Tobacco seedling full cycle monitoring management system and method based on Internet of Things

Abstract

The present invention proposes a tobacco seedling full-cycle monitoring and management system and method based on the Internet of Things, which relates to the technical field of tobacco production. The method comprises: collecting multi-dimensional data of each stage of the tobacco seedling cycle through an Internet of Things sensing device; uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data; uploading the preprocessed data to a central processor, and determining the health scores of tobacco seedlings at different stages based on a tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, outputting the type of abnormal health and abnormal location information; automatically adjusting the environmental conditions of a tobacco seedling shed and / or a nutrient pool according to the tobacco seedling health score result; establishing an early warning mechanism to warn of abnormal results; and realizing comprehensive and accurate monitoring and management of the tobacco seedling process by introducing Internet of Things technology, edge processing, machine learning and intelligent decision-making.

Description

Technical Field

[0001] The present application relates to the technical field of tobacco production, and in particular to a tobacco seedling cultivation full-cycle monitoring management system and method based on the Internet of Things. Background Art

[0002] In the field of tobacco production, the seedling stage is a crucial link, which directly affects the yield and quality of tobacco. The traditional tobacco seedling management method mainly relies on manual experience, through regular inspections, manual data recording and adjustment of the seedling environment accordingly. However, this method has many shortcomings: first, the frequency of manual inspections is limited, and it is difficult to achieve real-time and comprehensive monitoring of the seedling environment; second, manual data recording is prone to errors, and it is difficult to conduct accurate data analysis and prediction; finally, the experience-based adjustment method often lacks scientific basis, resulting in unsatisfactory adjustment effects, and may even have an adverse effect on the growth of tobacco seedlings.

[0003] With the rapid development of Internet of Things technology, its application in the agricultural field is becoming more and more extensive. Internet of Things technology can realize real-time monitoring and precise control of the agricultural production environment through sensors, smart devices and other means. However, in the field of tobacco seedling cultivation, although there are some monitoring and management systems based on the Internet of Things, most of these systems focus on data collection and display, lack in-depth analysis of data and intelligent decision-making capabilities, and cannot automatically adjust the seedling environment according to the growth status of tobacco seedlings; traditional experience-based adjustment methods often lack scientific basis, the adjustment effect is not ideal, and may even have an adverse effect on the growth of tobacco seedlings; most of the existing early warning mechanisms can only warn of a single parameter, and cannot comprehensively consider the situation of multiple parameters and multiple locations, and the early warning effect is limited.

[0004] Based on the above problems, this application realizes comprehensive and accurate monitoring and management of the tobacco seedling cultivation process by introducing advanced means such as Internet of Things technology, edge processing, machine learning and intelligent decision-making. Summary of the invention

[0005] In response to the above problems, the present invention proposes a tobacco seedling full-cycle monitoring management system and method based on the Internet of Things. Multi-dimensional data of each stage of the tobacco seedling cycle is collected through the Internet of Things sensing equipment, and the data is pre-processed and analyzed using edge processing and machine learning technology, thereby realizing real-time and accurate monitoring of the growth status of tobacco seedlings. At the same time, the method also establishes a tobacco seedling evaluation model, which can automatically determine the health score of the tobacco seedlings and automatically adjust the environmental conditions of the seedling shed and / or nutrient pool according to the scoring results. In addition, the method also establishes a multi-level early warning mechanism, which can promptly detect and handle abnormal situations to ensure the healthy growth of tobacco seedlings.

[0006] The purpose of this application is achieved by the following technical solutions:

[0007] In the first aspect, the present application provides a tobacco seedling cultivation full-cycle monitoring and management method based on the Internet of Things.

[0008] Collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices;

[0009] Uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data;

[0010] The pre-processed data is uploaded to the central processor, and the health scores of tobacco seedlings at different stages are determined based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the abnormal health type and abnormal location information are output;

[0011] Automatically adjust the environmental conditions of the tobacco seedling shed and / or nutrient pool according to the tobacco seedling health score results;

[0012] Establish an early warning mechanism to warn of abnormal results.

[0013] The present application provides a tobacco seedling cultivation full-cycle monitoring and management system based on the Internet of Things, the system comprising:

[0014] The data collection module is used to collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices;

[0015] A preprocessing module, used for uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data;

[0016] The evaluation module is used to upload the pre-processed data to the central processor and determine the health scores of tobacco seedlings at different stages based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the type of abnormal health and abnormal location information are output;

[0017] A control module, used to automatically adjust the environmental conditions of the tobacco seedling shed and / or nutrient pool according to the tobacco seedling health score result;

[0018] The early warning module is used to establish an early warning mechanism and issue early warnings for abnormal results.

[0019] The beneficial effects of the present invention include: comprehensively collecting multi-dimensional data of each stage of the tobacco seedling raising cycle through the Internet of Things sensing device, realizing the refined monitoring of the tobacco seedling growth environment; automatically adjusting the environmental conditions according to the tobacco seedling health score results, providing a more suitable growth environment for the tobacco seedlings, and helping to improve the quality and yield of the tobacco seedlings; using the tobacco seedling evaluation model and the correlation analysis between environmental parameters and abnormal categories, the health status of the tobacco seedlings can be accurately judged, and abnormalities can be discovered in time. According to the comprehensive evaluation value of the parameters, the adjustment parameters and adjustment amounts are determined, and the intelligent regulation of environmental conditions is realized, which reduces manual intervention and improves management efficiency. A perfect early warning mechanism is established, which can timely discover and warn potential problems, which helps managers to take measures in time to avoid the expansion of problems; different levels of early warning mechanisms can correspond to problems of different severity, so that managers can respond more targetedly. The influence of different tobacco leaf types, growth stages and environmental factors is taken into account, and it has strong adaptability and can be applied to different tobacco seedling raising scenarios. Through the parameter adjustment model, the mutual influence relationship of each preset target parameter and the energy consumption can be comprehensively considered to achieve the optimization adjustment of environmental conditions, which is conducive to energy conservation and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the whole-cycle monitoring and management method of tobacco seedling cultivation based on the Internet of Things provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] Below, the present application is further described in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0022] See also Figure 1 The present application embodiment provides a tobacco seedling cultivation full-cycle monitoring and management method based on the Internet of Things, the method comprising:

[0023] Collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices;

[0024] Uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data;

[0025] The pre-processed data is uploaded to the central processor, and the health scores of tobacco seedlings at different stages are determined based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the abnormal health type and abnormal location information are output;

[0026] Automatically adjust the environmental conditions of the tobacco seedling shed and / or nutrient pool according to the tobacco seedling health score results;

[0027] Establish an early warning mechanism to warn of abnormal results.

