Shiny-leaved yellowhorn cultivation regulation and control method, device and equipment based on digital twinning and medium

Abstract

The invention relates to a shiny-leaved yellowhorn cultivation regulation and control method, device and equipment based on digital twinning and a medium. The method comprises the steps that spectral signals of shiny-leaved yellowhorn plants are collected in real time through an optical fiber sensor, and a nutrient prediction model driven by machine learning is built in combination with historical data; generating an initial fertilization curve based on the current spectral signal and a preset fertilization function database, and driving the intelligent sprayer to execute precise fertilization; furthermore, a digital twinning technology is utilized to fuse and update a spectral signal and a shiny-leaved yellowhorn nutrient absorption model, the influence of different fertilization strategies on nutrient absorption is dynamically simulated, the element ratio is optimized, and a closed-loop regulation and control scheme is generated. According to the method, the problems of resource waste and yield fluctuation caused by fertilization decision lag, model stiffness and element antagonism effect in the traditional technology can be solved through real-time sensing of the plant nutrient state, dynamic adjustment of the fertilization strategy and multi-element collaborative optimization, and the precision and ecological economic benefits of shiny-leaved yellowhorn cultivation are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent plant cultivation regulation and control, and specifically to a Xanthoceras sorbifolia cultivation regulation and control method, device, equipment and medium based on digital twins. Background Art

[0002] Xanthoceras sorbifolia is a high-value-added economic tree unique to my country. Its fruit has a high oil content and great biomass energy potential, and it has important strategic value in the fields of ecological restoration and energy development. During the cultivation process, the precise regulation of key nutrients such as nitrogen, phosphorus, and potassium directly affects the growth rate, fruit yield, and quality of the plants. Traditional cultivation management mainly relies on manual experience judgment and regular sampling and testing, and adjusts the fertilization plan based on laboratory analysis results. However, this extensive management makes it difficult to dynamically capture the real-time changes in the growth status of the plants, and it is even more difficult to adapt to the differentiated nutrient requirements of different growth stages such as the seedling stage, rapid growth period, and maturity period.

[0003] In the existing technology, nutrient management methods have significant limitations: first, manual sampling and laboratory analysis have long cycles and high costs, resulting in delayed fertilization decisions, prone to problems of over-fertilization or insufficient nutrients, causing resource waste or reduced yields; second, static empirical models are difficult to predict complex environmental factors, such as soil moisture, climate change, and the dynamic relationship between the physiological state of the plant, resulting in a low match between fertilization plans and actual needs; third, excessive application of a single nutrient may cause antagonistic effects between elements, such as excessive potassium inhibiting magnesium absorption, and traditional methods lack the ability to synergistically optimize multiple elements, exacerbating the risk of nutrient imbalance. The above problems lead to low efficiency in Xanthoceras sorbifolia cultivation and large fluctuations in fruit quality, which seriously restrict the economic benefits and sustainable development of large-scale planting. Summary of the invention

[0004] Based on this, the purpose of the present invention is to provide a digital twin-based Xanthoceras sorbifolia cultivation regulation method, device, equipment and medium that can sense the nutrient status of the plant in real time, dynamically optimize the fertilization strategy and avoid the antagonistic effect of elements.

[0005] The purpose of the present invention is achieved by the following scheme: In a first aspect, the present invention provides a method for regulating and cultivating Xanthoceras sorbifolia based on digital twins, comprising the following steps: S1: Collect and preprocess the spectral signals of Xanthoceras sorbifolia plants based on optical fiber sensors to generate spectral data sets; S2: Based on machine learning technology, feature extraction and model training are performed on pre-input historical spectral data and spectral data sets to build a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations; S3: Based on the preset fertilization function database, the spectral signal of the current growth stage in the spectral data set is input into the nutrient prediction model for processing to generate an initial fertilization curve; S4: Processing the initial fertilization curve, generating an initial fertilization instruction, and sending the initial fertilization instruction to the intelligent sprayer, where the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation; S5: The initial fertilization curve is processed based on the feedback-based updated spectral signal set and digital twin technology, the Xanthoceras sorbifolia nutrient absorption model is updated and a nutrient regulation plan is generated, which includes the amount of fertilizer and daily spraying time points.

[0006] In one embodiment, S5 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S51: based on a preset time interval, re-collecting and pre-processing the spectral signals of the Xanthoceras sorbifolia plants to obtain an updated spectral signal set; S52: Based on the spectral analysis technology, the updated spectral signal set and the spectral data set are differentially processed to generate a nutrient concentration deviation vector. The calculation formula of the nutrient concentration deviation vector is: ; in, is the nutrient concentration deviation vector, , , They are the nitrogen, phosphorus and potassium concentrations of the plants in the updated spectral signal set, 、 、 are the nitrogen, phosphorus and potassium concentrations of the plants in the spectral signal concentration, is the time difference between the two acquisitions; S53: Based on the digital twin technology and virtual simulation engine, the nutrient concentration deviation vector is processed, and the nutrient concentration deviation vector is data-fused with the preset Xanthoceras sorbifolia nutrient absorption model to simulate the impact of different fertilization strategies on nutrient absorption and update the Xanthoceras sorbifolia nutrient absorption model; S54: Input the initial fertilization curve into the updated Xanthoceras sorbifolia nutrient absorption model for parameter optimization to generate a nutrient regulation plan.

[0007] In one embodiment, S53 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S531: Perform data fusion processing on the nutrient concentration deviation vector and the initial fertilization curve to generate a fused data set. The calculation formula of the fused data set is: ; in, To fuse the dataset, is the current output data of the Xanthoceras sorbifolia nutrient absorption model, i.e., the initial fertilization curve. is the preset fusion weight coefficient; S532: Based on the fused data set, the nutrient absorption efficiency under different fertilization strategies is simulated by a virtual simulation engine to generate a simulation result data set. The calculation formula of the simulation result data set is: ; in, is the simulation result data set, is the fertilizer input for the fused dataset, is the simulated current soil nutrient concentration, 、 、 are the model parameters of Xanthoceras sorbifolia nutrient absorption model, e is a natural constant, i.e. the base of the natural logarithm function; S533: updating the model parameters of the Xanthoceras sorbifolia nutrient absorption model according to the simulation result data set, and generating an updated Xanthoceras sorbifolia nutrient absorption model.

[0008] In one embodiment, S54 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S541: Dynamically adjust the weight of the initial fertilization curve to generate an optimized weight coefficient. The calculation formula of the optimized weight coefficient is: ; in, is the optimization weight coefficient of nitrogen, phosphorus and potassium, is the historical variance of nutrient concentration deviation; S542: Input the optimized weight coefficient into the updated Xanthoceras sorbifolia nutrient absorption model for processing to generate a dynamic fertilization amount. The calculation formula of the dynamic fertilization amount is: ; in, is the dynamic fertilization amount, is the baseline fertilizer amount in the initial fertilization curve, is the nutrient concentration deviation, is the dynamic adjustment coefficient; S543: The dynamic fertilizer application amount is processed based on fuzzy control technology, and a nutrient regulation plan is generated according to the relationship between the dynamic fertilizer application amount and the preset safety threshold, combined with the fertilizer requirement law of the Xanthoceras sorbifolia growth stage.

