In-situ multi-parameter sensing soil suitability evaluation and planting decision method and system

Abstract

The application discloses an in-situ multi-parameter sensing soil suitability evaluation and planting decision method and system, relates to the technical field of intelligent agriculture and artificial intelligence data analysis, and solves the problems of difficult data interpretation of existing portable detection equipment, great environmental interference of in-situ measurement and lack of targeted planting suggestions. The in-situ sensing terminal utilizes an integrated straight-in probe to collect soil temperature, humidity, conductivity, pH value and nitrogen, phosphorus and potassium indexes in real time to construct a measured vector; a nonlinear compensation model based on temperature and humidity coupling is constructed to correct data and vectorize; an ideal environment vector and a limit vector are constructed based on a crop growth database; a sensitivity weight vector is introduced to calculate the weighted generalized distance of the measured vector and the ideal vector, and a suitability score is generated in combination with a threshold truncation function; a reasoning engine analyzes the vector difference to generate water and fertilizer regulation and control decisions, and realizes the full-process intelligentization from data acquisition to suitability diagnosis and then to decision execution.

Description

Technical Field

[0001] This invention belongs to the field of smart agriculture and artificial intelligence data analysis technology, and in particular relates to an in-situ multi-parameter sensing method and system for soil suitability evaluation and planting decision-making. Background Technology

[0002] In the context of precision agriculture development, rapid perception and scientific decision-making regarding the soil environment are crucial for improving agricultural production efficiency. The physicochemical properties of soil directly determine the growth status, yield, and quality of crops. Arid regions, especially the Loess Plateau, have coarse, infertile soils with low nutrient levels, facing the dual challenges of soil erosion and drought, placing higher demands on crop suitability. Existing soil testing and evaluation technologies suffer from several problems. Traditional soil nutrient testing relies primarily on laboratory chemical analysis methods, such as the Kjeldahl method and flame photometry. While these methods are highly accurate, they require cumbersome steps such as soil sampling, drying, grinding, extraction, and color development, typically taking three to seven days to produce results. This time lag prevents farmers from making real-time adjustments to water and fertilizer management based on the current soil conditions, often causing them to miss the optimal window for agricultural activities. Most portable soil testers on the market can only detect a single indicator. Even rapid testers that integrate multiple sensors often simply list numbers, such as a nitrogen content of 63 mg / L. For ordinary farmers lacking professional agronomic knowledge, these cold numbers provide no useful information. Users don't know whether a nitrogen content of 63 mg / L is too high or too low for their crops, nor do they know which crops are suitable for this data combination. The data fails to be transformed into useful information. The soil environment is a complex, multi-factor coupled system where water, fertilizer, air, and heat interact. For example, even with sufficient nutrients, plants cannot absorb them in extremely dry conditions, or phosphorus can be fixed in highly acidic soils. Current technology lacks a mathematical model that can comprehensively consider multiple indicators and differentiate evaluations based on different crop characteristics. Judgments are often made based on human experience, which is highly subjective and prone to misjudgment. Furthermore, in-situ pin-type measurements are easily affected by environmental factors such as soil temperature, humidity, and compaction. For example, the mobility of soil ions decreases under low temperature or dry conditions, leading to significantly lower readings from electrochemical sensors. Existing equipment lacks a robust temperature-moisture coupling compensation algorithm, resulting in large discrepancies between field measurements and actual values, thus limiting its reference value. Therefore, achieving intelligent management of the entire process from data acquisition to suitability diagnosis and decision execution has become a pressing technical challenge. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes an in-situ multi-parameter sensing method and system for soil suitability evaluation and planting decision-making, thereby resolving the issues present in the prior art.

[0004] In a first aspect, to achieve the above objectives, the present invention provides an in-situ multi-parameter sensing method for soil suitability evaluation and planting decision-making, comprising the following steps: The integrated in-situ soil sensing probe was used to collect multiple soil physicochemical index data of the target detection point in real time, and a primary vector of measured soil characteristics was constructed. To address the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, nonlinear corrections were made to conductivity and nitrogen, phosphorus and potassium readings using real-time acquired temperature and moisture data. The corrected physical quantities were then mapped to a feature space of the same dimension to generate a standard soil feature vector. A pre-set crop growth database containing the optimal physicochemical environmental threshold ranges for various crops at different growth stages is provided. In response to the user's evaluation request, the set of crops to be evaluated is retrieved, and the ideal environment feature vector and tolerance limit feature vector corresponding to each crop are extracted. A sensitivity weight vector is introduced to calculate the weighted generalized distance between the standard soil feature vector and the ideal environment feature vector. At the same time, the risk penalty factor of the standard soil feature vector relative to the tolerance limit feature vector is calculated. Based on the weighted generalized distance and the risk penalty factor, a suitability evaluation function is constructed to quantify the current soil suitability for crops. The crop set is recommended and ranked based on suitability scores. For target crops, significant features are extracted from the vector differences and input into an inference engine built on rule chains and agronomic knowledge graphs to generate a precision planting decision scheme containing water and fertilizer regulation operation instructions.

