Intelligent orchard multi-source information fusion monitoring system and method

Abstract

The invention relates to the technical field of intelligent agriculture and industrial data processing, and particularly discloses an intelligent orchard multi-source information fusion monitoring system and method. According to the invention, the heterogeneous sensing terminal is deployed to collect meteorological, soil, pest and disease damage and fruit tree growth state multi-dimensional data, and through standardized analysis and time-space association fusion, a streaming and distributed calculation engine is called to carry out real-time cleaning and historical analysis, so that the problems of different data standards, insufficient fusion, weak processing capability and data islanding are solved; and the collaborative utilization rate and the intelligent management level of the orchard multi-source information are improved.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture and industrial data processing technology, and in particular to a smart orchard multi-source information fusion monitoring system and method. Background Technology

[0002] With the rapid development of modern agriculture towards intelligentization, the refined management of orchards places higher demands on multi-dimensional information monitoring. Currently, orchard monitoring mainly relies on manual inspections or single-function monitoring equipment to collect information on weather, soil, pests and diseases, and fruit tree growth status. The data standards of different collection terminals are inconsistent, and there is a lack of effective fusion mechanisms, making it difficult to collaboratively utilize monitoring data from different sources and creating data silos. At the same time, existing systems have weak processing capabilities at the industrial information and data processing levels. When faced with the massive amounts of heterogeneous data generated by large-scale orchards, computing resources are limited, and there is a lack of efficient real-time analysis architecture. The depth of data value mining is insufficient, failing to provide timely and reliable technical support for precision irrigation, fertilization decisions, and pest and disease early warning. This dual limitation of information fragmentation and processing capacity severely restricts the intelligence level and practical effect of orchard monitoring systems, urgently requiring technological breakthroughs through multi-source information fusion and improvements in industrial-grade data processing capabilities. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides an intelligent orchard multi-source information fusion monitoring system and method.

[0004] Firstly, the present invention provides a smart orchard multi-source information fusion monitoring system, the technical solution of which is as follows:

[0005] The multi-source data acquisition module is used to collect multi-dimensional monitoring data through multiple heterogeneous sensor terminals deployed in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data.

[0006] The data standardization module is used to parse the multi-dimensional monitoring data to obtain raw data fields containing monitoring values ​​and data labels, and based on the raw data fields, to perform format conversion and semantic normalization processing according to a preset unified data protocol, and output unified format data.

[0007] The multi-source information fusion module is used to align the unified format data using a data association method based on timestamps and spatial locations, and to perform feature-level fusion and decision-level fusion to generate multi-source fused data.

[0008] The industrial information and data processing module is used to call the streaming computing engine to perform real-time cleaning and anomaly detection on the multi-source fusion data, and to call the distributed computing framework to perform batch processing analysis on historical multi-source fusion data, outputting analysis results including real-time status and trend prediction.

[0009] The decision support module is used to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning based on the analysis results and by calling a preset decision rule library.

[0010] Furthermore, the multi-source data acquisition module includes a wireless sensor network, which consists of the heterogeneous sensor terminals and gateway nodes, and is used to upload the multi-dimensional monitoring data to the data standardization module.

[0011] Furthermore, the data standardization module is also used to validate the original data fields and attach data quality labels to the validated original data fields. The format conversion and semantic normalization processing are performed based on the data quality labels.

[0012] Furthermore, the data association method based on timestamps and spatial locations in the multi-source information fusion module specifically involves performing spatiotemporal matching on the unified format data according to the data collection time and the preset orchard geographic grid code.

[0013] Furthermore, the industrial information and data processing module performs real-time cleaning and anomaly detection on the multi-source fused data, including statistical outlier detection based on a sliding time window.

[0014] Furthermore, the decision support module also includes a decision optimization unit, which is used to dynamically update the decision rule base based on the feedback decision execution effect data.

[0015] Furthermore, the system also includes a visualization interaction module, which is used to receive and display the analysis results and the orchard management decision instructions, and to receive user operation feedback.

[0016] Furthermore, the visualization interaction module can be displayed in the following ways: a spatial distribution map of monitoring data based on an electronic map, historical data trend curves, and a decision command execution status panel.

[0017] Furthermore, when performing feature-level fusion, the multi-source information fusion module uses an adaptive fusion algorithm based on dynamic credibility and spatiotemporal correlation to calculate the fusion weights.