[0009] In one embodiment, S1 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S11: The spectral signals of the leaves and roots of Xanthoceras sorbifolia are collected by embedded optical fiber sensors. The spectral signals cover the growth stages of the Xanthoceras sorbifolia plants in the seedling stage, rapid growth stage and mature stage. S12: Perform wavelet denoising on the spectral signal to remove high-frequency noise components and generate a denoised data set; S13: normalize the denoised data set to generate a spectral data set.

[0010] In one embodiment, S2 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S21: performing principal component analysis on the spectral data set and the pre-input historical spectral data, retaining characteristic bands whose cumulative variance contribution rate is greater than a preset value, and generating a dimension reduction feature matrix; S22: Associating the dimension-reduced feature matrix with a pre-built nutrient concentration database, where the nutrient concentration database includes nitrogen, phosphorus, and potassium concentration data, to generate a training data set; S23: The random forest regression model is trained based on the training data set to construct a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations.

[0011] In one embodiment, S3 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S31: inputting the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing to generate real-time concentration values ​​of nitrogen, phosphorus and potassium; S32: performing difference calculation processing on the real-time concentration value and the preset concentration threshold value to generate a deviation vector including nitrogen deviation, phosphorus deviation and potassium deviation; S33: querying a pre-built fertilization function database according to the deviation vector to generate an initial fertilization curve, wherein the fertilization function database is used to store a linear mapping relationship between nitrogen, phosphorus, and potassium fertilization amounts and deviations.

[0012] In a second aspect, the present invention provides a Xanthoceras sorbifolia cultivation and regulation device based on digital twins, which is configured with the following modules: A data acquisition and processing module is used to collect and preprocess the spectral signals of Xanthoceras sorbifolia plants based on the optical fiber sensor to generate a spectral data set; A prediction model building module is used to extract features and train models based on pre-input historical spectral data and spectral data sets based on machine learning technology to build a nutrient prediction model for predicting nitrogen, phosphorus, and potassium concentrations; A fertilization curve generation module is used to input the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing based on a preset fertilization function database to generate an initial fertilization curve; An instruction generation and sending module is used to process the initial fertilization curve, generate an initial fertilization instruction, and send the initial fertilization instruction to the intelligent sprayer, where the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation; The nutrient control plan generation module is used to process the initial fertilization curve based on the feedback-based updated spectral signal set and digital twin technology, update the Xanthoceras sorbifolia nutrient absorption model and generate a nutrient control plan, which includes the amount of fertilizer and daily spraying time points.

[0013] In a third aspect, the present application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, any one of the above-mentioned digital twin-based Xanthoceras sorbifolia cultivation and regulation methods is implemented.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned digital twin-based Xanthoceras sorbifolia cultivation and regulation methods.

[0015] In summary, the digital twin-based Xanthoceras sorbifolia cultivation and regulation method provided by the present invention collects spectral signals in real time through optical fiber sensors, constructs a nutrient prediction model through preprocessing and machine learning technology, and can accurately predict the nutrient requirements of plants; and generates an initial fertilization curve in combination with a fertilization function database, and realizes precise fertilization through an intelligent sprayer; further utilizes the feedback spectral signal and digital twin technology to update the nutrient absorption model, optimizes the nutrient regulation scheme, and can realize dynamic monitoring and precise regulation of the entire growth process of Xanthoceras sorbifolia plants. The method can effectively solve the problems of delayed fertilization decision-making, waste of resources, and decreased yield caused by artificial experience judgment and regular sampling and detection in traditional cultivation management, improve the cultivation efficiency of Xanthoceras sorbifolia, stabilize the fruit quality, and provide a strong guarantee for the economic benefits and sustainable development of large-scale planting. In addition, the method can avoid the antagonistic effect between multiple elements, optimize the synergistic effect of nutrients such as nitrogen, phosphorus, and potassium, reduce the risk of excessive fertilization or insufficient nutrients, reduce environmental pollution and waste of resources, and provide an efficient, accurate, and sustainable solution for the large-scale planting of Xanthoceras sorbifolia.

[0016] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a process flow of a Xanthoceras sorbifolia cultivation and regulation method based on digital twins provided in an embodiment of the present application; Figure 2 A schematic diagram of a process for generating a nutrient regulation scheme provided in an embodiment of the present application; Figure 3A schematic structural diagram of a Xanthoceras sorbifolia cultivation and regulation device based on digital twin is provided in another embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. The preferred embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the invention more thorough and comprehensive.

[0019] Unless otherwise defined, all technical terms and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0020] In one embodiment, Figure 1 As shown, a method for regulating and controlling Xanthoceras sorbifolia cultivation based on digital twins is provided. This embodiment takes the method applied to a terminal as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps: S1: The spectral signals of Xanthoceras sorbifolia plants are collected and preprocessed based on the optical fiber sensor to generate a spectral dataset.

[0021] Specifically, the system uses high-precision fiber optic sensors to collect spectral signals of Xanthoceras sorbifolia plants in real time. The sensors are installed near the leaves, stems and roots of the plants to ensure that the collected spectral signals can fully reflect the physiological state of the plants. The fiber optic sensor uses spectral acquisition technology in a specific wavelength range to generate raw data containing the plant's reflection spectrum, transmission spectrum and absorption spectrum. The acquisition frequency of the spectral signal can be dynamically adjusted according to the growth stage and environmental conditions of the plant to ensure the timeliness and accuracy of the data. Preferably, the wavelength range of the fiber optic sensor can be 400nm to 1100nm.

[0022] The collected raw spectral signals undergo preprocessing steps, including noise filtering, baseline correction, spectral normalization, and data dimension reduction. The preprocessing process uses advanced signal processing algorithms, such as wavelet transform to remove high-frequency noise and multivariate scattering correction to eliminate interference caused by optical path differences. Through these technologies, the system can effectively improve the quality of spectral data and provide a reliable data foundation for subsequent model training.

[0023] The core of fiber optic sensor technology lies in its high sensitivity and wide wavelength coverage, which can capture subtle spectral changes of plants at different growth stages. Changes in spectral reflectance are closely related to the physiological state of the plant. For example, changes in chlorophyll content in leaves will directly affect the reflectance of a specific band. By collecting and processing spectral signals in real time, the system can dynamically perceive the nutrient needs of the plant and provide data support for precise fertilization.

[0024] S2: Based on machine learning technology, feature extraction and model training are performed on the pre-input historical spectral data and spectral data sets to build a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations.

[0025] Specifically, the system merges the previously collected historical spectral data with the currently generated spectral data set to form a training data set. These historical data contain spectral features at different growth stages and under different environmental conditions and the corresponding measured values ​​of nitrogen, phosphorus, and potassium concentrations. Through supervised learning algorithms (such as support vector machines, random forests, or deep neural networks), the system extracts features from the data set and trains the model.

[0026] The feature extraction process uses principal component analysis (PCA) or independent component analysis (ICA) to identify key feature dimensions in spectral data, and combines domain knowledge to extract the spectral absorption peak position, intensity, and shape features that are highly correlated with nutrient concentration. The machine learning model can accurately predict the nutrient concentration of the current spectral signal by learning the mapping relationship between these features and nutrient concentration.

