Organic carbon source data optimization method and system based on crop growth feedback

Abstract

The application provides an organic carbon source data optimization method and system based on crop growth feedback, data acquisition and data preprocessing are performed on crops, a crop growth data set and a crop environment data set are obtained, crop growth indicators and environmental stress indicators are obtained based on the crop growth data set and the crop environment data set, a causal correlation chain between the crop growth indicators and the environmental stress indicators is mined, and a growth environment correlation rule library is constructed according to the causal correlation chain, an organic carbon source regulation model is constructed based on the crop growth indicators, the environmental stress indicators and the growth environment correlation rule library, an organic carbon source regulation decision is generated according to the organic carbon source regulation model, an implementation effect of the organic carbon source regulation decision is obtained, and the organic carbon source regulation rule library and the organic carbon source regulation model are iteratively optimized according to the implementation effect of the organic carbon source regulation decision, so that the accuracy of the organic carbon source regulation scheme is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural technology, in particular to an organic carbon source data optimization method and system based on crop growth feedback. BACKGROUND

[0002] In current agricultural production, the application of organic carbon sources mainly relies on empirical formula, fixed cycle management or static model based on soil basic nutrients for decision-making. Due to the lack of direct response mechanism to crop growth, the traditional method often fails to establish a dynamic correlation between organic carbon source input and plant growth efficiency, resulting in mismatch of farmland resources and intensified environmental risks.

[0003] On the one hand, the traditional method often relies on static indicators such as soil organic carbon and pH for the optimization decision of organic carbon source regulation, ignoring the real-time monitoring of key parameters of crop growth, so that the actual influence of carbon source input on crop growth and physiological process cannot be reflected, thus causing the problem of mismatch of farmland resources; on the other hand, the traditional method often uses fixed threshold or linear rule to analyze the relationship between crops and carbon sources, which has certain limitations in analyzing the complex nonlinear coupling relationship among environment, crops and carbon sources, and the accuracy of organic carbon source regulation scheme cannot be guaranteed in the face of variety differences, soil heterogeneity and climate fluctuation scenarios. SUMMARY

[0004] In order to solve the above technical problems, the present application provides an organic carbon source data optimization method and system based on crop growth feedback to solve the technical problems that the actual influence of carbon source input on crop growth and physiological process cannot be reflected and the accuracy of organic carbon source regulation scheme cannot be guaranteed in the face of variety differences, soil heterogeneity and climate fluctuation scenarios in the prior art.

[0005] The purpose and effect of the organic carbon source data optimization method and system based on crop growth feedback of the present application are achieved by the following specific technical means:

[0006] The organic carbon source data optimization method based on crop growth feedback comprises:

[0007] S1: data acquisition and data preprocessing are performed on crops to obtain crop growth data set and crop environment data set;

[0008] S2: crop growth indicators and environmental stress indicators are obtained based on the crop growth data set and the crop environment data set, respectively;

[0009] S3: the causal correlation chain between the crop growth indicators and the environmental stress indicators is mined, and a growth environment correlation rule library is constructed according to the causal correlation chain;

[0010] S4: constructing an organic carbon source regulation model based on the crop growth index, the environmental stress index and the growth environment correlation rule base, and generating an organic carbon source regulation decision according to the organic carbon source regulation model;

[0011] S5: obtaining an implementation effect of the organic carbon source regulation decision, and iteratively optimizing the organic carbon source regulation rule base and the organic carbon source regulation model according to the implementation effect of the organic carbon source regulation decision.

[0012] As a further scheme of the present application, the crop growth index and the environmental stress index are respectively obtained based on a crop growth data set and a crop environment data set, including:

[0013] The crop growth data set is subjected to feature extraction to obtain stem image data, root projection area data, soil humidity data, canopy LAI, leaf chlorophyll content data and leaf texture density data;

[0014] The displacement change of the stem image data of three consecutive frames at a fixed time interval is analyzed to obtain the internal strain energy distribution of the stem, and the vascular strain energy density is generated based on the internal strain energy distribution;

[0015] The time series change acceleration of the root projection area data is analyzed, and the soil humidity data fluctuation characteristics are combined to generate a root anchoring decay rate;

[0016] The crop root and canopy growth signal transmission efficiency is obtained by analyzing the root projection area data and the canopy LAI, and the root canopy response entropy is generated based on the transmission efficiency;

[0017] The photosynthetic phase space is constructed based on the leaf chlorophyll content data and the leaf texture density data, and the photosynthetic phase space compression degree is generated by analyzing the motion trajectory of the leaf chlorophyll content data and the leaf texture density data in the photosynthetic phase space;

[0018] The crop structural index is constructed based on the vascular strain energy density and the root anchoring decay rate, and the functional synergy index is constructed based on the root canopy response entropy and the photosynthetic phase space compression degree, wherein the crop growth index includes the crop structural index and the functional synergy index;

[0019] The crop environment data set is subjected to feature extraction to obtain environmental feature data, and the environmental stress index is constructed based on the environmental feature data.

[0020] As a further scheme of the present application, the crop environment data set is subjected to feature extraction to obtain environmental feature data, and the environmental stress index is constructed based on the environmental feature data, including:

[0021] The environmental feature data includes photosynthetic active radiation data, canopy spectrum data, air temperature and humidity data, stem surface strain rate, soil oxidation-reduction potential data and light intensity data.

[0022] Generate light-carbon disorder degree data by analyzing the response relationship between photosynthetically active radiation data and canopy spectral data;

[0023] Generate water damage coupling data by analyzing the response relationship between air temperature and humidity data and stem surface strain rate;

[0024] Generate marginal load data by analyzing soil redox potential data, light intensity data and corresponding spatial position coordinates;

[0025] Generate environmental disturbance memory data by analyzing soil redox potential data in continuous time;

[0026] Construct a metabolic abnormal environment coupling index based on light-carbon disorder degree data and water damage coupling data, construct a spatial heterogeneity index based on marginal load data and environmental disturbance memory data, and the environmental stress index includes the metabolic abnormal environment coupling index and the spatial heterogeneity index.

[0027] As a further scheme of the present application, the causal correlation chain between the crop growth index and the environmental stress index is mined, and a growth environment correlation rule library is constructed according to the causal correlation chain, comprising:

[0028] Obtain the causal entropy between the crop growth index and the environmental stress index in a preset time window length, and quantify the time sequence coupling relationship between the crop growth index and the environmental stress index corresponding to different growth stages of the crop based on the causal entropy;

[0029] Calculate the hierarchical transition causal gain value from the organ level to the system level, and filter the macro-lag interference based on the hierarchical transition causal gain value;

[0030] Wherein, the organ level represents the structural and functional performance of each organ of the crop, the system level represents the comprehensive growth state of the crop, and the macro-lag interference represents the delay phenomenon of the organ and the system in response to environmental changes;

[0031] Under the premise of keeping the environmental stress index unchanged, the crop variety is virtually switched, the change difference rate of the crop growth index is observed, the universality of the time sequence coupling relationship is detected based on the change difference rate, and if the change difference rate reaches a preset difference rate threshold interval, it is determined that the time sequence coupling relationship has variety universality;

[0032] Synthesize virtual data of different soil conditions in the same plot by a generative adversarial network, and separate the interference of soil properties on the time sequence coupling relationship;

[0033] Generate a causal correlation chain between the crop growth index and the environmental stress index based on the time sequence coupling relationship, and construct a growth environment correlation rule library based on the causal correlation chain.

[0034] As a further scheme of the present application, the long- and short- term environmental correlation rule library is constructed based on the cause-effect correlation chain, comprising:

[0035] The rule extraction is performed on the cause-effect correlation chain to obtain three types of long- and short- term environmental correlation rules, i.e., atomic rules, combined rules and space-time constraint rules, the support, confidence and lift of the long- and short- term environmental correlation rules are obtained, and the long- and short- term environmental correlation rules that do not meet the preset support threshold, confidence threshold and lift threshold are removed to construct the long- and short- term environmental correlation rule library.