[0018] The adaptive fusion algorithm calculates the fusion weight W_i(x, t) of the i-th data source at time t for a specific monitoring parameter x using the following formula:

[0019]

[0020] in, This represents the fusion weight of the i-th data source for monitoring parameter x at time t; This represents the historical data reliability evaluation value of the i-th data source up to time t, which is obtained through data backtracking verification. This represents the inherent confidence coefficient for the i-th data source type; This is a dynamic weight adjustment factor, with a value ranging from 0 to 1; This represents the difference between the current time t and the latest data timestamp from the i-th data source; This is the time decay coefficient; This represents the average spatiotemporal data difference between the i-th data source and other data sources in this fusion. This is the difference penalty coefficient; This indicates the total number of data sources participating in the integration.

[0021] Secondly, this invention provides a smart orchard multi-source information fusion monitoring method, the technical solution of which is as follows:

[0022] Multi-dimensional monitoring data is collected by multiple heterogeneous sensor terminals deployed in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data.

[0023] The multi-dimensional monitoring data is parsed to obtain raw data fields containing monitoring values ​​and data tags. Based on the raw data fields, format conversion and semantic normalization are performed according to a preset unified data protocol to output unified format data.

[0024] The unified format data is aligned using a data association method based on timestamps and spatial locations, and feature-level fusion and decision-level fusion are performed to generate multi-source fused data.

[0025] The streaming computing engine is invoked to perform real-time cleaning and anomaly detection on the multi-source fusion data, and the distributed computing framework is invoked to perform batch processing analysis on the historical multi-source fusion data, outputting analysis results including real-time status and trend prediction.

[0026] Based on the analysis results, a preset decision rule base is invoked to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning.

[0027] The technical solution of this invention collects multi-dimensional data on meteorology, soil, pests and diseases, and fruit tree growth status by deploying heterogeneous sensing terminals. After standardized analysis and spatiotemporal correlation fusion, the data is cleaned and analyzed in real time by calling streaming and distributed computing engines. This solves the problems of inconsistent data standards, insufficient fusion, weak processing capabilities, and data silos, and improves the collaborative utilization rate of multi-source information in orchards and the level of intelligent management.

[0028] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0030] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0031] Figure 1 This is a schematic diagram of an embodiment of the intelligent orchard multi-source information fusion monitoring system of the present invention;

[0032] Figure 2 This is a flowchart illustrating an embodiment of a smart orchard multi-source information fusion monitoring method according to the present invention. Detailed Implementation

[0033] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0034] Figure 1 This diagram illustrates the structure of an embodiment of a smart orchard multi-source information fusion monitoring system provided by the present invention. Figure 1 As shown, the system includes:

[0035] The multi-source data acquisition module 110 is used to collect multi-dimensional monitoring data through multiple heterogeneous sensor terminals deployed in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data.

[0036] Heterogeneous sensing terminals refer to various sensor devices deployed in orchards that collect data based on different working principles or communication protocols; for example, temperature and humidity sensors, soil pH sensors, multispectral cameras, and trunk micro-change sensors installed simultaneously in an orchard, each collecting different physical quantities. Multidimensional monitoring data refers to a collection of various types of monitoring data reflecting the orchard environment and crop status from different perspectives; for example, air temperature and humidity, soil temperature and humidity, nutrient content, captured images of diseased and pest-infested leaves, and data on changes in fruit tree stem diameter collected by various sensing terminals in an orchard. Meteorological data refers to data describing the state and changes of the orchard's atmospheric environment; for example, data collected by a meteorological station in an orchard, including temperature, humidity, wind speed, wind direction, light intensity, and rainfall. Soil data refers to data describing the physical and chemical properties of orchard soil; for example, data collected by soil sensors in an orchard, including soil moisture, temperature, pH, and nitrogen, phosphorus, and potassium content. Pest and disease image data refers to visual data acquired through image acquisition equipment used to identify and assess the status of fruit tree pests and diseases; for example, images of fruit tree leaves and fruits regularly taken by high-definition cameras deployed in an orchard, which may contain lesions or pest characteristics. Fruit tree growth status data refers to data that directly or indirectly reflects the dynamic physiological growth of fruit trees; for example, data collected in an orchard using stem flow sensors, trunk diameter sensors, or chlorophyll fluorometers, used to indicate water transport, stem thickening, or photosynthetic efficiency in fruit trees.