[0027] The core of machine learning technology lies in its powerful pattern recognition ability, which can extract characteristic patterns related to nutrient concentration from complex spectral data. Support vector machines achieve classification or regression by finding the optimal hyperplane, random forests improve the generalization ability of the model by integrating multiple decision trees, and deep neural networks capture nonlinear relationships in data through multi-layer neuron structures. Through cross-validation strategies, the system ensures the stability and accuracy of the model on different data sets, and the prediction accuracy reaches the preset threshold.

[0028] S3: Based on the preset fertilization function database, the spectral signal of the current growth stage in the spectral data set is input into the nutrient prediction model for processing to generate an initial fertilization curve.

[0029] Specifically, the fertilization function database stores fertilization parameters for different growth stages of Xanthoceras sorbifolia, including nitrogen, phosphorus and potassium concentration thresholds, fertilization frequency and upper limit of fertilization amount. These parameters are based on long-term field tests and data analysis, and can accurately match the nutrient requirements of plants at different stages. The system generates the optimal fertilization curve based on the growth stage corresponding to the current spectral signal and the predicted nutrient concentration through dynamic programming algorithms or genetic algorithms.

[0030] The generation process of the fertilization curve takes into account factors such as fertilization efficiency, environmental impact, and plant health to ensure the scientificity and feasibility of the fertilization strategy. After the fertilization curve is generated, the system further converts it into specific fertilization instructions, including daily fertilization amount, fertilization time point, and spraying mode. The dynamic update mechanism of the fertilization function database can adjust fertilization parameters in real time according to environmental changes (such as soil moisture and climate conditions) to ensure the adaptability and effectiveness of the fertilization strategy. The application of dynamic programming algorithms in optimization problems has significant advantages. It can achieve the global optimal solution through staged decision-making, providing reliable theoretical support for the optimization of fertilization strategies.

[0031] S4: Processing the initial fertilization curve, generating an initial fertilization instruction, and sending the initial fertilization instruction to the intelligent sprayer, where the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation.

[0032] Specifically, the initial fertilization curve is further processed and converted into executable fertilization instructions. The processing process includes discretization of fertilizer amount, optimization of spraying time points, and adjustment of element ratios. The system determines the optimal time window for daily spraying (such as early morning or evening) based on the transpiration law of Xanthoceras sorbifolia plants and the soil moisture retention capacity to ensure maximum absorption efficiency of fertilizers.

[0033] The generated initial fertilization instruction is sent to the smart sprayer through a wireless communication module (such as LoRa, NB-IoT). After receiving the instruction, the smart sprayer controls the flow rate and spraying time of different fertilizer solutions through a high-precision solenoid valve to achieve precise fertilization operation. The working parameters of the sprayer (such as spray pressure, flow rate, coverage) can be dynamically adjusted according to plant density and weather conditions to ensure the accuracy and adaptability of fertilization operation.

[0034] Smart sprayer is a kind of equipment that uses advanced technology to realize automatic spraying. Its technical principles mainly include the following aspects. First, the smart sprayer monitors environmental information in real time through sensors, such as temperature, humidity, light, wind speed, etc. These sensors can collect environmental data and transmit it to the controller for processing. The controller analyzes the data transmitted by the sensor according to the preset parameters and algorithms, and makes decisions, such as adjusting the time, dosage and range of spraying, to achieve efficient and accurate spraying operations. Secondly, the spray device of the smart sprayer can convert the liquid into fine particles or mist according to the instructions of the controller, and spray it through the nozzle to achieve a uniform spray effect. In the field of agriculture, the smart sprayer can be combined with GPS and data acquisition systems to automatically adjust the spraying amount and spraying method according to the growth status of crops and the situation of pests and diseases, so as to improve the efficiency of pesticide use. In addition, the smart sprayer can also have remote control and intelligent management functions. The operator can remotely control the equipment through wireless transmission technology to achieve real-time monitoring, automatic adjustment and other functions.

[0035] The system dynamically adjusts the spraying rate and nozzle angle according to the amount of fertilizer and time point in the initial fertilization instruction to complete the fertilization operation. The control logic of the sprayer is based on a closed-loop feedback mechanism, which monitors the fertilization process in real time through flow sensors and pressure sensors to ensure that the amount of fertilizer is consistent with the instruction. The closed-loop feedback mechanism eliminates errors in the fertilization process through real-time monitoring and adjustment, and improves the accuracy and stability of fertilization. The multi-mode spraying function of the smart sprayer can adapt to the fertilization needs under different growth stages and environmental conditions. For example, atomization spraying is used in the rapid growth period to improve fertilizer utilization, and precision drip irrigation is used in the maturity period to reduce nutrient loss. The application of smart sprayer technology in precision agriculture has significantly improved fertilization efficiency, reduced resource waste, and provided important support for the sustainable development of agricultural production.

[0036] Preferably, the wireless communication technology can adopt LoRa (Long Range Radio) technology or NB-IoT (Narrow Band - Internet of Things) technology to ensure that the fertilization instructions can be transmitted to the smart sprayer in real time and reliably. LoRa technology is suitable for device communication in farmland environment with its long-distance transmission capability and low power consumption; while NB-IoT ensures the stable operation of the system in complex terrain with its high connection density and deep coverage capability. Through these technologies, the system can realize remote control of the smart sprayer and improve the automation level of fertilization operation.

[0037] S5: The initial fertilization curve is processed based on the feedback-based updated spectral signal set and digital twin technology, the Xanthoceras sorbifolia nutrient absorption model is updated and a nutrient regulation plan is generated, which includes the amount of fertilizer and daily spraying time points.

[0038] Specifically, after the fertilization operation is performed, the system continuously collects updated spectral signals through optical fiber sensors to form a feedback spectral signal set. These feedback signals are used to verify the effectiveness of the initial fertilization curve and serve as input data for the digital twin model. The digital twin technology constructs a virtual model of the Xanthoceras sorbifolia plant, which includes the plant's physiological structure, nutrient absorption dynamics model, and environmental response model.

[0039] By comparing the feedback spectral signal with the virtual model in real time, the system can use Kalman filtering or particle filtering algorithms to update the nutrient absorption model online. The updated model can more accurately reflect the nutrient absorption characteristics of the plant under the current environmental conditions. Based on the updated model, the system recalculates the optimal fertilization curve and generates a nutrient control plan. The control plan includes the adjusted fertilization amount and daily spraying time.

[0040] The core of digital twin technology lies in its ability to accurately model physical entities. By synchronizing data from the physical world and the virtual world in real time, the digital twin model can dynamically reflect the growth status of plants and environmental changes. The Kalman filter algorithm uses a recursive estimation method to update model parameters in real time to ensure that the model's prediction results are consistent with the actual observed data. In this way, the system can form a closed-loop control system to ensure that the plants are always in the best nutrient supply state, while avoiding element antagonism and improving fertilizer utilization.