[0036] As a further scheme of the present application, the organic carbon source regulation model is constructed based on the crop growth index, the environmental stress index and the long- and short- term environmental correlation rule library, and the organic carbon source regulation decision is generated according to the organic carbon source regulation model, comprising:

[0037] The physiological state of the crop is diagnosed in real time according to the crop growth index, the environmental impact on the crop growth process is quantified through the environmental stress index, the long- and short- term environmental correlation rules corresponding to the crop growth index and the environmental stress index in the long- and short- term environmental correlation rule library are matched and called, and the organic carbon source regulation decision is generated based on the matched long- and short- term environmental correlation rules to regulate the organic carbon source of the crop.

[0038] As a further scheme of the present application, the implementation effect of the organic carbon source regulation decision is obtained, and the organic carbon source regulation rule library and the organic carbon source regulation model are iteratively optimized according to the implementation effect of the organic carbon source regulation decision, comprising:

[0039] The organic carbon source regulation decision that does not improve the target crop growth index within a predetermined time is determined as a failed decision, the long- and short- term environmental correlation rule corresponding to the failed decision is marked as a gradually ineffective rule, and the confidence of the gradually ineffective rule is reduced;

[0040] The long- and short- term environmental correlation rule that is marked as a gradually ineffective rule for three times in succession is determined as an ineffective rule, and the ineffective rule is subjected to a freezing operation;

[0041] The organic carbon source regulation decision that improves the target crop growth index within a predetermined time is determined as a successful decision, the long- and short- term environmental correlation rule corresponding to the successful decision is marked as an effective rule, and the confidence of the effective rule is improved.

[0042] As a further scheme of the present application, the crop is subjected to data acquisition and data preprocessing to obtain a crop growth data set and a crop environmental data set, comprising:

[0043] The acquired crop data is subjected to data preprocessing including at least outlier filtering, time-space alignment, data standardization and data normalization to obtain the crop growth data set and the crop environmental data set;

[0044] The crop growth data set is used to describe the multi-dimensional dynamic growth characteristics of the crop, and at least includes crop canopy dimensional data, stem dimensional data, underground root dimensional data, and physiological dimensional data;

[0045] The crop environment data set is used to describe the growth environment of the crop, and at least includes surface environment data, soil layer environment data, and microenvironment data.

[0046] As a further scheme of the present application, the collected crop data is subjected to data preprocessing including at least outlier filtering, time-space alignment, data standardization and data normalization, including:

[0047] The outlier filtering is represented as outlier filtering and missing value filling of the crop data;

[0048] The time-space alignment is represented as unifying the crop data into the same geographical coordinate reference and time reference;

[0049] The data standardization is represented as data standardization operation on the crop growth data contained in the crop data;

[0050] The data normalization is represented as data normalization operation on the crop environment data contained in the crop data.

[0051] The organic carbon source data optimization system based on crop growth feedback, comprising:

[0052] A data processing module, the data processing module is used for data acquisition of the crop, data preprocessing of the collected data, generation of crop growth data set and crop environment data set;

[0053] An index generation module, the index generation module is used for generating corresponding crop growth index and environmental stress index according to the crop growth data set and the crop environment data set;

[0054] A rule extraction module, the rule extraction module is used for extracting the causal correlation chain between the crop growth index and the environmental stress index, and generating corresponding rules according to the causal correlation chain;

[0055] A model construction module, the model construction module is used for constructing an organic carbon source regulation model according to the crop growth index, the environmental stress index and the growth environment correlation rule base, and generating an organic carbon source regulation decision;

[0056] A feedback optimization module, the feedback optimization module is used for obtaining the implementation effect of the organic carbon source regulation decision, and optimizing and adjusting the organic carbon source regulation rule base and the organic carbon source regulation model according to the implementation effect.

[0057] Based on the above aspects, the embodiment of the present application realizes data collection and data preprocessing of crops, obtains crop growth data set and crop environment data set, eliminates interference noise and data offset generated in the collection process by preprocessing the collected data, and provides accurate data input for subsequent generation of quantitative indicators after preprocessing the data;

[0058] Based on the crop growth data set and the crop environment data set, crop growth indicators and environmental stress indicators are obtained, the causal correlation chain between the crop growth indicators and the environmental stress indicators is mined, and a growth environment correlation rule base is constructed according to the causal correlation chain. By constructing the crop growth indicators and the environmental stress indicators, the discrete environmental data and crop physiological data are accurately quantified, and comprehensive and accurate data input is provided for subsequent model generation of organic carbon source regulation decision. The causal coupling relationship between the environmental data and the crop physiological data is analyzed by mining the causal correlation chain between the indicators, and the corresponding quantitative rules are obtained, which provides a reference benchmark for subsequent model generation of organic carbon source regulation decision;

[0059] Based on the crop growth indicators, the environmental stress indicators and the growth environment correlation rule base, an organic carbon source regulation model is constructed, and an organic carbon source regulation decision is generated according to the organic carbon source regulation model. The implementation effect of the organic carbon source regulation decision is obtained, the organic carbon source regulation rule base and the organic carbon source regulation model are iteratively optimized according to the implementation effect of the organic carbon source regulation decision, the confidence of the efficient rules in the organic carbon source regulation rule base is strengthened, and the confidence of the inefficient rules is weakened by monitoring the implementation effect of the organic carbon source regulation decision, so as to improve the accuracy of the organic carbon source regulation scheme. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is the execution flow schematic diagram of the organic carbon source data optimization method based on crop growth feedback provided by the embodiment of the present application;

[0061] Figure 2 is the execution flow schematic diagram of step S2 in the organic carbon source data optimization method based on crop growth feedback provided by the embodiment of the present application;

[0062] Figure 3 is the execution flow schematic diagram of step S3 in the organic carbon source data optimization method based on crop growth feedback provided by the embodiment of the present application;

[0063] Figure 4 is the marginal two-dimensional plane coordinate system grid division schematic diagram provided by the embodiment of the present application;

[0064] Figure 5 is the schematic diagram of the organic carbon source data optimization system based on crop growth feedback provided by the embodiment of the present application. DETAILED DESCRIPTION

[0065] The embodiments of the present application will be further described in details with reference to the accompanying drawings and examples. The following examples are used to illustrate the technical solutions of the present application, but cannot be used to limit the protection scope of the present application.

[0066] Example 1:

[0067] As shown in the accompanying drawings and examples: Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 ,

[0068] The embodiment of the present application provides a method for optimizing organic carbon source data based on crop growth feedback, which is suitable for the field of agricultural technology and includes the following steps:

[0069] Step S1, data acquisition and data preprocessing of crops are performed to obtain a crop growth data set and a crop environment data set.

[0070] Specifically, the collected crop data is subjected to data preprocessing including at least outlier filtering, time-space alignment, data standardization and data normalization to obtain the crop growth data set and the crop environment data set.

[0071] It can be understood that the outlier filtering means that the crop data is subjected to outlier filtering and missing value filling; the time-space alignment means that the crop data is unified to the same geographical coordinate reference and time reference; the data standardization means that the crop growth data contained in the crop data is subjected to data standardization operation; and the data normalization means that the crop environment data contained in the crop data is subjected to data normalization operation.

[0072] It can be understood that the crop growth data set is used to describe the multi-dimensional dynamic growth characteristics of crops, and at least includes crop canopy dimensional data, stem dimensional data, underground root dimensional data and physiological dimensional data.