[0005] Optionally, the process of nonlinearly correcting the conductivity and nitrogen, phosphorus and potassium readings includes: correcting the difference between the measured temperature and the standard laboratory temperature using a temperature compensation coefficient, correcting the ratio of the measured volumetric water content to the reference value of saturated water content using a humidity influence factor, and eliminating the influence of temperature on ion mobility and the dilution or concentration effect of water on electrolyte concentration through the product of the temperature term and the humidity term.

[0006] Optionally, the process of constructing the suitability evaluation function includes: using the sensitivity weight vector to perform a weighted summation of the differences between the standard soil feature vector and the ideal environment feature vector in each dimension; setting a threshold cutoff function, whereby the threshold cutoff function outputs a minimum penalty coefficient when a certain indicator exceeds the survival red line defined by the crop tolerance limit vector, otherwise the output is 1; and multiplying the weighted summation result with the output of the threshold cutoff function to obtain the current soil suitability score for the crop.

[0007] Optionally, the process of obtaining the sensitivity weight vector includes: collecting multiple sets of historical planting case data, each set of data containing soil feature vectors and actual crop yield labels; constructing difference pairs for each crop, one set being high-yield samples and the other set being low-yield samples; constructing a loss function containing weight parameters, minimizing the loss function using stochastic gradient descent, so that the score of high-yield samples is significantly higher than that of low-yield samples under the weighted evaluation system, and automatically learning the weights of key indicators affecting crop yield.

[0008] Optionally, the process of generating a precision planting decision scheme includes: semantically mapping the vector difference to identify positive deviation, negative deviation, and the degree of deviation; retrieving entity nodes and relational edges that match the target crop and deviation features in the agronomic knowledge graph; combining the matched control measures into structured text using a pre-trained natural language generation template, and outputting decision suggestions containing water and fertilizer control operation instructions.

[0009] Secondly, this invention also provides a crop soil suitability assessment and intelligent planting decision-making system based on in-situ multi-parameter sensing in arid areas, used to implement the in-situ multi-parameter sensing soil suitability assessment and planting decision-making method, the system comprising: The in-situ sensing terminal is a handheld or fixed hardware device. The front end is equipped with a direct-insertion multi-parameter composite probe, which integrates a stainless steel five-pin electrode array, a thermistor and a pH glass electrode. The internal components include a signal conditioning circuit and a wireless communication module. The cloud computing platform is equipped with data processing services to receive raw data and execute environmental coupling compensation algorithms, and stores a crop growth database. The intelligent analysis engine runs in the cloud or at the edge and includes a weight training module and an inference engine; the interactive application client runs on the user's mobile terminal and is equipped with a visual human-computer interface.

[0010] Optionally, the visual human-computer interface of the interactive application client includes a multi-color block dashboard. The multi-color block dashboard presets the health threshold range of each indicator under the general agronomic standard. When real-time data is received, the background color of the color block is automatically rendered according to the range in which the value is located. The indicator name, real-time value and unit are displayed in the color block at the same time. Clicking on any color block will jump to the historical trend curve page of that indicator.

[0011] Optionally, the system further includes a closed-loop feedback optimization module, which receives the harvesting results input by the user in the interactive application client, constructs a new training sample pair with the early detection data and the later harvesting results, triggers the online incremental learning of the intelligent analysis engine, and fine-tunes and updates the sensitivity weight vector.

[0012] Thirdly, the present invention also provides a computer terminal device, comprising: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the in-situ multi-parameter sensing soil suitability evaluation and planting decision method in the first aspect described above.

[0013] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the in-situ multi-parameter sensing soil suitability evaluation and planting decision method in the first aspect described above.