[0041] In summary, the digital twin-based Xanthoceras sorbifolia cultivation and regulation method provided by the present invention collects spectral signals in real time through optical fiber sensors, constructs a nutrient prediction model through preprocessing and machine learning technology, and can accurately predict the nutrient requirements of plants; and generates an initial fertilization curve in combination with a fertilization function database, and realizes precise fertilization through an intelligent sprayer; further utilizes the feedback spectral signal and digital twin technology to update the nutrient absorption model, optimizes the nutrient regulation scheme, and can realize dynamic monitoring and precise regulation of the entire growth process of Xanthoceras sorbifolia plants. The method can effectively solve the problems of delayed fertilization decision-making, waste of resources, and decreased yield caused by artificial experience judgment and regular sampling and detection in traditional cultivation management, improve the cultivation efficiency of Xanthoceras sorbifolia, stabilize the fruit quality, and provide a strong guarantee for the economic benefits and sustainable development of large-scale planting. In addition, the method can avoid the antagonistic effect between multiple elements, optimize the synergistic effect of nutrients such as nitrogen, phosphorus, and potassium, reduce the risk of excessive fertilization or insufficient nutrients, reduce environmental pollution and waste of resources, and provide an efficient, accurate, and sustainable solution for the large-scale planting of Xanthoceras sorbifolia.

[0042] In one embodiment, if Figure 2 As shown, S5 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S51: Based on a preset time interval, the spectral signals of the Xanthoceras sorbifolia plants are collected and preprocessed again to obtain an updated spectral signal set.

[0043] Specifically, the fiber optic sensor is installed on the surface of plant leaves or near fruits, and the reflectance spectrum of leaves or fruits is captured through a multi-channel fiber optic probe. The system can use the principle of diffuse reflectance spectroscopy to reflect the physiological state of the plant through changes in spectral reflectance. The collected spectral signal is digitized through a signal amplifier and an analog-to-digital converter (ADC) and stored as raw spectral data.

[0044] In order to improve data quality, the system can preprocess the spectral signal, including baseline correction, smoothing and characteristic band extraction. Baseline correction eliminates background interference through polynomial fitting, smoothing can use wavelet transform to remove high-frequency noise, and characteristic band extraction extracts characteristic bands related to nutrient concentration through singular value decomposition (SVD) and normalization. The preprocessed spectral data set not only removes noise interference, but also enhances the interpretability of the data, providing high-quality input data for subsequent model updates. Fiber optic sensor technology is widely used in the field of agricultural monitoring due to its non-contact, high-precision and real-time characteristics, especially for plant status monitoring in complex environments.

[0045] S52: Perform differential processing on the updated spectral signal set and the spectral data set based on spectral analysis technology to generate a nutrient concentration deviation vector.

[0046] Specifically, spectral analysis technology is a method of obtaining information about the structure and chemical composition of a substance by measuring the interaction between matter and light. In this system, spectral analysis technology is used to analyze the spectral signals of Xanthoceras sorbifolia plants and extract characteristic bands related to nitrogen, phosphorus, and potassium concentrations. The core of spectral analysis is to establish a quantitative relationship between spectral reflectance and nutrient concentration, and to achieve accurate prediction of the nutrient status of the plant through feature extraction and model training. Among them, the calculation formula for the nutrient concentration deviation vector is: ; in, is the nutrient concentration deviation vector, , , They are the nitrogen, phosphorus and potassium concentrations of the plants in the updated spectral signal set, 、 、 are the nitrogen, phosphorus and potassium concentrations of the plants in the spectral signal concentration, The time difference between two acquisitions. Through differential processing, the system can dynamically capture changes in plant nutrient status and provide key data support for subsequent model updates and fertilization strategy optimization. The core of differential processing is to quantify the rate of change of nutrient concentration through time series analysis, thereby providing a scientific basis for the dynamic adjustment of fertilization strategies.

[0047] S53: Based on the digital twin technology and virtual simulation engine, the nutrient concentration deviation vector is processed, and the nutrient concentration deviation vector is fused with the preset Xanthoceras sorbifolia nutrient absorption model to simulate the impact of different fertilization strategies on nutrient absorption and update the Xanthoceras sorbifolia nutrient absorption model.

[0048] Specifically, the digital twin technology constructs a virtual model of the Xanthoceras sorbifolia plant to synchronize the spectral signals and fertilization status of the physical plant in real time. The virtual model adopts a multi-physical field coupling modeling method, which comprehensively considers the spectral characteristics of leaves, the nutrient absorption efficiency of the root system, and the dynamics of fruit development to ensure a high degree of consistency with the physical plant. The system uses a virtual simulation engine to simulate the impact of different fertilization strategies on plant nutrient absorption and evaluate the feasibility and effectiveness of fertilization strategies. Based on the simulation results, the system dynamically adjusts the parameters of the nutrient absorption model and optimizes the fertilization strategy to avoid the antagonistic effect of elements. The core advantage of digital twin technology is that it realizes dynamic optimization and precise control of complex systems through real-time interaction between virtual models and physical entities. Its application in the agricultural field provides a new technical path for plant growth regulation. Preferably, the Xanthoceras sorbifolia nutrient absorption model is updated through the following steps: S531: Perform data fusion processing on the nutrient concentration deviation vector and the initial fertilization curve to generate a fused data set.

[0049] Specifically, the system can use a linear weighted fusion method to perform weighted processing on the deviation vector and the initial fertilization curve to generate a fused data set. The calculation formula of the fused data set is: ; in, To fuse the dataset, is the current output data of the Xanthoceras sorbifolia nutrient absorption model, i.e., the initial fertilization curve. is the preset fusion weight coefficient; fusion weight coefficient Dynamic adjustments are made based on historical data and model performance to ensure that the fused data set can comprehensively reflect the actual nutrient requirements of the plants and the model prediction results. The application of data fusion technology in the agricultural field has significant advantages and can improve the scientificity and accuracy of decision-making by integrating multi-source data.

[0050] S532: Based on the fused data set, the nutrient absorption efficiency under different fertilization strategies is simulated by a virtual simulation engine to generate a simulation result data set.

[0051] Specifically, the system uses a virtual simulation engine to process the fused data set and simulate the nutrient absorption efficiency under different fertilization strategies. The virtual simulation engine constructs a virtual model of the Xanthoceras sorbifolia plant and combines soil nutrient concentration, fertilizer application amount and environmental conditions to simulate the nutrient absorption process of the plant under different fertilization strategies. The calculation formula for the simulation result data set is: ; in, is the simulation result data set, is the fertilizer input for the fused dataset, is the simulated current soil nutrient concentration, 、 、 are the model parameters of Xanthoceras sorbifolia nutrient absorption model, e is a natural constant, i.e., the base of the natural logarithm function. The virtual simulation engine dynamically adjusts the relationship between the amount of fertilizer applied and the soil nutrient concentration through numerical simulation methods to predict the nutrient absorption efficiency of plants. This method can effectively avoid the high cost and long cycle problems in traditional experimental methods, while providing high-precision prediction results and providing a scientific basis for optimizing fertilization strategies.

[0052] S533: updating the model parameters of the Xanthoceras sorbifolia nutrient absorption model according to the simulation result data set, and generating an updated Xanthoceras sorbifolia nutrient absorption model.