[0073] Specifically, the unmanned aerial vehicle carrying a multispectral camera is used to collect data of crop canopy, and the canopy LAI is obtained. The canopy LAI is a key parameter for measuring the structure of the vegetation canopy, and is used to quantify ecological processes such as photosynthesis and respiration. The obtained canopy LAI is corrected for terrain, such as inputting 1-meter DEM data. For the area with a terrain slope greater than 5 degrees, the canopy LAI is compensated according to "original LAI*(1+0.018*terrain slope value)". The terrain slope value is obtained according to the angle of the terrain slope and the relative position of the sun, that is, when the corresponding terrain slope of the collected canopy LAI is 15 degrees and the sun is facing, the corresponding terrain slope value is +15. The effective period of the corresponding canopy LAI is screened, such as excluding the corresponding data with a solar elevation angle less than 40 degrees. The spectral band of the corresponding multispectral camera is changed during the effective period of the rainy weather, to ensure the accuracy of the collected canopy LAI (here, only the terrain correction and effective period screening are exemplified, and the actual situation needs to be determined).

[0074] Further, the chlorophyll content of crop leaves and the corresponding leaf temperature are collected. The temperature drift compensation is performed on the collected leaf chlorophyll content data, such as data correction according to "SPAD correction value = measured value + 0.28 x (25-leaf temperature)". The leaf chlorophyll content data is compensated according to the variety of crops, such as increasing the SPAD of the tillering period of indica rice by 5%, that is, the measured value of the SPAD is 42, and the compensated SPAD value is 42*1.05=44.1 (here, only the temperature drift compensation and crop variety compensation are exemplified, and the actual situation needs to be determined).

[0075] Further, data collection is performed on the back of the crop leaves to obtain leaf texture density data. The collected leaf texture density data is subjected to noise elimination, such as using HSV color domain segmentation and near-infrared absorption detection to remove leaf lesion area data noise. The light intensity calibration and data normalization are performed according to the canopy irradiance map to obtain the leaf texture expansion, such as correcting the leaf texture density data according to "leaf texture expansion = leaf texture density data measured value*(1000 / actual light intensity)" (here, only the noise elimination and light intensity calibration are exemplified, and the actual situation needs to be determined).

[0076] In this embodiment, the pre-processed canopy LAI, leaf chlorophyll content data and leaf texture density data are spatio-temporally aligned to generate crop canopy dimensional data.

[0077] Specifically, for the marked monitoring points on the stem part of the crop, the stem part of the crop is photographed and monitored at fixed time intervals, and stem image data is obtained, and the temperature corresponding to the stem part of the crop in the photographing period is recorded. The displacement of the marked monitoring points in the stem image data is calculated by using a phase correlation algorithm to obtain the stem displacement, and the obtained stem displacement is subjected to noise removal. Based on the stem displacement, the strain rate of the stem surface is obtained. The preprocessed stem image data is spatio-temporally aligned to generate stem dimensional data.

[0078] Further, the root projection area original image, the soil humidity data of 20 cm underground and 40 cm underground, and the soil humidity data of 20 cm underground and 40 cm underground are obtained. Based on the soil humidity data of 20 cm underground and 40 cm underground and the characteristics of the crop, the soil water content is obtained by feature fusion, such as cotton using a structure of 0.3*20 cm underground soil humidity data+0.7*40 cm underground soil humidity data for weighted fusion to obtain the soil water content, and lettuce using a structure of 0.8*20 cm underground soil humidity data+0.2*40 cm underground soil humidity data for weighted fusion to obtain the soil water content. The data is frozen for 2 hours after irrigation of the crop, so as to shield the water infiltration disturbance; based on the convolutional neural network, the feature region corresponding to the new root in the root projection area original image is screened, such as the region with a daily increase of more than 5%, an aspect ratio of more than 3, a color of milky white, and reaching a preset RGB threshold, and combined with the geometric verification that the angle between the root tip extension direction and the gravity direction is less than 30 degrees, the soil crack false image interference is excluded; according to the soil water content, the root projection area original image is corrected in visibility to obtain the root projection area data, for example, for clay soil with soil clay content greater than 45%, the scanning area of the root projection area original image is divided by 0.89 times the negative 0.028 times the water content of e to obtain the visibility value for visibility correction, and for sandy soil with soil sand content greater than 70%, the scanning area is multiplied by 1.8 minus the reciprocal of the coefficient of 0.016 times the water content; the preprocessed root projection area data and the soil humidity data are spatio-temporally aligned to generate underground root dimensional data (here, only the soil water content, the new root region screening, and the root projection area original image visibility correction are exemplified, and the actual situation needs to be determined).

[0079] Further, the canopy spectral data of the crop is acquired, the canopy spectral red edge of the canopy spectral data is analyzed, the canopy spectral red edge displacement amount is acquired, for example, the canopy spectral red edge is located based on the first derivative peak value in the [680, 750] nanometer interval, the canopy spectral red edge displacement amount is acquired by analyzing the difference of the canopy spectral red edge in a fixed time period; the preprocessed leaf chlorophyll content data, the leaf texture density data and the canopy spectral red edge displacement amount are spatiotemporally aligned, the initial physiological dimension data of the crop is acquired based on the leaf texture density data, the initial physiological dimension data is corrected and compensated according to the canopy spectral red edge displacement amount, and the initial physiological dimension data is secondarily corrected through the leaf chlorophyll content data to generate the physiological dimension data, for example, the physiological dimension data is generated through The physiological dimension data of the crop is acquired, wherein The physiological dimension data is used to represent the photosynthetic efficiency of the crop, The leaf texture density data is used to represent the leaf texture density of the crop, The canopy spectral red edge displacement amount is used to represent the canopy spectral red edge displacement amount of the crop, and the maximum η value is limited to not more than 0.6 when the leaf chlorophyll content is less than 45 (here, only a possibility of the physiological dimension data is exemplified, and the actual content needs to be determined according to the specific situation).

[0080] It can be understood that the crop environment data set is used to describe the growth environment of the crop, and at least includes the ground surface environment data, the soil layer environment data and the microenvironment data.

[0081] Specifically, the sensor module group deployed 1 meter above the canopy is used to collect the original photosynthetically active radiation data, the air temperature and humidity data and the light intensity data, and the Beidou positioning space coordinates are synchronously bound. For the mutation of the photosynthetically active radiation data caused by cloud flashing, the sliding window median filtering with a sliding window length of 5 minutes is used for noise removal. For the photosynthetically active radiation data collection during the dewing period of the crop, the water droplet shielding correction is started. When the leaf temperature of the crop is lower than the dew point temperature and the corresponding humidity is greater than 80%, the original photosynthetically active radiation data is corrected according to “photosynthetically active radiation data correction value = photosynthetically active radiation data measured value * (1-0.4 * dewing coverage)”. (Here, only a possibility of the water droplet shielding correction is exemplified, and the actual content needs to be determined according to the specific situation).

[0082] Further, for air temperature and humidity data, preset physical threshold is adopted to eliminate abnormal data noise, and mutation of air temperature and humidity data is checked for mutation rationality. If temperature difference is greater than 3℃ or humidity difference is greater than 15% within 10 minutes, wind speed is used for auxiliary verification. If wind speed is lower than 1m / s, the mutation is regarded as failure, and the mutation data is eliminated and the standby sensor is switched to reacquire data to fill in the data loss. For direct exposure point without sunshade, temperature virtual high is compensated according to solar radiation intensity. If radiation is 1kW / m², 0.4℃ needs to be reduced. When the rainfall is greater than 5mm / h, the weight of air humidity data in the rainfall period is reduced to suppress humidity virtual high. For example, the confidence of air humidity data within 30 minutes after rain is reduced to 80%. During the dewing period of crops, the theoretical dew point temperature is obtained based on air temperature and humidity data. If the deviation between air temperature data and theoretical dew point temperature data exceeds the preset temperature deviation threshold, and the air humidity data exceeds the preset humidity threshold, it is determined that dewing failure occurs. The air temperature and humidity data in the corresponding period is replaced by the data in the adjacent period. For example, the mean value of air temperature and humidity data in the last 5 minutes before dewing failure period is used to replace the corresponding data in the dewing failure period.

