Animal and plant intelligent diagnosis system based on multi-source data

Abstract

The invention relates to the technical field of intelligent diagnosis and recognition, in particular to an animal and plant intelligent diagnosis system based on multi-source data. Identifying a geographic space risk map, and obtaining a historical state and an intervention record of the crop plant; performing data acquisition on crop plants through a remote sensing technology, and obtaining multi-source remote sensing time sequence data based on a multispectral sensor, a hyperspectral sensor, a thermal infrared sensor and a laser radar sensor; identifying the geographic space risk map, the multi-source remote sensing time sequence data and the historical state and intervention record, and determining a risk abnormal plant; identifying a risk abnormal plant through acoustic detection equipment and an annular electrode array to obtain acoustic characteristic data and a three-dimensional electrical characteristic spatial distribution map, and further identifying to obtain an abnormal state index of the crop plant; according to the invention, clustering analysis is carried out based on the abnormal state indexes and the three-dimensional coordinates of the crop plants, the infection center and the infection area are determined, and abnormal early warning is carried out.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent diagnosis and recognition, and particularly to an intelligent diagnosis system for animals and plants based on multi-source data. BACKGROUND

[0002] In the prior art, computer vision and image analysis methods have been preliminarily applied to the health monitoring of animals and plants. For example, by collecting visible light or multispectral images of crop plants, traditional image processing and machine learning algorithms are used to extract and classify the color, texture and other visual phenotype features of leaf lesions to identify specific diseases. Some solutions also introduce sensors such as laser radars to construct three-dimensional point cloud models of crop plants, and analyze spatial structure features such as canopy density to assess their growth status. These methods have certain effects in identifying diseases that have shown significant macroscopic features.

[0003] However, the analysis process of the prior art has significant limitations. First, its recognition objects are limited to the external optical or three-dimensional structural features of living organisms. When the disease occurs internally or in the early latent stage, visible light image information may not have changed significantly, making it difficult for traditional feature engineering and pattern recognition algorithms to effectively detect and classify. Second, different causes may lead to similar external visual patterns, and this phenomenon of different diseases with similar symptoms makes it easy to cause confusion by relying only on feature extraction of a single image modality, resulting in insufficient accuracy of scene understanding and decision-making.

[0004] In addition, although the prior art has also explored the application of different sensor data, it mostly stays at the level of independent analysis of data or simple superposition of results at the result layer, lacking a computing framework that can deeply fuse and cooperatively analyze features of external remote sensing images and other heterogeneous non-image data such as physiological and biochemical data, environmental parameters, etc.

[0005] Therefore, an intelligent diagnosis system for animals and plants based on multi-source data is proposed. SUMMARY

[0006] The purpose of the present application is to provide an intelligent diagnosis system for animals and plants based on multi-source data, which determines the infection center and infection area through cluster analysis of the abnormal state index and three-dimensional coordinates of crop plants, and performs abnormal early warning.

[0007] To achieve the above purpose, the present application provides an intelligent diagnosis system for animals and plants based on multi-source data, comprising:

[0008] A data acquisition module obtains a geographic spatial risk map based on stock parameters and soil parameters; obtains historical states and intervention records of crop plants to obtain historical states, intervention time, intervention measures and intervention effects of the crop plants;

[0009] The feature recognition module periodically collects data of crop plants by remote sensing technology to obtain multi-source remote sensing time series data; a deep risk identification model is constructed to identify geographical spatial risk atlas, multi-source remote sensing time series data and historical state and intervention records to determine risk abnormal plants;

[0010] The infection judgment module collects data of the risk abnormal plants by acoustic detection equipment to obtain acoustic feature data; a three-dimensional electrical characteristic spatial distribution map of root systems is generated by identifying the risk abnormal plants through a ring electrode array; and the abnormal state index of the crop plants is identified based on the acoustic feature data and the three-dimensional electrical characteristic spatial distribution map;

[0011] The abnormal early warning module performs clustering analysis based on the abnormal state index and three-dimensional coordinates of the crop plants to determine infection centers and infection areas and perform abnormal early warning.

[0012] The identification process of the geographical spatial risk atlas comprises:

[0013] A digital information map is obtained, and the stock parameters of the crop plants and the soil parameters of the plots are labeled;

[0014] According to a plant physiological characteristic database, a standardized stock resistance score is given to each stock, and according to the specific parameters of clay and sand in the soil parameters, a standardized soil susceptibility score is given to the plots with different soil parameters;

[0015] The stock resistance score and the soil susceptibility score on the digital information map are calculated by a weighted fusion algorithm to generate a geographical spatial risk atlas.

[0016] The multi-source remote sensing time series data are obtained by using a drone carrying a multi-spectral sensor, a hyperspectral sensor, a thermal infrared sensor and a laser radar sensor to periodically scan the crop plants by remote sensing;

[0017] The vegetation index, chlorophyll content and nitrogen status parameters are obtained by the multi-spectral sensor and the hyperspectral sensor; the canopy temperature is obtained by the thermal infrared sensor to evaluate the water stress state caused by root damage; the three-dimensional point cloud data of the canopy are obtained by the laser radar sensor to calculate the leaf area index and the canopy density physical structure parameters;

[0018] The multi-dimensional data obtained in each collection period are taken as time slices to form the multi-source remote sensing time series data in the collection order.

[0019] The identification process of the deep risk identification model comprises: taking the multi-source remote sensing time series data as the core, identifying the crop plants with persistent decline in health indicators by a time series analysis algorithm, and obtaining dynamic decline characteristics;

[0020] The historical state of the crop plant and intervention records are called, causal correlation analysis is performed based on the historical state, intervention time, intervention measures and intervention effect, and a historical state correlation coefficient is obtained; a geographical space risk map is called to obtain a geographical space risk coefficient of the crop plant; comprehensive analysis is performed based on the dynamic decline characteristics, the historical state correlation coefficient and the geographical space risk coefficient to obtain a comprehensive risk score; and a risk abnormal plant is determined based on the comprehensive risk score.

[0021] The process of obtaining the historical state correlation coefficient includes: obtaining historical state and intervention records to obtain the historical state, intervention time, intervention measures and intervention effect of the crop plant; based on the historical state, intervention time and intervention measures, querying the built-in agricultural knowledge base to dynamically generate a quantitative standard expected effect; comparing the intervention effect in the historical state and intervention records with the standard expected effect to calculate an effect difference degree score;

[0022] The historical state in the records is compared with a database of typical symptoms of the root knot nematode to calculate a symptom similarity score;

[0023] The effect difference degree score and the symptom similarity score are comprehensively calculated by a preset fusion function to obtain the historical state correlation coefficient.