[0037] The data standardization module 120 is used to parse the multi-dimensional monitoring data to obtain raw data fields containing monitoring values ​​and data labels, and based on the raw data fields, to perform format conversion and semantic normalization processing according to a preset unified data protocol, and output unified format data.

[0038] Among them, the monitored value refers to the specific value of a physical or chemical quantity directly measured by the sensing terminal or obtained through preliminary calculation; for example, the specific reading of "25.3" collected and reported by a soil moisture sensor in an orchard at a specific time. The data tag refers to the metadata identifier used to describe the source, attributes, unit and collection conditions of the monitored value; for example, the tag information attached to the monitored value "25.3" includes the sensor ID "SoilSensor_ZoneA_01", the monitoring parameter "soil volumetric water content", the unit "%" and the collection timestamp "2025-10-27 10:00:00". Raw data fields refer to structured data units composed of monitored values ​​and their associated data tags; for example, a data packet unit transmitted in an orchard might contain the following content: {Sensor ID: “SoilSensor_ZoneA_01”, Parameter: “Soil volumetric water content”, Value: 25.3, Unit: “%”, Time: “2025-10-27 10:00:00”}. A unified data protocol refers to a predefined set of standards covering data field definitions, encoding formats, unit systems, and transmission specifications to eliminate data heterogeneity; for example, a mandatory data specification used within an orchard monitoring system requires all soil moisture data to use “Soil_Moisture” as the field name, adopt a floating-point format, use a uniform percentage (%) unit, and use a UTC timestamp. Format conversion and semantic normalization refer to the process of standardizing the numerical format, units, and semantic descriptions of raw data fields according to a unified data protocol. For example, converting the raw field "{humidity: 72, unit: 'RH%'}" reported by a sensor in an orchard into the unified format "{Air_Humidity: 72.0, unit: '%'}". Unified format data refers to standard structured data that conforms to a preset unified data protocol after format conversion and semantic normalization. For example, all temperature and humidity data from various sources in an orchard are ultimately processed into a JSON format data stream with the same field names, data types, and units.

[0039] The multi-source information fusion module 130 is used to align the unified format data using a data association method based on timestamps and spatial locations, and to perform feature-level fusion and decision-level fusion to generate multi-source fused data.

[0040] Data association based on timestamps and spatial location refers to an algorithm that matches and binds data from different sources that are spatially and temporally similar, using data acquisition time information and sensor geographic coordinates. For example, pairing meteorological data and soil data from an orchard within one minute before and after 10:00:00 on October 27, 2025, all located within the geographic grid "Row 5 of Area A". Feature-level fusion refers to fusion at the data feature level, combining or weighting feature vectors from uniformly formatted data from different sources to generate new fused features. For example, combining soil moisture, air temperature, and leaf color features (extracted from images) at the same location and time in an orchard into a multi-dimensional feature vector. Decision-level fusion refers to fusion at the preliminary decision or judgment result level, integrating intermediate decisions derived from different data sources to arrive at a final joint decision. For example, combining the "suspected disease" conclusion derived from image analysis and the "high infection risk" conclusion derived from microclimate data analysis in an orchard to ultimately make a joint decision to "activate an early warning". Multi-source fusion data refers to data or decision results that integrate multi-dimensional information after feature-level fusion or decision-level fusion processing; for example, a data record generated in an orchard contains the "current water and fertilizer stress index" and the preliminary judgment of "recommended irrigation" calculated by soil, meteorological and image data.

[0041] The industrial information and data processing module 140 is used to call the streaming computing engine to perform real-time cleaning and anomaly detection on the multi-source fusion data, and to call the distributed computing framework to perform batch processing analysis on the historical multi-source fusion data, and output analysis results including real-time status and trend prediction.