[0083] Further, for light intensity data, effective data is dynamically selected according to solar elevation angle. For example, only light intensity data corresponding to solar elevation angle greater than 40 degrees is retained, and interference data with low signal-to-noise ratio at dawn and dusk is eliminated. For light intensity data on cloudy days, the weight of 20% light intensity data is increased. During heavy rainfall period, the weight of light intensity data is reduced to avoid light intensity virtual high caused by water droplet scattering. During the dewing period of crops, the liquid droplet coverage area is located by infrared thermal imager, the data of contaminated sensors in the liquid droplet coverage area is marked, and the corresponding confidence is reduced to 60%.

[0084] Then, the preprocessed photosynthetic active radiation data, air temperature and humidity data, and light intensity data are spatiotemporally aligned to generate ground surface environment data.

[0085] The soil oxidation-reduction potential data at the corresponding depths is obtained by burying a redox potential sensor at a depth of 20-40 cm in the soil, the collected soil oxidation-reduction potential data is temperature compensated, for example, a temperature compensation factor of "-0.7*(T-25)" is applied to the soil oxidation-reduction potential data, where T is the current soil temperature; the collected soil oxidation-reduction potential data is salinity compensated, for example, for every 0.1 mS / cm of conductivity exceeding the preset baseline, a compensation of +0.1 mV is applied accordingly; the data is updated 2 hours after the freezing and fertilization, so as to avoid the data disturbance caused by fertilization; the fused soil oxidation-reduction potential data is generated by weighted fusion according to the distribution characteristics of the crop root system, for example, for deep-rooted plants, the weighted fusion is performed by using a weight ratio of "0.3*20 cm depth corresponding soil oxidation-reduction potential data+0.7*40 cm depth corresponding soil oxidation-reduction potential data", and for shallow-rooted plants, the weighted fusion is performed by using a weight ratio of "0.8*20 cm depth corresponding soil oxidation-reduction potential data+0.2*40 cm depth corresponding soil oxidation-reduction potential data"; the fused soil oxidation-reduction potential data is normalized, for example, the soil oxidation-reduction potential data range of a rice field is [-200, 200] mV, the data in this interval range is mapped to the interval [0, 1], and for example, -120 mV, the normalized fused soil oxidation-reduction potential data value is [-120-(-200)] / [200-(-200)]=0.2.

[0086] Further, the preprocessed soil oxidation-reduction potential data and the soil humidity data are spatio-temporally aligned to generate soil layer environment data, and the canopy spectrum data and the stem surface strain rate are spatio-temporally aligned to generate microenvironment data.

[0087] In step S2, the crop growth indicators and environmental stress indicators are obtained based on the crop growth data set and the crop environment data set, respectively.

[0088] In this embodiment, step S2 includes:

[0089] In step S21, the stem image data, root projection area data, soil humidity data, canopy LAI, leaf chlorophyll content data, and leaf texture density data are obtained by feature extraction on the crop growth data set.

[0090] In step S22, the internal strain energy distribution of the stem is obtained by analyzing the displacement change of the stem image data of three consecutive frames at a fixed time interval, and the vascular strain energy density is generated based on the internal strain energy distribution.

[0091] In some possible embodiments, assuming that the maximum stem displacement of a certain corn stalk at a monitoring point is 0.35 mm within 3 hours in a 7-level gale environment, the maximum stem displacement is divided by the stem internode distance of 300 mm to obtain a vascular strain rate of 0.35 / 300≈0.12%, and the corresponding elastic modulus of the corn is 1.8 GPa, and the vascular strain energy density is obtained according to “0.5*elastic modulus*vascular strain rate squared” as 1.296 kJ / m³ (here, only the calculation process of the vascular strain energy density is exemplified, and the actual calculation data needs to be determined according to the specific situation).

[0092] In step S23, the root anchoring decay rate is generated by analyzing the time series acceleration of the root projection area data and combining the fluctuation characteristics of the soil humidity data.

[0093] In some possible embodiments, assuming that the root projection area data of a certain winter wheat at the jointing stage for three consecutive days are 650 cm², 590 cm², and 540 cm², respectively, the corresponding time series acceleration is -10 cm² / day², the soil moisture data is used to obtain the soil water content, the soil water content sequence corresponding to the above root projection area data sequence is 35%, 28%, and 41%, the mean value of the soil water content sequence is 34.67%, and the variance is 28.22%, and the standard deviation is 5.31%; the time series acceleration is corrected based on the standard deviation of the soil water content sequence, for example, the time series acceleration obtained in the above example is -10 cm² / day², and the standard deviation of the soil water content sequence is 5.31%, a “1+0.3*standard deviation” acceleration compensation amount is assigned to the time series acceleration, and the corrected time series acceleration is -10*(1+0.3*0.0531)≈-10.16 cm² / day²; the corrected time series acceleration is mapped to the root anchoring decay rate according to the preset decay rate mapping function, for example, with a preset decay rate mapping function of 1% root anchoring decay rate corresponding to every -10 cm² / day², the corrected time series acceleration obtained in the above example is mapped to the root anchoring decay rate, that is, -10.16 / 10=-1.016% (here, only the calculation process of the root anchoring decay rate is exemplified, and the actual calculation data needs to be determined according to the specific situation).

[0094] In step S24, the transmission efficiency of the crop root and canopy growth signals is obtained by analyzing the root projection area data and the canopy LAI, and the root canopy response entropy is generated based on the transmission efficiency.

[0095] In some possible embodiments, assuming that the root projection area data of a certain indica rice for 5 consecutive days are 550 cm², 610 cm², 690 cm², 780 cm² and 760 cm² in sequence, and the corresponding canopy LAI in the time period are 3.8, 4.2, 4.5, 4.7 and 4.6 in sequence, based on the root projection area data sequence, four groups of effective root projection area data daily increment data are extracted, which are 60 cm², 80 cm², 90 cm² and -20 cm² in sequence, based on the canopy LAI sequence, four groups of effective canopy LAI daily increment data are extracted, which are 0.4, 0.3, 0.2 and -0.1 in sequence, based on the above four groups of effective root projection area data daily increment data and effective canopy LAI daily increment data, the corresponding root-shoot ratio sequence is obtained, which is 150 cm² / LAI, 267 cm² / LAI, 450 cm² / LAI and 200 cm² / LAI; the mean and standard deviation of the root-shoot ratio sequence are calculated, and the quotient of the standard deviation and the mean is taken as the dispersion coefficient of the root-shoot ratio sequence, that is, the dispersion coefficient corresponding to the above root-shoot ratio sequence is 118.6 / 266.75*100%≈44.5%; the obtained dispersion coefficient is compared with the preset dispersion threshold value, if the dispersion coefficient is lower than the preset dispersion threshold value, the root-shoot response entropy value is mapped according to the root-shoot response entropy value mapping function in the form of “0.5*(dispersion coefficient / 20%)”, if the dispersion coefficient is higher than the preset dispersion threshold value, the root-shoot response entropy value is mapped according to the root-shoot response entropy value mapping function in the form of “0.8+0.5*[(dispersion coefficient-40%) / 30%], that is, after the root-shoot response entropy value mapping of the above dispersion coefficient, the obtained root-shoot response entropy value is 0.875 (here, only the calculation process of the root-shoot response entropy value is exemplified, and the actual calculation data needs to be determined according to the specific situation).

[0096] In step S25, a photosynthetic phase space is constructed based on the leaf chlorophyll content data and the leaf texture density data, and the photosynthetic phase space compression degree is generated by analyzing the motion trajectory of the leaf chlorophyll content data and the leaf texture density data in the photosynthetic phase space.