[0024] The process of generating a three-dimensional electrical characteristic spatial distribution map of the root system includes: deploying a ring electrode array in the soil at the base of the risk abnormal plant, applying weak and multi-frequency high-frequency alternating current between different electrode pairs in sequence by the host system, and synchronously measuring the voltage distribution on all the remaining electrodes to collect complete electrical response data sets;

[0025] The electrical response data sets are inversely operated by a reconstruction algorithm to generate a three-dimensional matrix in voxel units; a three-dimensional electrical characteristic spatial distribution map is obtained from the three-dimensional matrix, including the three-dimensional coordinates, electrical impedance value and capacitance characteristic parameters of each voxel obtained by multi-frequency scanning.

[0026] The process of identifying the abnormal state index of the crop plant includes: monitoring the elastic wave generated in the xylem vessel due to water column breakage under water stress of the crop plant to obtain acoustic characteristic data;

[0027] The acoustic characteristic data are matched with a preset root knot nematode threat sound spectrum library to obtain acoustic correlation characteristic data; the crop plant infected by the root knot nematode is judged based on comprehensive analysis of the acoustic correlation characteristic data and the three-dimensional coordinates, electrical impedance value and capacitance characteristic parameters of the voxels in the three-dimensional electrical characteristic spatial distribution map.

[0028] The three-dimensional image segmentation algorithm is used for processing the three-dimensional electrical characteristic space distribution map, and based on the resistance impedance value and the three-dimensional morphology of the lesion tissue, the total volume of the lesion tissue and the total volume of the healthy root system are recognized and calculated, and an abnormal state index is calculated.

[0029] The process of determining the infection center and the infection area includes: data binding the abnormal state index of the crop plant with the three-dimensional geographic coordinates to form a spatial point data set; using a density-based spatial clustering algorithm to analyze the spatial point data set, and identifying the crop plants that are adjacent in space and have infection indexes reaching a preset high-risk threshold as a cluster;

[0030] The geometric center point of the cluster with the highest density is determined as the infection center, and the geographic range of the cluster covered at the infection center is determined as the infection area.

[0031] Compared with the prior art, the present application has the following advantages:

[0032] 1、The present application integrates plant genetic resistance and soil environment data to construct a quantitative and visual geographic spatial risk map; realizes accurate prediction and spatial positioning of potential disease occurrence risk, and converts fuzzy agricultural experience into a standardized data model. Compared with traditional extensive management relying on manual patrol, this method can scientifically guide the optimization of monitoring resources, and preferentially focus attention on high-risk areas, significantly improving the probability of early disease detection and management efficiency; not only provides a key static risk benchmark for subsequent dynamic monitoring, but also provides a long-term decision basis for planting planning and soil improvement, thereby reducing the possibility of large-scale disease outbreaks from the source.

[0033] 2、The present application constructs a deep recognition model integrating dynamic time series data, historical records and static spatial risk; through time series analysis algorithm, the continuous decline trajectory of the health status of the crop plant is accurately captured, and through cross verification with historical records and geographic risk, the abnormal signal is deeply analyzed and confirmed, thereby significantly improving the accuracy and reliability of risk identification and effectively avoiding misjudgment caused by a single data source.

[0034] 3、The present application realizes deep mining and intelligent analysis of the historical health records of the crop plant by dynamically querying the built-in agricultural knowledge base; converts the qualitative historical description into quantitative and diagnostic correlation coefficients, solves the problem that historical data cannot be directly utilized by the model; by comparing the actual intervention effect with the standard expected effect, the crop plants that react abnormally to the conventional management measures can be accurately identified, and the possible deep causes under the surface symptoms can be revealed; combined with the similarity analysis of typical disease symptoms, strong historical evidence support is provided for early warning of root diseases, and the logical reasoning ability and accuracy of the diagnosis model are significantly enhanced.

[0035] 4、The present application creates a high-confidence root disease diagnosis method through acoustic detection and collaborative analysis of three-dimensional electrical characteristic spatial distribution map; the matching of acoustic signals provides functional evidence at the physiological level for the abnormal areas found in the three-dimensional electrical characteristic spatial distribution map, significantly improves the specificity and accuracy of diagnosis, and effectively avoids misdiagnosis; further, through intelligent segmentation and calculation of the three-dimensional image, the precision and non-destructive quantification of the degree of root infection are realized; a scientific basis is provided for evaluating the disease severity grade of single plant, and a unified data basis is provided for subsequent regional disease assessment and prevention decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The structure diagram of the intelligent diagnosis system for animals and plants based on multi-source data of the present application;

[0037] Figure 2 The data flow diagram of the intelligent diagnosis system for animals and plants based on multi-source data of the present application;

[0038] Figure 3 The historical state correlation coefficient acquisition process diagram of the present application. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0040] Embodiment one:

[0041] Referring to Figure 1 and Figure 2 , the present application proposes an intelligent diagnosis system for animals and plants based on multi-source data, wherein Figure 1 is a structure diagram of the system, Figure 2 is a data flow diagram of the system; further, the system includes:

[0042] The data acquisition module identifies the geographical space risk map based on the stock parameters and soil parameters; obtains the historical state and intervention record of the crop plant, and obtains the historical state, intervention time, intervention measure and intervention effect of the crop plant.

[0043] The identification process of the geographical space risk map includes:

[0044] Obtain the digital information map, and label the stock parameters of the crop plant and the soil parameters of the plot where the crop plant is located;

[0045] According to the plant physiological characteristics database, a standardized rootstock resistance score is given to each rootstock, and according to the specific parameters of clay and sandy soil in the soil parameters, a standardized soil susceptibility score is given to the plots with different soil parameters;

[0046] Through a weighted fusion algorithm, the rootstock resistance score and the soil susceptibility score on the digital information map are calculated to generate a geographical spatial risk map.

[0047] The occurrence of diseases is the result of the joint action of internal factors (crop plant resistance) and external factors (environmental suitability); by quantifying and fusing these two key static factors, a prospective risk assessment can be made before the disease occurs, providing a scientific basis for subsequent monitoring resource allocation.