[0053] Specifically, the system can use the Kalman filter algorithm to compare the simulation result data set with the model prediction results and update the model parameters in real time. The Kalman filter algorithm uses a recursive estimation method to adjust the model parameters in real time to ensure that the model's prediction results are consistent with the actual observation data. This method can effectively improve the dynamic adaptability of the model, especially under complex environmental conditions, and can quickly respond to changes in the nutrient status of the plant to ensure the scientificity and effectiveness of the fertilization strategy. In this way, the system can continuously optimize the Xanthoceras sorbifolia nutrient absorption model, improve the model's prediction accuracy and fertilization efficiency, and provide reliable technical support for precision fertilization.

[0054] S54: Input the initial fertilization curve into the updated Xanthoceras sorbifolia nutrient absorption model for parameter optimization to generate a nutrient regulation plan.

[0055] Specifically, the updated nutrient absorption model is based on the latest spectral signals and simulation results, and can more accurately reflect the nutrient requirements and absorption efficiency of the plants. The system can use the NSGA-II algorithm (Non-dominated SortingGenetic Algorithm II) to optimize parameters such as fertilizer amount, fertilization time and spraying mode, and generate a comprehensive control plan including daily fertilizer amount, fertilization time point and spraying mode. The optimization process takes into account multiple factors such as fertilization efficiency, environmental impact and plant health to ensure the scientificity and feasibility of the fertilization strategy. The final generated nutrient control plan is sent to the intelligent sprayer through the wireless communication module to guide the dynamic adjustment of the fertilization operation. Preferably, the nutrient control plan is generated through the following steps: S541: Perform dynamic weight adjustment processing on the initial fertilization curve to generate an optimized weight coefficient.

[0056] Specifically, the system dynamically adjusts the weights of different nutrients based on real-time feedback and historical data to improve the accuracy and adaptability of fertilization strategies. The calculation formula for the optimized weight coefficient is: ; in, is the optimization weight coefficient of nitrogen, phosphorus and potassium, is the historical variance of nutrient concentration deviation. Through dynamic weight adjustment, the system can dynamically adjust the fertilization strategy according to the stability of historical data to ensure that the amount of fertilizer is highly matched with the actual needs of the plants. The application of dynamic weight adjustment technology in the agricultural field has significant advantages. It can optimize fertilization strategies through real-time feedback, improve fertilization efficiency, and reduce resource waste.

[0057] S542: Input the optimized weight coefficient into the updated Xanthoceras sorbifolia nutrient absorption model for processing to generate a dynamic fertilization amount.

[0058] Specifically, the calculation formula for dynamic fertilization amount is: ; in, is the dynamic fertilization amount, is the baseline fertilizer amount in the initial fertilization curve, is the nutrient concentration deviation, The dynamic adjustment coefficient. Through the calculation of dynamic fertilization amount, the system can dynamically adjust the fertilization strategy according to the actual nutrient requirements of the plant and environmental changes to ensure that the fertilization amount is highly matched with the plant's needs. The calculation of dynamic fertilization amount not only takes into account the baseline fertilization amount, but also combines the nutrient concentration deviation and the dynamic adjustment coefficient, which improves the scientificity and adaptability of the fertilization strategy.

[0059] S543: The dynamic fertilizer application amount is processed based on fuzzy control technology, and a nutrient regulation plan is generated according to the relationship between the dynamic fertilizer application amount and the preset safety threshold, combined with the fertilizer requirement law of the Xanthoceras sorbifolia growth stage.

[0060] Specifically, the system generates a nutrient regulation plan based on the relationship between the dynamic fertilization amount and the preset safety threshold, combined with the fertilizer requirement law of Xanthoceras sorbifolia during its growth stage. Fuzzy control technology compares the dynamic fertilization amount with the safety threshold through a fuzzy rule base and reasoning mechanism to generate fertilization adjustment suggestions. For example, when the dynamic fertilization amount exceeds the safety threshold, the system will reduce the fertilization amount to avoid over-fertilization; when the fertilization amount is insufficient, the system will increase the fertilization amount to meet the nutrient needs of the plant.

[0061] The application of fuzzy control technology in the agricultural field has been widely verified. For example, in greenhouse crop management, fuzzy control technology is used to optimize fertilization and irrigation strategies to improve crop yield and quality. In this system, fuzzy control technology dynamically adjusts the amount of fertilizer to ensure that the plants are always in the best nutrient supply state, while avoiding element antagonism and improving fertilizer utilization.

[0062] The above-mentioned digital twin-based Xanthoceras sorbifolia cultivation and control method can realize real-time, accurate monitoring and dynamic control of the nutrient status of Xanthoceras sorbifolia plants. From the timed update of spectral signal acquisition, to the generation of nutrient concentration deviation vectors, to the use of digital twin technology to optimize the nutrient absorption model, and finally to the generation of personalized nutrient control solutions, the entire process forms a closed-loop intelligent control system, which can effectively solve the problems of inaccurate fertilization, waste of resources, and unstable yield quality caused by manual experience judgment and sampling and detection lag in traditional cultivation management, thereby significantly improving the cultivation efficiency and fruit quality of Xanthoceras sorbifolia, reducing production costs and environmental risks, and providing strong technical support for the large-scale and intelligent cultivation of Xanthoceras sorbifolia, and promoting the development of agricultural cultivation technology towards precision, efficiency and sustainability.

[0063] In one embodiment, S1 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S11: The spectral signals of the leaves and roots of Xanthoceras sorbifolia are collected by embedded optical fiber sensors. The spectral signals cover the growth stages of the Xanthoceras sorbifolia plants in the seedling stage, rapid growth stage and mature stage.

[0064] Specifically, the embedded optical fiber sensor is a sensor based on the fiber Bragg grating (FBG) principle, which is widely used in structural health monitoring, environmental monitoring and agricultural fields. The fiber Bragg grating sensor detects the wavelength deviation of light waves to achieve accurate measurement of physical quantities such as temperature and pressure. The high sensitivity and anti-electromagnetic interference characteristics of the sensor give it unique advantages in monitoring in complex environments. In the agricultural field, the embedded optical fiber sensor can monitor the physiological state of plants in real time, providing important technical support for precision agriculture.

[0065] In this embodiment, the system uses a high-precision embedded fiber optic sensor to collect the spectral signals of Xanthoceras sorbifolia plants in real time. The fiber optic sensor is installed near the leaves and roots of the plant to ensure that the collected spectral signals can fully reflect the physiological state of the plant. The sensor generates raw data including the plant's reflection spectrum, transmission spectrum and absorption spectrum through spectral acquisition technology in a specific wavelength range. The acquisition frequency of the spectral signal can be dynamically adjusted according to the growth stage and environmental conditions of the plant to ensure the timeliness and accuracy of the data.

[0066] S12: Perform wavelet denoising on the spectral signal to remove high-frequency noise components and generate a denoised data set.

[0067] Specifically, the collected spectral signals usually contain a certain amount of noise, which mainly comes from environmental interference, electronic noise of the sensor itself, and scattering effects in optical transmission. In order to improve the quality of the data, the system uses wavelet denoising technology to process the spectral signal. The core of wavelet denoising is to decompose the signal into sub-bands of different frequencies through multi-scale analysis, thereby effectively removing high-frequency noise components while retaining the main features of the signal.