[0028] The working principle and effect of the above technical solution are: using IoT sensing devices (such as temperature sensors, humidity sensors, light sensors, soil moisture sensors, etc.) deployed in tobacco seedling sheds and nutrient pools to collect multi-dimensional data at each stage of the tobacco seedling cycle in real time. These data cover key environmental factors that affect tobacco seedling growth, such as temperature, humidity, light intensity, soil moisture content, etc.

[0029] The collected multi-dimensional data is transmitted to the edge processing device via the wireless network. As the first checkpoint of data processing, the edge processing device is responsible for pre-processing operations such as preliminary screening, cleaning, compression and format conversion of the original data to reduce the amount of data transmission, improve data transmission efficiency, and preliminarily eliminate abnormal or invalid data.

[0030] The pre-processed data is uploaded to the central processing unit for further analysis. The central processing unit has a built-in tobacco seedling evaluation model, which is built based on machine learning or deep learning algorithms. It can comprehensively consider the physiological characteristics and environmental requirements of tobacco seedlings at different growth stages and conduct in-depth mining and analysis of the input data.

[0031] By comparing with the preset scoring thresholds for each stage, the tobacco seedling evaluation model can determine the health scores of tobacco seedlings at different stages. If the health score is lower than the scoring threshold for the corresponding stage, the model will output the type of abnormal health (such as malnutrition, pests and diseases, etc.) and the specific abnormal location information.

[0032] According to the tobacco seedling health score results, the system automatically triggers the environmental adjustment mechanism. By controlling the environmental adjustment equipment (such as heating equipment, humidification equipment, ventilation equipment, irrigation equipment, etc.) in the tobacco seedling shed and / or nutrient pool, the temperature, humidity, light and other environmental conditions are automatically adjusted to improve the growth environment of the tobacco seedlings and promote their healthy growth.

[0033] A multi-level early warning mechanism is established to deal with possible abnormal situations. Once abnormal healthy tobacco seedlings are detected or environmental conditions exceed the preset threshold, the system will immediately trigger an early warning and send early warning information to managers through sound and light alarms, SMS notifications, email reminders, etc. At the same time, the system will also provide corresponding improvement measures based on the type and severity of the abnormality to help managers respond and deal with problems quickly.

[0034] In summary, the Internet of Things-based tobacco seedling full-cycle monitoring and management method realizes comprehensive monitoring, precise analysis and intelligent management of the tobacco seedling cultivation process by integrating advanced technologies such as Internet of Things technology, edge computing, big data analysis and intelligent decision-making, providing a strong guarantee for improving tobacco yield and quality.

[0035] In some embodiments, the multi-dimensional data includes tobacco seedling images and environmental parameters; the multi-dimensional data of each stage of the tobacco seedling cultivation cycle collected by the IoT sensing device includes:

[0036] A plurality of positioning base stations are arranged in the seedling raising shed, and the seedling raising tray is positioned by using the positioning base stations and positioning tags; an intelligent system with positioning tags is used to trigger the track camera to locate and capture the tobacco seedling images in the seedling raising tray;

[0037] The environmental parameters of the nursery shed and each nutrient pool are obtained through sensors; the environmental parameters include the water depth, water temperature, water quality parameters of the nutrient pool and the environmental temperature, humidity, soil parameters, water quality parameters and light conditions of the nursery shed.

[0038] The working principle and effect of the above technical solution are as follows: multiple positioning base stations are set up in the seedling shed, and these base stations use wireless signals or RFID (radio frequency identification) technology to accurately locate the seedling trays equipped with positioning tags.

[0039] The intelligent system with positioning tags can trigger the track camera to move precisely to the corresponding position and capture the image of the tobacco seedlings in the seedling tray according to the positioning information. This image acquisition method is not only efficient, but also ensures that each tray of tobacco seedlings is carefully monitored. Through various sensors deployed in the seedling shed and each nutrient pool, key environmental parameters are obtained in real time. These parameters include but are not limited to the water depth, water temperature, water quality parameters (such as pH value, dissolved oxygen content, etc.) of the nutrient pool, as well as the ambient temperature, humidity, soil parameters (such as soil moisture, soil pH value, etc.), and light conditions in the seedling shed.

[0040] In some embodiments, uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data includes:

[0041] Clean the received data through edge processing devices to remove duplicate, abnormal or invalid data;

[0042] Compress the image;

[0043] The data is associated and integrated through positioning tags, including the tobacco seedling data of the seedling tray, the historical environmental parameters of the nutrient pool where the seedling tray is located, the historical environmental parameters of the seedling shed where the tobacco seedlings are located, and pictures of the growth process of the tobacco seedlings in the seedling tray and related information.

[0044] In some embodiments, the method for determining abnormal data includes:

[0045] Dividing a day into a first time period; dividing a day into a first time period, wherein the first time period includes early morning, daytime, first half of the night, and second half of the night;

[0046] The plurality of data of the same sensor in the same first time period at the same stage of the seedling raising cycle and the variation of the adjacent data are taken as the first cluster;

[0047] Taking multiple data of multiple adjacent sensors of the same type at the same time point as the second cluster; counting the difference between the sensors in the second cluster at the same time point as the second difference; taking multiple groups of second difference values ​​at multiple time points as the third cluster;

[0048] According to the first cluster and the third cluster, an abnormality determination model for data of the same type of sensor in the first time period is established through machine learning;

[0049] Determine abnormal data and the location of abnormal data through the data anomaly determination model; and send reminders to the monitoring terminal;

[0050] According to the feedback from the monitoring terminal, optimize the corresponding data anomaly judgment model.