[0053] First, high-precision geographic information system (GIS) data is used to mark the planting of each grape tree on the map, and the built-in plant physiological characteristics database is queried to give the rootstock its corresponding root aphid resistance score as the rootstock resistance score; grape rootstock is the root system part of specially bred grape crop plants, and the most core and original function is to control root aphids.

[0049] Through grid soil sampling, the physical and chemical properties of the soil are analyzed to obtain key soil parameters, especially the accurate percentage content of clay and sandy soil; according to soil science knowledge, sandy soil is more conducive to the survival and spread of root aphids, while plots with high clay content are not conducive to the survival and spread of root aphids; therefore, according to the specific parameters of clay and sandy soil in the soil parameters, such as the accurate percentage content of clay and sandy soil, a standardized soil susceptibility score is given to plots with different soil parameters.

[0050] Through a weighted fusion algorithm, the comprehensive static risk value of each grape tree is calculated as a geographical spatial risk coefficient, and is visualized on a digital information map with a color gradient to form a geographical spatial risk map.

[0051] The present application integrates plant genetic resistance and soil environmental data to construct a quantitative and visual geographical spatial risk map; it realizes accurate prediction and spatial positioning of potential disease occurrence risk, and converts vague agricultural experience into a standardized data model. Compared with traditional extensive management relying on manual patrol, this method can scientifically guide the optimization of monitoring resources, and preferentially focus attention on high-risk areas, significantly improving the probability of early disease detection and management efficiency; not only provides a key static risk benchmark for subsequent dynamic monitoring, but also provides a long-term decision-making basis for planting planning and soil improvement, reducing the possibility of large-scale disease outbreaks from the source.

[0052] A feature recognition module periodically collects data on crop plants through remote sensing technology to obtain multi-source remote sensing time series data; a deep risk identification model is constructed to identify geographic spatial risk maps, multi-source remote sensing time series data, and historical states and intervention records to determine risk abnormal plants.

[0053] During the infection of grape crop plants by the grapevine aphid, damage to the root system of grape crop plants will first affect water absorption, leading to abnormal increases in canopy temperature; subsequently, it will affect nutrient transport, leading to decreases in chlorophyll content and photosynthesis efficiency; finally, it will affect the growth of the aboveground part, leading to changes in physical structure parameters. Periodic monitoring by a combination of multiple sensors can capture a series of cascading physiological responses triggered by root problems, and a comprehensive and dynamic health profile can be constructed.

[0054] Specifically, a drone equipped with multispectral, hyperspectral, thermal infrared, and laser radar sensors is used to cover and scan the vineyard every week according to a preset flight route. After the flight is completed, the system automatically processes the data: normalized vegetation index, chlorophyll content, nitrogen status parameters, etc. are calculated using multispectral and hyperspectral data; canopy temperature maps are generated using thermal infrared data, and water stress status is further analyzed; leaf area index and canopy volume are calculated using laser radar point cloud data, and changes in canopy density physical structure are measured based on the canopy volume.

[0055] Spectral data processing: using multispectral and hyperspectral data, a variety of vegetation indices such as normalized difference vegetation index (NDVI) and chlorophyll index are calculated through standard formulas, and chlorophyll content and nitrogen status are retrieved based on specific band reflectance.

[0056] Thermal infrared data processing: align the thermal infrared image with the visible light image, extract the canopy temperature of individual crop plants, and compare it with the ambient temperature to calculate the canopy temperature stress index, which is used to assess the water stress status.

[0057] Laser radar data processing: process point cloud data through ground point filtering, single plant segmentation, and other steps to accurately calculate the canopy height, canopy volume, leaf area index, and canopy density of each crop plant.

[0058] Time series data construction: all the above parameters obtained in each collection period are taken as a high-dimensional data vector, i.e. a time slice, arranged in chronological order to form multi-source remote sensing time series data for each crop plant.

[0059] The present application realizes all-around, high-precision and periodical non-destructive monitoring of the growth state of crop plants by integrating an unmanned aerial vehicle platform and multiple source sensors; the traditional single-point and destructive sampling detection is improved to macroscopic and non-contact three-dimensional information perception, greatly expanding the data dimension and collection efficiency; the generated multi-dimensional time series data can not only reveal the subtle changes of crop plants in spectrum, temperature and physical structure, but also capture the dynamic decline characteristics of crop plants over time, providing a high-quality data basis for the training of deep learning models.

[0060] The feature recognition module can further include a multi-disease concurrent recognition unit based on hyperspectral fingerprints;

[0061] The multi-disease concurrent recognition unit is configured to: obtain the hyperspectral reflectance curve of the crop plant canopy covering hundreds of continuous narrow wave bands by using the hyperspectral sensor; pre-construct a hyperspectral fingerprint feature library containing multiple diseases (such as root aphids, powdery mildew, and downy mildew) and nutrient stresses (such as nitrogen deficiency and iron deficiency), each stress type in the library corresponding to one and / or multiple unique spectral absorption / reflection feature patterns; automatically extract the key features of the hyperspectral curve of the crop plant to be tested by using a deep learning algorithm, and perform high-dimensional feature matching with the fingerprint library, so as to identify multiple diseases concurrently occurring on a single crop plant at one time and give the confidence scores of various diseases; and integrate the confidence scores of various diseases into a deep risk identification model to more accurately determine the risk abnormal plants.

[0062] Hyperspectral fingerprint feature library construction: a crop plant known to be infected with a specific disease (such as root aphids, powdery mildew) or under a specific nutrient stress (such as iron deficiency) is measured by a hyperspectral instrument, and the measured spectral curve is feature-extracted and labeled to construct a hyperspectral fingerprint feature library.

[0063] Through the above design, the diagnostic breadth and depth of the system are greatly expanded, making it upgrade from a specialized diagnostic system mainly targeting root diseases to a system capable of simultaneously identifying multiple above-ground and underground diseases, pests and nutrient problems; the advantages of rich hyperspectral data information are fully utilized to achieve accurate decoupling and concurrent identification of different stress types.

[0064] This solves the pain points of traditional diagnostic methods that can only target one problem at a time and are prone to confusing similar symptoms, and provides agricultural managers with a more comprehensive and three-dimensional understanding of crop health, which can guide more complex and comprehensive field management decisions and significantly improve the fine-grained level of agricultural production.