[0042] Among them, a streaming computing engine refers to a software framework used for real-time computation and processing of continuously generated data streams; for example, a software framework deployed in an orchard monitoring system for continuously receiving and processing multi-source fusion data streams reported by sensors. Real-time cleaning and anomaly detection refers to the process of real-time noise filtering, invalid value removal, and identification of outliers exceeding the normal range in continuously flowing data; for example, a streaming computing engine in an orchard checks incoming soil temperature data in real time, marking obvious outliers that jump to 100℃ as invalid and removing them. A distributed computing framework refers to a software platform that uses multiple computer clusters for parallel computation to process massive amounts of data; for example, a software platform deployed in the backend of an orchard monitoring system for batch analysis and modeling of accumulated historical multi-source fusion data. Historical multi-source fusion data refers to a collection of multi-source fusion data accumulated and stored over a past period; for example, daily fusion water and fertilizer stress indices, pest and disease risk levels, etc., stored in an orchard database since January 2024. Real-time status refers to the analytical conclusions reflecting the latest current condition of the orchard; for example, an orchard monitoring system might calculate, based on data from the last 10 minutes, that "the average soil moisture in area A is currently 21.5%, indicating a slightly dry state." Trend prediction refers to inferences about changes in orchard conditions over a future period based on historical data and the current status; for example, an orchard monitoring system might predict that "the probability of pests and diseases in area B will rise to 65% within the next 24 hours." Analysis results refer to the comprehensive set of conclusions output by the industrial information and data processing module 140, including real-time status and trend predictions; for example, a report output by an orchard system might include the current water and fertilizer status of each zone, an assessment of meteorological impacts over the next 48 hours, and a prediction of pest and disease risks.

[0043] The decision support module 150 is used to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning by calling a preset decision rule library based on the analysis results.

[0044] The decision rule base refers to a collection that stores a series of preset logical judgment conditions and corresponding action instructions; for example, a set of rules in an orchard system may include condition-action pairs such as "If the soil moisture is below 23% and there is no rainfall in the next 12 hours, then trigger an irrigation instruction." Orchard management decision instructions refer to the specific executable operation commands ultimately generated by the decision support module 150 based on the analysis results and the decision rule base; for example, an orchard system may issue a command to the automatic irrigation controller to "Execute 15 minutes of drip irrigation on area A at 14:00 on October 27, 2025."

[0045] The technical solution of this embodiment collects multi-dimensional data on meteorology, soil, pests and diseases, and fruit tree growth status by deploying heterogeneous sensor terminals. After standardized parsing and spatiotemporal correlation fusion, streaming and distributed computing engines are called to perform real-time cleaning and historical analysis, which solves the problems of inconsistent data standards, insufficient fusion, weak processing capabilities and data silos, and improves the collaborative utilization rate of multi-source information in orchards and the level of intelligent management.

[0046] In one alternative approach, the multi-source data acquisition module 110 includes a wireless sensor network, which consists of the heterogeneous sensor terminals and gateway nodes, and is used to upload the multi-dimensional monitoring data to the data standardization module 120.

[0047] Wireless sensor networks refer to data acquisition and transmission networks composed of distributed, heterogeneous sensor terminals and centralized gateway nodes connected wirelessly. For example, in an orchard, all sensors transmit data wirelessly to a gateway device deployed at the center of the orchard, which then uploads the data to a cloud server via a cellular network. A gateway node is the central device in a wireless sensor network responsible for protocol conversion, data aggregation, and remote communication.

[0048] Among the above-mentioned optional methods, a wireless sensor network can be further constructed in the orchard, consisting of a data transmission system composed of heterogeneous sensor terminals and gateway nodes, to achieve flexible aggregation and secure uploading of multi-dimensional monitoring data, thereby enhancing the system's deployment adaptability and communication reliability in complex environments.

[0049] In an alternative approach, the data standardization module 120 is further configured to validate the original data fields and attach data quality labels to the validated original data fields, wherein the format conversion and semantic normalization processing are performed based on the data quality labels.

[0050] Data quality labels refer to tags attached to data to identify the level of reliability, completeness, or accuracy of the data. For example, a data standardization module 120 in an orchard attaches a "Quality: High" tag to the verified soil moisture data, and a "Quality: Estimated" tag to the data interpolated during the signal loss period.

[0051] In the above-mentioned optional methods, further validation operations are performed on the original data fields and quality labels are attached, so that format conversion and semantic normalization processing can be differentiated according to data credibility, abnormal data can be identified, and a quality assessment basis can be provided for subsequent fusion stages.

[0052] In one alternative approach, the data association method based on timestamps and spatial locations in the multi-source information fusion module 130 specifically involves performing spatiotemporal matching on the unified format data according to the data collection time and the preset orchard geographic grid code.