[0051] The working principle of the above technical solution is:

[0052] IoT sensing devices (such as sensors, cameras, etc.) continuously monitor environmental parameters (such as temperature, humidity, light intensity, water quality, etc.) and tobacco seedling images in the tobacco seedling shed. These devices are usually deployed in key locations, such as near the seedling trays, in the nutrient pool, and in different areas of the entire seedling shed to ensure comprehensive coverage.

[0053] The edge processing device first receives data from the IoT sensing device, including tobacco seedling images and environmental parameters. The received data is initially cleaned to remove duplicate, abnormal or invalid data. This usually involves checking the integrity and consistency of the data and whether it conforms to the preset data format and range.

[0054] For the collected tobacco seedling images, the edge processing device will compress them to reduce the bandwidth requirements for data transmission and the storage space occupied. The selection of compression algorithm should ensure that the quality and details of the image are maintained as much as possible while reducing the amount of data for subsequent analysis and identification.

[0055] Using the information from the positioning tags, edge processing devices can correlate and integrate data from different sources.

[0056] Specifically, the tobacco seedling data of the seedling tray, the historical environmental parameters of the nutrient pool where the seedling tray is located, the historical environmental parameters of the seedling shed where the tobacco seedlings are located, and the pictures and related information of the tobacco seedlings growing process in the seedling tray are linked to form a complete data chain. This kind of association and integration will help to conduct a comprehensive analysis of the tobacco seedlings' growth status and environmental factors in the future.

[0057] In order to effectively identify and process abnormal data, a machine learning model based on time clustering is adopted. The specific steps are as follows:

[0058] Time period division: Divide a day into early morning, daytime, first half of the night, second half of the night, etc., so as to consider the impact of time factors (light, temperature and humidity) on the data.

[0059] Data cluster analysis:

[0060] The first cluster: considers the data and its changes of the same sensor in the same stage and time period of the seedling cycle.

[0061] The second cluster: Analyze the data differences between multiple adjacent sensors of the same type at the same time point.

[0062] The third cluster: Count the differences in the second cluster data at multiple time points to form a higher-level data cluster.

[0063] Anomaly determination model establishment: Combining the first and third clusters, a machine learning algorithm is used to build an abnormal data determination model that can identify data fluctuations beyond the normal range.

[0064] Abnormal data determination and feedback: Once the model determines abnormal data, the system will immediately notify the monitoring terminal and dynamically optimize the abnormal determination model based on the terminal feedback results to improve accuracy and response speed. Once abnormal data is found, the system will immediately send a reminder message to the monitoring terminal so that the management personnel can take timely measures to deal with it.

[0065] The effect of the above technical solution is: by cleaning the received data through the edge processing device, duplicate, abnormal or invalid data can be effectively removed, thereby improving the accuracy and reliability of subsequent analysis and decision-making. Compressing and processing image data not only reduces the storage and transmission load, but also retains sufficient image information for subsequent analysis and identification. Data preprocessing on the edge processing device reduces the amount of data transmitted to the central processor or the cloud, thereby reducing the demand and latency of network bandwidth. The preprocessed data is more compact and orderly, which is conducive to subsequent rapid analysis and processing. The data is associated and integrated through positioning tags, and the tobacco seedling data, environmental parameters, growth process pictures and other information of the seedling tray are closely combined, providing a rich data source for comprehensive analysis of the growth status of tobacco seedlings. This correlation helps to reveal the complex relationship between tobacco seedling growth and environmental factors, and provides strong support for precise regulation.

[0066] Dividing a day into different time periods and establishing anomaly determination models based on the data characteristics in different time periods can more accurately identify abnormal data that may appear in different time periods. By considering the data changes of the same sensor during the seedling cycle and the data differences between adjacent sensors, the model can capture more subtle abnormal changes. Once abnormal data is found, the system can immediately send an alert message to the monitoring terminal, allowing managers to respond quickly and take measures. This real-time early warning mechanism helps reduce losses caused by the failure to detect abnormal situations in time.

[0067] Based on the feedback from the monitoring terminal, the system can continuously optimize the abnormality judgment model to improve its accuracy and stability. This continuous optimization mechanism helps the system to continuously adapt to the complex changes in the tobacco seedling cultivation process and improve the overall intelligent level of monitoring and management.

[0068] In some embodiments, the compressing the image comprises:

[0069] Evaluate the image quality and adjust the image compression ratio based on the quality evaluation results, illumination change results, and tobacco seedling evaluation model feedback results; including:

[0070] Image quality is obtained by calculating the entropy, edge strength, and texture complexity features of the image;

[0071] The image quality is:

[0072] ;

[0073] in, is the quality of the g-th image, is the score of the e-th element of the g-th image; The weight of the e-th element in the g-th image; f is the number of elements, including the entropy, edge strength, and texture complexity of the calculated image;

[0074] Determine the initial compression ratio based on the average quality of the image or the statistical results of historical data;

[0075] ;

[0076] Among them, YSBg is the compression ratio of the g-th image, YSBy is the preset compression ratio, Zy is the preset image quality; it can be the average quality of historical images; α is the adjustment coefficient, which is set according to the illumination change and feedback results. ; YSBmin is the historical minimum compression ratio, and YSBmax is the historical maximum compression ratio.

[0077] The working principle and effect of the above technical solution are:

[0078] Based on the results of image quality assessment, an initial compression ratio benchmark can be set; this benchmark can be determined based on the average quality of the image or the statistical results of historical data; for example, for most cases where the image quality is good, a lower compression ratio can be set; while for cases where the image quality is poor or there are fewer details, the compression ratio can be increased.