[0065] The identification process of the deep risk identification model includes: taking multi-source remote sensing time series data as the core, identifying crop plants with a slow and continuous decline trajectory of health indicators by using a time series analysis algorithm, and obtaining dynamic decline characteristics;

[0066] The historical state of the crop plant and the intervention record are called, a causal correlation analysis is carried out based on the historical state, intervention time, intervention measure and intervention effect, and a historical state correlation coefficient is obtained; a geographic space risk map is called, and a geographic space risk coefficient of the crop plant is obtained; comprehensive analysis is carried out based on the dynamic decline feature, the historical state correlation coefficient and the geographic space risk coefficient, and a comprehensive risk score is obtained; and the risk abnormal plant is determined based on the comprehensive risk score.

[0067] The deep risk identification model is constructed based on a long short-term memory network, is used for identifying a specific mode, that is, a slow but continuous decline trajectory of the health index in weeks or months, instead of a sharp fluctuation caused by short-term drought or improper management; and a dynamic decline feature vector is output, which quantifies the stability and severity of the decline trend.

[0068] The historical state correlation coefficient is obtained by calling the historical state and the intervention record, and the historical state correlation coefficient is calculated; the coefficient reflects the abnormal degree of the response of the crop plant to the past intervention measures, and the similarity of the symptoms and the target disease.

[0069] The geographic space risk coefficient is obtained by calling the generated geographic space risk map, and the geographic space risk coefficient of the location where the crop plant is located is extracted according to the accurate geographic coordinates of the crop plant.

[0070] The comprehensive risk score is calculated and the abnormal crop plant is determined: a comprehensive analysis model (such as a gradient boosting decision tree XGBoost) is designed, the features of the three dimensions are fused, and the comprehensive risk score is calculated.

[0071] The application constructs a deep identification model integrating dynamic time series data, historical records and static spatial risks; the continuous decline trajectory of the health state of the crop plant is accurately captured through the time series analysis algorithm, and the abnormal signal is deeply analyzed and confirmed through cross verification with the historical records and the geographic risk, the accuracy and reliability of risk identification are significantly improved, and the misjudgment caused by a single data source is effectively avoided.

[0072] The process of obtaining the historical state correlation coefficient includes: obtaining the historical state and the intervention record, obtaining the historical state, intervention time, intervention measure and intervention effect of the crop plant; based on the historical state, intervention time and intervention measure in the historical record, querying the built-in agricultural knowledge base, dynamically generating the quantitative standard expected effect; comparing the historical state and the intervention effect in the intervention record with the standard expected effect, and calculating the effect difference degree score between them;

[0073] The historical state in the record is compared with the typical symptom database of the root knot aphid, and a symptom similarity score is calculated;

[0074] The effect difference score and the symptom similarity score are integrated by a preset fusion function to obtain a historical state correlation coefficient. The acquisition process of the historical state correlation coefficient is as shown in Figure 3

[0075] The process is based on the principles of causal inference and knowledge graph. The core logic is that if the symptoms of a plant are highly similar to the infection of root knot nematodes, and the conventional intervention measures aimed at solving these surface symptoms do not produce the expected positive response, then these symptoms are likely not caused by common problems such as lack of fertilizer and lack of water, but by a deeper and more hidden cause such as root damage caused by root knot nematodes.

[0076] By quantifying the "difference in treatment effect" and "similarity of symptoms", the embodiment provides a logical framework for the computer model to evaluate the suspiciousness; and converts the scattered historical records into clear and quantified evidence, providing key historical background information for the deep risk identification model.

[0077] Specifically, a structured agricultural knowledge base is constructed in advance; the agricultural knowledge base includes: a symptom library storing typical above-ground symptom descriptions and quantified characteristics of diseases such as root knot nematodes, such as "leaf edge yellowing" and "new shoot growth stagnation";

[0078] Intervention-effect correlation library: stores standard farming operations and standard expected effects that should be produced on healthy crop plants. The expected effects are represented by quantified indicators.

[0079] Effect difference score calculation: retrieve the historical state and intervention records of the target crop plant, query the knowledge base according to the intervention measures in the records, obtain the standard expected effect, compare the actual intervention effect in the records with the standard expected effect, and calculate the effect difference score. The higher the score, the more sluggish the crop plant is to the conventional treatment, indicating a potential deep problem.

[0080] Symptom similarity score calculation: extract all symptom descriptions in the historical records, such as leaf yellowing and slow growth; compare these symptoms with the typical symptom database of root knot nematodes in the knowledge base; a natural language processing (NLP) text similarity algorithm can be used in combination with symptom image or symptom vector similarity to calculate a symptom similarity score, ranging from 0 to 1. The higher the score, the more the historical state conforms to the typical characteristics of root knot nematodes.

[0081] Fusion function calculation: the effect difference score and the symptom similarity score are integrated by a preset fusion function to obtain the final historical state correlation coefficient.

[0082] ​The application realizes deep mining and intelligent analysis of the historical health records of crop plants by dynamically querying the built-in agricultural knowledge base; converts qualitative historical descriptions into quantitative correlation coefficients with diagnostic value, solving the problem that historical data is difficult to be directly utilized by the model; through comparison of the actual intervention effect and the standard expected effect, the crop plants that react abnormally to the conventional management measures can be accurately identified, revealing the possible deep causes under the surface symptoms; combined with the similarity analysis of typical disease symptoms, it provides strong historical evidence support for early warning of root diseases, significantly enhancing the logical reasoning ability and accuracy of the diagnosis model.

[0083] The calculation process of the effect difference score can be further defined as:

[0084] The system is built-in a dynamic expected effect generation unit based on machine learning, which is used to replace the static standard expected effect in the built-in agricultural knowledge base; when the historical intervention records of the crop plants are obtained, this unit not only calls the type of intervention measures, but also synchronously obtains the physiological state represented by the multi-dimensional sensing data of the crop plants at the intervention time point, and the environmental meteorological data within a period of time after the intervention; based on the above multi-dimensional input, this unit dynamically predicts and generates an individualized expected health trajectory curve that the crop plant should have under certain conditions; the system forms an actual health trajectory curve with the actual health state data collected periodically subsequently, and compares it with the individualized expected trajectory curve as a whole, calculates the shape difference and separation degree of the two curves through time series similarity algorithm, and generates the final effect difference score.

[0085] Training of dynamic expectation model: a data set containing "pre-intervention state-intervention measure-post-intervention environment-actual intervention effect trajectory" is collected in advance. A time series prediction model is trained using these data; the model learns the dynamic effect that various intervention measures should produce under different initial conditions and different environments.