[0097] In some possible embodiments, a two-dimensional plane coordinate system can be constructed with the leaf chlorophyll content as the horizontal axis and the leaf expansion degree obtained based on the leaf texture density data as the vertical axis, the leaf chlorophyll content data and the leaf texture density data of the same plant are collected every 2 hours from 6:00 to 18:00 every day, the collected data is mapped to the constructed two-dimensional plane coordinate system to form dynamic trajectory points, a continuous motion trajectory path is generated through cubic spline interpolation, the phase space area enclosed by the motion trajectory is calculated, and the ratio of the minimum phase space area to the maximum phase space area is taken as the photosynthetic phase space compression degree (here, only one possibility of the photosynthetic phase space compression degree is exemplified, and the actual situation needs to be determined according to the specific situation).

[0098] Step S26, constructing a crop structure index based on the vascular strain energy density and the root anchoring decay rate, and constructing a functional synergy index based on the root crown response entropy and the photosynthetic phase space compression degree, wherein the crop growth index includes the crop structure index and the functional synergy index.

[0099] In some possible embodiments, the quantified value of the crop structure index can be obtained in the form of "√(strain energy2+ decay rate2)", and when the quantified value of the crop structure index exceeds a preset structure imbalance threshold, it is determined that the crop has structure imbalance, and the physical strength and stability of the crop organs are monitored according to the crop structure index.

[0100] In some possible embodiments, the quantified value of the functional synergy index can be obtained in the form of "root crown response entropy*photosynthetic phase space compression degree", the photosynthesis efficiency of the crop is diagnosed through the quantified value of the functional synergy index, and the optimization of the organic carbon source injection strategy is driven, and the photosynthesis material transfer and energy conversion efficiency between crop organs are quantified according to the functional synergy index.

[0101] Step S27, feature extraction is performed on the crop environment data set to obtain environment feature data, and an environment pressure index is constructed based on the environment feature data.

[0102] Specifically, the environment feature data includes photosynthetic active radiation data, crown spectrum data, air temperature and humidity data, stem surface strain rate, soil oxidation-reduction potential data, and light intensity data.

[0103] In this embodiment, step S27 further includes the following steps:

[0104] Step S271, generating photosynthetic carbon disorder degree data by analyzing the response relationship between the photosynthetic active radiation data and the crown spectrum data.

[0105] In some possible embodiments, a photosynthetically active radiation data mutation event is identified, such as a sudden increase in photosynthetically active radiation data exceeding 200 pmol / m-2 / s-1 within 1 minute due to cloud movement, a response delay time of the red edge displacement amount of the canopy spectral data is captured, such as a red edge displacement amount of 15 nm of the canopy spectral data after a sudden increase in photosynthetically active radiation data, a mean value of a normal response delay time is obtained according to historical data analysis, for example, the mean value of the normal response delay time is 5 minutes; when the actual response delay time continuously exceeds the preset response delay threshold value, a ratio of a total duration of the accumulated response delay time to a total duration of the effective illumination period is obtained, for example, the total duration of the effective illumination period is 6 hours, and the total duration of the accumulated response delay time is 36 minutes, so the ratio is 0.1; a photo-carbon disorder degree data is obtained in the form of "final disorder degree = delay ratio * red edge displacement abnormal amplitude coefficient", for example, the photo-carbon disorder degree data increases by 0.2 of the red edge displacement abnormal amplitude coefficient for each 1 nm deviation from the theoretical value, in the above example, the delay ratio is 0.1, and the corresponding red edge displacement offset is 8 nm, so the corresponding photo-carbon disorder degree data is 0.1*(1+0.2*8)=0.26 (here, only the calculation process of the photo-carbon disorder degree data is exemplified, and the actual calculation data needs to be determined according to the specific situation).

[0106] In step S272, water damage coupling data is generated by analyzing the response relationship between air temperature and humidity data and stem surface strain rate.

[0107] In some possible embodiments, based on the air temperature and humidity data, the saturated vapor pressure difference VPD is obtained, the stem surface strain rate is compensated in real time for temperature drift, and the stem surface strain rate correction value is obtained, such as applying a "-0.0000125*air temperature mutation" to the stem surface strain rate, that is, when the air temperature suddenly rises by 8°C, 0.01% of the thermal expansion pseudo-strain is automatically deducted; when the VPD and the stem surface strain rate exceed the corresponding preset threshold value at the same time, the water damage coupling data is calculated, for example, a set of data is VPD=2.8 kPa, and the strain rate of the stem surface strain rate is 0.13% / min, since the VPD of the set of data is 2.8 kPa, which is greater than the preset VPD threshold value of 2.5 kPa, and the stem surface strain rate is 0.13% / min, which is greater than the preset stem surface strain rate threshold value of 0.1% / min, at this time, the water damage coupling data is calculated in the form of "(VPD-2.0)*stem surface strain rate*10" to obtain the water damage coupling data of (2.8-2.0)*0.13*10=1.04 (here, only the calculation process of the water damage coupling data is exemplified, and the actual calculation data needs to be determined according to the specific situation).

[0108] In step S273, marginal load data is generated by analyzing the soil oxidation-reduction potential data, the light intensity data, and the corresponding spatial position coordinates.

[0109] Furthermore, a two-dimensional planar coordinate system is constructed at the boundary, and a grid is generated.

[0110] For details, see attached. Figure 4 As shown, the coordinate origin is the intersection of multiple field ridges, the X-axis is the inner boundary line of the field ridges, and the Y-axis is the direction perpendicular to the field ridges. The field is divided into square grid units with a side length of 0.5 meters, and then... Figure 3 The system divides the field into core and edge zones as shown. Using the center point of each grid as a data collection point, soil redox potential, light intensity, and distance from the field ridge are collected simultaneously within the grid. The collected data undergoes preprocessing. The collected light intensity data is normalized by dividing the mean light intensity data of the core zone by the mean light intensity data of the core zone, yielding relative light intensity data. When the mean light intensity data of the core zone is greater than the light saturation point of the corresponding crop, nonlinear correction is applied to the relative light intensity data. For example, in a field planted with rice, the light intensity of the core zone... The average light intensity data is 1850 μmol. The light intensity data collected in a certain unit block is 1295 μmol. The calculated relative light intensity data is 1295 / 1850 = 0.7. Assuming the light saturation point of rice is 1800 μmol, and since the average light intensity data in the core area is greater than the corresponding crop's light saturation point, a correction factor of "1 - 0.2*(1850-1800) / 1000 = 0.99" is applied to the obtained relative light intensity. Therefore, the corrected relative light intensity data is 0.7 * 0.9. 9 = 0.693; The distance attenuation coefficient is obtained based on the spatial coordinates. For example, in the above example, the corresponding spatial coordinates are (0,2). From these coordinates, we know that the current data collection point is 1 meter away from the field ridge. According to "0.7 + 0.15 * distance from the field ridge", the corresponding distance attenuation coefficient is calculated to be 0.7 + 0.15 * 1 = 0.85; The marginal load data is obtained based on the soil redox potential data, relative light intensity data, and distance attenuation coefficient. For example, in the above example, the corresponding soil redox potential data, relative light intensity data, and distance attenuation coefficient are calculated to be 0.7 + 0.15 * 1 = 0.85. The distance attenuation coefficients are -120mV, 0.693, and 0.85, respectively. After normalizing the soil redox potential data to 0.2, the marginal load data is calculated in the form of "0.4 * soil redox potential data + 0.3 * relative light intensity data + 0.3 * distance attenuation coefficient". The marginal load data is 0.4 * 0.2 + 0.3 * 0.693 + 0.3 * 0.85 = 0.5429 (this is just an example of the calculation process for the marginal load data; the actual calculation data needs to be determined based on the specific situation).