[0014] Compared with the prior art, the present invention has the following advantages and technical effects: This invention provides an in-situ multi-parameter sensing method and system for soil suitability evaluation and planting decision-making. By integrating an in-situ soil sensing probe to collect multiple physicochemical indicators in real time, it solves the problems of lagging traditional detection methods and their inability to meet real-time control requirements. Addressing the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, a multi-parameter coupling compensation model is constructed to nonlinearly correct the original data, eliminating systematic errors introduced by environmental factors and improving the reliability of field measurement data. By using a pre-set crop growth database, the environmental requirements of crops are transformed into ideal environmental feature vectors and tolerance limit feature vectors. A sensitivity weight vector is introduced to calculate the weighted generalized distance between the measured vector and the ideal vector, and a suitability score is generated by combining this with a threshold truncation function. This achieves comprehensive consideration and differentiated evaluation of multi-dimensional indicators, overcoming the shortcomings of relying on subjective experience. Based on the suitability score, the crop set is recommended and ranked, and significant features are extracted from the vector differences. These features are input into an inference engine built on an agronomic knowledge graph to generate precise water and fertilizer control decisions, transforming dry multi-dimensional data into intuitive planting suggestions and lowering the entry barrier for precision agriculture. Attached Figure Description

[0015] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the structure of the in-situ multi-parameter sensing soil suitability evaluation and planting decision-making system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the in-situ multi-parameter sensing method for soil suitability evaluation and planting decision-making according to an embodiment of the present invention. Detailed Implementation

[0016] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0017] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0018] Example 1 like Figure 2 As shown, this embodiment provides an in-situ multi-parameter sensing method for soil suitability evaluation and planting decision-making, including: The integrated in-situ soil sensing probe was used to collect multiple soil physicochemical index data of the target detection point in real time, and a primary vector of measured soil characteristics was constructed. To address the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, nonlinear corrections were made to conductivity and nitrogen, phosphorus and potassium readings using real-time acquired temperature and moisture data. The corrected physical quantities were then mapped to a feature space of the same dimension to generate a standard soil feature vector. A pre-set crop growth database containing the optimal physicochemical environmental threshold ranges for various crops at different growth stages is provided. In response to the user's evaluation request, the set of crops to be evaluated is retrieved, and the ideal environment feature vector and tolerance limit feature vector corresponding to each crop are extracted. A sensitivity weight vector is introduced to calculate the weighted generalized distance between the standard soil feature vector and the ideal environment feature vector. At the same time, the risk penalty factor of the standard soil feature vector relative to the tolerance limit feature vector is calculated. Based on the weighted generalized distance and the risk penalty factor, a suitability evaluation function is constructed to quantify the current soil suitability for crops. The crop set is recommended and ranked based on suitability scores. For target crops, significant features are extracted from the vector differences and input into an inference engine built on rule chains and agronomic knowledge graphs to generate a precision planting decision scheme containing water and fertilizer regulation operation instructions.

[0019] Specifically, the implementation process of this embodiment includes: Step S1: Multidimensional Heterogeneous Data Perception and Vectorization Using an integrated in-situ soil sensing probe, real-time soil physical and electrochemical field data at the target detection point are acquired. The data includes at least: soil temperature (T), volumetric water content (H), electrical conductivity (EC), hydrogen ion concentration index (pH), available nitrogen (N), available phosphorus (P), available potassium (K), and comprehensive fertility index (F). After analog-to-digital conversion and denoising, the above-mentioned raw simulated signals are used to construct a primary vector of measured soil characteristics. ; in, This is the primary vector of measured soil characteristics; Let be the measured value of the i-th soil index. This represents the dimension of soil indicators.

[0020] Step S2, Environmental Coupling Compensation and Feature Space Mapping: To address the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, a multi-parameter coupled compensation model was constructed. The conductivity and nitrogen, phosphorus, and potassium readings were nonlinearly corrected using real-time acquired temperature (T) and moisture content (H) to obtain the corrected physical quantities. Subsequently, the Z-Score normalization method was used to map the data from each dimension to a feature space of the same dimension, generating a standard soil feature vector. .

[0021] Step S3: Construction and retrieval of the crop demand database: A pre-built crop growth database contains optimal physicochemical environmental threshold ranges for various crops at different growth stages; in response to user evaluation requests, the database retrieves the set of crops to be evaluated. Extract the ideal environmental feature vector for each crop. and tolerance limit eigenvector .

