Tea garden management method based on intelligent sensing

Abstract

The invention discloses a tea garden management method based on intelligent sensing, and the method comprises the steps: collecting the temperature, humidity, illumination and soil parameters of a tea garden through an intelligent sensor, and generating a multi-dimensional environment data set; transmitting the multi-dimensional environment data stream to a cloud platform by adopting an Internet of Things protocol, and generating a cloud storage data set; according to the smooth environment data set, fusing multi-source data, and constructing a data set containing time sequence features; training the data set containing the time sequence features by adopting a long-short-term memory network to generate an environment change prediction model; analyzing a future environment trend through the environment change prediction model, and generating environment change trend data; aiming at the environment change trend data, performing classification processing by applying a random forest model to generate a scientific planting suggestion data set; and according to the scientific planting suggestion data set, integrating a decision system, and generating an automatic planting management instruction.

Description

Technical Field

[0001] The present invention belongs to the technical field of tea garden management, and in particular relates to a tea garden management method based on intelligent sensing. Background Art

[0002] Currently, tea tree planting and management mainly adopts a combination of traditional manual experience and basic monitoring equipment. In order to improve the scientific nature and accuracy of tea tree planting and realize "smart planting" (i.e., intelligent decision-making based on real-time environmental data), the Internet of Things monitoring system is gradually being promoted and applied in tea garden management, but there are still significant shortcomings. Although the existing automated management solutions have introduced some sensor technologies, their functions are single and their data processing capabilities are limited, making it difficult to provide comprehensive and scientific planting recommendations. In addition, the data processing capabilities are weak and there is a lack of in-depth analysis based on algorithms. It is impossible to combine the growth characteristics of tea trees with the tea planting scenarios to provide decision-making solutions, and it is unable to meet the complex needs of climate-changing regions.

[0003] Therefore, how to integrate the Internet of Things, multi-source data fusion and artificial intelligence technology to build a tea garden management method that can collect multi-dimensional environmental data in real time, conduct in-depth analysis and generate scientific planting suggestions has become a key issue in promoting the precision and intelligence of tea garden management. Summary of the Invention

[0004] The present invention proposes a tea garden management method based on intelligent sensing to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above object, the present invention provides a tea garden management method based on intelligent sensing, comprising the following steps:

[0006] Collecting a multi-dimensional environmental data set of the tea garden through smart sensors, the multi-dimensional environmental data set includes temperature, humidity, light, and soil parameters;

[0007] The multi-dimensional environmental data stream is transmitted to the cloud platform using the Internet of Things protocol to generate a cloud storage data set;

[0008] Performing data denoising on the cloud storage data set using a Kalman filter algorithm to generate a smoothed environment data set;

[0009] Based on the smoothed environment dataset, multi-source data is fused to construct a dataset containing time series features;

[0010] Build an environmental change prediction model based on a dataset containing time series features and a long short-term memory network;

[0011] Analyze future environmental trends through environmental change prediction models and generate environmental change trend data;

[0012] Targeting, classifying and processing the environmental change trend data through a random forest model to generate a scientific planting recommendation dataset;

[0013] Based on scientific planting recommendation data sets, the decision-making system is integrated to generate automated planting management instructions.

[0014] Optionally, the multi-dimensional environmental dataset collected from the tea garden includes:

[0015] Intelligent sensors are used to obtain temperature data, humidity data, light intensity, and soil parameters from the tea garden environment to form a preliminary set of environmental information;

[0016] Use data integration technology to clean and format the preliminary environmental information collection to obtain a structured multidimensional data set;

[0017] For structured multidimensional data sets, use pre-established classification models to perform parameter analysis to determine whether each environmental parameter is within the normal range;

[0018] If the parameter analysis results show that a certain environmental parameter exceeds the preset threshold, the latest data of the parameter is re-obtained through sensing and the data is updated.

[0019] Optionally, generating a cloud storage data set includes:

[0020] Acquire multi-dimensional data through IoT protocols to form environmental data streams;

[0021] Adopt the transmission protocol standard to format the environmental data stream to obtain a standardized data stream;

[0022] Segment the standardized data stream using data stream management technology to obtain segmented data stream fragments;

[0023] If the segmented data stream fragment is interrupted during transmission, the corresponding environmental data stream fragment is re-acquired;

[0024] The recovered data stream segments are integrated and processed through the cloud platform, and a unified cloud repository is built using cloud data integration technology to obtain the integrated data set.

[0025] The integrated data set is stored through the data storage architecture.

[0026] Optionally, generating a smoothed environment dataset includes:

[0027] Obtain environmental data from cloud storage, perform preliminary classification on the data through a pre-established data screening mechanism, and obtain classified environmental data groups;

[0028] The Kalman filter algorithm is used to perform data denoising on the classified environmental data group to generate denoised data units;

[0029] According to the denoised data units, different categories of data are merged through data integration technology to obtain the integrated data framework;

[0030] For the integrated data framework, if data is missing or abnormal, the corresponding data fragments are retrieved from the environmental monitoring source through cloud technology to obtain the completed data structure;

[0031] According to the completed data structure, the smoothing technology is applied to perform secondary optimization on the data to generate an optimized data set;

[0032] The optimized data set is stored in layers through the cloud storage mechanism to obtain the final stored data archive;

[0033] According to the final stored data file, the stored content is checked for consistency through data verification to obtain the verified data record.

