A method for rapid detection and fertilization of dryland soil fertility

Abstract

The present invention relates to the technical field of analyzing the physical and chemical properties of soil, and in particular to a method for rapid detection and fertilization of dryland soil fertility. The method uses a physical field collaborative separation technology of gradient centrifugation and dielectrophoresis enrichment to separate the colloidal phase and dissolved phase of humic acid complex based on density differences and dielectric property differences. A genetic algorithm is used to optimize the dual-channel detection timing, and a non-overlapping timing signal stream is generated with the signal interference suppression ratio as the fitness function to suppress time-domain cross-interference. A programmable resistance-capacitance compensation model is constructed, and the parameters of the parallel capacitor array are dynamically adjusted to correct the electrochemical phase angle offset. The three-dimensional fertility distribution surface and residual map are generated in combination with the Kriging interpolation model. Based on the spatial distribution characteristics of the residual map, a cross-linked slow-release system is deployed to improve the organic matter binding capacity in the humic acid-deficient area, a mesoporous zeolite-based passivator is applied to regulate the ion migration path, and a closed-loop feedback mechanism is formed through dynamic phase response monitoring to achieve accurate soil fertility detection and directional remediation.

Description

Technical Field

[0001] The present invention relates to the technical field of analyzing the physical and chemical properties of soil, and in particular to a method for quickly detecting and improving the fertility of dryland soil. Background Art

[0002] Existing rapid detection and fertilization technology for dryland soil fertility is based on physical and chemical analysis methods. Through spectral analysis technology combined with electrochemical sensors, multi-parameter simultaneous measurement of soil organic matter, nitrogen, phosphorus, potassium content and pH value is carried out to establish a correlation model between soil nutrient deficiency and microbial community structure. In the analysis of the test results, bioinformatics algorithms are used to identify limiting fertility factors, and then a targeted fertilization plan is designed: humic acid improvers are applied to areas with organic matter depletion, and molecular membrane slow-release technology is used to extend the fertilizer effect; mineral-based passivators are applied to soils with ion imbalance to adjust the cation exchange capacity, and at the same time, the soil microecology is reconstructed through inoculation of composite bacterial agents. This technical system realizes in-situ data collection and dynamic evaluation through Internet of Things nodes, and optimizes the fertilization cycle in combination with the soil carbon and nitrogen cycle kinetics model, forming a detection-diagnosis-regulation closed-loop mechanism, which significantly improves the sustainable productivity of soil in arid areas.

[0003] The core pain point of rapid soil fertility detection and fertilization technology for dryland farming lies in the signal cross-interference between spectral analysis technology and electrochemical sensors during the simultaneous multi-parameter measurement process. This is manifested in the coupling effect between the characteristic spectral absorption peaks of organic matter and the electrochemical response signals of metal ions in the time and frequency domains, leading to systematic deviations in the inversion model of nitrogen, phosphorus, and potassium content. This interference originates from the heterogeneous adsorption of humic acid and mineral complexes in the soil heterogeneous medium on the multi-sensor probes. The resulting parasitic capacitance effect further distorts the phase angle of the electrochemical impedance spectrum, reducing the spatial consistency between the in-situ detection data and the laboratory calibration values, ultimately affecting the accurate identification of limiting fertility factors and the dynamic optimization efficiency of targeted fertilization strategies. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a method for rapid detection and fertilization of dryland soil fertility. The present invention solves the problem of phase angle distortion of electrochemical impedance spectroscopy caused by cross-interference between spectrum and electrochemical signals and parasitic capacitance effect caused by heterogeneous adsorption of humic acid and mineral complexes in soil fertility determination.

[0005] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0006] The present invention provides a method for rapid detection and fertilization of dryland soil fertility, comprising:

[0007] Step S1: Perform multi-physical field synergistic separation on the soil heterogeneous medium, and obtain the colloidal phase and dissolved phase of the humic acid complex through the sorting action of gradient centrifugation and dielectrophoresis enrichment. The colloidal phase particle size >100 nm is used for spectral detection, and the dissolved phase particle size <100 nm is used for electrochemical detection;

[0008] Step S2: Using a preset dual-channel detection probe, optimizing the preset dual-channel detection timing based on a genetic algorithm, taking the signal interference suppression ratio as the objective function, and generating a non-overlapping timing scanning sequence of the spectrum and electrochemical signal;

[0009] Step S3: Perform multimodal signal decomposition on the spectral signal to extract the humic acid characteristic spectrum, and simultaneously perform neural network transfer learning on the electrochemical signal to separate the organic functional group vibration spectrum and the metal ion migration effect through reverse gradient operation, and output the decoupled spectral absorption coefficient and ion concentration vector;

[0010] Step S4: constructing a programmable resistance-capacitance compensation model, taking the impedance phase angle offset as the optimization target, and adjusting the capacitance compensation parameters through a genetic algorithm to correct the conductivity phase distortion to obtain the corrected conductivity;

[0011] Step S5: inputting the decoupled spectral absorption coefficient and the corrected conductivity into the spatial interpolation model, and simultaneously optimizing the variance function parameters to generate a fertility distribution surface and a residual map including a humic acid content gradient and a metal ion concentration gradient;

[0012] Step S6: Based on the residual map of step S5, an adsorption compensation model is constructed, and a surface-modified cross-linked sustained-release system is deployed in areas where the humic acid content gradient is lower than the preset threshold to inhibit heterogeneous adsorption. In areas where the metal ion concentration gradient exceeds the standard range, a microbial-loaded mesoporous structure passivator is applied to regulate the ion mass transfer path, and fertility compensation data is generated through dynamic phase response monitoring.

[0013] Furthermore, in the method for rapid detection and fertilization of dryland soil fertility of the present invention, in step S1, the multi-physical field collaborative separation includes:

[0014] The soil heterogeneous medium was initially separated by gradient centrifugation, and the colloidal phase and dissolved phase of the humic acid complex were obtained under a preset centrifugal force threshold.

[0015] The colloidal phase and dissolved phase separated in the step are respectively input into the dielectrophoresis enrichment device, and the colloidal particles are caused to undergo dielectrophoretic migration by applying an alternating electric field, and a spatially sorted pretreated sample is formed according to the difference in migration rate, wherein the ratio of the migration rate of the colloidal phase to the migration rate of the dissolved phase is greater than 2:1.

[0016] Furthermore, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes step S2:

[0017] The spectral detection channel of the dual-channel detection probe is loaded with the colloidal phase, and the electrochemical detection channel is loaded with the dissolved phase;

[0018] The genetic algorithm optimizes the timing allocation logic to generate non-overlapping timing signal streams by dynamically allocating spectral pulse widths and electrochemical scanning intervals.

[0019] Furthermore, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes step S3:

[0020] The signal collected by the spectral detection channel in step S2 is input into the multimodal signal decomposition unit, and the humic acid characteristic spectrum is extracted after removing high-frequency noise through wavelet packet decomposition;

[0021] The signal collected by the electrochemical detection channel in step S2 is input into the transfer learning network, and the carboxyl vibration peak and the iron ion diffusion peak in the humic acid characteristic spectrum are separated by reverse gradient operation to generate a decoupled spectral absorption coefficient and an ion concentration vector.

[0022] Furthermore, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes step S4:

[0023] Inputting the ion concentration vector generated in step S3 into a programmable resistance-capacitance compensation model to establish a correlation mapping between the metal ion migration effect and the phase angle offset;

[0024] Taking the impedance phase angle offset as the optimization target, the compensation parameters of the parallel capacitor array are iteratively adjusted through the genetic algorithm so that the attenuation of the parasitic capacitance effect in the corrected conductivity data is ≥30dB.