[0022] Step S4: Suitability measurement based on weighted generalized distance: Introducing a sensitivity weight vector Calculate the standard soil feature vector With the feature vector of the ideal environment Weighted generalized distance between Simultaneous calculation Compared to Risk penalty factor Based on distance With penalty factor Constructing a suitability evaluation function Quantify the current soil conditions for crops The degree of suitability.

[0023] Step S5: Knowledge Graph-Driven Decision Generation The crop set is recommended and ranked based on suitability scores; for the target crop, vector differences are extracted. The salient features are input into a reasoning engine built on rule chains and agronomic knowledge graphs to generate a precision planting decision scheme that includes water and fertilizer regulation operation instructions.

[0024] Furthermore, the process of nonlinearly correcting the conductivity and nitrogen, phosphorus and potassium readings includes: correcting the difference between the measured temperature and the standard laboratory temperature using a temperature compensation coefficient; correcting the ratio of the measured volumetric water content to the reference value of saturated water content using a humidity influence factor; and eliminating the influence of temperature on ion mobility and the dilution or concentration effect of water on electrolyte concentration through the product of the temperature and humidity terms.

[0025] Specifically, the implementation process of this embodiment includes: ; ; ; in, These are the corrected ion parameter values; These are temperature correction items and humidity correction items, respectively. This represents the sensor's raw reading; T represents the soil temperature. The standard laboratory temperature is 25℃. Here, H is the temperature compensation coefficient; H is the moisture content. This is a reference value for saturated moisture content. This is the humidity compensation coefficient; this formula is used to eliminate the effect of temperature on ion mobility and the dilution or concentration effect of moisture on electrolyte concentration.

[0026] Furthermore, the process of constructing the suitability evaluation function includes: using the sensitivity weight vector to perform a weighted summation of the differences between the standard soil feature vector and the ideal environment feature vector in each dimension; setting a threshold cutoff function, whereby the threshold cutoff function outputs a minimum penalty coefficient when a certain indicator exceeds the survival red line defined by the crop tolerance limit vector, otherwise the output is 1; and multiplying the weighted summation result with the output of the threshold cutoff function to obtain the current soil suitability score for the crop.

[0027] Specifically, the implementation process of this embodiment includes: Suitability evaluation function The specific calculation model is as follows: ; ; in, The weighted generalized distance corresponding to the j-th crop; Let be the sensitivity weight of the j-th crop to the k-th soil index; This represents the value of the k-th index in the standard soil feature vector; It is the ideal value of the kth index in the ideal environmental feature vector of the jth crop; This is a threshold truncation function; This represents the limit value of the k-th indicator in the environmental tolerance limit feature vector of the j-th crop. When an indicator exceeds the tolerance limit vector of the crop... When defining the survival red line, The output is a minimum penalty coefficient (e.g., 0.1), otherwise the output is 1; this achieves the mathematical expression of the "one-vote veto" limiting factor.

[0028] Furthermore, the process of obtaining the sensitivity weight vector includes: collecting multiple sets of historical planting case data, each set of data containing soil feature vectors and actual crop yield labels; constructing difference pairs for each crop, one set being high-yield samples and the other set being low-yield samples; constructing a loss function containing weight parameters, and minimizing the loss function using stochastic gradient descent, so that the score of high-yield samples is significantly higher than that of low-yield samples under the weighted evaluation system, and automatically learning the weights of key indicators affecting crop yield.

[0029] Specifically, the implementation process of this embodiment includes: Sensitivity weight vector The data is obtained using a gradient iterative method based on few-sample difference learning, including: Collect M sets of historical planting case data, each set containing soil feature vectors. Compared with actual crop yield / quality labels (High yield / Low yield); For each crop, construct differential pairs. ,in For high-yield samples, This is a low-yield sample; Construct a loss function L that includes weight parameters W: ; in, Here, W is the loss function; W is the sensitivity weight vector. Index for sample pairs; This represents summing over all sample pairs; It is a function for maximizing the value; For interval parameters; For high-yield samples, This is a low-yield sample; and Samples and Suitability score under weight vector W.

[0030] By using stochastic gradient descent (SGD) to minimize the loss function L, the score of high-yield samples is always significantly higher than that of low-yield samples under the weighted evaluation system, thereby automatically learning the weights of key indicators affecting crop yield.

