Finally, the acceptance quality matrix and the stage quality matrix are mapped and analyzed to identify gaps and implement feedback optimization management. Since the optimization of arable land resources may be affected by external environmental factors (such as climate change and soil natural remediation cycles), some objectives may not be fully achieved, requiring real-time adjustments through feedback optimization management. For example, gap analysis identifies indicators in the acceptance quality matrix that have not met stage objectives, and analyzes the reasons for non-achievement, such as reduced fertilization effectiveness due to climate conditions. By adjusting the optimization plan, the optimization strategy for the current stage is adjusted based on the gap indicators, and the implementation time of this stage is extended if necessary to ensure that the optimization effect meets the expected goals. Through feedback optimization, the adjusted strategy is passed to the next stage, providing a basis for correction in subsequent strategy implementation and forming a closed-loop management system.

[0068] Feedback-based optimization management allows for dynamic responses to external uncertainties, ensuring that the quality optimization results at each stage meet expectations. The phased management approach provides refined support for resource optimization, while the feedback optimization mechanism enhances the adaptability of management, guaranteeing the continuous improvement of arable land resources.

[0069] In summary, the embodiments of this application have at least the following technical effects:

[0070] This application constructs a quality relationship matrix of the main controlling factors and the driving mechanism of decline in arable land quality. Combined with a strategy generator with a built-in conditional probability matrix, it dynamically generates optimization strategies. Based on real-time quality matrix information collection and grey relational projection analysis, it realizes feedback optimization and adjustment. According to the optimization strategy, it conducts phased management and acceptance of target arable land to ensure efficient resource utilization and the achievement of optimization goals.

[0071] It has achieved the technical effect of precise and dynamic adaptation in optimizing arable land quality through multi-source data analysis and feedback optimization mechanisms.

[0072] Example 2, based on the same inventive concept as the method for optimizing arable land resource quality under multi-source data analysis in the foregoing examples, such as... Figure 2 As shown, this application provides a device for optimizing arable land resource quality through multi-source data analysis. The device and method embodiments in this application are based on the same inventive concept. The device includes:

[0073] The quality relationship matrix mining module 11 is used to obtain the target arable land attributes and mine the quality relationship matrix. The quality relationship matrix includes a first matrix based on the main control factors of arable land quality and a second matrix based on the driving mechanism of arable land quality decline.

[0074] A strategy generator construction module 12 is configured to construct a strategy generator based on the quality relation matrix, wherein the strategy generator is built-in with a conditional probability matrix for strategy generation constraint, and the strategy generator is a loop structure.

[0075] An optimization strategy generation module 13 is configured to collect information of target farmland, determine a real-time quality matrix, transmit the real-time quality matrix to the strategy generator, generate an optimized strategy under the constraint of the conditional probability matrix, and determine a quality optimization strategy based on feedback optimization of grey relational projection analysis.

[0076] A farmland resource management module 14 is configured to manage farmland resources of the target farmland based on the quality optimization strategy.

[0077] Further, the quality relation matrix mining module 11 is further configured to perform the following steps:

[0078] Target farmland attributes are acquired, and agricultural big data retrieval is performed with the region of the target farmland as a constraint to call farmland monitoring data of the same attributes; the farmland monitoring data is traversed to mine a matrix element range of the quality relation matrix; and quality optimization constraint is performed on the target farmland based on the matrix element range, including natural optimization that meets the matrix element range and non-natural condition optimization that does not meet the matrix element range.

[0079] Further, the strategy generator construction module 12 is further configured to perform the following steps:

[0080] The quality main control factors are traversed, and farmland resource optimization samples are called to perform factor-by-factor decomposition to determine N groups of optimization samples, wherein the N groups of optimization samples correspond to the quality main control factors one by one; the N groups of optimization samples are traversed to perform in-group co-reference fitting to determine a baseline adjustment scheme, wherein the co-reference fitting refers to fitting of optimization samples of the same mode; the baseline adjustment scheme is built into the strategy generator, and the strategy generator is supervised and trained based on the farmland resource optimization samples.

[0081] Further, the optimization strategy generation module 13 is further configured to perform the following steps:

[0082] A real-time quality matrix and an ideal quality matrix of target farmland are acquired, a matrix difference is measured, and an optimization target is determined; the optimization target is input into the strategy generator, a plurality of single-factor strategies are determined based on a conditional probability matrix, a first optimization strategy is determined by combination, feedback optimization is performed with the first optimization strategy as a benchmark, and the quality optimization strategy is output until a preset iteration number is met.

[0083] Further, the optimization strategy generation module 13 is further configured to perform the following steps:

[0084] The first optimization strategy is simulated and a grey correlation projection analysis is performed to determine a quality coefficient, wherein the quality coefficient comprises a projection angle and a projection value, the projection angle being a difference in direction from an ideal quality matrix, and the projection value being a degree of excellence or inferiority from the ideal quality matrix; first feedback information is generated based on the quality coefficient; and a second optimization strategy is generated and combined based on the first feedback information and the conditional probability matrix.

[0085] Further, the optimization strategy generation module 13 is further configured to perform the following steps:

[0086] The first matrix is initialized with weights in accordance with an optimization target to determine a first distribution weight, wherein the first distribution weight is positively correlated with the optimization target; and the second matrix is initialized with weights in accordance with the optimization target to determine a second distribution weight, wherein the second distribution weight is negatively correlated with the optimization target.

[0087] Further, the arable land resource management module 14 is further configured to perform the following steps:

[0088] The quality optimization strategy is identified, and M strategy stages are divided, wherein each strategy stage is identified with a stage quality matrix; the quality optimization acceptance of the target arable land is performed in stages based on the M strategy stages to determine an acceptance quality matrix; and the acceptance quality matrix and the stage quality matrix are mapped to perform feedback optimization management.