[0068] The wavelet denoising process specifically includes the following steps: First, select the appropriate wavelet basis function and decomposition level to decompose the spectral signal at multiple scales. Commonly used basis functions include Daubechies, Symlets, and Coiflets. The number of decomposition levels is usually determined dynamically based on the length of the signal and the noise level. Secondly, threshold processing is performed on the decomposed high-frequency sub-bands, and the noise component is removed using a soft threshold or hard threshold method. Finally, the processed sub-bands are recombined through a wavelet reconstruction algorithm to generate a denoised spectral signal.

[0069] S13: normalize the denoised data set to generate a spectral data set.

[0070] Specifically, normalization is a technique that scales data to a specific range (such as 0 to 1), which can eliminate the dimensional differences between different data and improve the comparability and consistency of the data. In this system, normalization is achieved through linear transformation, mapping the intensity value of the spectral signal to the range of 0 to 1. Normalization can not only improve the interpretability of the data, but also enhance the efficiency and stability of model training. Through normalization, the system can ensure the comparability of spectral data at different growth stages and environmental conditions, providing high-quality input data for subsequent model training and prediction.

[0071] In one embodiment, S2 of a digital twin-based Xanthoceras sorbifolia cultivation regulation method provided by the present invention specifically includes the following steps: S21: Perform principal component analysis on the spectral data set and the pre-input historical spectral data, retain the characteristic bands whose cumulative variance contribution rate is greater than a preset value, and generate a dimensionality reduction feature matrix.

[0072] Specifically, principal component analysis is a commonly used statistical method that projects the original data into a new feature space through linear transformation, thereby extracting the characteristic bands that can explain the maximum variance of the data. In this step, the system first merges the spectral data set and the historical spectral data to form a comprehensive data set. Then, the covariance matrix of the data set is calculated, and its eigenvalues ​​and eigenvectors are found. By selecting eigenvectors whose cumulative variance contribution rate is greater than the preset value, the system can retain the main characteristic bands in the data, remove redundant information, and generate a reduced-dimensional feature matrix.

[0073] The core of principal component analysis is that it can effectively reduce the data dimension while retaining the main features of the data. In spectral data analysis, this method can remove the correlation in spectral data, improve the interpretability of data and the training efficiency of the model. In this way, the system can extract characteristic bands that are highly correlated with the nutrient status of the plant, providing high-quality input data for subsequent model training.

[0074] S22: Associating the dimension-reduced feature matrix with a pre-constructed nutrient concentration database, where the nutrient concentration database includes nitrogen, phosphorus, and potassium concentration data, to generate a training data set.

[0075] Specifically, the system associates the reduced dimension feature matrix with a pre-built nutrient concentration database. The nutrient concentration database contains nitrogen, phosphorus, and potassium concentration data of Xanthoceras sorbifolia plants at different growth stages and environmental conditions. These data are obtained through laboratory analysis and field sampling to ensure the accuracy and representativeness of the data. The association process generates a training data set containing spectral features and nutrient concentrations by matching spectral data with corresponding nutrient concentration data.

[0076] Specifically, the system matches the timestamps and sample identifiers of the reduced dimension feature matrix and the nutrient concentration database to ensure that each spectral sample corresponds to the correct nutrient concentration data. Secondly, through data cleaning and preprocessing steps, outliers and missing values ​​are removed to ensure the integrity and consistency of the data. Finally, the matched data is integrated into a structured training data set to provide high-quality input for subsequent model training.

[0077] This correlation processing method can ensure the correspondence between spectral data and nutrient concentration data, providing a basis for building an accurate nutrient prediction model. In this way, the system can make full use of historical data and real-time data to improve the generalization ability and prediction accuracy of the model.

[0078] S23: The random forest regression model is trained based on the training data set to construct a nutrient prediction model for predicting nitrogen and potassium concentrations.

[0079] Specifically, random forest regression is a supervised learning algorithm based on ensemble learning. It improves the accuracy and generalization ability of the model by constructing multiple decision trees and integrating their prediction results. During the training process, the system randomly extracts samples and features from the training data set to construct multiple decision trees. Each decision tree is trained by minimizing the mean square error (MSE), and finally generates the final prediction value by majority voting or averaging the prediction results. The advantage of the random forest regression model is that it can handle high-dimensional data, avoid overfitting, and provide feature importance assessment to help identify the features that have the greatest impact on the prediction results. Through the random forest regression model, the system can accurately predict the nitrogen, phosphorus, and potassium concentrations of the plants, providing a scientific basis for the optimization of fertilization strategies.

[0080] In one embodiment, the present invention provides a twin Xanthoceras sorbifolia cultivation regulation method S3 specifically comprises the following steps: S31: The spectral data is concentrated; the spectral signals of the growth stage are input into the nutrient prediction model for processing to generate real-time concentration values ​​of nitrogen, phosphorus and potassium.

[0081] Specifically, the nutrient prediction model is built based on the random forest regression algorithm, which can accurately predict the nutrient status of plants through the nonlinear mapping relationship between spectral features and nutrient concentrations. During the processing, the system first pre-processes the input spectral signal, including baseline correction, smoothing, and feature extraction to ensure the quality and consistency of the input data. Then, the spectral features are comprehensively analyzed through multiple decision trees of the random forest regression model. Each decision tree predicts the input signal based on the feature weights learned from the training data, and finally generates the real-time concentration values ​​of nitrogen, phosphorus, and potassium through majority voting or average prediction results.

[0082] S32: Perform difference calculation on the real-time concentration value and the preset concentration threshold value to generate a deviation vector including nitrogen deviation, phosphorus deviation and potassium deviation Specifically, the preset concentration thresholds are determined based on the optimal nutrient requirements of Xanthoceras sorbifolia plants at different growth stages. These thresholds are set through long-term field trials and expert experience to ensure that the plants can grow under the optimal nutrient conditions. The calculation formula for the deviation vector is: ; in, is the deviation vector, 、 、 is the real-time concentration value, , , The deviation vector reflects the difference between the current nutrient status of the plant and the ideal status, providing key data support for subsequent fertilization strategy adjustments. Through the calculation of the deviation vector, the system can dynamically capture changes in plant nutrient requirements and ensure the real-time and scientific nature of the fertilization strategy.

[0083] S33: querying a pre-built fertilization function database according to the deviation vector to generate an initial fertilization curve, wherein the fertilization function database is used to store a linear mapping relationship between nitrogen, phosphorus, and potassium fertilization amounts and deviations.

[0084] Specifically, the fertilization function database stores the linear mapping relationships between nitrogen, phosphorus, and potassium fertilizer amounts and deviations. These relationships are established through long-term field trials and data analysis to ensure that fertilizer amounts can be reasonably adjusted based on deviations.

[0085] Specifically, the fertilization function database contains multiple linear equations, each of which describes the relationship between the fertilizer amount and the deviation. For example, for nitrogen, the fertilizer amount With deviation The relationship can be expressed as: ; in, and is the linear mapping coefficient, obtained by fitting historical data. Similarly, the amount of fertilizer for phosphorus and potassium and It is also calculated by the corresponding linear equation.