[0034] Optionally, constructing a dataset containing time series features includes:

[0035] Acquiring environmental information from the smoothed data, and preliminarily sorting the environmental information using a pre-established classification rule to obtain sorted environmental units;

[0036] Obtain relevant external data sources, perform correlation processing based on the sorted environmental units using data matching technology, and determine the correlated data groups;

[0037] For the associated data group, the time series analysis method is used to extract sequence features, build a feature framework including time trends, and determine the integrity of the feature framework;

[0038] According to the integrity of the feature framework, data fusion technology is applied to deeply combine external related data with environmental information to obtain a fused data set.

[0039] Optionally, building an environmental change prediction model includes:

[0040] The long short-term memory network is used to process the time series related data set, extract the features of the environmental information contained therein, and obtain the preliminary feature combination;

[0041] Based on the preliminary feature combination, the long-term memory and short-term memory characteristics in the time series are separated and processed to determine the separated long-term features and short-term features;

[0042] By separating the long-term and short-term characteristics, we can analyze the changing trends in the environment, obtain the pattern information related to the changing trends, and determine the potential fluctuation patterns.

[0043] Based on the potential fluctuation patterns and combined with key elements in environmental information, sequence analysis is performed to obtain sequence patterns that are closely related to environmental changes;

[0044] If the fluctuation pattern in the sequence pattern does not conform to the preset threshold range, a secondary feature extraction is performed on the data set to obtain an adjusted feature combination;

[0045] Based on the adjusted feature combination, the initial framework of the prediction model is constructed, and the prediction parameters corresponding to the change trend in the framework are obtained to obtain the prediction structure;

[0046] Through the prediction structure, combined with the characteristics of long-term memory and short-term memory, the future trend of environmental changes is simulated to determine the final prediction output.

[0047] Optionally, generating environmental change trend data includes:

[0048] Using the monitoring data of environmental changes and pre-established prediction models, we can make preliminary deductions of future trends and obtain initial trend analysis results.

[0049] Based on the initial trend analysis results, core data is screened and processed to extract key information related to environmental monitoring and determine the characteristic set of change patterns;

[0050] By combining the characteristic set of changing rules with the output of trend analysis, data integration is structured to obtain pattern information corresponding to future trends;

[0051] Based on the pattern information, we use the long short-term memory network to conduct in-depth analysis of time series data in response to potential fluctuations in environmental changes, and determine the distribution characteristics of the fluctuation trend.

[0052] If the distribution characteristics of the fluctuation trend are inconsistent with the preset threshold range, the core data will be cleaned twice to obtain an adjusted data combination;

[0053] Through the adjusted data combination, calibration analysis is performed on the intermediate results of trend extraction to obtain prediction parameters that are closely related to environmental changes;

[0054] Based on the prediction parameters and the final structure of the model output, a dynamic simulation of future trends is performed to determine the trend data set of environmental changes.

[0055] Optionally, generating a scientific planting suggestion dataset includes:

[0056] Through preliminary collation of environmental change trend data and removal of redundant information through data cleaning, a collated environmental data set is obtained;

[0057] Based on the sorted environmental data set, the data is classified using the random forest model to determine the category distribution related to the planting conditions and obtain the classified feature set;

[0058] If the distribution of some categories in the classified feature set does not match the preset threshold range, the feature set will be screened again to extract the core indicators that are highly relevant to agricultural planting and determine the core feature combination;

[0059] Based on the core feature combination, data mapping is performed according to the needs of agricultural planting, matching information adapted to the planting conditions is obtained, and an adapted data framework is obtained;

[0060] Through the adapted data framework and combined with the business rules of scientific planting, the matching information is structured and integrated to determine the guiding parameters related to planting recommendations;

[0061] According to the guidance parameters, information is reorganized according to the generation requirements of the recommendation dataset to obtain the final scientific planting recommendation dataset.

[0062] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0063] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method when executed by a processor.

[0064] Compared with the prior art, the present invention has the following advantages and technical effects:

[0065] The present invention discloses an intelligent tea garden management method based on the Internet of Things and artificial intelligence. The method collects multi-dimensional environmental data of the tea garden through intelligent sensors, and uses the Internet of Things technology to transmit the data to a cloud platform and store it. The present invention applies a Kalman filter algorithm to denoise the data, fuses multi-source data to construct a time series feature data set, and uses a long and short-term memory network to train an environmental change prediction model. Based on the prediction model, future environmental trends are analyzed, and classification processing is combined with a random forest model to generate scientific planting recommendations. Finally, the present invention integrates a decision-making system to automatically generate planting management instructions. The method realizes intelligent monitoring, data analysis, trend prediction and scientific management of the tea garden environment, improves the precision and automation level of tea planting, provides intelligent decision-making support for tea farmers, and effectively improves tea yield and quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0067] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0068] Figure 2 This is a structural diagram of a data acquisition system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0069] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0070] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0071] Example 1

[0072] like Figure 1 As shown, this embodiment provides a tea garden management method based on intelligent sensing, including the following steps:

[0073] Smart sensors are used to collect multi-dimensional environmental data sets from tea gardens, including temperature, humidity, light, and soil parameters.

[0074] The multi-dimensional environmental data stream is transmitted to the cloud platform using the Internet of Things protocol to generate a cloud storage data set;

[0075] The Kalman filter algorithm is used to perform data denoising on the cloud storage data set to generate a smooth environment data set;

[0076] Based on the smoothed environment dataset, multi-source data is fused to construct a dataset containing time series features;

[0077] Build an environmental change prediction model based on a dataset containing time series features and a long short-term memory network;

[0078] Analyze future environmental trends through environmental change prediction models and generate environmental change trend data;

[0079] Targeting, using random forest models to classify environmental change trend data and generate scientific planting recommendation datasets;

[0080] Based on scientific planting recommendation data sets, the decision-making system is integrated to generate automated planting management instructions.