[0079] When an image is captured and prepared for compression, its quality is first evaluated; based on the image quality evaluation result, the compression ratio is dynamically calculated; for example, if the image quality evaluation result is higher than a preset threshold, the compression ratio is reduced to retain more details; if it is lower than the threshold, the compression ratio is increased to reduce the amount of data.

[0080] If subsequent analysis (such as through the tobacco seedling evaluation model) finds that too much image information is lost or the data volume is still too large due to improper compression ratio setting, the compression ratio can be adjusted again based on the feedback results.

[0081] A time window can be set to periodically re-evaluate image quality and adjust compression ratio. This can be based on a specific time period of the day (such as when light conditions change) or when a specific event occurs according to the seedling cycle (such as environmental changes exceeding a threshold, seedling tray position changes, etc.), triggering the process of re-evaluating and adjusting the compression ratio; stage by stage. Adjust the compression ratio by adjusting the coefficient.

[0082] For example, the initial compression ratio is set to 50% (i.e. the amount of data after compression is 50% of the original image), and is dynamically adjusted through the following strategies:

[0083] If the image quality assessment results indicate that the image is rich in detail and important (such as critical growth stages or anomalies), adjust the compression ratio to 30% or lower.

[0084] If image quality assessment results indicate that the image has little detail or changes (such as at night or during a period of stable growth), increase the compression ratio to 70% or higher.

[0085] As the system runs and data accumulates, the compression ratio adjustment algorithm and strategy can be continuously optimized based on historical data and actual results to achieve the best compression effect and data transmission efficiency.

[0086] In some embodiments, uploading the preprocessed data to a central processor and determining a tobacco seedling health score based on a tobacco seedling evaluation model includes:

[0087] A first classification is performed according to the types of tobacco leaves; under the first classification, each growth stage is divided by a preset growth cycle; the historical data is first divided according to the time series of each growth stage to obtain a first group; wherein, the start of each sowing preparation is regarded as the beginning of a new time series;

[0088] Divide the data of the first group into the second group according to each day in the time series;

[0089] Cluster the data of the second group according to natural light, temperature and humidity to obtain the third group;

[0090] Counting the characteristics of the tobacco seedlings in each third group and the corresponding environmental characteristics, including the parameter mean, standard deviation and image characteristics of multiple samplings; the image characteristics include the height, color, number of leaf petals and size of leaf petals of the tobacco seedlings;

[0091] Comparing the similarity of tobacco seedling characteristics of the same third group in a plurality of second groups adjacent to each other in time series;

[0092] ;

[0093] ;

[0094] Among them, Sp is the total similarity of the two groups of tobacco seedling features, Sj is the similarity of the jth feature; xj is the weight of the jth feature; n is the total number of tobacco seedling features;

[0095] Auj and Buj are the component vectors of the j-th eigenvectors of the two groups respectively; v is the number of components of the j-th eigenvector;

[0096] If the similarity is greater than a preset similarity threshold, the adjacent multiple second groups are taken as a time series window, and the mean, maximum, minimum, and standard deviation values ​​of the tobacco seedling characteristics of the consecutive and identical third groups and the mean, maximum, minimum, and standard deviation values ​​of the corresponding environmental parameters are input as features into the tobacco seedling evaluation model; the environmental parameters include nutrient pool environmental information and seedling shed environmental information;

[0097] According to the current data and the corresponding time series, based on the tobacco seedling evaluation model, the tobacco seedling health scores at different stages are determined.

[0098] The working principle and effect of the above technical solution are as follows: First, a first classification is performed according to the types of tobacco leaves to ensure that data of different types of tobacco leaves can be processed separately to adapt to their respective unique growth characteristics and needs.

[0099] Under the first classification, the growth process of tobacco seedlings is divided into different growth stages through the preset growth cycle; this helps the system to more accurately identify the growth stage of tobacco seedlings and adjust the evaluation criteria and parameters accordingly; then, according to the time series of each growth stage, the historical data is divided into the first grouping. Starting from each sowing preparation, as the beginning of a new time series, the continuity, integrity and referenceability of the data are ensured.

[0100] The data of the first group is divided into a second group according to each day in the time series, so as to analyze and process the data of each day in more detail. Then, the data of the second group is clustered according to environmental parameters such as natural light, temperature and humidity to form a third group. Clustering the data according to environmental parameters (such as natural light, temperature and humidity) to form a third group helps to identify data sets with similar environmental conditions, which can more accurately evaluate the impact of the environment on the growth of tobacco seedlings, reduce the amount of data processed by the model, improve the efficiency of data processing, and optimize the utilization of system resources.

[0101] The characteristics of the tobacco seedlings in each third group and the corresponding environmental characteristics are statistically analyzed, including the parameter mean, standard deviation and image characteristics (such as the height, color, number of petals, size of petals, etc.) of multiple samplings; these characteristics together form the basis for evaluating the health status of tobacco seedlings.

[0102] Compare the similarity of tobacco seedling features of the same third group in multiple adjacent second groups in the time series. By calculating the similarity between feature vectors (such as using the cosine similarity formula), the system can evaluate the stability and change trend of tobacco seedling features at different time points. By comparing the similarity of tobacco seedling features of the same third group in multiple adjacent second groups in the time series and constructing a time series window, the system can focus on relatively stable data segments that are less affected by environmental factors for analysis, thereby reducing unnecessary calculations and data redundancy.

[0103] If the similarity of tobacco seedling characteristics of the same third group in the adjacent plurality of second groups is greater than a preset similarity threshold, these adjacent groups are taken as a time series window, which means that during these time periods, the growth state of the tobacco seedlings is relatively stable and less affected by environmental factors.

[0104] The mean, maximum, minimum and standard deviation of the tobacco seedling characteristics of the third consecutive group and the corresponding environmental parameters are input into the tobacco seedling evaluation model as features. These features fully reflect the growth status and environmental conditions of the tobacco seedlings, providing rich input data for the model and reducing the amount of model data.

[0105] According to the current data and the corresponding time series, based on the pre-trained tobacco seedling evaluation model, the system can determine the tobacco seedling health score at different stages. This score combines information on tobacco seedling growth characteristics, environmental parameters, historical data, and other aspects, providing managers with intuitive assessment results of tobacco seedling health status.