[0111] Step S274: By analyzing soil redox potential data over a continuous period of time, environmental disturbance memory data is generated.

[0112] In some possible embodiments, assuming that the arithmetic mean of the soil oxidation-reduction potential data of the last 6 hours is taken as the initial historical memory value, after a new fused Eh value is collected every hour, a new memory value is generated according to the weight ratio of “historical memory value: current soil oxidation-reduction potential data = 8:2”; the deviation standard deviation between the current soil oxidation-reduction potential data and the new memory value is calculated, and the deviation standard deviation is taken as the environmental disturbance memory data.

[0113] In step S275, the metabolic abnormality environment coupling index is constructed based on the photo-carbon disorder degree data and the moisture damage coupling data, and the spatial heterogeneity index is constructed based on the marginal load data and the environmental disturbance memory data, and the environmental stress index includes the metabolic abnormality environment coupling index and the spatial heterogeneity index.

[0114] In some possible embodiments, the metabolic abnormality environment coupling index quantitative value can be obtained in the form of “√ photo-carbon disorder degree data * moisture damage coupling data”, and the crop metabolism is diagnosed according to the metabolic abnormality environment coupling index quantitative value.

[0115] In some possible embodiments, the spatial heterogeneity index quantitative value can be obtained in the form of “marginal load data / 10+0.2*environmental disturbance memory data”, and the generation of the organic carbon source regulation decision is optimized according to the spatial heterogeneity index quantitative value.

[0116] In step S3, the causal correlation chain between the crop growth index and the environmental stress index is mined, and the growth environment correlation rule library is constructed according to the causal correlation chain.

[0117] In this embodiment, step S3 includes:

[0118] In step S31, the causal entropy between the crop growth index and the environmental stress index in a preset time window length is obtained, and the time sequence coupling relationship between the crop growth index and the environmental stress index corresponding to different growth stages of the crop is quantified based on the causal entropy.

[0119] In some possible embodiments, the conduction efficiency of the functional synergy index is analyzed in probability for the change of the metabolic abnormality environment coupling index, for example, when the value of the metabolic abnormality environment coupling index calculated is greater than 0.6, the probability of deterioration of the root crown response entropy in the functional synergy index is increased to 73%, which means that the risk of the functional synergy ability of the crop will be expanded by 18.7% accordingly when the intensity of the environmental stress suffered by the crop is increased by 0.1 unit;

[0120] Assuming that the average causal entropy of the crop structural index and the environmental stress index is 0.28 at the tillering stage of the crop, the crop structural index and the environmental stress index at the growth stage present a weak coupling characteristic, and the environmental stress can only explain part of the crop structural change, so the corresponding decision generated for the crop at this stage is mainly to repair the carbon source of the root system; after entering the heading stage, the causal entropy of the functional synergy index and the metabolic abnormal environment coupling index jumps to 0.79, and the functional synergy index and the metabolic abnormal environment coupling index present a strong decoupling state, so the corresponding decision generated for the crop at this stage is mainly to optimize the carbon assimilation of the leaf surface of the crop.

[0121] Step S32, calculate the level transition causal gain value from the organ level to the system level, filter the macro-lag interference based on the level transition causal gain value.

[0122] Specifically, the organ level represents the structural and functional performance of each organ of the crop, the system level represents the comprehensive growth state of the crop, and the macro-lag interference represents the delay phenomenon of the organ and the system in response to environmental changes.

[0123] In some possible embodiments, when the organ layer index triggers the system layer response, such as the sudden drop of leaf stomatal conductance leading to the attenuation of photosynthetic rate, the response time difference of the organ layer index triggering the system layer response, the occurrence probability value of the organ layer index triggering the system layer response, and the occurrence probability value of the system layer response under natural conditions without organ layer index variation are obtained, and the causal gain value of the variation of the response time difference and the occurrence probability value is generated, such as the photosynthetic attenuation probability caused by stem damage is 80%, and the photosynthetic attenuation probability under natural conditions is 35%, and the corresponding causal gain value is 0.8-0.35=0.45; if the causal gain value is greater than the preset effective correlation strength threshold value, it is determined as an effective correlation conduction; if the response time difference is greater than the crop response time threshold value or the causal gain value is lower than the preset effective correlation strength threshold value, it is considered as a lag interference, such as the transient closure of stomata caused by short-term cloud layer, and the interference signals caused by environmental noise or physiological buffer, such as the interference signals caused by short-term vibration of stem caused by gust and short-term water shortage of root system, are shielded by actively shielding the mutation nodes of the dynamic tracking causal gain value.

[0124] Step S33, under the premise of keeping the environmental stress index unchanged, virtually switching the crop varieties, observing the change difference rate of the crop growth index, detecting the universality of the time sequence coupling relationship based on the change difference rate, and if the change difference rate reaches the preset difference rate threshold interval, it is determined that the time sequence coupling relationship has the variety universality.

[0125] In some possible embodiments, on the premise that the time series coupling relationship is an atomic rule involving root anchoring decay rate, and the recorded root anchoring decay rate of the stress-resistant variety crop is -4.8% and the root anchoring decay rate of the sensitive variety crop reaches -12.3%, the change difference rate is calculated as | -4.8% - -12.3% | / | -4.8% | = 156%, the change difference rate falls within the preset change difference rate threshold interval [80%, 200%], and it is determined that the current time series coupling relationship has variety universality.

[0126] In step S34, virtual data of different soil conditions in the same plot is synthesized by a generative adversarial network to separate the interference of soil properties on the time series coupling relationship.

[0127] A double-channel adversarial model is constructed to generate virtual data of different soil conditions in the same plot based on the time series data corresponding to the crop environment data set of the real plot; the distribution difference between the real data and the virtual data is compared, the statistical feature error of the generated virtual data and the real data is made lower than a preset error threshold through iterative adversarial training, and it is ensured that the virtual data meets the physiological rationality;

[0128] The virtual data is imported into the time series coupling relationship for verification. If a time series coupling relationship still maintains strong correlation after injecting virtual data for verification, it is determined that the time series coupling relationship is weakly interfered by soil conditions; otherwise, if the correlation collapses, it is marked that the relationship has soil-specific interference, the soil property interference term is analyzed through the difference between the virtual data and the real data, and the interference term is removed from the corresponding time series coupling relationship based on the soil property interference term, thereby improving the universality of the time series coupling relationship.

[0129] In step S35, a causal association chain between the crop growth index and the environmental stress index is generated based on the time series coupling relationship, and a growth environment association rule library is constructed based on the causal association chain.

[0130] Specifically, the causal association chain is subjected to rule extraction to obtain three types of growth environment association rules, i.e., atomic rules, combined rules, and spatiotemporal constraint rules, the support, confidence, and lift of the growth environment association rules are obtained, and the growth environment association rules that do not meet the preset support threshold, confidence threshold, and lift threshold are removed to construct the growth environment association rule library.

[0131] It can be understood that the atomic rule represents a single-index direct driving longness environment association rule, the combination rule represents a multi-index cooperative longness environment association rule, the space-time constraint rule represents a longness environment association rule associated with the growth stage of the crop and the geographical location, the support represents the frequency of the longness environment association rule in the historical data, the confidence represents the probability of the longness environment association rule successfully taking effect when the corresponding established condition is triggered, and the lift represents the performance gain rate of the rule prediction ability of the longness environment association rule relative to random guessing.

[0132] In step S4, an organic carbon source regulation model is constructed based on the crop growth index, the environmental stress index, and the longness environment association rule library, and an organic carbon source regulation decision is generated according to the organic carbon source regulation model.

[0133] Specifically, the physiological state of the crop is diagnosed in real time according to the crop growth index, the environmental impact on the growth of the crop is quantified through the environmental stress index, the longness environment association rule corresponding to the crop growth index and the environmental stress index in the longness environment association rule library is matched and called, and the organic carbon source regulation decision is generated based on the matched longness environment association rule to regulate the organic carbon source of the crop.