[0081] The specific steps include:

[0082] S101. Collect tea garden temperature, humidity, light, and soil parameters through smart sensors to generate a multi-dimensional environmental data set.

[0083] Specifically, smart sensors collect temperature, humidity, light intensity, and soil parameters from the tea garden environment to form a preliminary environmental information set. Data integration technology is used to cleanse and format this preliminary environmental information set, resulting in a structured multidimensional dataset. A pre-established classification model is then used to perform parameter analysis on this structured multidimensional dataset to determine whether each environmental parameter is within a normal range. If the parameter analysis results indicate that a particular environmental parameter exceeds a preset threshold, the latest data for that parameter is retrieved through sensors and the data is updated.

[0084] In this embodiment, the acquisition system is as follows Figure 2 As shown, it uses the STM32F103C8T6 as the main control chip, connects to the GY-302 light sensor and OSA-60 soil pH sensor via the I2C interface, connects to the DHT11 temperature and humidity sensor via the GPIO interface, and collects soil moisture analog signals through the ADC channel, thus building a complete agricultural environmental monitoring network. For communication, it innovatively adopts the ESP8266 Wi-Fi and SX1278 LoRa dual-mode design, automatically switching through a signal quality assessment algorithm to ensure data transmission reliability. The power management system integrates solar power and lithium battery backup, using the AMS1117 and MT3608 chipsets to achieve precise voltage regulation at multiple voltage levels: 3.3V, 5V, and 12V.

[0085] In this embodiment, temperature, humidity, light intensity and soil parameters are obtained from the tea garden environment through smart sensors to form a preliminary environmental information set.

[0086] For example, sensors are deployed in different areas of a tea garden to collect real-time data. Temperature sensors record ambient temperature, such as 28°C; humidity sensors measure air humidity, such as 70%; light intensity sensors measure light levels, such as 50,000 lux; and soil sensors detect soil moisture content, such as 30%, and soil pH, such as 5.5. This data is wirelessly transmitted to a central processing system, forming a preliminary data set. The advantage of this approach is that it comprehensively captures the real-time environmental conditions of the tea garden, providing a rich data foundation for subsequent analysis.

[0087] For example, the preliminary environmental information set is cleaned and formatted to obtain a structured multidimensional dataset.

[0088] Specifically, the cleaning process involves removing outliers. For example, if a sensor records a temperature of 100°C due to a fault, this is clearly outside the acceptable range and should be removed or marked as invalid. Formatting unifies data from different sensors into a standard format, such as standardizing temperature units to degrees Celsius and humidity to percentages. The data is then organized by timestamp and geographic location to form a multidimensional dataset. This structured data facilitates subsequent analysis, improves data consistency and accuracy, and effectively reduces analytical errors caused by formatting errors.

[0089] In this embodiment, a pre-established classification model is used to perform parameter analysis on a structured multi-dimensional data set to determine whether each environmental parameter is within a normal range.

[0090] For example, a classification model trained on historical data assumes a normal temperature range of 15-30°C, humidity of 60-80%, light intensity of 30,000-60,000 lux, soil moisture content of 25-35%, and soil pH of 5.0-6.5. During analysis, the model compares real-time data with pre-set thresholds.

[0091] For example, if the temperature in a data set is 32°C, which is outside the normal range, the model will mark it as an anomaly. The benefit of this method is that it can quickly identify potential problems, provide accurate basis for tea garden management, and protect the tea growing environment.

[0092] For example, if the parameter analysis shows that a certain environmental parameter exceeds a threshold, such as a temperature of 32°C, the latest data of the parameter is re-obtained through the sensor for updating.

[0093] For example, if the temperature data after recollection is 29°C, it may indicate that the temperature is abnormal due to a brief period of direct sunlight. After the update, the data returns to normal. This dynamic update mechanism can effectively avoid misjudgments and improve data reliability.

[0094] It's important to note that the recollected data is cleaned and formatted again before being integrated into a multidimensional dataset to ensure the continuity and accuracy of the analysis. The advantage of this approach is that it reduces the impact of environmental fluctuations on analytical results through real-time verification, ensuring scientifically sound tea plantation management decisions.

[0095] In this embodiment, the above technical process forms a closed-loop management through continuous updating of multidimensional data sets and optimization of classification models.

[0096] For example, if long-term monitoring reveals persistently low soil pH in a particular area, model analysis can be combined to adjust irrigation or fertilization strategies. This approach not only enhances the intelligence level of tea garden environmental monitoring but also optimizes tea growing conditions through data-driven methods, significantly improving yield and quality.

[0097] S102: Using the Internet of Things protocol to transmit the multi-dimensional environmental data stream to the cloud platform to generate a cloud storage data set.

[0098] Specifically, multi-dimensional data is acquired from environmental monitoring technology through the Internet of Things (IoT) protocol to form an environmental data stream. The environmental data stream is formatted using a transmission protocol standard to obtain a standardized data stream. The standardized data stream is segmented using data stream management technology to determine the segmented data stream fragments. If the segmented data stream fragments are interrupted during transmission, the corresponding environmental data stream fragments are retrieved through the cloud platform system. The recovered data stream fragments are integrated and processed through the cloud platform system, and a unified cloud repository is constructed using cloud data integration technology to obtain an integrated data set. Based on the integrated data set, it is classified and stored using a data storage architecture.