[0025] Furthermore, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes step S5:

[0026] The decoupled spectral absorption coefficient and the conductivity data corrected by resistance-capacitance compensation were input into the spatial interpolation model. Based on the spatial correlation between the spectral absorption characteristics of the functional groups of humic acid molecules and the migration paths of metal ions, the base value and range parameters in the variation function were optimized to generate a three-dimensional residual map including the gradient distribution of humic acid content and the gradient of metal ion concentration.

[0027] Furthermore, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes step S6:

[0028] Based on the humic acid content gradient distribution in the three-dimensional residual map, a cross-linked sustained-release system modified with polyacrylamide was deployed in areas where the gradient value was below the preset threshold, inhibiting heterogeneous adsorption at the mineral interface through carboxyl coordination.

[0029] Based on the spatial distribution characteristics of the metal ion concentration gradient, a mesoporous zeolite-based passivator loaded with Bacillus subtilis was applied to the concentration exceeding the standard area, and its pore size selectivity was used to regulate the migration path of the target ions.

[0030] By real-time monitoring of the phase angle change rate of the electrochemical impedance spectroscopy, the degree of elimination of the residual effect of parasitic capacitance is verified and fertility compensation data is generated.

[0031] Beneficial effects of the present invention:

[0032] The present invention uses the physical field collaborative separation technology of gradient centrifugation and dielectrophoresis enrichment to sort the colloidal phase and dissolved phase of humic acid complex based on density differences and dielectric property differences, effectively reducing the heterogeneous adsorption of heterogeneous media on sensors; combines the non-overlapping time series signal stream distribution strategy optimized by genetic algorithm to suppress the cross-interference of spectral and electrochemical signals in the time domain; utilizes wavelet packet decomposition and neural network gradient inversion operation to decouple the vibration characteristics of humic acid functional groups and metal ion migration effects, eliminating frequency domain signal coupling; dynamically corrects the phase angle offset caused by parasitic capacitance through a programmable resistance-capacitance compensation model, and improves the spatial consistency of conductivity data; based on the spatial interpolation and closed-loop feedback mechanism of the three-dimensional residual map, accurately deploys the cross-linked slow-release system and mesoporous passivator to achieve directional repair and dynamic regulation of soil fertility parameters, significantly improving the detection accuracy and fertilization efficiency of dryland soil fertility. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0034] Figure 1 The present invention provides a flowchart of a method for rapid detection and fertilization of dryland soil fertility. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described in detail below.

[0036] See also Figure 1 The present invention provides a method for rapid detection and fertilization of dryland soil fertility, comprising:

[0037] Step S1: Perform multi-physical field synergistic separation on the soil heterogeneous medium, and obtain the colloidal phase and dissolved phase of the humic acid complex through the sorting action of gradient centrifugation and dielectrophoresis enrichment. The colloidal phase particle size >100 nm is used for spectral detection, and the dissolved phase particle size <100 nm is used for electrochemical detection;

[0038] During the primary separation of heterogeneous soil media using gradient centrifugation, a centrifugal force threshold is set based on the density difference between the humic acid complex and the dissolved phase. By controlling the centrifugal speed and time, the colloidal phase settles to the bottom of the centrifuge tube, while the dissolved phase remains suspended in the upper liquid layer. The centrifugal force threshold is set based on the Stokes sedimentation coefficient range of the target components, ensuring that the complex colloids formed by humic acid and mineral particles preferentially settle in the centrifugal field, while free metal ions remain in the dissolved phase. The separated colloidal phase is transferred via a siphon to the inlet tank of a dielectrophoresis enrichment device. The dissolved phase is then filtered through a membrane to remove residual particulate matter before being introduced into an electrophoresis cell.

[0039] A non-uniform alternating electric field is applied to the dielectrophoresis enrichment device. By adjusting the frequency and intensity of the electric field, the colloidal particles are induced to produce a positive dielectrophoretic effect. Humic acid complexes, with a higher dielectric constant than the surrounding medium, migrate toward regions of high field strength. Ions in the dissolved phase diffuse along the electric field gradient due to electroosmosis. The difference between the migration rates of the colloidal and dissolved phases is calibrated using particle image velocimetry, forming spatially sorted enrichment zones. The duty cycle of the high-frequency alternating electric field is controlled in the range of 30%-50%, balancing the dielectrophoretic migration efficiency with the Joule heating effect and avoiding thermal denaturation of the colloidal phase. The enriched colloidal phase is deposited on the surface of the collecting electrode, and the dissolved phase is discharged through a microfluidic channel, completing the physical field-coordinated sorting of the pretreated sample.

[0040] This step, through the synergistic effect of gradient centrifugation and dielectrophoresis enrichment, separates a colloidal phase with particle sizes greater than 100 nanometers and a dissolved phase with particle sizes less than 100 nanometers. The surface functional group spectral characteristics of the humic acid and mineral complexes in the colloidal phase are distinct, making them suitable for near-infrared spectroscopy analysis. The high concentration of free metal ions in the dissolved phase is well-suited for voltammetric detection using electrochemical sensors. Physical field separation technology reduces heterogeneous adsorption of heterogeneous media onto multi-sensor probes, paving the way for the independent acquisition of subsequent spectral and electrochemical signals.

[0041] Step S2: Using a preset dual-channel detection probe, optimizing the preset dual-channel detection timing based on a genetic algorithm, taking the signal interference suppression ratio as the objective function, and generating a non-overlapping timing scanning sequence of the spectrum and electrochemical signal;

[0042] In step S2, the dual-channel detection probe's spectral detection channel is equipped with a broadband light source and a spectroscopic grating to receive the reflected spectral signal from the colloidal sample. The electrochemical detection channel integrates a three-electrode system, with the working electrode surface modified with an ion-selective membrane to detect the voltammetric response of metal ions in the dissolved phase. The two channels are spatially isolated to prevent physical coupling between the spectral excitation optical path and the electrochemical cell.

[0043] When optimizing timing allocation using a genetic algorithm, a randomized parameter combination, including spectral pulse width and electrochemical scan interval, is initialized as the initial population. The fitness function is based on the signal-to-interference suppression ratio, quantifying the timing configuration's interference resistance by calculating the ratio of the power spectral density of the electrochemical background noise during the spectral excitation period to the effective signal bandwidth. A roulette wheel selection strategy selects individuals with high fitness, a single-point crossover operator restructures the timing parameter sequence, and an adaptive mutation operator fine-tunes the pulse width and scan interval.

[0044] During the iterative optimization process, a non-dominated sorting strategy selects timing configurations in which the spectral pulse emission period and the electrochemical scan interval are completely staggered. The hardware synchronization trigger module receives the optimized timing parameters and controls the pulse triggering of the spectral light source and the start and stop timing of the electrochemical scan. The spectral detection channel activates the light source within the allocated pulse width. The spectroscopic grating decomposes the reflected light from the colloidal phase into a characteristic spectrum, and the detector array synchronously collects absorbance data. The electrochemical detection channel applies a linear sweep voltage within the set scan interval, and the current-voltage response curve of the dissolved phase is recorded using a potentiostat. The timing allocation strategy reduces signal coupling between sensors through time-domain isolation, forming complementary detection windows and achieving data independence for simultaneous multi-parameter measurements.