[0106] In summary, this working principle achieves accurate assessment and prediction of the health status of tobacco seedlings through sophisticated data classification, grouping, feature statistics and similarity comparison, as well as the construction of time series windows and the application of tobacco seedling evaluation models.

[0107] In some embodiments, the automatic adjustment of the environmental conditions of the tobacco seedling shed and / or the nutrient pool according to the tobacco seedling health score result includes:

[0108] If the tobacco seedlings have abnormal health scores, the abnormal category is determined;

[0109] Based on the correlation analysis between the environmental parameters and the abnormal categories; obtaining the correlation coefficient between the environmental parameters and the abnormal categories;

[0110] If there are multiple abnormal positions in the same nursery shed and the abnormal categories are the same; then calculate the ratio of the total area of ​​the multiple abnormal positions to the total area of ​​the nursery shed as the first ratio, and obtain the distribution of the multiple abnormal positions;

[0111] If the first ratio is less than or equal to a preset ratio threshold; or, multiple abnormal locations are concentrated;

[0112] Then, the current parameters of the multiple first abnormal positions are taken as an abnormal set, and the current parameters of other positions are taken as a normal set; the difference between the current parameters in the abnormal set and the normal set is obtained as the first difference; wherein the first difference is:

[0113] ;

[0114] Where, Delk is the first difference of the k-th parameter; Cak is the mean of the k-th parameter in the abnormal set; Cnk is the mean of the k-th parameter in the normal set; ;

[0115] Obtain a comprehensive parameter evaluation value based on the abnormal health conditions that affect the growth cycle or tobacco seedlings and the first difference;

[0116] The comprehensive evaluation value of the parameters is: ;

[0117] Where Wk is the correlation coefficient between the kth category and the abnormal health category; abs() is the absolute value;

[0118] According to the comprehensive evaluation value of the parameters, the corresponding adjustment parameters and adjustment amounts are determined.

[0119] The working principle and effect of the above technical solution are as follows: first, the health score of the tobacco seedlings is monitored, and when an abnormal health score is found, the abnormal category is automatically determined. These abnormal categories may include but are not limited to abnormal leaf color, abnormal development, and abnormal height development.

[0120] Next, based on historical data and professional knowledge, the correlation analysis between environmental parameters (such as light, temperature, humidity, etc.) and abnormal categories is carried out. Through statistical analysis or machine learning algorithms, the correlation coefficients (Wk) between each environmental parameter and different abnormal categories are calculated. These coefficients reflect the closeness between environmental parameters and abnormal categories.

[0121] If there are multiple abnormal locations in the same nursery shed, and the abnormal categories of these locations are the same, the distribution and impact range of these abnormal locations will be further analyzed; by calculating the ratio of the total area of ​​multiple abnormal locations to the total area of ​​the nursery shed (the first ratio), and observing whether the distribution of the abnormal locations is concentrated, the degree and scope of the impact of the abnormality can be preliminarily judged.

[0122] If the first ratio is less than or equal to the preset ratio threshold, or multiple abnormal locations are concentrated, the current environmental parameters of these abnormal locations are taken as an abnormal set, and the current environmental parameters of other locations are taken as a normal set; in order to more accurately identify the key factors causing the abnormality.

[0123] The system will calculate the difference between the current environmental parameters in the abnormal set and the normal set (the first difference Delk). This difference reflects the difference between the environmental parameters of the abnormal location and the normal location. , the system is able to normalize these differences for subsequent comparison and analysis.

[0124] Next, the system will calculate the comprehensive parameter evaluation value (Pk) based on the first difference, the correlation coefficient (Wk) between the environmental parameters and the abnormality category, and the degree of impact of the abnormality on the growth cycle or the health of the tobacco seedlings. The comprehensive parameter evaluation value (Pk) calculated by combining the first difference, the correlation coefficient (Wk) and the degree of impact of the abnormality on the growth cycle or the health of the tobacco seedlings can more comprehensively reflect the impact of environmental parameters on the health of tobacco seedlings.

[0125] Finally, according to the comprehensive evaluation value of the parameters, the environmental parameters that need to be adjusted and the corresponding adjustment amounts are determined. These adjustments are aimed at restoring the environmental conditions in abnormal locations to normal levels to promote the healthy growth of tobacco seedlings.

[0126] In summary, this technical solution automatically detects abnormal health scores of tobacco seedlings, analyzes the association between environmental parameters and abnormal categories, evaluates the degree and scope of impact of abnormal locations, calculates the difference and comprehensive evaluation value of environmental parameters, and finally determines and implements targeted environmental adjustment measures to achieve intelligent management of environmental conditions in tobacco seedling sheds and / or nutrient pools.

[0127] In some embodiments, determining the corresponding adjustment parameter and parameter adjustment amount according to the comprehensive parameter evaluation value includes:

[0128] Select the parameters corresponding to the top N comprehensive evaluation values ​​or the parameters greater than the preset evaluation value threshold as the preset target parameters for adjustment; Use the mean of each parameter in the normal set as the expected value of the corresponding parameter adjustment; N ≥ 1 and is an integer;

[0129] A parameter adjustment model is established through the expected values ​​of each preset target parameter, the mutual influence relationship of each preset target parameter, and the energy consumption;

[0130] The adjustment amount of each target parameter is obtained through the parameter adjustment model;

[0131] If the adjustment amount of a certain parameter is less than or equal to the absolute value of the corresponding adjustment threshold;

[0132] Then the parameter at the position is adjusted directly by the adjustment amount;

[0133] If the adjustment amount of a certain parameter is greater than the absolute value of the corresponding adjustment threshold;

[0134] The absolute value of the parameter threshold is used as a single adjustment amount to adjust the parameter at the position.