[0086] The system queries the fertilization function database, substitutes each deviation value in the deviation vector into the corresponding linear equation, and calculates the corresponding fertilization amount. Then, these fertilization amounts are integrated into the initial fertilization curve, which describes the change trend of fertilization amount at each time point in the future time window.

[0087] The generation process of the initial fertilization curve is realized through the numerical integration method to ensure the continuity and rationality of the fertilization amount in time. In this way, the system can generate a scientific and reasonable fertilization strategy based on the current nutrient deviation to ensure that the plants can obtain the appropriate amount of nutrients at different growth stages, thereby improving the growth efficiency of the plants and the quality of the fruits.

[0088] Preferably, if Figure 3 As shown, the present invention provides a Xanthoceras sorbifolia cultivation and regulation device 600 based on digital twin, which is configured with the following modules: The data acquisition and processing module 610 is used to collect and pre-process the spectral signals of the Xanthoceras sorbifolia plants based on the optical fiber sensor to generate a spectral data set; A prediction model building module 620 is used to perform feature extraction and model training on pre-input historical spectral data and spectral data sets based on machine learning technology to build a nutrient prediction model for predicting nitrogen, phosphorus, and potassium concentrations; A fertilization curve generation module 630 is used to input the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing based on a preset fertilization function database to generate an initial fertilization curve; The instruction generation and sending module 640 is used to process the initial fertilization curve, generate an initial fertilization instruction, and send the initial fertilization instruction to the intelligent sprayer, where the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation; The nutrient control scheme generation module 650 is used to process the initial fertilization curve based on the feedback updated spectral signal set and digital twin technology, update the Xanthoceras sorbifolia nutrient absorption model and generate a nutrient control scheme, which includes the amount of fertilizer and daily spraying time points.

[0089] In summary, the Xanthoceras sorbifolia cultivation and control device based on digital twin provided by the present invention collects spectral signals in real time through optical fiber sensors, constructs a nutrient prediction model through preprocessing and machine learning technology, and can accurately predict the nutrient requirements of plants; and generates an initial fertilization curve in combination with a fertilization function database, and realizes precise fertilization through an intelligent sprayer; further uses the feedback spectral signal and digital twin technology to update the nutrient absorption model, optimizes the nutrient regulation scheme, and can realize dynamic monitoring and precise regulation of the entire growth process of Xanthoceras sorbifolia plants. This method can effectively solve the problems of delayed fertilization decision-making, waste of resources, and decreased yield caused by artificial experience judgment and regular sampling and detection in traditional cultivation management, improve the cultivation efficiency of Xanthoceras sorbifolia, stabilize the quality of fruits, and provide a strong guarantee for the economic benefits and sustainable development of large-scale planting. In addition, the device can avoid the antagonistic effect between multiple elements, optimize the synergistic effect of nutrients such as nitrogen, phosphorus, and potassium, reduce the risk of excessive fertilization or insufficient nutrients, reduce environmental pollution and waste of resources, and provide an efficient, accurate, and sustainable solution for the large-scale planting of Xanthoceras sorbifolia.

[0090] Preferably, the nutrient regulation scheme generation module 650 provided by the present invention is configured with the following components: A spectral signal update collection unit is used to re-collect and pre-process the spectral signals of the Xanthoceras sorbifolia plants based on a preset time interval to obtain an updated spectral signal set; A deviation vector generating unit, used for performing differential processing on the updated spectral signal set and the spectral data set based on spectral analysis technology to generate a nutrient concentration deviation vector; The nutrient absorption model update unit is used to process the nutrient concentration deviation vector based on the digital twin technology and virtual simulation engine, fuse the nutrient concentration deviation vector with the preset Xanthoceras sorbifolia nutrient absorption model, simulate the impact of different fertilization strategies on nutrient absorption, and update the Xanthoceras sorbifolia nutrient absorption model; Preferably, the nutrient absorption model updating unit is configured with the following subunits: The data fusion subunit is used to perform data fusion processing on the nutrient concentration deviation vector and the initial fertilization curve to generate a fused data set; The simulation subunit is used to simulate the nutrient absorption efficiency under different fertilization strategies through a virtual simulation engine based on the fusion data set to generate a simulation result data set; The model updating subunit is used to update the model parameters of the Xanthoceras sorbifolia nutrient absorption model according to the simulation result data set to generate an updated Xanthoceras sorbifolia nutrient absorption model.

[0091] The nutrient control scheme generation unit is used to input the initial fertilization curve into the updated Xanthoceras sorbifolia nutrient absorption model for parameter optimization processing to generate a nutrient control scheme.

[0092] Preferably, the nutrient regulation scheme generation unit is configured with the following subunits: The weight adjustment subunit is used to dynamically adjust the initial fertilization curve and generate an optimized weight coefficient; A fertilizer application amount generating subunit is used to input the optimized weight coefficient into the updated Xanthoceras sorbifolia nutrient absorption model for processing and generate a dynamic fertilizer application amount; The control scheme generation subunit is used to process the dynamic fertilization amount based on fuzzy control technology, and generate a nutrient control scheme according to the relationship between the dynamic fertilization amount and the preset safety threshold, combined with the fertilizer requirement law of the Xanthoceras sorbifolia growth stage.

[0093] Preferably, the data acquisition and processing module 610 provided by the present invention is configured with the following units: A spectral signal acquisition unit, used to collect spectral signals of Xanthoceras sorbifolia leaves and roots through an embedded optical fiber sensor, the spectral signals covering the growth stages of the Xanthoceras sorbifolia plants in the seedling stage, rapid growth stage and mature stage; A signal denoising unit is used to perform wavelet denoising on the spectral signal, remove high-frequency noise components, and generate a denoised data set; The data normalization unit is used to normalize the denoised data set to generate a spectral data set.

[0094] Preferably, the prediction model building module 620 provided by the present invention is configured with the following units: A feature dimension reduction unit is used to perform principal component analysis on the spectral data set and the pre-input historical spectral data, retain the feature bands whose cumulative variance contribution rate is greater than a preset value, and generate a dimension reduction feature matrix; A training data generation unit is used to associate the dimension-reduced feature matrix with a pre-built nutrient concentration database, where the nutrient concentration database includes nitrogen, phosphorus, and potassium concentration data, to generate a training data set; The model training unit is used to train the random forest regression model based on the training data set and build a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations.

[0095] Preferably, the fertilization curve generation module 630 provided by the present invention is configured with the following units: A real-time concentration value generating unit is used to input the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing to generate real-time concentration values ​​of nitrogen, phosphorus and potassium; A deviation vector calculation unit, used for performing difference calculation processing between the real-time concentration value and the preset concentration threshold value, and generating a deviation vector including nitrogen deviation, phosphorus deviation and potassium deviation; The curve mapping generation unit is used to query a pre-built fertilization function database according to the deviation vector to generate an initial fertilization curve. The fertilization function database stores the linear mapping relationship between the nitrogen, phosphorus and potassium fertilizer amounts and the deviations.