[0045] Step S3: Perform multimodal signal decomposition on the spectral signal to extract the humic acid characteristic spectrum, and simultaneously perform neural network transfer learning on the electrochemical signal to separate the organic functional group vibration spectrum and the metal ion migration effect through reverse gradient operation, and output the decoupled spectral absorption coefficient and ion concentration vector;

[0046] In step S3, the raw signal collected by the spectral detection channel is input into the multimodal signal decomposition unit, and the spectral data is subjected to multi-scale analysis using wavelet packet decomposition. The Daubechies wavelet basis function is selected to decompose the spectral signal. The high-frequency subband corresponds to the instrument noise and environmental interference components. By setting an energy threshold, high-frequency noise is filtered out and retained. The low-frequency effective signal is retained. The denoised spectral data is input into the empirical mode decomposition module, which generates a set of intrinsic mode functions based on the local extreme value characteristics of the signal. The modal components that match the vibration frequency of the carboxyl and hydroxyl functional groups of humic acid are selected to extract the intrinsic spectral characteristics that characterize the molecular structure of humic acid.

[0047] The voltammetric response signal from the electrochemical detection channel is fed into a pretrained two-channel convolutional neural network. The network architecture incorporates a residual connection module to enhance deep feature representation. During transfer learning, a gradient reversal layer acts on the shallow feature extraction stage. During backpropagation, a negative weight coefficient is applied to the gradient of humic acid-related features to suppress the interference of carboxyl vibration peaks on the resolution of iron ion diffusion peaks. The decoupled electrochemical signal is then passed through a convolutional layer to extract joint time-frequency domain features. A pooling layer compresses redundant information and outputs a metal ion concentration vector.

[0048] The output paths for the spectral absorption coefficient and ion concentration vector are independently configured, forming a decoupled multi-parameter detection dataset. The humic acid characteristic spectrum generates a standardized spectral absorption coefficient by quantifying the absorbance at characteristic wavelengths. The metal ion concentration vector is inversely calculated based on the calibration relationship between the integrated area of ​​the voltammetric curve and the standard solution. This step, through the coordinated processing of signal decomposition and feature decoupling, eliminates cross-interference between spectral and electrochemical signals, providing high-precision input parameters for subsequent spatial interpolation.

[0049] Step S4: constructing a programmable resistance-capacitance compensation model, taking the impedance phase angle offset as the optimization target, and adjusting the capacitance compensation parameters through a genetic algorithm to correct the conductivity phase distortion to obtain the corrected conductivity;

[0050] In step S4, the programmable RC compensation model receives the ion concentration vector output from step S3 and establishes a mapping relationship by analyzing the correlation between the metal ion migration effect and the phase angle offset of the electrochemical impedance spectroscopy. The time-frequency domain characteristic parameters in the ion concentration vector and the phase angle offset are correlated using a multivariate linear regression model to quantify the impact of parasitic capacitance caused by heterogeneous medium adsorption. The compensation model has a built-in parallel capacitor array, and the capacitance value of each branch is adjusted by a digital potentiometer to form a programmable capacitive reactance compensation network.

[0051] During the multi-objective optimization process of the genetic algorithm, the initial population consists of randomly generated combinations of capacitor compensation parameters. The phase angle offset fitness function is determined by calculating the absolute phase difference between the impedance spectra before and after correction at characteristic frequencies. The Nyquist plot integrity fitness function is based on a comprehensive score of the semicircular arc goodness of fit and high-frequency linearity. The algorithm uses a non-dominated sorting strategy to select the Pareto optimal solution set, combines crowding distance to maintain population diversity, and iteratively updates the compensation value combinations for the parallel capacitor array.

[0052] The optimized capacitance parameters are converted into analog voltage signals by a digital-to-analog conversion module, driving digital potentiometers to adjust the capacitive reactance values ​​of each branch. In the electrochemical sensor impedance spectrum after dynamic compensation, the phase angle offset caused by parasitic capacitance is significantly reduced, and the capacitive reactance arc in the high-frequency region of the Nyquist plot has restored its standard semicircular shape. The corrected conductivity data is then inverted to calculate the corresponding relationship between metal ion concentration and migration rate, eliminating measurement deviations caused by heterogeneous medium adsorption and providing high-fidelity input parameters for subsequent spatial interpolation.

[0053] Step S5: inputting the decoupled spectral absorption coefficient and the corrected conductivity into the spatial interpolation model, and simultaneously optimizing the variance function parameters to generate a fertility distribution surface and a residual map including a humic acid content gradient and a metal ion concentration gradient;

[0054] In step S5, the decoupled spectral absorption coefficient and the conductivity data corrected by resistance-capacitance compensation are input into the spatial interpolation model. The spectral absorption coefficient characterizes the vibration characteristics of the carboxyl and hydroxyl functional groups in the humic acid molecule, and the absorbance value at the characteristic wavelength is used to quantify the gradient distribution of the humic acid content. The corrected conductivity data reflects the migration path and concentration differences of metal ions in the soil medium. The spatial correlation parameter is established by combining the ion mobility and conductivity correlation model. The spatial interpolation algorithm constructs a semivariogram function based on the Kriging method to characterize the spatial autocorrelation characteristics of the humic acid content and ion concentration.

[0055] The sill value in the variogram parameter represents the extreme value of spatial variation in the data, while the range parameter characterizes the extent of spatial autocorrelation. A multi-objective genetic algorithm initializes a random parameter combination containing the sill value and the range, and iteratively optimizes the parameters through crossover and mutation operations. The fitness function comprehensively evaluates the goodness-of-fit of the root mean square error (RMSE) and semivariogram function between the interpolated predicted values ​​and the measured points, selecting parameter combinations that ensure that the spatial distribution model simultaneously maintains global smoothness and local detail. The optimized variogram parameters drive kriging interpolation to generate a three-dimensional fertility distribution surface, quantifying the spatial heterogeneity of nitrogen, phosphorus, and potassium nutrients.

[0056] The residual map is generated by calculating the standardized difference between the interpolated surface prediction and the measured value. Low-value areas of humic acid gradient are determined by the residual threshold of the spectral absorption coefficient, and areas of abnormal metal ion concentration gradient are identified based on the residual dispersion of the conductivity. The three-dimensional residual map uses thermal layer rendering technology to map the spatial coordinates of humic acid-deficient areas and ion imbalance areas to the geographic information system, forming a visual positioning benchmark. This step establishes a quantitative association between fertility distribution characteristics and soil remediation strategies through multi-source data fusion and spatial modeling, providing a precise spatial reference for subsequent targeted fertilization.

[0057] Step S6: Based on the residual map of step S5, an adsorption compensation model is constructed, and a surface-modified cross-linked sustained-release system is deployed in areas where the humic acid content gradient is lower than the preset threshold to inhibit heterogeneous adsorption. In areas where the metal ion concentration gradient exceeds the standard range, a microbial-loaded mesoporous structure passivator is applied to regulate the ion mass transfer path, and fertility compensation data is generated through dynamic phase response monitoring.

[0058] In step S6, based on the gradient distribution of humic acid content in the three-dimensional residual map, a cross-linked slow-release system modified with polyacrylamide is deployed in areas where the gradient value is lower than a preset threshold. Polyacrylamide is synthesized by free radical polymerization, and the carboxyl functional group is modified at the end of the molecular chain to form a coordination bond with the hydroxyl group on the mineral surface in the humic acid-deficient area, thereby inhibiting heterogeneous adsorption at the mineral interface. The degree of cross-linking is controlled by adjusting the molar ratio of glutaraldehyde to polyvinyl alcohol to form a three-dimensional network with a pore gradient structure, which encapsulates the humic acid molecules to form slow-release particles. The slow-release particles are embedded in the soil surface by mechanical spreading, and the swelling-diffusion mechanism triggered by soil moisture is used to gradually release humic acid, thereby increasing the organic matter binding capacity in the deficient area.