[0099] For example, in the field of tea garden environmental monitoring, acquiring multi-dimensional data through IoT protocols is a critical step. IoT protocols can be understood as a system of rules for inter-device communication, ensuring that sensor devices can transmit real-time data such as temperature, humidity, and light intensity within the tea garden to a data processing center. For example, imagine multiple sensor nodes deployed in a tea garden, each collecting environmental data every five minutes—for example, a temperature of 25.5 degrees Celsius and a humidity of 70%. This data is transmitted in data packets using the protocol, forming the initial environmental data stream. This approach ensures the real-time and integrity of the data, laying the foundation for subsequent processing.

[0100] For example, when formatting environmental data streams, transmission protocol standards serve the purpose of unifying heterogeneous data collected by different devices into a standard format. For example, suppose different sensor models in a tea garden collect data in varying formats, some in text format and some in binary. Protocol standards can unify this data into a common structured format for easier analysis. This formatting improves data compatibility and prevents data loss or misinterpretation due to inconsistent formats.

[0101] For example, in data stream segmentation, data stream management technology can segment continuous data streams by time period or data type. For example, if a tea garden collects a large amount of data per day, this data can be segmented into 24 data segments per hour, with each segment containing all environmental parameter data for that period. This segmentation approach facilitates efficient data management and rapid retrieval, significantly improving processing efficiency, especially when dealing with large amounts of data.

[0102] For example, if a data stream segment is interrupted during transmission, the cloud platform system can re-acquire the lost data segment. Imagine that data for a certain period of time fails to transmit due to network fluctuations. The cloud system will automatically detect the loss and re-request the data for that segment through an alternate communication channel. This mechanism ensures data integrity and prevents data loss caused by transmission issues.

[0103] For example, in cloud-based data integration, building a unified repository through cloud-based integration technology can reassemble scattered data fragments into a complete dataset. For example, if data from multiple monitoring points in a tea garden is stored on separate servers, the cloud system will integrate this data into a unified database, forming an environmental dataset covering the entire tea garden. This integration approach facilitates centralized data management and subsequent access.

[0104] For example, when categorizing and storing integrated data sets, the data storage architecture can be categorized by data type or time range. Suppose temperature data is stored in one sub-database and humidity data in another, and the storage paths are divided by date, such as data from October 2023 being archived separately. This categorized storage method can improve data query efficiency and facilitate the quick location of environmental data at a specific time or type. Through the above methods, data collection, transmission, processing, and storage in tea garden environmental monitoring form a complete closed-loop system, ensuring that every link from data collection to storage is efficient and reliable, providing solid support for subsequent environmental analysis and management.

[0105] S103: Applying a Kalman filter algorithm to the cloud storage data set for data denoising to generate a smoothed environment data set.

[0106] Specifically, environmental data is obtained from cloud storage, and the data is preliminarily classified through a pre-established data screening mechanism to obtain a classified environmental data group. For the classified environmental data group, a Kalman filter algorithm is used to perform data denoising to generate denoised data units. Based on the denoised data units, data integration technology is used to merge data of different categories to determine the integrated data framework. For the integrated data framework, if data is missing or abnormal, the corresponding data fragments are retrieved from the environmental monitoring source through cloud technology to obtain a completed data structure. Based on the completed data structure, smoothing processing technology is applied to perform secondary optimization on the data to generate an optimized data set. The optimized data set is stored in layers through a cloud storage mechanism to determine the final stored data archive. Based on the final stored data archive, a data verification tool is used to perform consistency testing on the stored content to obtain a verified data record.

[0107] For example, when retrieving environmental data from cloud storage, a pre-set interface can be used to access the cloud database and filter out environmental data for a specific time period or area. For example, in a city air quality monitoring project, the system extracts data from the cloud daily, including temperature, humidity, and PM2.5 concentration, and preliminarily categorizes it into air quality and meteorological conditions groups. This classification mechanism can be based on the data's source sensor type or monitoring target, ensuring more targeted subsequent processing.

[0108] For example, applying a Kalman filter to denoise classified environmental data can be understood as a dynamic estimation method designed to reduce noise generated by interference during sensor acquisition. For example, if PM2.5 concentration data fluctuates abnormally during certain periods due to device jitter, a Kalman filter can smooth these fluctuations by taking a weighted average of historical data and current measurements, generating more stable data units. This approach is particularly suitable for continuous monitoring scenarios.

[0109] For example, when data integration technology merges data from different categories, it can align the air quality and meteorological conditions data by timestamp, forming an integrated framework that encompasses multi-dimensional information. For example, during a given day's monitoring, the system correlates the PM2.5 concentration data at 8:00 AM with the temperature and humidity data at the same time, forming a complete data record at that point in time. This integration facilitates subsequent analysis of correlations between environmental factors.

[0110] For example, if missing data is detected in the integrated data framework, such as humidity data for a certain hour not being uploaded, the system can communicate with the environmental monitoring equipment through cloud technology to retrieve the data fragment for that time period. This completion mechanism ensures the integrity of the data framework, which is particularly important in long-term monitoring projects.