[0086] Generation of individualized expected health trajectory: when the system analyzes the historical records, the dynamic expected effect generation unit is triggered; the following are obtained:

[0087] Intervention measure: application of nitrogen fertilizer;

[0088] Initial state: multi-dimensional data of crop plant A on June 1 (such as NDVI = 0.65, canopy temperature = 28℃, leaf area index = 2.1);

[0089] Environmental data: weather forecast data from June 1 to June 15 (such as light, temperature, rainfall).

[0090] The model, based on the above inputs, predicts a future 14-day vegetation index expected change curve; for example, the expected vegetation index will start to steadily rise after 3 days, reach a peak of 0.72 on the 10th day, and then slowly stabilize.

[0091] Actual trajectory acquisition and comparison: The system uses subsequent remote sensing monitoring to obtain the actual vegetation index change curve of crop plant A from June 1 to June 15. For example, the actual vegetation index rises slightly to 0.67 in the first 5 days, but then begins to turn downward.

[0092] Difference score calculation: The dynamic time warping algorithm is used to calculate the warping path distance between the personalized expected health trajectory curve and the actual health trajectory curve; the dynamic time warping algorithm can effectively measure the similarity between two unequal length time series, even if there is a certain delay or stretching on the time axis. The distance calculated after normalization is the final effectiveness difference score; a larger score means that the actual recovery is significantly deviated from the most ideal recovery under the current conditions.

[0093] The design creatively upgrades the evaluation of effectiveness difference from a static, single-point numerical comparison to a dynamic, multi-point, and full-process trajectory comparison. Its core advantages are:

[0094] Highly contextual and personalized: By dynamically generating expected results, the system takes into account key factors such as the health of the crop plant itself and environmental variables during the intervention period, and customizes highly scientific and personalized benchmarks for each intervention.

[0095] Improved diagnostic information dimension: By comparing the trajectory of the entire recovery process, the system can capture more rich and subtle abnormal signals. For example, it can distinguish between different patterns of abnormal reactions such as "completely ineffective", "first rising and then falling", or "extremely slow recovery", which may correspond to different underlying causes, providing deeper insights for diagnosis.

[0096] Robustness and accuracy significantly enhanced: This method can effectively filter out crop plants with poor results due to poor weather or poor crop plant state, providing reliable evidence for identifying deep and chronic diseases.

[0097] The infection judgment module acquires data from the risk abnormal plant through the acoustic detection device to obtain acoustic feature data; it identifies the risk abnormal plant through the ring electrode array to generate a three-dimensional electrical characteristic spatial distribution map of the root system; based on the acoustic feature data and the three-dimensional electrical characteristic spatial distribution map, the abnormal state index of the crop plant is identified.

[0098] The process of generating the three-dimensional electrical characteristic spatial distribution map of the root system includes: deploying a ring electrode array in the soil around the base of the risk abnormal plant, sequentially applying weak, multi-frequency high-frequency alternating current between different electrode pairs by a host system, and synchronously measuring the voltage distribution on all the remaining electrodes to collect a complete electrical response data set;

[0099] The electrical response data set is solved by inverse operation through a reconstruction algorithm to generate a three-dimensional matrix in voxel units; a three-dimensional electrical characteristic spatial distribution map is obtained from the three-dimensional matrix, including the three-dimensional coordinates, electrical impedance value and capacitance characteristic parameters obtained by multi-frequency scanning of each voxel.

[0100] The process is based on bioelectrical impedance tomography (EIT) technology. The core physical principle is that different biological tissues have different electrical characteristics, such as electrical resistivity and dielectric constant. Healthy plant root tissue, diseased tissue after being invaded by pests such as root aphids, and surrounding soil medium all have different cell structures, water content and ion concentrations, and therefore exhibit significantly different electrical impedance values. By applying an excitation current around the root system and measuring the boundary voltage, the EIT system can collect information reflecting the internal conductivity distribution.

[0101] The reconstruction algorithm is a mathematical "inverse problem" solving process that can invert the internal attribute distribution from external measurements. Multi-frequency scanning is selected to obtain more abundant tissue information, because different frequencies of current have different sensitivities to cell membranes (capacitance characteristics) and cell fluids (resistance characteristics), which helps to more accurately distinguish healthy and diseased tissues.

[0102] The "adjacent excitation-adjacent measurement" mode is adopted, and a pair of adjacent electrodes is selected as the excitation electrodes to apply a weak, multi-frequency high-frequency sinusoidal alternating current. At the same time of applying the excitation, the voltage distribution on all adjacent electrode pairs is synchronously measured.

[0103] The excitation electrode pairs are automatically switched in sequence, and the above process is repeated until all possible excitation measurement combinations are completed. The entire process forms a complete data set containing multi-frequency, multi-angle electrical responses.

[0104] Image reconstruction: input the collected electrical response data set into a computer equipped with an image reconstruction algorithm; adopt an inverse problem solving algorithm based on the finite element method, such as the Gauss-Newton algorithm. First, a three-dimensional finite element mesh model matching the actual detection area is established. The reconstruction algorithm calculates the electrical impedance value of each element (voxel) in the mesh through iterative optimization, so that the boundary voltage calculated by the positive problem under the electrical impedance distribution can best match the actual measured voltage data.

[0105] Finally, the algorithm outputs a three-dimensional matrix in voxel units. In this matrix, each voxel contains its three-dimensional spatial coordinates, the electrical impedance value at this location, and the electrical capacitance characteristic parameters (such as phase angle) obtained from multi-frequency scanning.

[0106] Three-dimensional image visualization: using three-dimensional visualization software, the reconstructed three-dimensional matrix is rendered into an intuitive three-dimensional electrical property spatial distribution map. In the image, different electrical impedance values can be represented by different colors and transparencies, clearly showing the spatial distribution of the electrical properties of the root system and the surrounding soil medium.

[0107] The present application applies electrical impedance tomography technology to non-destructive testing of plant root systems, achieving three-dimensional, visualized and accurate diagnosis of the health status of underground root systems. The generated three-dimensional electrical property spatial distribution map can intuitively display the location, morphology and approximate range of the lesion area, with an accuracy of centimeters; the electrical capacitance characteristic parameters obtained through multi-frequency scanning further enhance the ability to distinguish different tissue types; it breaks through the limitations of traditional root system detection, which relies on excavation sampling, is highly destructive and cannot be repeatedly observed, and provides a dynamic and effective observation method.