[0059] The spatial distribution characteristics of metal ion concentration gradients were analyzed using a Kriging interpolation model. A mesoporous zeolite-based passivator loaded with Bacillus subtilis was applied to areas where concentrations exceeded the standard. After acid activation and pore expansion, the mesoporous zeolite was loaded with a Bacillus subtilis spore suspension via an impregnation method. Iron carriers produced by bacterial metabolism compete with organic acids for adsorption, reducing the activity of heavy metal ions. The zeolite's mesoporous structure has a pore size distribution ranging from 2 to 50 nanometers. This pore-size sieving effect selectively blocks the migration of sodium and aluminum ions while promoting the free diffusion of potassium and calcium ions. The passivator was evenly applied in powder form to the areas of ion imbalance and, combined with irrigation water osmosis, regulated the migration paths of the target ions.

[0060] The dynamic phase response monitoring module collects phase angle data from the electrochemical impedance spectroscopy in real time and evaluates the degree of elimination of the residual effect of parasitic capacitance by calculating the rate of change of the phase angle before and after correction. When the rate of change of the phase angle is lower than the set threshold, the compensation is deemed effective, triggering the generation of fertility compensation data. If the rate of change exceeds the threshold, feedback is fed back to the resistance-capacitance compensation model to iteratively optimize the parameters of the parallel capacitor array. The spatial distribution of the monitoring data and the residual map is superimposed and analyzed to form a dynamic adjustment basis for the humic acid release rate and the intensity of passivation agent application, establishing a closed-loop control system for detection, repair, and verification. This step achieves precise spatial regulation of soil fertility parameters through targeted material deployment and dynamic monitoring feedback.

[0061] In step S1, multi-physical field collaborative separation is performed on the soil heterogeneous medium, including sorting operations of gradient centrifugation and dielectrophoresis enrichment. Gradient centrifugation is based on the density difference of different components, and separates the colloidal phase and the dissolved phase of the humic acid complex under preset centrifugal force conditions. The colloidal phase is composed of particles with a particle size greater than 100 nanometers, and the dissolved phase includes ions and soluble components with a particle size less than 100 nanometers. The separated colloidal phase and dissolved phase are loaded into the dielectrophoresis enrichment device respectively, and the dielectrophoretic migration behavior of the colloidal particles is adjusted by a high-frequency alternating electric field, so that the colloidal phase is aggregated in the high field strength area in the non-uniform electric field, and the dissolved phase is separated in the low field strength area due to the electroosmotic flow, forming a pre-treated sample for spatial dimension sorting. This process reduces the interference of heterogeneous media on subsequent detection through the action of physical fields, and provides adapted samples for spectral and electrochemical detection.

[0062] In step S2, a genetic algorithm is used to optimize the timing allocation strategy for dual-channel detection, generating a non-overlapping timing scan sequence with the signal-interference suppression ratio as the optimization objective. The timing parameters of the spectral and electrochemical detection channels are dynamically adjusted to achieve staggered acquisition. The signal-interference suppression ratio is quantified by calculating the ratio of the power spectral density of the electrochemical background noise to the effective signal bandwidth during the spectral excitation period. After the genetic algorithm initializes the timing parameter population, it iteratively searches for the optimal timing combination through roulette wheel selection, single-point crossover, and adaptive mutation operators to ensure complete temporal isolation of the spectral signal and electrochemical response, thereby suppressing cross-coupling effects between sensors.

[0063] In step S3, the data collected by the spectral detection channel is processed jointly using wavelet packet decomposition and empirical mode analysis. Wavelet packet decomposition extracts and filters high-frequency noise components, while empirical mode decomposition generates a set of humic acid intrinsic mode functions based on the local extremum characteristics of the signal, screening for modal components that match the vibrational characteristics of humic acid functional groups. The electrochemical signal is input into a pre-trained dual-channel convolutional neural network, which uses a residual connection structure to enhance feature expression. A gradient reversal layer applies negative gradient weights to humic acid-related features, separating the organic functional group vibration spectrum from the metal ion migration effect, and outputting the decoupled spectral absorption coefficient and ion concentration vector.

[0064] In step S4, a programmable resistance-capacitance compensation model is constructed to perform phase correction on the electrochemical detection signal. After the ion concentration characteristic vector is input into the compensation network, the compensation value of the parallel capacitor array is dynamically adjusted through a genetic algorithm multi-objective optimization. The phase angle offset is calculated by the phase difference between the characteristic frequency points of the impedance spectrum before and after correction. The integrity of the Nyquist spectrum is comprehensively evaluated based on the semicircular arc fit and high-frequency linearity. The optimized capacitance parameters are driven by a digital potentiometer through the digital-to-analog conversion module to adjust the equivalent capacitive reactance, suppressing the phase angle distortion caused by parasitic capacitance and improving the spatial consistency of the conductivity data.

[0065] In step S5, the decoupled spectral absorption coefficient and corrected conductivity are input into a spatial interpolation model to generate a three-dimensional fertility distribution surface. The Kriging interpolation algorithm constructs a spatial autocorrelation matrix based on the semivariogram function, integrating the spatial distribution characteristics of spectral and electrochemical data. The sill value and range in the variogram parameters are simultaneously optimized using a multi-objective genetic operator, ensuring that the interpolation results meet both global smoothness and local detail preservation requirements. A residual map is generated by calculating the standardized difference between the interpolated predicted value and the measured value, identifying areas with low humic acid content gradients and areas with abnormal metal ion concentration gradients.

[0066] In step S6, the Pareto front analysis method is used to determine the ratio of humic acid modifier and passivator based on the spatial distribution characteristics of the residual map. A surface-modified cross-linked sustained-release system is deployed in the humic acid-deficient area, and a pore gradient structure is formed by regulating the cross-linking degree of the polymer to control the release rate of humic acid. A mesoporous zeolite-based passivator loaded with microorganisms is applied to the ion imbalance area, and the pore size selectivity of the mesoporous structure is used to regulate the migration path of the target ions. The metabolites of the bacterial agent reduce the bioavailability of excess metal ions through competitive adsorption. The dynamic phase response monitoring module collects the phase angle change rate of the electrochemical impedance spectrum in real time to verify the degree of elimination of the residual effect of the parasitic capacitance, forming a closed-loop feedback mechanism for the fertility restoration effect.

[0067] Specifically, in the method for rapid detection and fertilization of dryland soil fertility of the present invention, in step S1, the multi-physical field collaborative separation includes:

[0068] The soil heterogeneous medium was initially separated by gradient centrifugation, and the colloidal phase and dissolved phase of the humic acid complex were obtained under a preset centrifugal force threshold.

[0069] The colloidal phase and dissolved phase separated in the step are respectively input into the dielectrophoresis enrichment device, and the colloidal particles are caused to undergo dielectrophoretic migration by applying an alternating electric field, and a spatially sorted pretreated sample is formed according to the difference in migration rate, wherein the ratio of the migration rate of the colloidal phase to the migration rate of the dissolved phase is greater than 2:1.

[0070] During the primary separation of heterogeneous soil media using gradient centrifugation, a centrifugal force threshold is set based on the density difference between the humic acid complex and the dissolved phase. By controlling the centrifugal speed and time, the colloidal phase settles to the bottom of the centrifuge tube, while the dissolved phase remains suspended in the upper liquid layer. The centrifugal force threshold is set based on the Stokes sedimentation coefficient range of the target components, ensuring that the complex colloids formed by humic acid and mineral particles preferentially settle in the centrifugal field, while free metal ions remain in the dissolved phase. The separated colloidal phase is transferred via a siphon to the inlet tank of a dielectrophoresis enrichment device. The dissolved phase is then filtered through a membrane to remove residual particulate matter before being introduced into an electrophoresis cell.