[0031] Furthermore, the process of generating precision planting decision-making schemes includes: semantic mapping of vector differences to identify positive deviations, negative deviations, and the degree of deviation; retrieving entity nodes and relational edges that match the target crop and deviation features in the agronomic knowledge graph; and using pre-trained natural language generation templates to combine the matched control measures into structured text and output decision suggestions containing water and fertilizer control operation instructions.

[0032] Example 2 Based on the same general inventive concept, this invention also provides a crop soil suitability assessment and intelligent planting decision-making system based on in-situ multi-parameter sensing in arid regions. The following describes the crop soil suitability assessment and intelligent planting decision-making system based on in-situ multi-parameter sensing provided by this invention. The crop soil suitability assessment and intelligent planting decision-making system based on in-situ multi-parameter sensing described below can be referred to in conjunction with the in-situ multi-parameter sensing soil suitability assessment and planting decision-making method described above. The system includes: The in-situ sensing terminal is a mobile hardware device with a front-end plug-in multi-parameter composite probe, which integrates a stainless steel five-pin electrode array, a thermistor and a pH glass electrode. Internally, it has a signal conditioning circuit and a wireless communication module. The cloud computing platform is equipped with data processing services to receive raw data and execute environmental coupling compensation algorithms, and stores a crop growth database. The intelligent analysis engine runs in the cloud or at the edge and includes a weight training module and an inference engine; the interactive application client runs on the user's mobile terminal and is equipped with a visual human-computer interface.

[0033] Furthermore, the interactive application client's visual human-computer interface includes a multi-color block dashboard. The multi-color block dashboard presets the health threshold range of each indicator under the general agronomic standard. When real-time data is received, the background color of the color block is automatically rendered according to the range in which the value is located. The indicator name, real-time value and unit are displayed simultaneously in the color block. Clicking on any color block will jump to the historical trend curve page of that indicator.

[0034] Furthermore, the system also includes a closed-loop feedback optimization module, which receives the harvesting results input by the user in the interactive application client, constructs a new training sample pair with the early detection data and the later harvesting results, triggers the online incremental learning of the intelligent analysis engine, and fine-tunes and updates the sensitivity weight vector.

[0035] Example 3 In this embodiment, a computer terminal device is provided, including: processor; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described in-situ multi-parameter sensing soil suitability evaluation and planting decision method.

[0036] In this embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described in-situ multi-parameter sensing soil suitability evaluation and planting decision method.

[0037] like Figure 1 As shown, the in-situ sensing terminal includes a probe body and a handheld display.

[0038] Probe structure design: Five 316L medical-grade stainless steel probes are arranged in a straight line, each 10cm long. Stainless steel is chosen for its resistance to acid and alkali corrosion, and the five-needle structure is used to eliminate the influence of contact resistance.

[0039] Sensor measurement principle: Moisture measurement: Utilizing the FDR (Frequency Domain Reflectometry) principle. The probe acts as the capacitor plate, and the soil as the dielectric. The circuit generates a 100MHz high-frequency signal to measure changes in the soil's dielectric constant, thereby retrieving the volumetric water content. The high-frequency signal effectively reduces interference from soil salinity in moisture measurement.

[0040] Electrical conductivity (EC) and nutrient (N / P / K) estimation: Soil complex impedance is measured using an AC excitation signal (1kHz-10kHz, to prevent electrode polarization). Based on the Nernst equation and ion selectivity coefficient matrix, an empirical mapping model between electrical conductivity and available nitrogen, phosphorus, and potassium ions is established for estimation. While it cannot completely replace laboratory ion chromatography, it has extremely high relevance in guiding fertilization in the field.

[0041] Temperature measurement: A high-precision NTC thermistor is encapsulated in the metal tube wall inside the probe, closely attached to the metal wall to quickly respond to soil temperature, with a response time of <5 seconds.

[0042] pH measurement: The probe integrates a robust planar antimony electrode and a reference electrode in the center to detect the hydrogen ion potential difference.

[0043] Data acquisition process: The MCU (microcontroller unit, such as STM32 series) polls each sensor channel every 1 second to acquire analog voltage signals. After 12-bit ADC conversion and Kalman filtering for noise reduction, the signals are temporarily stored in a buffer and sent to the host computer via Bluetooth or LoRa module.

[0044] In actual field measurements, environmental factors have a significant impact on readings, and directly obtained values ​​are often unusable. This embodiment uses a compensation formula for correction.

[0045] Scenario setting: Early autumn morning, soil temperature is low (10℃) and relatively dry (moisture content 15%). At this time, ion activity is low and migration is slow, resulting in a falsely low conductivity reading.