[0108] The process of identifying the abnormal state index of the crop plant includes: monitoring the elastic wave generated in the xylem conduit due to the breakage of the water column under water stress of the crop plant to obtain acoustic characteristic data;

[0109] According to the matching of the acoustic characteristic data and the preset root aphid threat sound spectrum library, acoustic correlation characteristic data is obtained; based on the acoustic correlation characteristic data and the three-dimensional coordinates, electrical impedance values and electrical capacitance characteristic parameters of the voxels in the three-dimensional electrical property spatial distribution map, the crop plant infected by the root aphid is judged; the root aphid threat sound spectrum library is obtained by acoustic detection of the plant infected by the root aphid.

[0110] The three-dimensional electrical property spatial distribution map is processed by a three-dimensional image segmentation algorithm, and based on the electrical impedance values and the three-dimensional morphology of the lesion tissue, the total volume of the lesion tissue and the total volume of the healthy root system are identified and calculated, and the abnormal state index is calculated.

[0111] This process is based on the principle of multi-physical field coupling diagnosis, and combines two independent physical detection methods to verify each other, improving the accuracy of diagnosis.

[0112] Acoustic detection: based on plant physiology, root damage leads to water absorption disorders, which will cause xylem embolism under water stress, producing unique acoustic emission signals; this is a functional detection method to determine whether the root system is abnormal.

[0113] EIT detection: Based on biophysics, directly imaging the root system in three dimensions, observing its structure; this is a morphological detection, judging where the root system has abnormalities.

[0114] Combining the functional abnormality evidence of acoustic detection with the structural abnormality evidence of EIT can form a complete diagnostic closed loop, which greatly eliminates the interference of other stress factors. Finally, through three-dimensional image segmentation and volume calculation, the qualitative diagnostic result is converted into a precise, standardized quantitative index - abnormality index, which is crucial for evaluating the disease, guiding treatment and predicting yield loss.

[0115] Specifically, acoustic feature data acquisition: Install a high-sensitivity piezoelectric acoustic emission sensor at the base of the main stem of the risk abnormal plant, which is responsible for monitoring and recording the weak elastic wave (stress wave) signals generated by the rupture of water column in the xylem vessels of the crop plant. The collection frequency is 20-200 kHz, forming the original acoustic feature data.

[0116] Acoustic data analysis and matching: Filter and feature extraction on the collected acoustic signals, calculate their amplitude, duration, energy and other parameters; match the extracted signal features with a pre-set root aphid threat sound spectrum library. This library pre-stores a large number of experimentally confirmed air pocket acoustic signal patterns specific to root aphid infestation-induced root water absorption dysfunction.

[0117] Through pattern recognition algorithms such as dynamic time warping (DTW), the matching degree of the measured signal and the feature signal in the library is calculated, and the acoustic correlation feature data is obtained. High matching degree indicates that the detected water stress is most likely caused by root aphids.

[0118] Multi-modal data comprehensive analysis and infection judgment: Acoustic analysis results as key verification information, if the crop plant shows high electrical impedance abnormal regions (from EIT) and high matching degree of stress acoustic signals at the same time, the system judges that the crop plant is infected with root aphids with high confidence.

[0119] Abnormality index calculation: Process the three-dimensional electrical property spatial distribution map of the diagnosed crop plant. First, use three-dimensional image segmentation algorithms (such as threshold-based segmentation or more advanced deep learning segmentation models such as U-Net), according to the difference in electrical impedance value between healthy roots and diseased tissues (diseased tissues usually have higher electrical impedance), accurately divide the voxels in the three-dimensional image into three categories: healthy roots, diseased tissues and soil background.

[0120] Based on the segmentation results, automatically calculate the total volume of all voxels identified as diseased tissues and the total volume of all voxels identified as healthy roots, and finally quantify the abnormality index.

[0121] The present application creates a high-confidence root disease diagnosis method through acoustic detection and collaborative analysis of three-dimensional electrical characteristic spatial distribution map; the matching of acoustic signals provides functional evidence at the physiological level for the abnormal regions found in the three-dimensional electrical characteristic spatial distribution map, significantly improves the specificity and accuracy of diagnosis, and effectively avoids misdiagnosis; further, through intelligent segmentation and calculation of the three-dimensional image, the precision and non-destructive quantification of the infection degree of the root system are realized; a scientific basis is provided for evaluating the disease severity grade of single plant crop plants, and a unified data basis is provided for subsequent regional disease assessment and prevention decision-making.

[0122] The abnormal early warning module performs cluster analysis based on the abnormal state index and three-dimensional coordinates of the crop plants to determine the infection center and the infection area and perform abnormal early warning.

[0123] The process of determining the infection center and the infection area includes: data binding of the abnormal state index of all diagnosed crop plants and three-dimensional geographic coordinates to form a spatial point data set; using a density-based spatial clustering algorithm to analyze the spatial point data set, and identifying crop plants that are adjacent in space and have infection indexes reaching a preset high-risk threshold as a cluster;

[0124] The geometric center point of the cluster with the highest density is determined as the infection center, and the geographic range of the cluster covered by the infection center is determined as the infection area.

[0125] This process is based on the principles of geographic information science and spatial statistics, especially the spatial autocorrelation theory, i.e., geographically adjacent things are more likely to have similar properties. The spread of diseases is usually not random, but centered on one or more source points and distributed outward in a clustered manner; the core advantage of the density-based spatial clustering algorithm is to find density-based clustering patterns; through this algorithm, scattered single plant diagnosis information can be upgraded to regional disease situation insight with spatial pattern.

[0126] Determining the infection center is to find the core point of the most severe disease, which is most likely to be the source of the spread, for traceability analysis and centralized processing; determining the infection area is to define a clear management boundary to guide the scope of agricultural operations (such as isolation and precise pesticide application) and prevent further spread of the disease.

[0127] The abnormal early warning module further includes a spatio-temporal diffusion prediction unit.

[0128] The spatio-temporal diffusion prediction unit is used for: comprehensive call of the determined infection center and infection area, geographical space risk atlas, external acquisition of real-time meteorological data and soil parameter distribution map; based on a preset disease diffusion model, simulating and predicting the most likely diffusion direction, speed and range of the infection area in the future period of time; finally, in addition to marking the current infection area on the early warning map, the potential high-risk spread area in the future is visualized in the form of a dynamic, gradual risk layer.