[0134] In some possible embodiments, assuming that the vascular strain energy density of the rice collected at a certain time is 0.22 kJ / m³ and the root anchoring decay rate is -7%, the quantized value of the corresponding crop structural index is √(0.22² +(-0.07)²) ≈0.23; the root crown response entropy is 0.75, and the photosynthetic phase space compression degree is 0.32, so the quantized value of the corresponding functional synergy index is 0.75*0.32=0.24; the light carbon disorder degree data is 0.29, and the water damage coupling data is 1.3, so the quantized value of the corresponding metabolic abnormal environment coupling index is √(0.29*1.3) ≈0.61; the marginal load data is 0.43, and the environmental disturbance memory data is 9.1, so the quantized value of the corresponding spatial heterogeneity index is (0.43 / 10)+0.2*9.1=1.863; the quantized values corresponding to each index are compared with the corresponding preset threshold, the priorities are divided according to the proportion of the indexes exceeding the preset threshold, and the weights of each priority index are allocated; in this example, the priorities from high to low are the functional synergy index, the crop structural index, the metabolic abnormal environment coupling index, and the spatial heterogeneity index, the organic carbon source regulation model matches the corresponding rules from the longness environment association rule library, generates the corresponding regulation decisions for each index, and fuses the decisions according to the weights of each index. When the decisions corresponding to each index conflict, the decision corresponding to the index with a higher weight is preferentially retained (here, only the process of generating the organic carbon source regulation decision by the organic carbon source regulation model according to the crop growth index and the environmental stress index is exemplified, and the actual process needs to be determined according to the specific situation).

[0135] Step S5, obtain the effect of the organic carbon source regulation decision implementation, and iteratively optimize the organic carbon source regulation rule base and the organic carbon source regulation model according to the effect of the organic carbon source regulation decision implementation.

[0136] Specifically, the organic carbon source regulation decision that does not improve the growth index of the target crop within a predetermined time is determined as a failed decision, and the growth environment association rule corresponding to the failed decision is marked as a gradually ineffective rule, and the confidence of the gradually ineffective rule is reduced. The growth environment association rule marked as a gradually ineffective rule for three times in succession is determined as an ineffective rule, and a freezing operation is performed on the ineffective rule. The organic carbon source regulation decision that improves the growth index of the target crop within a predetermined time is determined as a successful decision, and the growth environment association rule corresponding to the successful decision is marked as an effective rule, and the confidence of the effective rule is increased.

[0137] In some possible embodiments, the weight of the index corresponding to the gradually ineffective rule is increased according to the contribution rate of the crop growth stage, and if the failed decision does not change to a successful decision within three days after the adjustment, the index weight is automatically rolled back, and iterative adjustment is performed again until the failed decision changes to a successful decision or the gradually ineffective rule is determined as an ineffective rule. For a scene where the decision misjudgment rate exceeds 20%, the preset threshold values of the indexes corresponding to the climate characteristics of the region where the crop is located are adjusted, and the adjusted threshold values of the indexes need to reduce the misjudgment rate to below 8% within 7 days, otherwise the original index threshold values are restored, and iterative adjustment is performed again until the misjudgment rate is reduced to below 8%.

[0138] The embodiment of the present application also provides an organic carbon source data optimization system based on crop growth feedback, which is suitable for the field of agricultural technology and includes:

[0139] A data processing module, which is configured to collect data of crops, pre-process the collected data, and generate a crop growth data set and a crop environment data set;

[0140] An index generation module, which is configured to generate corresponding crop growth indexes and environmental stress indexes according to the crop growth data set and the crop environment data set;

[0141] A rule extraction module, which is configured to extract a causal association chain between the crop growth indexes and the environmental stress indexes, and generate corresponding rules according to the causal association chain;

[0142] A model construction module, which is configured to construct an organic carbon source regulation model according to the crop growth indexes, the environmental stress indexes and a growth environment association rule base, and generate an organic carbon source regulation decision;

[0143] A feedback optimization module is configured to obtain the implementation effect of the organic carbon source regulation decision, and to optimize and adjust the organic carbon source regulation rule library and the organic carbon source regulation model according to the implementation effect.

[0144] The specific use mode and role of the embodiment are as follows:

[0145] First, the crop is subjected to data acquisition and data preprocessing to obtain a crop growth data set and a crop environment data set. By means of data preprocessing for the collected various data, interference noise and data deviation generated in the collection process are eliminated, the accuracy of the data is improved, and the data support of high accuracy is provided for subsequent generation of quantitative indicators.

[0146] Then, crop growth indicators and environmental stress indicators are obtained based on the crop growth data set and the crop environment data set, the causal correlation chain between the crop growth indicators and the environmental stress indicators is mined, and a growth environment correlation rule library is constructed according to the causal correlation chain. By constructing the crop growth indicators and the environmental stress indicators, the discrete environmental data and crop physiological data are accurately quantified, the causal coupling relationship between the environmental data and the crop physiological data is analyzed by mining the causal correlation chain between the indicators, the corresponding quantitative rules are obtained, and data support and decision reference benchmarks are provided for subsequent model generation of organic carbon source regulation decisions.

[0147] Finally, an organic carbon source regulation model is constructed based on the crop growth indicators, the environmental stress indicators and the growth environment correlation rule library, and an organic carbon source regulation decision is generated according to the organic carbon source regulation model. The implementation effect of the organic carbon source regulation decision is obtained, the organic carbon source regulation rule library and the organic carbon source regulation model are iteratively optimized according to the implementation effect of the organic carbon source regulation decision, the confidence of the efficient rules in the organic carbon source regulation rule library is improved, and the confidence of the inefficient rules is reduced by monitoring the implementation effect of the organic carbon source regulation decision, so as to improve the accuracy of the organic carbon source regulation scheme.

[0148] In addition, the embodiment of the present application also provides an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method in the above-mentioned embodiment one.

[0149] The various constituent components of the electronic device will be specifically introduced as follows:

[0150] The processor is the control center of the electronic device, and can be one processor or a combination of multiple processing elements. For example, the processor is one or more central processing units (CPUs), application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement an embodiment of the present application, such as one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0151] The processor can execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0152] The memory is used to store software programs for implementing the present application, and is controlled by the processor to execute. The specific implementation manner can refer to the above method embodiments, and will not be described here.

[0153] The memory can be a real-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only (CD-ROM), or other optical disk storage, optical disk storage (including compact disks, laser disks, optical disks, digital versatile disks, Blu-ray disks, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory can be integrated with the processor or exist independently and coupled to the processor through the interface circuit of the electronic device, and the present application is not limited in this regard.

[0154] The above-described embodiments can be implemented in whole or in part by software, hardware (such as a circuit), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server or data center to another website, computer, server or data center by limited (for example, infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0155] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B, and the existence of B alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after are an "or" relationship, but can also represent an "and / or" relationship, which can be understood according to the context before and after.