[0046] Assume the original collected data is: conductivity .

[0047] If this data is used directly, the system may misjudge it as "nutrient deficiency". The system then invokes the compensation algorithm: Temperature correction: 25℃ is used as the standard temperature. Take the empirical value of 0.02.

[0048] ; in, This is a temperature correction term, indicating that at 10℃, the ion activity is only 70% of that at 25℃, and needs to be compensated back.

[0049] Humidity correction: set saturated moisture content =30%, Correction Factor .

[0050] ; in, This is a humidity correction term, indicating that in a semi-arid state, the conductive pathways of the solution decrease, the resistance increases, and a significant compensation is required.

[0051] Final correction calculation: ; After compensation, it was found that the actual soil salinity (2418) was actually sufficient or even high. Without this algorithm, users might mistakenly continue to fertilize, leading to seedling burn. This step is key to achieving "precision" in this embodiment.

[0052] Crop suitability projection and field application in typical areas of the Loess Plateau: 1. Test Environment and Background: A county in the heart of the Loess Plateau (hereinafter referred to as "XX County") was selected as a typical test area. The terrain in this area is complex, and it faces the dual challenges of soil erosion and drought. Traditional experience-based planting methods often fail to accurately match the best economic crops.

[0053] 2. Data Acquisition (Sensor End): Using the multi-parameter soil sensor in this embodiment, gridded multi-point sampling is performed on the target plot in XX County.

[0054] Detection indicators: The sensor acquired the soil pH value (shown as weakly alkaline), organic matter content, electrical conductivity (EC value), and soil moisture content of the plot in real time.

[0055] Special parameters: In view of the loose soil characteristics of this region, the sensor accurately obtains soil compaction and aeration data through a specific probe structure.

[0056] 3. Data processing and analysis (software): The measured data collected by the sensors are uploaded to the crop suitability analysis program via a wireless transmission module.

[0057] Multidimensional matching: The program compares the measured soil data with the built-in "crop growth model database".

[0058] Exclusion and Selection: The algorithm first excludes planting recommendations for water-intensive crops based on soil moisture content and local rainfall conditions; then, based on the characteristics of slightly alkaline soil and the advantages of diurnal temperature difference, it identifies crop categories that are light-loving, drought-resistant, and require nutrient accumulation.

[0059] 4. Output Results and Verification: The software ultimately outputs a list of recommended crop suitability, with the top-ranked crops being: bell peppers (chili peppers), tomatoes, eggplants, etc.

[0060] Significance: The results are highly consistent with the actual growth of high-yield economic crops in the local area (or where they are currently grown), verifying the reliability of this system in accurately matching soil and crops under complex geological conditions.

[0061] This invention provides an in-situ multi-parameter sensing method and system for soil suitability evaluation and planting decision-making. By integrating an in-situ soil sensing probe to collect multiple physicochemical indicators in real time, it solves the problems of lagging traditional detection methods and their inability to meet real-time control requirements. Addressing the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, a multi-parameter coupling compensation model is constructed to nonlinearly correct the original data, eliminating systematic errors introduced by environmental factors and improving the reliability of field measurement data. By using a pre-set crop growth database, the environmental requirements of crops are transformed into ideal environmental feature vectors and tolerance limit feature vectors. A sensitivity weight vector is introduced to calculate the weighted generalized distance between the measured vector and the ideal vector, and a suitability score is generated by combining this with a threshold truncation function. This achieves comprehensive consideration and differentiated evaluation of multi-dimensional indicators, overcoming the shortcomings of relying on subjective experience. Based on the suitability score, the crop set is recommended and ranked, and significant features are extracted from the vector differences. These features are input into an inference engine built on an agronomic knowledge graph to generate precise water and fertilizer control decisions, transforming dry multi-dimensional data into intuitive planting suggestions and lowering the entry barrier for precision agriculture.