[0129] Multi-source data integration: after the spatio-temporal diffusion prediction unit is activated, the current determined infection center is first acquired, then the GSR geographical space risk atlas is loaded as the "resistance / attraction" background of diffusion. Finally, the wind speed / direction data of the local meteorological station and the humidity data of the soil moisture monitoring network are acquired in real time through the API interface.

[0130] Diffusion model simulation: the entire plot is gridded using a cellular automaton model, and the state (healthy / latent / infection) of each grid cell is iteratively updated according to its own state, neighbor state, and a series of transition rules.

[0131] Rule definition: the probability of diffusion in the wind direction is much higher than that in the opposite direction; in areas with high GSR values (such as sandy soil), the diffusion speed is faster; in areas with excessively dry or excessively wet soil, the diffusion speed will slow down.

[0132] Prediction and visualization: the model iteratively simulates 14 cycles (representing 14 days) to generate daily spatial distribution prediction maps of the disease. In the final early warning map, the prediction results of the 14 days are superimposed to represent the potential infection range at different time points in the future, forming a dynamic risk corridor.

[0133] This design greatly improves the forward-looking and initiative of the early warning system, upgrading the traditional static warning to dynamic prediction. By integrating multi-physical field data and simulating disease transmission dynamics, it predicts where the disease will go. It provides active intervention strategies to effectively break the disease transmission chain and realizes the intelligent upgrade from risk management to future risk prediction.

[0134] The present application combines the diagnosis data of single crop plants with geographical space information, and uses a density-based clustering algorithm to realize the leap from point diagnosis to surface state perception; it can automatically and objectively identify the spatial distribution pattern of diseases in the field, accurately locate the infection center and delineate the infection area boundary; it implements a zoning management strategy, realizes the transition from passive and comprehensive prevention and control to active and precise regional prevention and control, significantly improves the prevention and control efficiency and reduces resource waste.

[0135] Example two:

[0136] The application provides a plant and animal intelligent diagnosis system based on multi-source data, and specific implementation details are given based on embodiment one, including:

[0137] A data acquisition module identifies a geographic space risk atlas based on stock parameters and soil parameters, obtains historical states and intervention records of crop plants, and obtains historical states, intervention time, intervention measures and intervention effects of the crop plants;

[0138] A feature recognition module periodically collects data of the crop plants by using remote sensing technology, obtains multi-source remote sensing time series data, constructs a deep risk identification model, identifies the geographic space risk atlas, the multi-source remote sensing time series data and the historical states and intervention records, and determines risk abnormal plants;

[0139] An infection judgment module collects data of the risk abnormal plants by using acoustic detection equipment, obtains acoustic characteristic data, identifies the risk abnormal plants by using a ring electrode array, generates a three-dimensional electrical characteristic space distribution map of root systems, and identifies an abnormal state index of the crop plants based on the acoustic characteristic data and the three-dimensional electrical characteristic space distribution map;

[0140] An abnormal early warning module performs clustering analysis based on the abnormal state index of the crop plants and three-dimensional coordinates, determines an infection center and an infection area, and performs abnormal early warning.

[0141] Specifically, in the GIS platform, a digital orthographic image map, a vector planting point layer containing stock types and a vector land layer of soil types (for example, sandy loam and clay loam) are spatially registered and superimposed.

[0142] According to an embedded plant physiological characteristic database, different stocks are valued. According to soil physical and chemical parameters, different soil types are valued. For example, the sandy loam (high risk) is set to 8.5, and the clay loam (low risk) is set to 3.0.

[0143] The geographic space risk coefficient of each crop plant point on the map is calculated by using a weighted fusion algorithm.

[0144] A Kriging spatial interpolation method is used to generate a smooth risk level grid layer covering the entire monitoring area based on the geographic space risk coefficients of all points; the layer directly shows the low-to-high static risk distribution by using a color spectrum (green-yellow-red).

[0145] An unmanned aerial vehicle platform is adopted, multi-spectral sensors, hyperspectral sensors, thermal infrared sensors and laser radar sensors are carried on the unmanned aerial vehicle, the crop plants are periodically scanned, data is collected in a 7-day cycle, and the data collection is performed in a key growth period of the crop plants.

[0146] Vegetation index, chlorophyll content, and nitrogen status parameters were obtained through multispectral and hyperspectral sensors; canopy temperature was obtained through thermal infrared sensors to assess water stress status caused by root damage; three-dimensional point cloud data of the canopy were obtained through a laser radar sensor to calculate leaf area index and canopy density physical structure parameters.

[0147] The multi-dimensional data obtained in each collection cycle was taken as a time slice, and multi-source remote sensing time series data were formed in the collection order.

[0148] The time series data were input into a pre-trained deep risk identification model, and the model output a high score of dynamic decline characteristics for crops showing persistent slow decline.

[0149] The deep risk identification model was constructed with a long short-term memory network as the core, containing two hidden layers with 64 units each. The input data were continuous multi-collection cycle multi-source remote sensing data feature vectors. The output of the LSTM was a feature value quantifying the decline trend; the feature value, together with the historical state correlation coefficient and the geographical space risk coefficient, formed a 3-dimensional feature vector, which was input into a pre-trained XGBoost classifier; the classifier contained 150 estimators with a maximum depth of 4; the model output a comprehensive risk score between 0 and 1, and when the score was greater than 0.8, the plant was determined to be a risk abnormal plant.

[0150] Historical record causal correlation analysis: the system automatically retrieves historical intervention records, compares the records with the agricultural knowledge base, calculates the difference score, and finally generates the historical state correlation coefficient. The fusion model integrates the information in each dimension to calculate the comprehensive risk score.

[0151] Three-dimensional electrical impedance tomography (EIT): a ring-shaped electrode array was deployed on the high-risk crop plants for multi-frequency scanning from 1 kHz to 100 kHz; the reconstruction algorithm generated a three-dimensional electrical impedance distribution voxel model of the root region. The model showed that the electrical impedance value of healthy root tissue was about 500 Ω·m, while a large area of abnormal high impedance region with a value exceeding 1500 Ω·m appeared in the model.

[0152] Acoustic emission (AE) signal cross-validation: AE sensors were placed on the main stems of the crop plants to monitor several acoustic signals caused by water stress-induced xylem embolism. The peak frequency (about 150 kHz) and energy distribution of the signals matched 92% of the root aphid stress sound spectrum library, providing functional evidence for the structural abnormalities found by EIT.

[0153] Abnormal state index calculation: a 3DU-Net segmentation algorithm was used to intelligently segment the EIT three-dimensional model, and the total volume of diseased tissue, the total volume of healthy root tissue, and the final quantitative index were automatically calculated.