[0111] For example, when applying smoothing techniques to secondary optimization of the completed data structure, a moving average method can be used to process the data. For example, if the temperature data shows a small jump over a short period of time, the system can smooth the data by calculating the average of the five preceding and succeeding time points, generating a more continuous optimization set. This processing method helps reduce occasional errors in the data.

[0112] For example, when optimizing data sets for tiered storage, they can be stored in different cloud-based directories by data type and time range. For example, PM2.5 data can be stored monthly in the air quality directory, while temperature data can be stored in the meteorological directory. This hierarchical structure facilitates quick retrieval and access.

[0113] For example, when using data verification tools to perform consistency checks on stored content, the system can compare the upload and storage times to ensure data has not been tampered with or lost. If the system detects that the storage time of a particular piece of data is inconsistent with the expected time, it will automatically flag it and notify the administrator for verification. This verification method ensures the reliability of the data archive and provides a solid foundation for subsequent analysis.

[0114] S104: Based on the smoothed environment dataset, multi-source data is integrated to construct a dataset containing time series features.

[0115] Specifically, environmental information is obtained from smoothed data and initially organized using pre-established classification rules to generate organized environmental units. Based on these organized environmental units, external data sources are obtained to meet the needs of multi-source integration. Data matching techniques are used to perform correlation processing and determine the associated data sets. Time series analysis methods are used to extract sequence features from these associated data sets, construct a feature framework that includes temporal trends, and determine the integrity of the feature framework. Based on the integrity of the feature framework, data fusion techniques are applied to deeply combine the multi-source information with the environmental information to generate a fused dataset.

[0116] For example, in the field of environmental monitoring, by obtaining environmental information from smoothed data, the data can first be structured. Pre-established classification rules can be used to group data based on environmental parameters such as temperature, humidity, or air quality index, with thresholds set. Assuming that the smoothed data from a monitoring station contains temperatures of 20-35°C and humidity of 40-80%, the classification rules can classify temperatures below 25°C as a low-temperature group and temperatures above 30°C as a high-temperature group. The organized environmental units facilitate subsequent analysis and avoid data redundancy. This classification method helps to quickly locate data features under specific environmental conditions.

[0117] In this embodiment, in order to meet the needs of multi-source integration, the external relevant data sources may include real-time wind speed data from weather stations, historical data, or satellite remote sensing data. Data matching technology performs association processing based on timestamps or geographic locations.

[0118] For example, time series data from a city's monitoring points and wind speed data from a weather station can be aligned through timestamps to generate a linked data set containing temperature, humidity, and wind speed. This linking process ensures the temporal and spatial consistency of the data, providing a more comprehensive information foundation for subsequent analysis.

[0119] Specifically, time series analysis methods can extract sequence features through sliding window technology.

[0120] For example, for temperature data in a related data set, a 24-hour sliding window is set to calculate the mean, variance, and rate of change to construct a feature framework that captures temporal trends. If the temperature change rate is abnormally high during a certain period, such as a 5°C increase per hour, the feature framework is considered incomplete, and further verification of the data source reliability is required. This approach can capture dynamic data trends and provide a basis for environmental anomaly monitoring.

[0121] It should be noted that the integrity of the feature framework can be determined by checking the proportion of missing values ​​or data continuity.

[0122] For example, if more than 20% of temperature data is missing within a certain time period, the framework is considered incomplete and requires retrieving data from the cloud to complete the analysis. This checking mechanism ensures the reliability of subsequent analysis and avoids bias caused by missing data.

[0123] S105. Use a long short-term memory network to train a data set containing time series features to generate an environmental change prediction model.

[0124] Specifically, a long-short-term memory (LSTM) network is used to process a time series data set, extracting features from the environmental information contained therein to obtain a preliminary feature combination. Based on this preliminary feature combination, the long-term and short-term memory characteristics of the time series are separated to determine the separated long-term and short-term features. The separated long-term and short-term features are used to analyze the trends in environmental change, obtain pattern information related to these trends, and identify potential fluctuation patterns. Sequential analysis is then performed on these potential fluctuation patterns, combined with key elements of the environmental information, to obtain sequence patterns closely related to environmental change. If the fluctuation pattern in the sequence pattern does not meet a preset threshold, a secondary feature extraction is performed on the data set to determine an adjusted feature combination. Based on this adjusted feature combination, an initial framework for the prediction model is constructed, and prediction parameters corresponding to the change trends are obtained from the framework to obtain a prediction structure. Using this prediction structure, the future trends of environmental change are simulated by combining the characteristics of long-term and short-term memory to determine the final prediction output.

[0125] It's important to note that this method, when processing environmental data, can effectively distinguish patterns of change across different timescales, improving forecast accuracy and providing more targeted reference information for environmental management. By analyzing data from both long-term and short-term perspectives, it not only captures overall trends but also highlights local fluctuations, providing multi-layered insights for environmental monitoring and governance.

[0126] S106. Analyze future environmental trends through an environmental change prediction model to generate environmental change trend data.

[0127] Specifically, using environmental change monitoring data and a pre-established forecasting model, a preliminary deduction of future trends is performed to obtain initial trend analysis results. Based on these initial trend analysis results, core data is screened and processed to extract key information related to environmental monitoring and determine a feature set that reflects patterns of change. Using this feature set of patterns, combined with the output of the trend analysis, data integration is structured to obtain pattern information corresponding to future trends. Based on this pattern information, a long-short-term memory network is used to conduct in-depth analysis of time series data to identify potential fluctuations in environmental change and determine the distribution characteristics of the fluctuation trend. If the distribution characteristics of the fluctuation trend do not conform to the preset threshold range, the core data is cleaned again to obtain an adjusted data set. Using this adjusted data set, the intermediate results of trend extraction are calibrated and analyzed to obtain forecast parameters closely related to environmental change. Based on the forecast parameters and the final structure of the model output, a dynamic simulation of future trends is performed to determine the trend dataset for environmental change.