[0071] A non-uniform alternating electric field is applied to the dielectrophoresis enrichment device. By adjusting the frequency and intensity of the electric field, the colloidal particles are induced to produce a positive dielectrophoretic effect. Humic acid complexes, with a higher dielectric constant than the surrounding medium, migrate toward regions of high field strength. Ions in the dissolved phase diffuse along the electric field gradient due to electroosmosis. The difference between the migration rates of the colloidal and dissolved phases is calibrated using particle image velocimetry, forming spatially sorted enrichment zones. The duty cycle of the high-frequency alternating electric field is controlled in the range of 30%-50%, balancing the dielectrophoretic migration efficiency with the Joule heating effect and avoiding thermal denaturation of the colloidal phase. The enriched colloidal phase is deposited on the surface of the collecting electrode, and the dissolved phase is discharged through a microfluidic channel, completing the physical field-coordinated sorting of the pretreated sample.

[0072] Specifically, the method for rapid detection and fertilization of dryland soil fertility of the present invention comprises the following steps:

[0073] The spectral detection channel of the dual-channel detection probe is loaded with the colloidal phase, and the electrochemical detection channel is loaded with the dissolved phase;

[0074] The genetic algorithm optimizes the timing allocation logic to generate non-overlapping timing signal streams by dynamically allocating spectral pulse widths and electrochemical scanning intervals.

[0075] In step S2, the spectral detection channel of the dual-channel detection probe receives the colloidal phase sample sorted in step S1. This channel, equipped with a broadband light source, a spectroscopic grating, and a photodetector array, collects spectral data of the colloidal phase via a reflective optical path. The electrochemical detection channel loads the dissolved phase sample and uses a three-electrode system with an ion-selective membrane modified on the working electrode surface to record the voltammetric response curve of the target ion under constant potential control. The physical layout of the two channels adopts a parallel and independent structure to avoid physical interference between the spectral excitation optical path and the electrochemical cell.

[0076] When optimizing the timing allocation logic using a genetic algorithm, the population is initialized with random combinations of spectral pulse widths and electrochemical scan intervals. The fitness function is based on the signal-to-interference suppression ratio, calculated by calculating the ratio of the power spectral density of the electrochemical background noise to the effective signal bandwidth during the spectral excitation period. A roulette wheel selection strategy selects individuals with high fitness, a single-point crossover operator restructures the timing parameters, and an adaptive mutation operator fine-tunes the pulse widths and intervals. During the iterative optimization process, a non-dominated sorting strategy selects timing configurations in which the spectral pulse emission period and the electrochemical scan interval are completely offset, generating a non-overlapping timing signal stream.

[0077] The hardware synchronization trigger module receives optimized timing parameters to control the pulse triggering of the spectral light source and the start and stop timing of the electrochemical scan. The spectral detection channel activates the light source within the allocated pulse width. The spectroscopic grating decomposes the colloidal phase reflected light into a characteristic spectrum, and the detector array synchronously collects absorbance data. The electrochemical detection channel applies a linear sweep voltage within the set scan interval and records the current-voltage response curve of the dissolved phase using a potentiostat. The timing allocation strategy reduces signal coupling between sensors through time-domain isolation. The independent acquisition cycles of spectral and electrochemical data form complementary detection windows, improving the accuracy of simultaneous multi-parameter measurements.

[0078] Specifically, the method for rapid detection and fertilization of dryland soil fertility of the present invention comprises the following steps in step S3:

[0079] The signal collected by the spectral detection channel in step S2 is input into the multimodal signal decomposition unit, and the humic acid characteristic spectrum is extracted after removing high-frequency noise through wavelet packet decomposition;

[0080] The signal collected by the electrochemical detection channel in step S2 is input into the transfer learning network, and the carboxyl vibration peak and the iron ion diffusion peak in the humic acid characteristic spectrum are separated by reverse gradient operation to generate a decoupled spectral absorption coefficient and an ion concentration vector.

[0081] In step S3, the signal collected by the spectral detection channel is input into the multimodal signal decomposition unit, and the original spectrum is subjected to multi-scale decomposition using the selected wavelet basis function. During the decomposition process, the high-frequency subband corresponds to the instrument noise and environmental interference components. By setting an energy threshold to filter and remove high-frequency noise, the low-frequency effective signal reflecting the characteristics of humic acid is retained. The denoised spectral data is input into the empirical mode decomposition module, which adaptively generates a set of intrinsic mode functions based on the local extreme point characteristics of the signal. The modal components that match the vibration frequencies of the carboxyl and hydroxyl groups of humic acid are selected to extract the intrinsic spectral characteristics that characterize the humic acid functional groups.

[0082] The scanning signal from the electrochemical detection channel is fed into a pretrained two-channel convolutional neural network. The network architecture includes a residual connection module to enhance deep feature representation. During transfer learning, a gradient reversal layer acts on the shallow feature extraction stage. By applying negative weight coefficients to the gradients of humic acid-related features during backpropagation, it suppresses the interference of the carboxyl vibration peak on the resolution of the iron ion diffusion peak. The decoupled electrochemical signal is then passed through a convolutional layer to extract joint time-frequency domain features. A pooling layer compresses redundant information and outputs a metal ion concentration vector. The independent output paths of the spectral absorption coefficient and the ion concentration vector form a decoupled multi-parameter detection dataset, providing high-precision input for subsequent spatial interpolation.

[0083] Specifically, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes the following steps in step S4:

[0084] Inputting the ion concentration vector generated in step S3 into a programmable resistance-capacitance compensation model to establish a correlation mapping between the metal ion migration effect and the phase angle offset;

[0085] Taking the impedance phase angle offset as the optimization target, the compensation parameters of the parallel capacitor array are iteratively adjusted through the genetic algorithm so that the attenuation of the parasitic capacitance effect in the corrected conductivity data is ≥30dB.

[0086] In step S4, the programmable RC compensation model receives the ion concentration vector output from step S3 and establishes a mapping relationship by analyzing the correlation between the metal ion migration effect and the phase angle offset of the electrochemical impedance spectroscopy. The time-frequency domain characteristic parameters in the ion concentration vector and the phase angle offset are correlated using a multivariate linear regression model to quantify the impact of parasitic capacitance caused by heterogeneous medium adsorption. The compensation model has a built-in parallel capacitor array, and the capacitance value of each branch is adjusted by a digital potentiometer to form a programmable capacitive reactance compensation network.

[0087] During the multi-objective optimization process of the genetic algorithm, the initial population consists of randomly generated combinations of capacitor compensation parameters. The phase angle offset fitness function is determined by calculating the absolute phase difference between the impedance spectra before and after correction at characteristic frequencies. The Nyquist plot integrity fitness function is based on a comprehensive score of the semicircular arc goodness of fit and high-frequency linearity. The algorithm uses a non-dominated sorting strategy to select the Pareto optimal solution set, combines crowding distance to maintain population diversity, and iteratively updates the compensation value combinations for the parallel capacitor array.

[0088] The optimized capacitance parameters are converted into analog voltage signals by a digital-to-analog conversion module, driving digital potentiometers to adjust the capacitive reactance values ​​of each branch. In the electrochemical sensor impedance spectrum after dynamic compensation, the phase angle offset caused by parasitic capacitance is significantly reduced, and the capacitive reactance arc in the high-frequency region of the Nyquist plot has restored its standard semicircular shape. The corrected conductivity data is then inverted to calculate the corresponding relationship between metal ion concentration and migration rate, eliminating measurement deviations caused by heterogeneous medium adsorption and providing high-fidelity input parameters for subsequent spatial interpolation.