[0089] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0090] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0091] The present specification and drawings are merely exemplary of the present application, and any and all modifications, variations or combinations thereof which fall within the scope of the present application should be considered covered by the present application. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is intended to include these modifications and variations.

Claims

1. A method for optimizing quality of cultivated land resources under multi-source data analysis, characterized in that, The method comprises: acquiring target farmland attributes and mining a quality relationship matrix, wherein the quality relationship matrix comprises a first matrix based on farmland quality main control factors and a second matrix based on farmland quality decline driving mechanisms; based on the quality relationship matrix, constructing a strategy generator, wherein the strategy generator is internally provided with a conditional probability matrix, the probability matrix is used for strategy generation constraints, and the strategy generator is a loop structure; information collection is performed on the target farmland to determine a real-time quality matrix, which is transmitted to the strategy generator to generate an optimized strategy under the constraint of the conditional probability matrix and feedback optimization based on grey correlation projection analysis to determine a quality optimization strategy; based on the quality optimization strategy, farmland resource management is performed on the target farmland; constructing a strategy generator comprises: traversing the quality main control factors, calling farmland resource optimization samples for factor-by-factor decomposition to determine N groups of optimization samples, wherein the N groups of optimization samples correspond one-to-one to the quality main control factors; traversing the N groups of optimization samples, performing in-group coreference fitting to determine a baseline adjustment scheme, wherein coreference fitting refers to fitting optimization samples of the same mode; the baseline adjustment scheme is internally provided in the strategy generator, and the strategy generator is supervised and trained based on the farmland resource optimization samples.

2. The farmland resource quality optimization method under multi-source data analysis of claim 1, wherein, After mining the quality relationship matrix, the following steps are included: acquiring target farmland attributes, performing agricultural big data retrieval with the region of the target farmland as a constraint, and calling farmland monitoring data of the same attributes; traversing the farmland monitoring data to mine the matrix element range of the quality relationship matrix; based on the matrix element range, performing quality optimization constraints on the target farmland, including natural optimization that meets the matrix element range and non-natural condition optimization that does not meet the matrix element range.

3. The farmland resource quality optimization method under multi-source data analysis of claim 1, wherein, generating an optimized strategy under the constraint of a conditional probability matrix and feedback optimization based on grey correlation projection analysis comprises: acquiring a real-time quality matrix and an ideal quality matrix of the target farmland, measuring the matrix difference to determine an optimization target; inputting the optimization target into the strategy generator to determine a plurality of single-factor strategies based on a conditional probability matrix, and combining to determine a first optimization strategy; taking the first optimization strategy as a benchmark, performing generation feedback optimization until a preset iteration number is met, and outputting the quality optimization strategy.

4. The farmland resource quality optimization method under multi-source data analysis of claim 3, wherein, The generation feedback optimization comprises: simulating the first optimization strategy and performing grey correlation projection analysis to determine a quality coefficient, wherein the quality coefficient comprises a projection angle and a projection value, the projection angle is a directional difference from the ideal quality matrix, and the projection value is a good-bad degree from the ideal quality matrix; based on the quality coefficient, generating first feedback information; based on the first feedback information and the conditional probability matrix, generating and combining to determine a second optimization strategy.

5. The farmland resource quality optimization method under multi-source data analysis of claim 3, wherein, After determining the optimization target, the following steps are included: directing to the optimization target, performing weight initialization on the first matrix to determine a first distribution weight, wherein the first distribution weight is positively correlated with the correlation degree of the optimization target. The second matrix is initialized with weights according to the optimization target, and a second distribution weight is determined, wherein the second distribution weight is negatively correlated with the correlation degree of the optimization target.

6. The farmland resource quality optimization method under multi-source data analysis of claim 1, wherein, The target farmland is managed according to the farmland resource management. The quality optimization strategy is identified, and M strategy stages are divided, wherein each strategy stage is identified with a stage quality matrix. Based on the M strategy stages, the quality optimization acceptance of the target farmland is carried out in stages, and an acceptance quality matrix is determined. The acceptance quality matrix and the stage quality matrix are mapped, and feedback optimization management is carried out.

7. An arable land resource quality optimization device under multi-source data analysis, characterized in that, The device comprises: A quality relationship matrix mining module is configured to obtain target farmland attributes and mine a quality relationship matrix, wherein the quality relationship matrix comprises a first matrix based on farmland quality main control factors and a second matrix based on farmland quality decline driving mechanisms; A strategy generator construction module is configured to construct a strategy generator based on the quality relationship matrix, wherein the strategy generator is internally provided with a conditional probability matrix, the probability matrix is used for strategy generation constraints, and the strategy generator is a loop structure; An optimization strategy generation module is configured to collect information of the target farmland, determine a real-time quality matrix, and transmit it to the strategy generator to generate an optimization strategy under the constraint of the conditional probability matrix and feedback optimization based on gray correlation projection analysis, and determine a quality optimization strategy; A farmland resource management module is configured to manage farmland resources of the target farmland based on the quality optimization strategy; The strategy generator is constructed, comprising: Traverse the quality main control factors, call the farmland resource optimization samples for factor-by-factor decomposition, and determine N groups of optimization samples, wherein the N groups of optimization samples correspond one-to-one to the quality main control factors; Traverse the N groups of optimization samples, perform group-intra co-reference fitting, and determine a baseline adjustment scheme, wherein co-reference fitting refers to fitting optimization samples of the same mode; The baseline adjustment scheme is internally provided in the strategy generator, and the strategy generator is supervised trained based on the farmland resource optimization samples.

Images