[0128] For example, in environmental change monitoring, pre-established forecasting models can be used to deduce future temperature trends based on time series data on air quality. Assuming the monitoring data includes variables such as temperature and humidity, initial trend analysis may indicate an upward trend in temperature over the next seven days. When filtering core data, temperature data is prioritized because it is directly correlated with temperature trends. After filtering, the feature set may include the daily mean temperature and the temperature fluctuation range.

[0129] Specifically, for in-depth analysis of potential fluctuations, long short-term memory networks can be used to process time series data.

[0130] In this embodiment, when dynamically simulating future trends, a PM2.5 concentration dataset for the next seven days can be generated by combining long-term memory (historical pollution trends) and short-term memory (recent humidity changes).

[0131] For example, simulations show that concentrations could reach 65 micrograms per cubic meter on the third day and 72 micrograms per cubic meter on the seventh day. This dataset can provide a reference for environmental management, providing early warning of pollution peaks and preventing environmental pollution from impacting leaf photosynthesis and respiration. This method uses multi-level data processing to ensure accurate predictions. It also combines long-term and short-term characteristics to capture complex environmental changes and facilitate the development of precise pollution control strategies.

[0132] S107. Based on the environmental change trend data, a random forest model is applied to perform classification processing to generate a scientific planting recommendation dataset.

[0133] Specifically, through the preliminary collation of environmental change trend data, redundant information is removed using data cleaning tools to obtain a collated environmental data set. Based on the collated environmental data set, the random forest model is applied to classify the data, determine the category distribution related to the planting conditions, and obtain a classified feature set. If the distribution of certain categories in the classified feature set does not match the preset threshold range, the feature set is screened a second time to extract core indicators that are highly relevant to agricultural planting and determine the core feature combination. Based on the core feature combination, data mapping is performed according to the needs of agricultural planting, matching information adapted to the planting conditions is obtained, and an adapted data framework is obtained. Through the adapted data framework, combined with the business rules of scientific planting, the matching information is structured and integrated to determine the guidance parameters related to planting recommendations. Based on the guidance parameters, information is reorganized according to the generation requirements of the recommendation data set to obtain the final scientific planting recommendation data set.

[0134] In this embodiment, the preliminary organization of the environmental change trend data requires the use of data cleaning tools to remove redundant information.

[0135] For example, soil moisture, temperature, and rainfall data from a tea plantation scenario may contain missing or outliers. Missing values ​​can be addressed using mean imputation, and outliers can be detected and removed using boxplots to ensure dataset integrity. This cleaning method preserves data authenticity and provides a reliable foundation for subsequent analysis.

[0136] For example, when applying a random forest model for classification, soil nutrient content, pH, and light intensity from an environmental dataset can be used as input features to determine their suitability for tea cultivation. Using a voting mechanism among multiple decision trees, the random forest model outputs a categorical distribution, such as "suitable for cultivation" or "needs improvement." For example, if the soil pH of a particular plot of farmland is 4.5, below the suitable range of 5.5-7.5, the model will classify it as "needs improvement," thus clarifying the direction of subsequent adjustments. This classification method can quickly identify key issues with growing conditions.

[0137] It should be noted that if the distribution of certain categories in the classified feature set does not meet the preset threshold, a secondary screening is required.

[0138] For example, in tea cultivation, core indicators might include fertilizer content and soil moisture. If a feature set indicates excessive nitrogen content, exceeding a preset threshold of 20 mg / kg, then the anomalous sample is removed and a new feature set meeting the criteria is extracted. This screening approach focuses on the key needs of agricultural production and improves analytical accuracy.

[0139] In this embodiment, when data mapping is performed based on the core feature combination, soil moisture and light duration can be matched with the requirements of the tea growth period.

[0140] For example, if soil moisture in a certain area is 30%, while the ideal humidity for tea seedlings is 40%-50%, data mapping will indicate the need for increased irrigation. This mapping method directly links environmental data with planting requirements, providing a basis for scientific decision-making.

[0141] For example, when integrating structured matching information, a planting recommendation dataset can be generated based on guiding parameters. For example, if a field receives six hours of sunlight daily, which is less than the eight hours required during the tea growing season, business rules might suggest supplementing artificial lighting or adjusting planting times. This integration approach transforms dispersed environmental data into specific planting guidance, making it easier for farmers to implement.

[0142] In this embodiment, soil, meteorological, and crop characteristics may be integrated when information is reorganized to generate the final scientific planting recommendation dataset.

[0143] For example, for a tea field, if guidance parameters indicate moderate soil nitrogen content and low moisture, the dataset will recommend irrigation twice a week, 500 mm per session, along with an appropriate amount of organic fertilizer. This reorganization approach can provide specific, actionable planting plans and optimize agricultural production efficiency.

[0144] S108. Based on the scientific planting suggestion data set, the decision-making system is integrated to generate automated planting management instructions.