[0154] Spatial point dataset generation: bind the geographic coordinates of all diagnosed crop plants with the abnormal state index to form a spatial point dataset; based on the density-based clustering analysis to calculate the geometric center of the cluster as the core point of the disease outbreak. Infection area: generate the minimum convex hull that surrounds all points of the cluster, define it as the core infection area, and based on this generate a final control management area containing a safety buffer zone.

[0155] The system displays the infection center with a highlighted icon on the GIS early warning platform, displays the control management area as a semi-transparent surface layer, and attaches clear control recommendations to complete the entire intelligent diagnosis and early warning closed loop.

[0156] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent diagnosis system for animals and plants based on multi-source data, characterized in that, The method comprises the following steps: a data acquisition module obtains a geographical space risk map based on stock parameters and soil parameters; obtain the history state and intervention record of the crop plant, and obtain the history state, intervention time, intervention measure and intervention effect of the crop plant; a feature recognition module periodically collects data of the crop plant through remote sensing technology to obtain multi-source remote sensing time series data; a deep risk identification model is constructed to identify the geographical space risk map, multi-source remote sensing time series data and history state and intervention record, and determine the risk abnormal plant; an infection judgment module collects data of the risk abnormal plant through an acoustic detection device to obtain acoustic feature data; a ring electrode array is used to identify the risk abnormal plant, and a three-dimensional electrical characteristic space distribution map of the root system is generated; based on the acoustic feature data and the three-dimensional electrical characteristic space distribution map, the abnormal state index of the crop plant is identified; an abnormal early warning module performs clustering analysis based on the abnormal state index and three-dimensional coordinates of the crop plant to determine the infection center and infection area, and performs abnormal early warning.

2. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, wherein: the identification process of the geographical space risk map comprises: obtaining a digital information map, and marking the stock parameters of the crop plant and the soil parameters of the plot where the crop plant is located; according to a plant physiological characteristic database, a standardized stock resistance score is given to each stock, and according to the specific parameters of clay and sand in the soil parameters, a standardized soil susceptibility score is given to the plot with different soil parameters; the stock resistance score and the soil susceptibility score on the digital information map are calculated by a weighted fusion algorithm to generate a geographical space risk map.

3. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, wherein: the acquisition method of the multi-source remote sensing time series data comprises using a drone carrying a multi-spectral sensor, a hyperspectral sensor, a thermal infrared sensor and a laser radar sensor to periodically scan the crop plant by remote sensing; wherein the vegetation index, chlorophyll content and nitrogen status parameter are obtained by the multi-spectral sensor and the hyperspectral sensor; the crown temperature is obtained by the thermal infrared sensor to evaluate the water stress state caused by root damage; the crown three-dimensional point cloud data is obtained by the laser radar sensor to calculate the leaf area index and the crown density physical structure parameter; the multi-dimensional data obtained in each collection period is taken as a time slice to form multi-source remote sensing time series data in the collection order.

4. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, wherein: the identification process of the deep risk identification model comprises: taking the multi-source remote sensing time series data as the core, identifying the crop plant with persistent decline of health indicators by a time series analysis algorithm, and obtaining dynamic decline characteristics; The historical state of the crop plant and the intervention record are called, causal correlation analysis is performed based on the historical state, intervention time, intervention measure and intervention effect, and a historical state correlation coefficient is obtained; a geographical space risk map is called, and a geographical space risk coefficient of the crop plant is obtained; comprehensive analysis is performed based on the dynamic decline characteristics, the historical state correlation coefficient and the geographical space risk coefficient, and a comprehensive risk score is obtained; and the risk abnormal plant is determined based on the comprehensive risk score.

5. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 4, characterized in that: The process of obtaining the historical state correlation coefficient comprises: obtaining the historical state and the intervention record, and obtaining the historical state, intervention time, intervention measure and intervention effect of the crop plant; based on the historical state, intervention time and intervention measure, querying the built-in agricultural knowledge base to dynamically generate a quantitative standard expected effect; comparing the intervention effect in the historical state and intervention record with the standard expected effect to calculate an effect difference score; characteristic comparison is performed between the historical state in the record and a typical symptom database of the root aphid to calculate a symptom similarity score; The effect difference score and the symptom similarity score are comprehensively calculated by a preset fusion function to obtain the historical state correlation coefficient.

6. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, characterized in that: The process of generating the three-dimensional electrical characteristic space distribution map of the root system comprises: deploying a ring electrode array in the soil at the base of the risk abnormal plant, applying weak and multi-frequency high-frequency alternating current between different electrode pairs in sequence by the host system, and synchronously measuring the voltage distribution on all the remaining electrodes to collect complete electrical response data sets; a three-dimensional matrix in voxel units is solved by inverse operation of the electrical response data sets by a reconstruction algorithm; and a three-dimensional electrical characteristic space distribution map is obtained from the three-dimensional matrix, including the three-dimensional coordinates, electrical impedance value and capacitance characteristic parameters of each voxel.

7. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, characterized in that: The process of identifying the abnormal state index of the crop plant comprises: monitoring the elastic wave generated in the xylem vessel due to water column fracture under water stress of the crop plant to obtain acoustic characteristic data; acoustic correlation characteristic data are obtained by matching the acoustic characteristic data with a preset root aphid threat sound spectrum library; the crop plant infected by the root aphid is determined based on comprehensive analysis of the acoustic correlation characteristic data and the three-dimensional coordinates, electrical impedance value and capacitance characteristic parameters of the voxels in the three-dimensional electrical characteristic space distribution map; the three-dimensional electrical characteristic space distribution map is processed by a three-dimensional image segmentation algorithm, the total volume of the diseased tissue and the total volume of the healthy root system are identified and calculated based on the electrical impedance value and the three-dimensional morphology of the diseased tissue, and the abnormal state index is calculated.

8. The intelligent diagnosis system for animals and plants based on multi-source data according to claim 1, characterized in that: The process of determining the infection center and the infection area comprises: data binding of the abnormal state index of the crop plants and three-dimensional geographic coordinates to form a spatial point data set; adopting a density-based spatial clustering algorithm to analyze the spatial point data set, and identifying crop plants that are adjacent in space and have infection indexes reaching a preset high-risk threshold as a cluster; determining the geometric center point of the cluster with the highest density as the infection center, and determining the geographic range of the cluster covered at the infection center as the infection area.

Images