[0089] Specifically, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes the following steps in step S5:

[0090] The decoupled spectral absorption coefficient and the conductivity data corrected by resistance-capacitance compensation were input into the spatial interpolation model. Based on the spatial correlation between the spectral absorption characteristics of the functional groups of humic acid molecules and the migration paths of metal ions, the base value and range parameters in the variation function were optimized to generate a three-dimensional residual map including the gradient distribution of humic acid content and the gradient of metal ion concentration.

[0091] In step S5, the decoupled spectral absorption coefficient and the conductivity data corrected by resistance-capacitance compensation are input into the spatial interpolation model. The spectral absorption coefficient characterizes the vibration characteristics of functional groups such as carboxyl and hydroxyl groups in humic acid molecules, and the humic acid content gradient distribution is quantified by the absorbance at the characteristic wavelength. The corrected conductivity data reflects the migration path and concentration differences of metal ions in the soil medium. The spatial correlation parameters are established by combining the ion mobility and conductivity correlation model. The spatial interpolation algorithm constructs a semi-variance function based on the Kriging method to characterize the spatial autocorrelation characteristics of humic acid content and ion concentration.

[0092] The sill value in the variogram parameter represents the extreme value of spatial variation in the data, while the range parameter characterizes the extent of spatial autocorrelation. A multi-objective genetic algorithm initializes a random parameter combination, including the sill value and the range, and iteratively optimizes the parameters through crossover and mutation operations. The fitness function comprehensively evaluates the goodness-of-fit of the root mean square error (RMSE) and semivariogram function between the interpolated predicted values ​​and the measured points, selecting parameter combinations that ensure that the spatial distribution model simultaneously maintains global smoothness and local detail. The optimized variogram parameters drive kriging interpolation to generate a three-dimensional fertility distribution surface, quantifying the spatial heterogeneity of nitrogen, phosphorus, and potassium nutrients.

[0093] Residual maps are generated by calculating the standardized difference between the interpolated surface predictions and the measured values. Low-value areas of humic acid gradient are identified using the residual threshold of the spectral absorption coefficient, and areas of anomalous metal ion concentration gradients are identified based on the dispersion of the conductivity residual. Three-dimensional residual maps utilize thermal layer rendering technology to map the spatial coordinates of humic acid-deficient areas and ion imbalance areas to a geographic information system, providing a visual positioning benchmark for targeted fertilization. This step establishes a quantitative correlation between fertility distribution characteristics and soil remediation strategies through multi-source data fusion and spatial modeling.

[0094] Specifically, the method for rapid detection and fertilization of dryland soil fertility of the present invention includes the following steps in step S6:

[0095] Based on the humic acid content gradient distribution in the three-dimensional residual map, a cross-linked sustained-release system modified with polyacrylamide was deployed in areas where the gradient value was below the preset threshold, inhibiting heterogeneous adsorption at the mineral interface through carboxyl coordination.

[0096] Based on the spatial distribution characteristics of the metal ion concentration gradient, a mesoporous zeolite-based passivator loaded with Bacillus subtilis was applied to the concentration exceeding the standard area, and its pore size selectivity was used to regulate the migration path of the target ions.

[0097] By real-time monitoring of the phase angle change rate of the electrochemical impedance spectroscopy, the degree of elimination of the residual effect of parasitic capacitance is verified and fertility compensation data is generated.

[0098] In step S6, based on the gradient distribution of humic acid content in the three-dimensional residual map, a cross-linked slow-release system modified with polyacrylamide is deployed in areas where the gradient value is lower than a preset threshold. Polyacrylamide is synthesized by free radical polymerization, and the carboxyl functional group is modified at the end of its molecular chain to form a coordination bond with the hydroxyl group on the mineral surface in the humic acid-deficient area, thereby inhibiting heterogeneous adsorption at the mineral interface. The degree of cross-linking is controlled by adjusting the molar ratio of glutaraldehyde to polyvinyl alcohol to form a three-dimensional network structure with adjustable pore size, which encapsulates the humic acid molecules to form slow-release particles. The slow-release particles are embedded in the soil surface by mechanical spreading, and the humic acid is gradually released through the swelling-diffusion mechanism triggered by soil moisture, thereby increasing the organic matter binding capacity in the deficient area.

[0099] The spatial distribution characteristics of metal ion concentration gradients were analyzed using a Kriging interpolation model. A mesoporous zeolite-based passivator loaded with Bacillus subtilis was applied to areas where concentrations exceeded the standard. After acid activation and pore expansion, the mesoporous zeolite was loaded with a Bacillus subtilis spore suspension by impregnation. Iron carriers produced by bacterial metabolism compete with organic acids for adsorption, reducing the activity of heavy metal ions. The zeolite's mesoporous structure has a pore size distribution of 2-50 nanometers. This pore-size sieving effect selectively blocks the migration of sodium and aluminum ions, while promoting the free diffusion of potassium and calcium ions. The passivator was evenly applied in powder form to the ion imbalance zone, and combined with irrigation water infiltration, the migration path of the target ions was regulated.

[0100] The dynamic phase response monitoring module collects phase angle data from the electrochemical impedance spectroscopy in real time and evaluates the extent to which the residual effect of parasitic capacitance has been eliminated by calculating the rate of change of the phase angle before and after correction. When the rate of change falls below a threshold, compensation is considered effective, triggering the generation of fertility compensation data. When the rate of change exceeds the threshold, feedback is fed back to the resistance-capacitance compensation model in step S4 to iteratively optimize the parameters of the parallel capacitor array. The monitoring data is then overlaid and analyzed with the spatial residual map to form a basis for dynamic adjustment of the humic acid release rate and passivator application intensity.

[0101] The names of the technical features involved in the technical solution of the present invention are explained as follows:

[0102] A genetic algorithm optimizes timing allocation: Using the signal interference suppression ratio as the fitness function, the algorithm iteratively searches for the optimal time interval combination between spectral pulses and electrochemical scans by simulating the selection, crossover, and mutation mechanisms of biological evolution. The initial population consists of randomized timing parameters. Each iteration selects individuals with low interference ratios for genetic recombination, ultimately generating non-overlapping timing signal streams that completely isolate the spectral excitation period from the electrochemical scan window, thus resolving the issue of time-domain signal crosstalk.

[0103] Wavelet packet decomposition and empirical mode processing are combined: the Daubechies wavelet basis is used to perform multi-scale decomposition of the spectral signal. The high-frequency subband corresponds to instrument noise and environmental interference, and the noise component is filtered out by energy threshold. The low-frequency effective signal is input into the empirical mode decomposition module, which adaptively generates the intrinsic mode function based on the local extreme value characteristics of the signal. The modal components that match the vibration frequency of the carboxyl / hydroxyl groups of humic acid are selected to extract the intrinsic spectral characteristics after denoising.

[0104] Neural Network Gradient Reversal: A gradient reversal layer is introduced into a dual-channel convolutional neural network. During backpropagation, a negative weight coefficient is applied to the humic acid-related feature gradients, suppressing their contribution to metal ion concentration prediction. The network reuses the generalization capabilities of the pre-trained model through transfer learning. Combined with a residual connection structure, the network enhances deep feature representation and effectively decouples the vibrational spectra of organic functional groups from the effects of metal ion migration.

[0105] A genetic algorithm multi-objective optimization approach for RC compensation uses phase angle offset and Nyquist plot integrity as dual fitness functions to initialize the compensation parameter population for the parallel capacitor array and select the Pareto optimal solution set using a non-dominated sorting strategy. The algorithm dynamically adjusts the capacitor value combination to restore the high-frequency capacitive reactance arc of the corrected impedance spectrum to a standard semicircular shape, eliminating phase distortion caused by heterogeneous adsorption.