[0145] Specifically, based on the scientific planting recommendation data set, a data analysis tool is used to pre-process the data, extract key features related to planting management, and obtain structured planting condition data. Through the structured planting condition data, a decision tree model is applied to perform classification analysis to determine the degree of match between various features and planting management requirements, and obtain a classified feature set. If the distribution of certain categories in the classified feature set does not match the preset threshold range, a feature screening tool is used to perform a secondary analysis of the feature set to extract core indicators that are highly relevant to automated management and obtain a core indicator combination. Based on the core indicator combination, logical mapping rules are used to perform structured transformation on the data to generate matching information adapted to the planting management requirements and obtain an adapted data framework. Through the adapted data framework, combined with the business rules of automated management, a data integration tool is used for structured processing to generate guidance parameters related to management instructions and obtain management guidance parameters. Based on the management guidance parameters, a data reorganization tool is used to reorganize the information to generate an automated planting management instruction set that meets the requirements of automated planting management.

[0146] For example, during the preprocessing phase of a scientific planting recommendation dataset, the selection of data parsing tools is crucial. For example, suppose an agricultural base collects environmental data including soil moisture, temperature, light intensity, and nitrogen, phosphorus, and potassium levels. Data parsing tools can generate structured planting condition data by extracting key features, such as records with soil moisture between 20% and 30% or light intensity exceeding 5000 lux. This screening ensures that the data is highly relevant to planting management needs, laying the foundation for subsequent analysis.

[0147] In this embodiment, a decision tree model is applied to perform classification analysis on the structured planting condition data.

[0148] For example, in the context of tea cultivation, a decision tree can determine the suitability of a particular tea variety based on characteristics such as soil moisture, temperature, and light intensity. For example, if a dataset shows that tea yields are higher when temperatures are between 25°C and 30°C and soil moisture is above 25%, the decision tree will classify these conditions as "suitable for cultivation." This classified feature set clearly distinguishes the degree of compatibility between different planting requirements, providing a basis for precise management.

[0149] Specifically, if the distribution of certain categories in the classified feature set does not match the preset threshold, such as if the proportion of records with light intensity below 4000 lux is too high, a feature screening tool can be used for secondary analysis.

[0150] For example, using correlation analysis tools, we screened out core indicators highly correlated with tea growth, such as soil nitrogen content and daylight duration, and eliminated irrelevant variables such as wind speed to obtain a core indicator combination. This screening improved data quality and ensured the targeted nature of subsequent analysis.

[0151] In this embodiment, the core indicator combination is converted into matching information adapted to the planting management requirements through logical mapping rules.

[0152] For example, for the core indicator of soil nitrogen content, the mapping rule could be set as follows: when the nitrogen content is above 200 mg / kg, a low-nitrogen fertilizer is recommended; when it is below 100 mg / kg, an increased nitrogen fertilizer is recommended. This mapping generates an adapted data framework that clearly reflects the correspondence between management needs and environmental conditions.

[0153] For example, the adapted data framework, combined with automated management business rules, generates management guidance parameters through data integration tools. For example, if the business rule prioritizes tea irrigation efficiency, the integration tool can map records of soil moisture below 20% as instructions to increase irrigation frequency, generating specific parameters such as two 30-minute irrigations per day. This structured approach makes management instructions more actionable.

[0154] Specifically, the management guidance parameters are used to generate an automated planting management instruction set through a data reorganization tool.

[0155] For example, for tea fields with low-nitrogen soil, the instruction set might include "Apply 10 kg / hectare of nitrogen fertilizer weekly for three weeks" or "Activate supplemental lighting for four hours daily when sunlight is insufficient." These instructions, based on core indicators and business rules, have clear logic and can be directly applied to automated equipment, ensuring precise and efficient cultivation management.

[0156] In this embodiment, the generation of the automated planting management instruction set can also be extended to different crops.

[0157] For example, for tea cultivation, the instruction set can generate instructions based on temperature and soil moisture to reduce irrigation to 20 minutes per day during high-temperature periods. This expansion solution aligns with tea management logic, demonstrating the versatility of the instruction set while also adapting to the needs of different crops through specific parameters, enhancing the system's flexibility and practicality.

[0158] It's important to note that the above-described embodiment, through a complete process from data parsing to command generation, closely addresses the needs of scientific planting management, ensuring that the output of each step provides reliable input for the next. This logically rigorous processing approach not only improves data utilization efficiency but also provides actionable guidance for automated agricultural management, significantly optimizing the scientific nature and practicability of planting decisions.

[0159] This embodiment further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0160] This embodiment further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method when executed by a processor.

[0161] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A tea garden management method based on intelligent sensing, characterized in that: The following steps are involved: Collecting a multi-dimensional environmental data set of the tea garden through smart sensors, the multi-dimensional environmental data set includes temperature, humidity, light, and soil parameters; The multi-dimensional environmental data stream is transmitted to the cloud platform using the Internet of Things protocol to generate a cloud storage data set; Performing data denoising on the cloud storage data set using a Kalman filter algorithm to generate a smoothed environment data set; Based on the smoothed environment dataset, multi-source data is fused to construct a dataset containing time series features; Build an environmental change prediction model based on a dataset containing time series features and a long short-term memory network; Analyze future environmental trends through environmental change prediction models and generate environmental change trend data; Targeting, classifying and processing the environmental change trend data through a random forest model to generate a scientific planting recommendation dataset; Based on scientific planting recommendation data sets, the decision-making system is integrated to generate automated planting management instructions.