[0053] Orchard geogrid coding refers to a coding system that divides an orchard area into several cells according to certain rules and assigns a unique identifier to each cell. For example, an orchard might be divided into 20m x 20m grids, and a code like "ZN-05" might be used to uniquely identify the fifth row of planting areas. Spatiotemporal matching refers to the process of finding and associating data items from different data sources based on temporal and spatial consistency conditions. For example, in processing orchard data, the system might find and associate the "ZN-05 area weather station data" that is temporally and spatially closest to the "2025-10-27 10:00:00 ZN-05 area image data".

[0054] Among the above-mentioned optional methods, a spatiotemporal matching method combining data collection time and orchard geographic grid coding is further adopted to accurately align data in a unified format, establish a correlation benchmark between orchard spatial areas and monitoring time series, and support the spatiotemporal consistency fusion of multi-source information.

[0055] In one alternative approach, the industrial information and data processing module 140 performs real-time cleaning and anomaly detection on the multi-source fused data, including statistical outlier detection based on a sliding time window.

[0056] Among them, statistical outlier detection based on sliding time windows refers to: using statistical methods to identify data points that significantly deviate from the overall data distribution within a time period that slides forward over time; for example, an orchard system calculates the mean and standard deviation of soil moisture in the most recent hour and identifies newly arrived data points that exceed the mean ± 3 times the standard deviation as outliers.

[0057] Among the above-mentioned optional methods, a sliding time window mechanism is further introduced into real-time cleaning and anomaly detection. By dynamically analyzing the fluctuation characteristics of streaming data through statistical outlier detection methods, short-term anomalies in the monitoring data can be detected in a timely manner, thereby improving the response speed of real-time data quality control.

[0058] In an alternative embodiment, the decision support module 150 further includes a decision optimization unit, which dynamically updates the decision rule base based on feedback data on the decision execution effect.

[0059] Among them, decision execution effect data refers to data that quantifies or qualitatively evaluates the actual effects of executed orchard management decision instructions; for example, data on changes in chlorophyll content of fruit tree leaves or fruit growth rate obtained by subsequent sensor monitoring after an orchard executes a variable fertilization instruction.

[0060] In the above-mentioned optional methods, a decision optimization unit is further set up to automatically adjust the rule base parameters based on the feedback data of the decision execution effect, so that the orchard management decision model can learn from the actual application results and continuously iterate, thereby enhancing the adaptability of decision recommendations to environmental changes.

[0061] In an alternative embodiment, the system further includes a visualization and interaction module for receiving and displaying the analysis results and the orchard management decision instructions, and for receiving user feedback.

[0062] Operational feedback refers to the user's input response to the system's displayed content or recommended decisions through the human-computer interaction interface; for example, after seeing a system warning on the visualization interface, the orchard administrator clicks the confirmation button or manually modifies the system's recommended irrigation duration parameters.

[0063] Among the above-mentioned optional methods, a visual interactive function can be further added to receive and display analysis results and decision-making instructions, while collecting user operation feedback information, opening up a two-way information channel between people and the system, and improving the transparency and user participation of orchard monitoring and management.

[0064] In one alternative approach, the visualization interaction module may display a spatial distribution map of monitoring data based on an electronic map, a historical data trend curve, and a decision command execution status panel.

[0065] Among them, the monitoring data spatial distribution map refers to a graphic that visualizes monitoring data according to the collection location, using an electronic map as the base map; for example, on the monitoring screen of an orchard, a color-filled map intuitively shows the distribution of soil moisture in different areas, with dry areas displayed in red and wet areas displayed in blue. The historical data trend curve refers to a curve graph formed by connecting historical data of specific monitoring parameters or analytical indicators with time as the horizontal axis; for example, a line graph displayed in an orchard system showing the "average daily change in soil moisture in area A over the past 30 days" clearly reflects the decreasing trend of moisture over time. The decision instruction execution status panel refers to a graphical interface component that centrally displays various generated, pending, executing, and completed orchard management decision instructions and their current status; for example, a panel in an orchard system lists items such as "Irrigation Instruction - Issued," "Fertilization Instruction - Executing," and "Early Warning Instruction - Confirmed," along with their statuses.

[0066] Among the above optional methods, information is further displayed in three forms: electronic map spatial distribution map, historical trend curve and execution status panel, which transforms abstract data into intuitive graphics and helps users to fully grasp the spatial pattern, temporal evolution and decision-making progress of orchard monitoring elements.