[0106] Kriging interpolation spatial modeling: The spatial autocorrelation between humic acid content and ion concentration is characterized using a variogram. Sill values ​​quantify the data's variability limits, while range parameters define the spatial correlation range. Multi-objective genetic operators simultaneously optimize the variogram parameters, ensuring that the interpolation surface simultaneously captures global trends while preserving local details. This generates high-resolution three-dimensional fertility distribution maps, supporting precise fertilization decisions.

[0107] The application of the genetic algorithm in the technical solution of the present invention achieves functional collaboration in the following ways:

[0108] Timing optimization in step S2: Using the signal interference suppression ratio as the fitness function, a population of random parameters, including spectral pulse width and electrochemical scan interval, is initialized. A roulette wheel selection strategy is used to select individuals with low interference ratios. Single-point crossover is then used to restructure the timing sequence, and adaptive mutation is used to fine-tune the time interval. After iterative optimization using a genetic algorithm, a non-overlapping timing signal stream is generated, completely isolating the spectral excitation period from the electrochemical scan window and eliminating time-domain signal crosstalk. This process mimics the mechanisms of biological evolution, approaching the optimal timing allocation through selection, crossover, and mutation.

[0109] Step S4: Optimizing the RC compensation parameters: Using phase angle offset and Nyquist plot integrity as dual fitness objectives, the compensation parameter population for the parallel capacitor array is initialized. A non-dominated sorting strategy selects the Pareto front solution set, and combined with crowding distance, it maintains population diversity and avoids local optimality. The algorithm dynamically adjusts the capacitor value combination through multiple iterations to restore the high-frequency capacitive reactance arc of the corrected impedance spectrum to a standard semicircular shape, eliminating phase angle distortion caused by heterogeneous adsorption. This multi-objective optimization mechanism balances phase accuracy and morphological integrity of the electrochemical impedance spectroscopy.

[0110] Step S5: Optimizing the variogram parameters: A genetic algorithm initializes a population of random parameter combinations based on the sill value and range parameter in the kriging interpolation model. The fitness function integrates the interpolation prediction error and the semivariogram goodness-of-fit, using a crossover mutation operation to simultaneously optimize spatial autocorrelation and the ability to preserve local detail. The algorithm outputs the optimal parameter combination to generate a three-dimensional fertility distribution surface, ensuring that the interpolation results reflect both the humic acid content gradient and the spatial heterogeneity of metal ion concentrations.

[0111] Genetic algorithms employ a population evolution mechanism in all of the above steps: an initial population randomly generates candidate solutions, a fitness function quantifies the degree of achievement of technical objectives, and selection, crossover, and mutation operations drive the population's evolution toward optimization. Their core advantage lies in their global search capabilities, effectively solving multi-parameter nonlinear optimization problems in timing allocation, phase compensation, and spatial modeling, avoiding the pitfalls of traditional gradient descent methods that often fall into local optima. Based on this mechanism, those skilled in the art can clearly understand the synergistic logic of genetic algorithms in signal decoupling, phase correction, and spatial interpolation.

[0112] The application of the Nyquist plot achieves functional synergy through the following methods: The Nyquist plot, a graphical representation of the electrochemical impedance spectrum (EIS), with the real impedance component (Z') on the horizontal axis and the imaginary impedance component (Z'') on the vertical axis, is used to analyze the impedance characteristics of electrochemical sensors. In step S4, the impact of parasitic capacitance effects is assessed by the capacitive arc morphology in the high-frequency region of the Nyquist plot, using the semicircular arc goodness of fit and high-frequency linearity as fitness function indicators. Before correction, parasitic capacitance caused by adsorption of heterogeneous media causes the high-frequency capacitive arc to deviate from the standard semicircular shape. After optimizing the parallel capacitor array parameters using a genetic algorithm, the plot morphology returns to a regular semicircular shape, indicating that phase angle distortion has been suppressed. The dynamic phase response monitoring module in step S6 collects Nyquist plot data in real time and quantifies the residual parasitic capacitance effect by the rate of change of the phase angle. If the rate of change exceeds a threshold, it triggers iterative optimization of compensation parameters. The morphological integrity of the map is directly related to the spatial consistency of the conductivity data. Its high-frequency region characteristics provide a quantitative basis for capacitance compensation, and the low-frequency region characteristics reflect changes in the migration path of metal ions, ultimately supporting the spatial adaptability of soil fertility detection and remediation strategies.

[0113] The specific implementation of the present invention is as follows: First, a multi-physical field synergistic separation is performed on the heterogeneous soil medium, using gradient centrifugation to separate the colloidal and dissolved phases of the humic acid complex at a preset centrifugal force threshold. Components in the colloidal phase with a particle size greater than 100 nanometers settle to the bottom of the centrifuge tube via the Stokes sedimentation principle, while free metal ions in the dissolved phase with a particle size less than 100 nanometers are retained in the upper liquid phase. The separated colloidal and dissolved phases are then introduced into a dielectrophoresis enrichment device, where an alternating electric field with a frequency of 1-10 MHz is applied. The positive dielectrophoretic effect causes the colloidal phase to migrate toward regions of high field strength, while ions in the dissolved phase diffuse to regions of low field strength due to electroosmosis, forming a spatially sorted sample with a migration rate ratio greater than 2:1. This process uses physical field separation technology to reduce the heterogeneous adsorption of humic acid and mineral complexes on the sensor, reducing the initial source of parasitic capacitance.

[0114] A genetic algorithm was used to optimize the timing allocation strategy for dual-channel detection, initializing a random parameter combination consisting of spectral pulse width and electrochemical scan interval. The signal interference suppression ratio was quantified by calculating the ratio of the electrochemical background noise power spectral density to the effective signal bandwidth during the spectral excitation period, which served as a fitness function to drive the iterative optimization of the genetic algorithm. A roulette wheel selection strategy was used to select individuals with high fitness. Single-point crossover and adaptive mutation operations generated non-overlapping timing signal streams, completely isolating the spectral pulse emission period from the electrochemical scan interval in the temporal domain. The optimized timing parameters were controlled by a hardware synchronization trigger module to independently operate the dual-channel probes. The spectral detection channel activated a broadband light source with a pulse width of 50-200 ms to acquire colloidal phase reflectance spectra, while the electrochemical detection channel applied a linear sweep voltage of 0.1-1 V / s during the interval to record the solution phase voltammetry curve, achieving signal decoupling for simultaneous multi-parameter measurement.

[0115] The spectral signal was subjected to multimodal decomposition, using a Daubechies wavelet basis for multiscale decomposition. After filtering out high-frequency noise, the vibrational signature of the carboxyl groups of humic acid was extracted using empirical mode methods. The electrochemical signal was fed into a pretrained two-channel convolutional neural network. A gradient reversal layer applied negative weights to humic acid-related features, separating the iron ion diffusion peak from the vibrational spectrum of organic functional groups. The decoupled spectral absorption coefficient and ion concentration vector were input into a programmable resistance-capacitance compensation model. The parameters of the parallel capacitor array were optimized using a genetic algorithm with multiple objectives. The phase angle offset and Nyquist plot semicircle arc fit were used as fitness functions, and the compensation value was dynamically adjusted to suppress parasitic capacitance effects. The corrected conductivity data and spectral absorption coefficient were fed into a kriging interpolation model. The sill value and range parameters of the variogram were optimized to generate a three-dimensional fertility distribution surface. Humic acid-deficient areas and ion imbalance regions were identified based on the residual map. A polyacrylamide cross-linked slow-release system was deployed to increase organic matter binding capacity. A mesoporous zeolite passivator loaded with Bacillus subtilis was applied to regulate ion migration pathways. The dynamic phase response monitoring module verifies the degree of elimination of the residual effect of parasitic capacitance in real time, forming a detection, repair, and feedback closed-loop control, and ultimately achieving high-precision analysis and spatial adaptive repair of soil fertility parameters.