[0135] The working principle of the above technical solution is: first, select the parameters that need to be adjusted first from all parameters. This can be done in two ways: one is to select the parameters with the top N comprehensive evaluation values ​​(Pk); the other is to select the parameters with comprehensive evaluation values ​​greater than the preset evaluation value threshold. Both methods can ensure that the parameters that have the greatest impact on the health of tobacco seedlings are the priority objects for adjustment.

[0136] The mean of each parameter in the normal set is taken as the adjusted expected value of the corresponding parameter; this is because the normal set represents the environmental conditions of most healthy tobacco seedlings in the nursery shed or nutrient pool, so its mean can be used as the ideal environmental parameter value.

[0137] Establish a parameter adjustment model (multi-objective optimization model); this model considers the expected values ​​of each preset target parameter, the mutual influence relationship between each preset target parameter (for example, increasing light may increase temperature, so it is necessary to consider the adjustment of temperature and light at the same time), and energy consumption (because the adjustment of environmental parameters often requires the consumption of energy, such as electricity or fuel). By comprehensively considering these factors, the model can calculate a parameter adjustment plan that is both economical and effective. A mixed integer linear programming (MILP) model or a nonlinear programming (NLP) model may be used in combination with a heuristic search strategy to solve it. Using the parameter adjustment model, calculate the adjustment amount of each target parameter; these adjustment amounts represent the specific amount that each parameter needs to be adjusted to achieve the desired environmental conditions.

[0138] After obtaining the adjustment amount, decide how to implement the adjustment based on the size of the adjustment amount.

[0139] If the adjustment amount of a certain parameter is less than or equal to the absolute value of the corresponding adjustment threshold, it means that the adjustment amount is relatively small, and the parameter at the position can be adjusted directly by the adjustment amount, thereby ensuring the accuracy and timeliness of the adjustment.

[0140] If the adjustment amount of a parameter is greater than the absolute value of the corresponding adjustment threshold, it means that the adjustment amount is relatively large. In order to avoid adverse effects on tobacco seedlings caused by excessive one-time adjustment, the absolute value of the parameter threshold can be used as a single adjustment amount. The desired environmental conditions can be gradually achieved through multiple small adjustments.

[0141] After implementing parameter adjustments, it is necessary to continuously monitor the health of tobacco seedlings and the environmental conditions of the seedling shed / nutrient pool. If the adjustment effect is not good or new problems arise, it is necessary to adjust the strategy in time and recalculate the adjustment amount. Through this process, intelligent and refined management of the environmental conditions of the tobacco seedling shed and / or nutrient pool can be achieved, thereby improving the growth quality and yield of tobacco seedlings.

[0142] The effect of the above technical solution is: by selecting the parameters with the top N comprehensive evaluation values ​​or the parameters greater than the preset evaluation value threshold as the adjustment object, the environmental parameters with the greatest impact on the health of tobacco seedlings can be accurately located, so as to give priority to solving key problems. The mean value of each parameter in the normal set is used as the adjustment expectation value, which provides a scientific and reasonable target value for parameter adjustment, and helps to restore environmental conditions to a state that is conducive to the growth of tobacco seedlings.

[0143] When establishing the parameter adjustment model, various factors such as the expected value of each preset target parameter, the mutual influence relationship, and energy consumption were considered to ensure the comprehensiveness and economy of the adjustment plan. By setting the adjustment threshold and adopting the strategy of phased adjustment when the adjustment amount exceeds the threshold, the adverse effects on the tobacco seedlings caused by excessive one-time adjustment are avoided. This helps to protect the tobacco seedlings from the impact of sudden environmental changes. Energy consumption was taken into account when establishing the adjustment model, so that the adjustment plan can reduce energy consumption as much as possible while achieving the goal. It helps to reduce production costs and improve economic benefits. By comprehensively considering multiple factors and setting adjustment thresholds, the system's adaptability and robustness to environmental changes are enhanced. Even in the face of complex and changing environmental conditions, a stable adjustment effect can be maintained.

[0144] In summary, this method achieves intelligent and refined management of the environmental conditions of tobacco seedling sheds and / or nutrient pools by accurately locating problem parameters, scientifically setting adjustment targets, comprehensively considering multiple factors, and avoiding excessive adjustments, thereby promoting the healthy growth of tobacco seedlings and improving economic benefits.

[0145] In some embodiments, the method further comprises:

[0146] If the first ratio is greater than a preset ratio threshold and the first abnormal positions are dispersedly distributed;

[0147] Then send an early warning to remind the terminal to conduct analysis, and recommend improvement measures based on the terminal analysis results;

[0148] If the abnormality categories of multiple abnormal locations are different;

[0149] Then, the influencing parameters of the abnormal health categories of multiple abnormal positions are obtained respectively;

[0150] and respectively calculate the difference between each parameter and the corresponding mean in the normal set as the second difference;

[0151] ;

[0152] Where Dik is the second difference of the k-th parameter at the ith abnormal position; Cik is the current value of the k-th parameter at the ith abnormal position; Cnk is the mean value of the k-th parameter in the normal set; ;

[0153] According to the second difference and the influence coefficient of the corresponding parameter; determining a second comprehensive evaluation value of the parameter;

[0154] ;

[0155] Select the parameter corresponding to the second comprehensive evaluation value ranked in the top N or the parameter greater than the preset evaluation value threshold as the preset target parameter for adjustment;

[0156] If multiple locations contain the same preset target parameter, the same preset target parameter is adjusted first.

[0157] The working principle of the above technical solution is: when it is detected that the first ratio (which may represent a certain degree of abnormality or the degree of deviation from the normal state) is greater than the preset ratio threshold, and these abnormal positions are dispersed in space, an early warning reminder is sent to the terminal (such as the administrator's mobile device or monitoring center).

[0158] After receiving the warning, the terminal will analyze it to determine the specific cause and nature of the abnormality. Based on the analysis results, the terminal will recommend corresponding improvement measures. These measures may include adjusting specific environmental parameters, checking for equipment failures, or performing other maintenance operations.