[0096] In one embodiment, the present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned digital twin-based Xanthoceras sorbifolia cultivation and regulation method when executing the computer program.

[0097] In one embodiment, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned digital twin-based Xanthoceras sorbifolia cultivation and regulation method.

[0098] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0099] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can refer to the partial description of the method embodiments. The device embodiments described above are only schematic, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0100] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for regulating and cultivating Xanthoceras sorbifolia based on digital twins, characterized in that: The following steps are involved: S1: Collect and preprocess the spectral signals of Xanthoceras sorbifolia plants based on optical fiber sensors to generate spectral data sets; S2: extracting features and training models on the pre-input historical spectral data and the spectral data set based on machine learning technology to construct a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations; S3: Based on the preset fertilization function database, the spectral signal of the current growth stage in the spectral data set is input into the nutrient prediction model for processing to generate an initial fertilization curve; S4: Processing the initial fertilization curve to generate an initial fertilization instruction, and sending the initial fertilization instruction to the intelligent sprayer, wherein the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation; S5: The initial fertilization curve is processed based on the feedback-based updated spectral signal set and digital twin technology, the Xanthoceras sorbifolia nutrient absorption model is updated and a nutrient regulation plan is generated, wherein the nutrient regulation plan includes the amount of fertilizer and the daily spraying time points.

2. The method according to claim 1, characterized in that The S5 includes: S51: based on a preset time interval, re-collecting and pre-processing the spectral signals of the Xanthoceras sorbifolia plants to obtain an updated spectral signal set; S52: performing differential processing on the updated spectral signal set and the spectral data set based on spectral analysis technology to generate a nutrient concentration deviation vector. The calculation formula of the nutrient concentration deviation vector is: ； in, is the nutrient concentration deviation vector, , , They are the nitrogen, phosphorus and potassium concentrations of the plants in the updated spectral signal set, 、 、 are the nitrogen, phosphorus and potassium concentrations of the plants in the spectral signal concentration, is the time difference between the two acquisitions; S53: processing the nutrient concentration deviation vector based on the digital twin technology and the virtual simulation engine, fusing the nutrient concentration deviation vector with a preset Xanthoceras sorbifolia nutrient absorption model, simulating the effects of different fertilization strategies on nutrient absorption, and updating the Xanthoceras sorbifolia nutrient absorption model; S54: Inputting the initial fertilization curve into the updated Xanthoceras sorbifolia nutrient absorption model for parameter optimization processing to generate a nutrient regulation scheme.

3. The method according to claim 2, characterized in that The S53 includes: S531: Perform data fusion processing on the nutrient concentration deviation vector and the initial fertilization curve to generate a fused data set. The calculation formula of the fused data set is: ； in, To fuse the dataset, is the current output data of the Xanthoceras sorbifolia nutrient absorption model, i.e., the initial fertilization curve. is the preset fusion weight coefficient; S532: Based on the fused data set, the nutrient absorption efficiency under different fertilization strategies is simulated by a virtual simulation engine to generate a simulation result data set. The calculation formula of the simulation result data set is: ； in, is the simulation result data set, is the fertilizer input for the fused dataset, is the simulated current soil nutrient concentration, 、 、 are the model parameters of Xanthoceras sorbifolia nutrient absorption model, e is a natural constant, i.e. the base of the natural logarithm function; S533: updating the model parameters of the Xanthoceras sorbifolia nutrient absorption model according to the simulation result data set to generate an updated Xanthoceras sorbifolia nutrient absorption model.

4. The method according to claim 3, characterized in that The S54 includes: S541: Perform dynamic weight adjustment processing on the initial fertilization curve to generate an optimized weight coefficient. The calculation formula of the optimized weight coefficient is: ； in, is the optimization weight coefficient of nitrogen, phosphorus and potassium, is the historical variance of nutrient concentration deviation; S542: Input the optimized weight coefficient into the updated Xanthoceras sorbifolia nutrient absorption model for processing to generate a dynamic fertilization amount. The calculation formula of the dynamic fertilization amount is: ； in, is the dynamic fertilization amount, is the baseline fertilizer amount in the initial fertilization curve, is the nutrient concentration deviation, is the dynamic adjustment coefficient; S543: The dynamic fertilization amount is processed based on fuzzy control technology, and a nutrient regulation plan is generated according to the relationship between the dynamic fertilization amount and a preset safety threshold value, combined with the fertilizer requirement law of the Xanthoceras sorbifolia growth stage.

5. The method according to claim 1, characterized in that The S1 includes: S11: collecting spectral signals of Xanthoceras sorbifolia leaves and roots through an embedded optical fiber sensor, wherein the spectral signals cover the growth stages of the Xanthoceras sorbifolia plants in the seedling stage, the rapid growth stage, and the mature stage; S12: performing wavelet denoising processing on the spectral signal to remove high-frequency noise components and generate a denoised data set; S13: performing normalization processing on the denoised data set to generate a spectral data set.

6. The method according to claim 1, characterized in that The S2 includes: S21: performing principal component analysis on the spectral data set and the pre-input historical spectral data, retaining characteristic bands whose cumulative variance contribution rate is greater than a preset value, and generating a dimension reduction feature matrix; S22: Associating the dimension reduction feature matrix with a pre-constructed nutrient concentration database, wherein the nutrient concentration database includes nitrogen, phosphorus, and potassium concentration data, to generate a training data set; S23: Training the random forest regression model based on the training data set to construct a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations.

7. The method according to any one of claims 1 to 6, characterized in that: The S3 includes: S31: inputting the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing to generate real-time concentration values ​​of nitrogen, phosphorus and potassium; S32: performing difference calculation processing on the real-time concentration value and the preset concentration threshold value to generate a deviation vector including nitrogen deviation, phosphorus deviation and potassium deviation; S33: querying a pre-built fertilization function database according to the deviation vector to generate the initial fertilization curve, wherein the fertilization function database is used to store a linear mapping relationship between nitrogen, phosphorus and potassium fertilization amounts and deviations.

8. A Xanthoceras sorbifolia cultivation and control device based on digital twin, characterized in that: The device comprises: A data acquisition and processing module, used for collecting and preprocessing the spectral signals of the Xanthoceras sorbifolia plant based on an optical fiber sensor to generate a spectral data set; A prediction model building module, for performing feature extraction and model training on the pre-input historical spectral data and the spectral data set based on machine learning technology, and building a nutrient prediction model for predicting nitrogen, phosphorus and potassium concentrations; A fertilization curve generation module is used to input the spectral signal of the current growth stage in the spectral data set into the nutrient prediction model for processing based on a preset fertilization function database to generate an initial fertilization curve; An instruction generation and sending module is used to process the initial fertilization curve, generate an initial fertilization instruction, and send the initial fertilization instruction to the intelligent sprayer, wherein the initial fertilization instruction is used to control the intelligent sprayer to perform a fertilization operation; A nutrient control scheme generation module is used to process the initial fertilization curve based on the feedback updated spectral signal set and digital twin technology, update the Xanthoceras sorbifolia nutrient absorption model and generate a nutrient control scheme, wherein the nutrient control scheme includes the amount of fertilizer and the daily spraying time points.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images