[0062] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An in-situ multi-parameter sensing method for soil suitability evaluation and planting decision-making, characterized in that, Includes the following steps: The integrated in-situ soil sensing probe was used to collect multiple soil physicochemical index data of the target detection point in real time, and a primary vector of measured soil characteristics was constructed. To address the cross-sensitivity effect of temperature and humidity changes on electrochemical sensors in in-situ measurements, nonlinear corrections were made to conductivity and nitrogen, phosphorus and potassium readings using real-time acquired temperature and moisture data. The corrected physical quantities were then mapped to a feature space of the same dimension to generate a standard soil feature vector. A pre-set crop growth database containing the optimal physicochemical environmental threshold ranges for various crops at different growth stages is provided. In response to the user's evaluation request, the set of crops to be evaluated is retrieved, and the ideal environment feature vector and tolerance limit feature vector corresponding to each crop are extracted. A sensitivity weight vector is introduced to calculate the weighted generalized distance between the standard soil feature vector and the ideal environment feature vector. At the same time, the risk penalty factor of the standard soil feature vector relative to the tolerance limit feature vector is calculated. Based on the weighted generalized distance and the risk penalty factor, a suitability evaluation function is constructed to quantify the current soil suitability for crops. The crop set is recommended and ranked based on suitability scores. For target crops, significant features are extracted from the vector differences and input into an inference engine built on rule chains and agronomic knowledge graphs to generate a precision planting decision scheme containing water and fertilizer regulation operation instructions.

2. The method according to claim 1, characterized in that, The process of nonlinearly correcting the conductivity and nitrogen, phosphorus and potassium readings includes: correcting the difference between the measured temperature and the standard laboratory temperature using a temperature compensation coefficient; correcting the ratio of the measured volumetric water content to the reference value of saturated water content using a humidity influence factor; and eliminating the influence of temperature on ion mobility and the dilution or concentration effect of water on electrolyte concentration by multiplying the temperature and humidity terms.

3. The method according to claim 1, characterized in that, The process of constructing the suitability evaluation function includes: using the sensitivity weight vector to perform a weighted summation of the differences between the standard soil feature vector and the ideal environment feature vector in each dimension; setting a threshold cutoff function, which outputs a minimum penalty coefficient when a certain indicator exceeds the survival red line defined by the crop tolerance limit vector, otherwise outputting 1; and multiplying the weighted summation result with the output of the threshold cutoff function to obtain the current soil suitability score for the crop.

4. The method according to claim 1, characterized in that, The process of obtaining the sensitivity weight vector includes: collecting multiple sets of historical planting case data, each set of data containing soil feature vectors and actual crop yield labels; constructing difference pairs for each crop, one set being high-yield samples and the other set being low-yield samples; constructing a loss function containing weight parameters, and minimizing the loss function using stochastic gradient descent, so that the score of high-yield samples is significantly higher than that of low-yield samples under the weighted evaluation system, and automatically learning the weights of key indicators affecting crop yield.

5. The method according to claim 1, characterized in that, The process of generating precision planting decision-making schemes includes: semantic mapping of vector differences to identify positive deviations, negative deviations, and the degree of deviation; retrieving entity nodes and relational edges that match the target crop and deviation features in the agronomic knowledge graph; and combining the matched control measures into structured text using pre-trained natural language generation templates to output decision suggestions containing water and fertilizer control operation instructions.

6. A crop soil suitability evaluation and intelligent planting decision-making system based on in-situ multi-parameter sensing in arid areas, characterized in that, The system for implementing the method of any one of claims 1-5 comprises: The in-situ sensing terminal is a handheld or fixed hardware device. The front end is equipped with a direct-insertion multi-parameter composite probe, which integrates a stainless steel five-pin electrode array, a thermistor and a pH glass electrode. The internal components include a signal conditioning circuit and a wireless communication module. The cloud computing platform is equipped with data processing services to receive raw data and execute environmental coupling compensation algorithms, and stores a crop growth database. The intelligent analysis engine runs in the cloud or at the edge and includes a weight training module and an inference engine; the interactive application client runs on the user's mobile terminal and is equipped with a visual human-computer interface.

7. The system according to claim 6, characterized in that, The interactive application client's visual human-computer interface includes a multi-color block dashboard. The multi-color block dashboard presets the health threshold range of each indicator under the general agronomic standard. When real-time data is received, the background color of the color block is automatically rendered according to the range in which the value is located. The indicator name, real-time value and unit are displayed in the color block at the same time. Clicking on any color block will jump to the historical trend curve page of that indicator.

8. The system according to claim 6, characterized in that, The system also includes a closed-loop feedback optimization module, which receives the harvesting results input by the user in the interactive application client, constructs a new training sample pair with the early detection data and the later harvesting results, triggers the online incremental learning of the intelligent analysis engine, and fine-tunes and updates the sensitivity weight vector.

9. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-5.

Images