[0159] If the system detects anomalies at multiple locations, and the anomalies are of different categories (e.g., insufficient light, high temperature, low humidity, etc.), these anomalies are handled separately. For each anomaly location, the system identifies the influencing parameters of the health category (e.g., light, temperature, etc.) that caused its anomaly.

[0160] By formula Calculate the difference between each influencing parameter and the corresponding mean in the normal set (second difference). Here, Cik is the current value of the k-th parameter at the i-th abnormal position, and Cnk is the mean of the k-th parameter in the normal set; this difference reflects the degree of deviation between the current value and the normal value.

[0161] Calculate the second comprehensive evaluation value: Based on the second difference and the influence coefficient of the corresponding parameter, use the formula Calculate the second comprehensive evaluation value of each parameter. This value comprehensively considers the deviation degree and importance of the parameter.

[0162] The parameters corresponding to the second comprehensive evaluation values ​​ranked in the top N or the parameters greater than the preset evaluation value threshold are selected as the preset target parameters that need to be adjusted.

[0163] If multiple locations contain the same preset target parameters, the system will prioritize adjusting these common parameters. This is because these parameters may have a significant impact on the health of the entire system or multiple locations, and prioritizing adjustments can restore the system to normal status more quickly.

[0164] The entire workflow achieves intelligent management and adjustment of complex environmental conditions by accurately identifying anomalies, classifying and processing, calculating differences and comprehensive evaluation values, and prioritizing the adjustment of important parameters. This method not only improves the response speed and accuracy of the system, but also effectively reduces the adverse effects of environmental anomalies on tobacco seedling growth.

[0165] The effect of the above technical solution is: when the system detects an abnormality and the abnormal location is scattered, it can quickly send an early warning reminder to the terminal; this ensures that the administrator or monitoring center can obtain the abnormal information in the first time, so as to respond quickly and take corresponding measures.

[0166] Through classification processing, different types of abnormal problems can be solved more targetedly, improving the efficiency and accuracy of processing. By calculating the second difference and the second comprehensive evaluation value, the importance of each parameter and the degree of deviation from the normal state can be scientifically evaluated, which helps to determine the preset target parameters that need to be adjusted first and ensure that resources are reasonably allocated and used. By giving priority to adjusting the parameters with the greatest impact, limited resources can be used more effectively. This helps to reduce energy consumption and operating costs and improve overall economic benefits.

[0167] In summary, this technical solution achieves comprehensive and efficient management of complex environmental conditions through timely warning, precise positioning, scientific assessment, intelligent management and priority adjustment, providing strong guarantees for the healthy growth of tobacco seedlings.

[0168] In some embodiments, the early warning mechanism includes:

[0169] If any parameter at any location is outside the preset threshold range, a first warning is issued;

[0170] If the same parameter is outside the preset threshold range at multiple locations in the same nursery shed, a second warning will be issued;

[0171] If multiple parameters are outside the preset threshold range at multiple locations in the same nursery shed, a third warning will be issued.

[0172] The working principle and effect of the above technical solution are as follows: when the system detects that any environmental parameter (such as temperature, humidity, light, etc.) at any location in the seedling shed exceeds the preset threshold range, the first warning will be triggered immediately. This warning usually indicates that there may be local environmental problems at that location, which requires the attention of the administrator and may require preliminary inspection or adjustment.

[0173] If the same parameter is found to exceed the preset threshold range in multiple locations in the same nursery, the system will trigger the second warning. This means that there may be a common problem with the parameter in the entire nursery, such as the temperature in the entire shed is too high or the humidity is too low. This warning prompts the administrator to pay special attention to and adjust the parameter to ensure the consistency and stability of the shed environment.

[0174] When multiple environmental parameters exceed the preset threshold range at multiple locations in the same seedling shed, the system will trigger the third warning. This is the highest level of warning, indicating that there are multiple environmental factors in the seedling shed that are abnormal, which may seriously affect the growth of tobacco seedlings. This warning requires the administrator to take immediate action to comprehensively check and adjust the environment in the shed to ensure the healthy growth of tobacco seedlings.

[0175] This early warning mechanism helps administrators quickly identify the severity and scope of environmental problems in the seedling shed by distinguishing different levels of warnings. From the preliminary inspection of a single point and a single parameter, to the general problem attention of a single point and multiple parameters, to the comprehensive response of multiple points and multiple parameters, a step-by-step early warning and processing process is formed, which not only improves the response speed and accuracy of the system, but also helps to reduce the adverse effects of environmental problems on the growth of tobacco seedlings.

[0176] The present application embodiment provides a tobacco seedling cultivation full-cycle monitoring and management system based on the Internet of Things, the system comprising:

[0177] The data collection module is used to collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices;

[0178] A preprocessing module, used for uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data;

[0179] The evaluation module is used to upload the pre-processed data to the central processor and determine the health scores of tobacco seedlings at different stages based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the type of abnormal health and abnormal location information are output;

[0180] A control module, used to automatically adjust the environmental conditions of the tobacco seedling shed and / or nutrient pool according to the tobacco seedling health score result;

[0181] The early warning module is used to establish an early warning mechanism and issue early warnings for abnormal results.

[0182] In some embodiments, the data acquisition module includes:

[0183] The positioning unit is used to set up multiple positioning base stations in the seedling shed, and use the positioning tags to locate the seedling tray through the positioning base stations; use the intelligent system with the positioning tags to trigger the track camera to locate and capture the tobacco seedling images in the seedling tray;

[0184] The collection unit is used to obtain the environmental parameters of the seedling shed and each nutrient pool through sensors; the environmental parameters include the water depth, water temperature, water quality parameters of the nutrient pool and the environmental temperature, humidity, soil parameters, water quality parameters and light conditions of the seedling shed.

[0185] In some embodiments, the preprocessing module comprises:

[0186] A first processing unit, configured to clean the received data through an edge processing device to remove duplicate, abnormal or invalid data;

[0187] An image compression unit, used for compressing images;

[0188] The association unit is used to associate and integrate data through positioning tags, and associate the tobacco seedling data of the seedling tray, the historical environmental parameters of the nutrient pool where the seedling tray is located, the historical environmental parameters of the seedling shed where the tobacco seedlings are located, and pictures of the growth process of the tobacco seedlings in the seedling tray and related information.