[0156] It should be understood that in the embodiments of the present application, the size of the sequence number of the above-mentioned processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0157] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for optimizing organic carbon source data based on crop growth feedback, characterized in that, The method includes: S1: Collect and preprocess data on crops to obtain crop growth datasets and crop environment datasets; S2: Obtain crop growth indicators and environmental stress indicators based on crop growth dataset and crop environment dataset, respectively. S3: Discover the causal relationship between crop growth indicators and environmental stress indicators, and construct a growth-environment association rule base based on the causal relationship. The mining of the causal relationship chain is represented by obtaining the causal entropy between crop growth indicators and environmental pressure indicators within a preset time window, and quantifying the temporal coupling relationship between crop growth indicators and environmental pressure indicators at different growth stages based on the causal entropy. Calculate the hierarchical transition causal gain value from the organ level to the system level, and filter macroscopic hysteresis interference based on the hierarchical transition causal gain value; Among them, the organ level represents the structural and functional performance of each organ of the crop, the system level represents the overall growth status of the crop, and the macroscopic lag interference represents the delayed phenomenon of organs and systems responding to environmental changes. While keeping environmental stress indicators constant, crop varieties are virtually switched, and the rate of change in crop growth indicators is observed. The universality of the temporal coupling relationship is detected based on the rate of change. If the rate of change reaches the preset threshold range, the temporal coupling relationship is determined to have variety universality. By synthesizing virtual data of different soil conditions in the same plot using generative adversarial networks, the interference of soil properties on temporal coupling relationships can be separated. Based on the temporal coupling relationship, a causal relationship chain between crop growth indicators and environmental stress indicators is generated, and a growth-environment association rule base is constructed based on the causal relationship chain. The construction of the growth environment association rule base based on the causal association chain means extracting rules from the causal association chain to obtain three types of growth environment association rules: atomic rules, combined rules, and spatiotemporal constraint rules. The support, confidence, and lift of the growth environment association rules are obtained, and growth environment association rules that do not meet the preset support threshold, confidence threshold, and lift threshold are removed to construct the growth environment association rule base. S4: Construct an organic carbon source regulation model based on crop growth indicators, environmental stress indicators and a rule base for the association between growth and environment, and generate organic carbon source regulation decisions based on the organic carbon source regulation model. S5: Obtain the implementation effect of organic carbon source regulation decisions, and iteratively optimize the organic carbon source regulation rule base and organic carbon source regulation model based on the implementation effect of organic carbon source regulation decisions.

2. The method for optimizing organic carbon source data based on crop growth feedback according to claim 1, characterized in that, Crop growth indicators and environmental stress indicators were obtained based on crop growth datasets and crop environment datasets, respectively, including: Feature extraction was performed on the crop growth dataset to obtain stem image data, root projection area data, soil moisture data, canopy LAI, leaf chlorophyll content data, and leaf texture density data. By analyzing the displacement changes of three consecutive frames of stem image data at a fixed time interval, the strain energy distribution inside the stem tissue is obtained, and the vascular strain energy density is generated based on the strain energy distribution inside the tissue. By analyzing the temporal change acceleration of root projection area data and combining it with the fluctuation characteristics of soil moisture data, the root anchorage attenuation rate is generated. By analyzing root projection area data and canopy LAI, the transmission efficiency of crop root and canopy growth signals is obtained, and root-canopy response entropy is generated based on the transmission efficiency. A photosynthetic phase space is constructed based on leaf chlorophyll content data and leaf texture density data. The compressibility of the photosynthetic phase space is generated by analyzing the movement trajectory of leaf chlorophyll content data and leaf texture density data in the photosynthetic phase space. Crop structural indicators are constructed based on vascular strain energy density and root anchorage attenuation rate, and functional synergy indicators are constructed based on root-crown response entropy and photosynthetic phase spatial compressibility. The crop growth indicators include crop structural indicators and functional synergy indicators. Feature extraction is performed on crop environmental datasets to obtain environmental feature data, and environmental stress indicators are constructed based on the environmental feature data.

3. The method for optimizing organic carbon source data based on crop growth feedback according to claim 2, characterized in that, Feature extraction is performed on the crop environment dataset to obtain environmental feature data. Based on this environmental feature data, environmental stress indicators are constructed, including: Environmental characteristic data include photosynthetically active radiation data, canopy spectral data, air temperature and humidity data, stem surface strain rate, soil redox potential data, and light intensity data; Photocarbon disorder data were generated by analyzing the response relationship between photosynthetically active radiation data and canopy spectral data. By analyzing the response relationship between air temperature and humidity data and stem surface strain rate, moisture damage coupled data are generated. Marginal load data are generated by analyzing soil redox potential data, light intensity data, and corresponding spatial coordinates. By analyzing soil redox potential data over a continuous period of time, environmental disturbance memory data is generated. A metabolic abnormality environment coupling index is constructed based on photocarbon disorder data and water damage coupling data, and a spatial heterogeneity index is constructed based on marginal load data and environmental disturbance memory data. The environmental pressure index includes the metabolic abnormality environment coupling index and the spatial heterogeneity index.

4. The method for optimizing organic carbon source data based on crop growth feedback according to claim 1, characterized in that, An organic carbon source regulation model is constructed based on crop growth indicators, environmental stress indicators, and a rule base relating crop growth and environment. Based on this model, organic carbon source regulation decisions are generated, including: The physiological state of crops is diagnosed in real time based on crop growth indicators, and the environmental impact on crop growth is quantified through environmental stress indicators. The growth environment association rules corresponding to crop growth indicators and environmental stress indicators in the growth environment association rule library are matched and called. Based on the matched growth environment association rules, organic carbon source regulation decisions are generated to regulate the organic carbon source of crops.

5. The method for optimizing organic carbon source data based on crop growth feedback according to claim 1, characterized in that, To obtain the implementation effect of organic carbon source regulation decisions, and to iteratively optimize the organic carbon source regulation rule base and organic carbon source regulation model based on the implementation effect, including: Organic carbon source regulation decisions that fail to improve the growth indicators of target crops within the predetermined time are judged as failed decisions. At the same time, the growth environment association rules corresponding to the failed decisions are marked as gradually failing rules, and the confidence of the gradually failing rules is reduced. The growth environment association rule that has been marked as a gradually failing rule three times in a row will be determined as a failing rule and the failing rule will be frozen. The organic carbon source regulation decision that improves the growth indicators of the target crop within the predetermined time is judged as a successful decision. At the same time, the growth environment association rule corresponding to the successful decision is marked as a valid rule, and the confidence of the valid rule is increased.

6. The method for optimizing organic carbon source data based on crop growth feedback according to claim 1, characterized in that, Data collection and preprocessing of crops were performed to obtain crop growth datasets and crop environment datasets, including: The collected crop data is preprocessed with at least outlier filtering, spatiotemporal alignment, data standardization and data normalization to obtain crop growth datasets and crop environment datasets. The crop growth dataset is used to describe the multi-dimensional dynamic growth characteristics of crops, and includes at least crop canopy dimension data, stem dimension data, underground root system dimension data, and physiological dimension data. The crop environment dataset is used to describe the growth environment of crops and includes at least surface environment data, soil layer environment data, and microenvironment data.

7. The method for optimizing organic carbon source data based on crop growth feedback according to claim 6, characterized in that, The collected crop data undergoes data preprocessing, including at least outlier filtering, spatiotemporal alignment, data standardization, and data normalization, including: The outlier filtering refers to filtering outliers and imputing missing values ​​in crop data. The spatiotemporal alignment refers to unifying crop data to the same geographic coordinate reference and time reference. The data standardization refers to the data standardization operation performed on the crop growth data contained in the crop data. The data normalization refers to the data normalization operation performed on the crop environment data contained in the crop data.

8. An organic carbon source data optimization system based on crop growth feedback, used to implement the method described in any one of claims 1 to 7, characterized in that, include: The data processing module is used to collect data on crops, preprocess the collected data, and generate crop growth datasets and crop environment datasets. The indicator generation module is used to generate corresponding crop growth indicators and environmental stress indicators based on the crop growth dataset and the crop environment dataset. The rule extraction module is used to extract the causal relationship chain between crop growth indicators and environmental stress indicators, and generate corresponding rules based on the causal relationship chain. The model building module is used to build an organic carbon source regulation model based on crop growth indicators, environmental pressure indicators and a growth environment association rule base, and generate organic carbon source regulation decisions. The feedback optimization module is used to obtain the implementation effect of organic carbon source regulation decision and optimize and adjust the organic carbon source regulation rule base and organic carbon source regulation model according to the implementation effect.

Images