2. The method according to claim 1, characterized in that The multi-dimensional environmental dataset collected from the tea garden includes: Intelligent sensors are used to obtain temperature data, humidity data, light intensity, and soil parameters from the tea garden environment to form a preliminary set of environmental information; Use data integration technology to clean and format the preliminary environmental information collection to obtain a structured multidimensional data set; For structured multidimensional data sets, use pre-established classification models to perform parameter analysis to determine whether each environmental parameter is within the normal range; If the parameter analysis results show that a certain environmental parameter exceeds the preset threshold, the latest data of the parameter is re-obtained through sensing and the data is updated.

3. The method according to claim 1, characterized in that Generating a cloud storage data set includes: Acquire multi-dimensional data through IoT protocols to form environmental data streams; Adopt the transmission protocol standard to format the environmental data stream to obtain a standardized data stream; Segment the standardized data stream using data stream management technology to obtain segmented data stream fragments; If the segmented data stream fragment is interrupted during transmission, the corresponding environmental data stream fragment is re-acquired; The recovered data stream segments are integrated and processed through the cloud platform, and a unified cloud repository is built using cloud data integration technology to obtain the integrated data set. The integrated data set is stored through the data storage architecture.

4. The method according to claim 1, wherein Generating a smooth environment data set comprises: Obtain environmental data from cloud storage, perform preliminary classification on the data through a pre-established data screening mechanism, and obtain classified environmental data groups; The Kalman filter algorithm is used to perform data denoising on the classified environmental data group to generate denoised data units; According to the denoised data units, different categories of data are merged through data integration technology to obtain the integrated data framework; For the integrated data framework, if data is missing or abnormal, the corresponding data fragments are retrieved from the environmental monitoring source through cloud technology to obtain the completed data structure; According to the completed data structure, the smoothing technology is applied to perform secondary optimization on the data to generate an optimized data set; The optimized data set is stored in layers through the cloud storage mechanism to obtain the final stored data archive; According to the final stored data file, the stored content is checked for consistency through data verification to obtain the verified data record.

5. The method according to claim 1, wherein The construction of a dataset containing time series features includes: Acquiring environmental information from the smoothed data, and preliminarily sorting the environmental information using a pre-established classification rule to obtain sorted environmental units; Obtain relevant external data sources, perform correlation processing based on the sorted environmental units using data matching technology, and determine the correlated data groups; For the associated data group, the time series analysis method is used to extract sequence features, build a feature framework including time trends, and determine the integrity of the feature framework; According to the integrity of the feature framework, data fusion technology is applied to deeply combine external related data with environmental information to obtain a fused data set.

6. The method according to claim 1, characterized in that Building an environmental change prediction model includes: The long short-term memory network is used to process the time series related data set, extract the features of the environmental information contained therein, and obtain the preliminary feature combination; Based on the preliminary feature combination, the long-term memory and short-term memory characteristics in the time series are separated and processed to determine the separated long-term features and short-term features; By separating the long-term and short-term characteristics, we can analyze the changing trends in the environment, obtain the pattern information related to the changing trends, and determine the potential fluctuation patterns. Based on the potential fluctuation patterns and combined with key elements in environmental information, sequence analysis is performed to obtain sequence patterns that are closely related to environmental changes; If the fluctuation pattern in the sequence pattern does not conform to the preset threshold range, a secondary feature extraction is performed on the data set to obtain an adjusted feature combination; Based on the adjusted feature combination, the initial framework of the prediction model is constructed, and the prediction parameters corresponding to the change trend in the framework are obtained to obtain the prediction structure; Through the prediction structure, combined with the characteristics of long-term memory and short-term memory, the future trend of environmental changes is simulated to determine the final prediction output.

7. The method according to claim 1, characterized in that Generating environmental change trend data includes: Using the monitoring data of environmental changes and pre-established prediction models, we can make preliminary deductions of future trends and obtain initial trend analysis results. Based on the initial trend analysis results, core data is screened and processed to extract key information related to environmental monitoring and determine the characteristic set of change patterns; By combining the characteristic set of changing rules with the output of trend analysis, data integration is structured to obtain pattern information corresponding to future trends; Based on the pattern information, we use the long short-term memory network to conduct in-depth analysis of time series data in response to potential fluctuations in environmental changes, and determine the distribution characteristics of the fluctuation trend. If the distribution characteristics of the fluctuation trend are inconsistent with the preset threshold range, the core data will be cleaned twice to obtain an adjusted data combination; Through the adjusted data combination, calibration analysis is performed on the intermediate results of trend extraction to obtain prediction parameters that are closely related to environmental changes; Based on the prediction parameters and the final structure of the model output, a dynamic simulation of future trends is performed to determine the trend data set of environmental changes.

8. The method according to claim 1, characterized in that Generating a scientific planting recommendation dataset includes: Through preliminary collation of environmental change trend data and removal of redundant information through data cleaning, a collated environmental data set is obtained; Based on the sorted environmental data set, the data is classified using the random forest model to determine the category distribution related to the planting conditions and obtain the classified feature set; If the distribution of some categories in the classified feature set does not match the preset threshold range, the feature set will be screened again to extract the core indicators that are highly relevant to agricultural planting and determine the core feature combination; Based on the core feature combination, data mapping is performed according to the needs of agricultural planting, matching information adapted to the planting conditions is obtained, and an adapted data framework is obtained; Through the adapted data framework and combined with the business rules of scientific planting, the matching information is structured and integrated to determine the guiding parameters related to planting recommendations; According to the guidance parameters, information is reorganized according to the generation requirements of the recommendation dataset to obtain the final scientific planting recommendation dataset.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images