[0067] In one alternative approach, the multi-source information fusion module 130 employs an adaptive fusion algorithm based on dynamic credibility and spatiotemporal correlation to calculate the fusion weights when performing feature-level fusion.

[0068] The adaptive fusion algorithm calculates the fusion weight W_i(x, t) of the i-th data source at time t for a specific monitoring parameter x using the following formula:

[0069]

[0070] in, This represents the fusion weight of the i-th data source for monitoring parameter x at time t; This represents the historical data reliability evaluation value of the i-th data source up to time t, which is obtained through data backtracking verification. This represents the inherent confidence coefficient for the i-th data source type; This is a dynamic weight adjustment factor, with a value ranging from 0 to 1; This represents the difference between the current time t and the latest data timestamp from the i-th data source; This is the time decay coefficient; This represents the average spatiotemporal data difference between the i-th data source and other data sources in this fusion. This is the difference penalty coefficient; This indicates the total number of data sources participating in the integration.

[0071] The above formula comprehensively considers the dynamic historical performance, inherent reliability, data timeliness, and spatial consistency of the data sources. It dynamically calculates the fusion weight of each data source for a specific monitoring parameter at a specific time using a weighted harmonic method. The numerator linearly weights the historical data reliability evaluation value with the inherent confidence coefficient of the data source type to reflect the long-term reliability and inherent reliability of the data source. Then, a time decay exponential function with a base of the natural constant is introduced to quantify the freshness of the data, ensuring that older data contributes less. The denominator first sums the numerators of all data sources to achieve normalization, ensuring that the sum of all weights is theoretically 1. Simultaneously, an additional penalty term based on the difference between this data source and the average data from other sources is introduced. This penalty term is multiplied by a difference penalty coefficient, which significantly reduces the weight of abnormal data sources that are inconsistent with the group data in the spatiotemporal dimensions, thus mathematically achieving adaptive fusion.

[0072] The above formula, through a complete mathematical framework, transforms multi-dimensional evaluation criteria into a computable weight value, enabling the fusion process to automatically favor high-quality data sources that are historically reliable, of trustworthy type, with fresh data, and consistent with other sources, thereby improving the accuracy and robustness of the fusion results.

[0073] Among the above-mentioned optional methods, an adaptive algorithm is further adopted in feature-level fusion, which combines dynamic credibility, spatiotemporal correlation, time decay and data difference to calculate fusion weight, so that the contribution of different data sources is intelligently adjusted according to data quality and spatiotemporal relationship, thereby optimizing the accuracy of fusion results.

[0074] Figure 2 This diagram illustrates a flowchart of an embodiment of a smart orchard multi-source information fusion monitoring method provided by the present invention. Figure 2 As shown, it includes the following steps:

[0075] S1. Collect multi-dimensional monitoring data by deploying multiple heterogeneous sensor terminals in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data.

[0076] S2. The multi-dimensional monitoring data is parsed to obtain the original data field containing the monitoring values ​​and data tags. Based on the original data field, format conversion and semantic normalization are performed according to the preset unified data protocol to output unified format data.

[0077] S3. Align the unified format data using a data association method based on timestamps and spatial locations, and perform feature-level fusion and decision-level fusion to generate multi-source fused data;

[0078] S4. Call the streaming computing engine to perform real-time cleaning and anomaly detection on the multi-source fusion data, and call the distributed computing framework to perform batch processing analysis on the historical multi-source fusion data, outputting analysis results including real-time status and trend prediction.

[0079] S5. Based on the analysis results, call the preset decision rule base to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning.

[0080] The technical solution of this embodiment collects multi-dimensional data on meteorology, soil, pests and diseases, and fruit tree growth status by deploying heterogeneous sensor terminals. After standardized parsing and spatiotemporal correlation fusion, streaming and distributed computing engines are called to perform real-time cleaning and historical analysis, which solves the problems of inconsistent data standards, insufficient fusion, weak processing capabilities and data silos, and improves the collaborative utilization rate of multi-source information in orchards and the level of intelligent management.