[0116] The present invention solves the problems of signal interference and parasitic capacitance effect caused by heterogeneous adsorption of humic acid and mineral complexes in soil fertility determination through the following technical solutions:

[0117] First, a physical field synergistic separation technique combining gradient centrifugation and dielectrophoresis enrichment was employed to separate the colloidal and dissolved phases of the humic acid complex based on differences in density and dielectric properties. Humic acid-mineral complexes with particle sizes greater than 100 nanometers in the colloidal phase were analyzed using a spectral detection channel, while free metal ions with particle sizes less than 100 nanometers in the dissolved phase were measured using an electrochemical detection channel. This approach reduces heterogeneous adsorption of heterogeneous media to the sensor during the sample pretreatment stage, thus reducing the initial source of parasitic capacitance.

[0118] Secondly, a genetic algorithm was used to optimize the timing allocation strategy for dual-channel detection, dynamically generating non-overlapping time-series signal streams using the signal interference suppression ratio as the fitness function. The spectral pulse width and electrochemical scan interval were completely isolated in the time domain to avoid synchronous interference between spectral excitation and electrochemical scanning. Furthermore, wavelet packet decomposition and empirical mode processing were combined to extract the intrinsic spectral characteristics of humic acid. The gradient inversion layer of a dual-channel convolutional neural network was combined to decouple the vibrational characteristics of humic acid functional groups from the effects of metal ion migration, thus eliminating the frequency-domain coupling between the spectral and electrochemical signals at the signal processing level.

[0119] Furthermore, a programmable resistance-capacitance compensation model was constructed. Parallel capacitor array parameters were optimized using a genetic algorithm with multiple objectives. Phase angle offset and Nyquist plot integrity were used as fitness functions to dynamically correct for the parasitic capacitance effects of the electrochemical sensor. The optimized capacitance compensation parameters suppressed phase angle distortion in the impedance spectrum. Combined with spatial interpolation of the three-dimensional residual plot, a cross-linked slow-release system and mesoporous passivators were deployed to selectively control fertility parameters. Dynamic phase response monitoring formed a closed-loop feedback loop, ultimately enabling the analysis of spectral and electrochemical signals and the spatial adaptation of soil remediation strategies.

Claims

1. A method for rapid detection and fertilization of dryland soil fertility, characterized in that: include: Step S1: Perform multi-physical field synergistic separation on the soil heterogeneous medium, and obtain the colloidal phase and dissolved phase of the humic acid complex through the sorting action of gradient centrifugation and dielectrophoresis enrichment. The colloidal phase particle size >100 nm is used for spectral detection, and the dissolved phase particle size <100 nm is used for electrochemical detection; Step S2: Using a preset dual-channel detection probe, optimizing the preset dual-channel detection timing based on a genetic algorithm, taking the signal interference suppression ratio as the objective function, and generating a non-overlapping timing scanning sequence of the spectrum and electrochemical signal; Step S3: Perform multimodal signal decomposition on the spectral signal to extract the humic acid characteristic spectrum, and simultaneously perform neural network transfer learning on the electrochemical signal to separate the organic functional group vibration spectrum and the metal ion migration effect through reverse gradient operation, and output the decoupled spectral absorption coefficient and ion concentration vector; Step S4: constructing a programmable resistance-capacitance compensation model, taking the impedance phase angle offset as the optimization target, and adjusting the capacitance compensation parameters through a genetic algorithm to correct the conductivity phase distortion to obtain the corrected conductivity; Step S5: inputting the decoupled spectral absorption coefficient and the corrected conductivity into the spatial interpolation model, and simultaneously optimizing the variance function parameters to generate a fertility distribution surface and a residual map including a humic acid content gradient and a metal ion concentration gradient; Step S6: Based on the residual map of step S5, an adsorption compensation model is constructed, and a surface-modified cross-linked sustained-release system is deployed in areas where the humic acid content gradient is lower than the preset threshold to inhibit heterogeneous adsorption. In areas where the metal ion concentration gradient exceeds the standard range, a microbial-loaded mesoporous structure passivator is applied to regulate the ion mass transfer path, and fertility compensation data is generated through dynamic phase response monitoring.

2. The method for rapid detection and fertilization of dryland soil fertility according to claim 1, characterized in that: In step S1, the multi-physical field collaborative separation includes: The soil heterogeneous medium was initially separated by gradient centrifugation, and the colloidal phase and dissolved phase of the humic acid complex were obtained under a preset centrifugal force threshold. The colloidal phase and dissolved phase separated in the step are respectively input into the dielectrophoresis enrichment device, and the colloidal particles are caused to undergo dielectrophoretic migration by applying an alternating electric field, and a spatially sorted pre-processed sample is formed according to the difference in migration rate. The ratio of the migration rate of the colloidal phase to the migration rate of the dissolved phase is greater than 2:

1.

3. The method for rapid detection and fertilization of dryland soil fertility according to claim 2, characterized in that: The step S2 includes: The spectral detection channel of the dual-channel detection probe is loaded with the colloidal phase, and the electrochemical detection channel is loaded with the dissolved phase; The genetic algorithm optimizes the timing allocation logic to generate non-overlapping timing signal streams by dynamically allocating spectral pulse widths and electrochemical scanning intervals.

4. The method for rapid detection and fertilization of dryland soil fertility according to claim 3, characterized in that: The step S3 includes: The signal collected by the spectral detection channel in step S2 is input into the multimodal signal decomposition unit, and the humic acid characteristic spectrum is extracted after removing high-frequency noise through wavelet packet decomposition; The signal collected by the electrochemical detection channel in step S2 is input into the transfer learning network, and the carboxyl vibration peak and the iron ion diffusion peak in the humic acid characteristic spectrum are separated by reverse gradient operation to generate a decoupled spectral absorption coefficient and an ion concentration vector.

5. The method for rapid detection and fertilization of dryland soil fertility according to claim 4, characterized in that: The step S4 includes: Inputting the ion concentration vector generated in step S3 into a programmable resistance-capacitance compensation model to establish a correlation mapping between the metal ion migration effect and the phase angle offset; Taking the impedance phase angle offset as the optimization target, the compensation parameters of the parallel capacitor array are iteratively adjusted through the genetic algorithm so that the attenuation of the parasitic capacitance effect in the corrected conductivity data is ≥30dB.

6. The method for rapid detection and fertilization of dryland soil fertility according to claim 5, characterized in that: The step S5 includes: The decoupled spectral absorption coefficient and the conductivity data corrected by resistance-capacitance compensation were input into the spatial interpolation model. Based on the spatial correlation between the spectral absorption characteristics of the functional groups of humic acid molecules and the migration paths of metal ions, the base value and range parameters in the variation function were optimized to generate a three-dimensional residual map including the gradient distribution of humic acid content and the gradient of metal ion concentration.

7. The method for rapid detection and fertilization of dryland soil fertility according to claim 6, characterized in that: The step S6 includes: Based on the humic acid content gradient distribution in the three-dimensional residual map, a cross-linked sustained-release system modified with polyacrylamide was deployed in areas where the gradient value was below the preset threshold, inhibiting heterogeneous adsorption at the mineral interface through carboxyl coordination. Based on the spatial distribution characteristics of the metal ion concentration gradient, a mesoporous zeolite-based passivator loaded with Bacillus subtilis was applied to the concentration exceeding the standard area, and its pore size selectivity was used to regulate the migration path of the target ions. By real-time monitoring of the phase angle change rate of the electrochemical impedance spectroscopy, the degree of elimination of the residual effect of parasitic capacitance is verified and fertility compensation data is generated.

Images