[0081] Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0082] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0083] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and do not imply a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0084] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A smart orchard multi-source information fusion monitoring system, characterized in that, The system includes: The multi-source data acquisition module is used to collect multi-dimensional monitoring data through multiple heterogeneous sensor terminals deployed in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data. The data standardization module is used to parse the multi-dimensional monitoring data to obtain raw data fields containing monitoring values ​​and data labels, and based on the raw data fields, to perform format conversion and semantic normalization processing according to a preset unified data protocol, and output unified format data. The multi-source information fusion module is used to align the unified format data using a data association method based on timestamps and spatial locations, and to perform feature-level fusion and decision-level fusion to generate multi-source fused data. The industrial information and data processing module is used to call the streaming computing engine to perform real-time cleaning and anomaly detection on the multi-source fusion data, and to call the distributed computing framework to perform batch processing analysis on historical multi-source fusion data, outputting analysis results including real-time status and trend prediction. The decision support module is used to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning based on the analysis results and by calling a preset decision rule library.

2. The intelligent orchard multi-source information fusion monitoring system according to claim 1, characterized in that, The multi-source data acquisition module includes a wireless sensor network, which consists of heterogeneous sensor terminals and gateway nodes, and is used to upload the multi-dimensional monitoring data to the data standardization module.

3. The intelligent orchard multi-source information fusion monitoring system according to claim 2, characterized in that, The data standardization module is also used to validate the original data fields and attach data quality labels to the validated original data fields. The format conversion and semantic normalization processing are performed based on the data quality labels.

4. The intelligent orchard multi-source information fusion monitoring system according to claim 3, characterized in that, The data association method based on timestamps and spatial locations in the multi-source information fusion module specifically involves spatiotemporal matching of the unified format data according to the data collection time and the preset orchard geographic grid code.

5. The intelligent orchard multi-source information fusion monitoring system according to claim 4, characterized in that, The industrial information and data processing module performs real-time cleaning and anomaly detection on the multi-source fused data, including statistical outlier detection based on a sliding time window.

6. The intelligent orchard multi-source information fusion monitoring system according to claim 1, characterized in that, The decision support module also includes a decision optimization unit, which dynamically updates the decision rule base based on feedback data on decision execution effects.

7. The intelligent orchard multi-source information fusion monitoring system according to claim 1, characterized in that, The system also includes a visualization and interaction module, which is used to receive and display the analysis results and the orchard management decision instructions, and to receive user operation feedback.

8. The intelligent orchard multi-source information fusion monitoring system according to claim 7, characterized in that, The visualization and interactive module displays monitoring data spatial distribution maps based on electronic maps, historical data trend curves, and decision command execution status panels.

9. The intelligent orchard multi-source information fusion monitoring system according to claim 4, characterized in that, When performing feature-level fusion, the multi-source information fusion module uses an adaptive fusion algorithm based on dynamic credibility and spatiotemporal correlation to calculate the fusion weight. The adaptive fusion algorithm calculates the fusion weight W_i(x, t) of the i-th data source at time t for a specific monitoring parameter x using the following formula: in, This represents the fusion weight of the i-th data source for monitoring parameter x at time t; This represents the historical data reliability evaluation value of the i-th data source up to time t, which is obtained through data backtracking verification. This represents the inherent confidence coefficient for the i-th data source type; This is a dynamic weight adjustment factor, with a value ranging from 0 to 1; This represents the difference between the current time t and the latest data timestamp from the i-th data source; This is the time decay coefficient; This represents the average spatiotemporal data difference between the i-th data source and other data sources in this fusion. This is the penalty coefficient for the degree of difference; This indicates the total number of data sources participating in the integration.

10. A method for intelligent orchard multi-source information fusion monitoring, characterized in that, The method includes: Multi-dimensional monitoring data is collected by multiple heterogeneous sensor terminals deployed in the orchard. The multi-dimensional monitoring data includes meteorological data, soil data, pest and disease image data, and fruit tree growth status data. The multi-dimensional monitoring data is parsed to obtain raw data fields containing monitoring values ​​and data tags. Based on the raw data fields, format conversion and semantic normalization are performed according to a preset unified data protocol to output unified format data. The unified format data is aligned using a data association method based on timestamps and spatial locations, and feature-level fusion and decision-level fusion are performed to generate multi-source fused data. The streaming computing engine is invoked to perform real-time cleaning and anomaly detection on the multi-source fusion data, and the distributed computing framework is invoked to perform batch processing analysis on the historical multi-source fusion data, outputting analysis results including real-time status and trend prediction. Based on the analysis results, a preset decision rule base is invoked to generate orchard management decision instructions for precision irrigation, variable fertilization, and pest and disease early warning.

Images