[0189] In some embodiments, the method for determining abnormal data includes:

[0190] Dividing a day into a first time period; dividing a day into a first time period, wherein the first time period includes early morning, daytime, first half of the night, and second half of the night;

[0191] The plurality of data of the same sensor in the same first time period at the same stage of the seedling raising cycle and the variation of the adjacent data are taken as the first cluster;

[0192] Taking multiple data of multiple adjacent sensors of the same type at the same time point as the second cluster; counting the difference between the sensors in the second cluster at the same time point as the second difference; taking multiple groups of second difference values ​​at multiple time points as the third cluster;

[0193] According to the first cluster and the third cluster, an abnormality determination model for data of the same type of sensor in the first time period is established through machine learning;

[0194] Determine abnormal data and the location of abnormal data through the data anomaly determination model; and send a reminder to the monitoring terminal;

[0195] According to the feedback from the monitoring terminal, optimize the corresponding data anomaly judgment model.

[0196] In some embodiments, the assessment module includes:

[0197] The first grouping unit is used to perform a first classification according to the type of tobacco leaves; under the first classification, the growth stages are divided according to the preset growth cycle; the historical data is first divided according to the time series of each growth stage to obtain a first group; wherein, the start of each sowing preparation is regarded as the beginning of a new time series;

[0198] A second grouping unit is used to divide the data of the first group into second groups according to each day in the time series;

[0199] A third grouping unit, used for clustering the data of the second group according to natural light, temperature and humidity to obtain a third group;

[0200] A comparison unit, which counts the characteristics of the tobacco seedlings in each third group and the corresponding environmental characteristics, including the parameter mean, standard deviation and image characteristics of multiple samplings; the image characteristics include the height, color, number of petals and size of the tobacco seedlings; and compares the similarity of tobacco seedling characteristics of the same third group in multiple second groups adjacent in time series;

[0201] ;

[0202] Among them, Sp is the total similarity of the two groups of tobacco seedling features, Sj is the similarity of the jth feature; xj is the weight of the jth feature; n is the total number of tobacco seedling features;

[0203] A model input unit, for if the similarity is greater than a preset similarity threshold, treating multiple adjacent second groups as a time series window, and treating the mean, maximum, minimum, and standard deviation values ​​of tobacco seedling characteristics of consecutive identical third groups as features to input into a tobacco seedling evaluation model; the environmental parameters include nutrient pool environmental information and seedling shed environmental information;

[0204] The evaluation unit is used to determine the health scores of tobacco seedlings at different stages based on the tobacco seedling evaluation model according to the current data and the corresponding time series.

[0205] In some embodiments, the control unit comprises:

[0206] An abnormality determination unit, used to determine the abnormality category if the tobacco seedling has an abnormal health score;

[0207] A correlation coefficient determination unit, used for obtaining the correlation coefficient between the environmental parameter and the abnormal category based on the correlation analysis between the environmental parameter and the abnormal category;

[0208] A first distribution determination unit is used for calculating the ratio of the total area of ​​the multiple abnormal positions to the total area of ​​the seedling raising shed as a first ratio if there are multiple abnormal positions in the same seedling raising shed and the abnormal categories are the same, and obtaining the distribution of the multiple abnormal positions;

[0209] The first determination unit, if the first ratio is less than or equal to a preset ratio threshold; or, multiple abnormal positions are concentratedly distributed; then the current parameters of multiple first abnormal positions are taken as an abnormal set, and the current parameters of other positions are taken as a normal set; the difference between the current parameters in the abnormal set and the normal set is obtained as a first difference; wherein the first difference is:

[0210] ;

[0211] Where, Delk is the first difference of the k-th parameter; Cak is the mean of the k-th parameter in the abnormal set; Cnk is the mean of the k-th parameter in the normal set; ;

[0212] Obtain a comprehensive parameter evaluation value based on the abnormal health conditions that affect the growth cycle or tobacco seedlings and the first difference;

[0213] The comprehensive evaluation value of the parameters is: ;

[0214] Where Wk is the correlation coefficient between the kth category and the abnormal health category; abs() is the absolute value;

[0215] The first adjustment unit is used to determine the corresponding adjustment parameter and adjustment amount according to the comprehensive evaluation value of the parameter.

[0216] In some embodiments, the first adjustment unit includes:

[0217] The first sorting subunit is used to select the parameters corresponding to the top N comprehensive evaluation values ​​or the parameters greater than the preset evaluation value threshold as the preset target parameters for adjustment; the mean value of each parameter in the normal set is used as the expected value of the corresponding parameter adjustment;

[0218] The adjustment model establishment unit establishes a parameter adjustment model through expected values ​​of various preset target parameters, mutual influence relationships of various preset target parameters, and energy consumption;

[0219] A target adjustment amount acquisition subunit is used to obtain the adjustment amount of each target parameter through the parameter adjustment model;

[0220] A first adjustment subunit is used to directly adjust the parameter at the position by the adjustment amount if the adjustment amount of a certain parameter is less than or equal to the absolute value of the corresponding adjustment threshold;

[0221] The second adjustment subunit is used for adjusting the parameter at the position by taking the absolute value of the parameter threshold as a single adjustment amount if the adjustment amount of a certain parameter is greater than the absolute value of the corresponding adjustment threshold.

[0222] In some embodiments, the control unit further comprises:

[0223] The second determination unit sends an early warning to the terminal for analysis if the first ratio is greater than a preset ratio threshold and the first abnormal positions are dispersed, and recommends improvement measures based on the terminal analysis results;

[0224] The third determination unit, if the abnormal categories of the multiple abnormal positions are different, obtains the influencing parameters of the abnormal health categories of the multiple abnormal positions respectively; and calculates the difference between each parameter and the corresponding mean value in the normal set as the second difference;

[0225] ;

[0226] Where Dik is the second difference of the k-th parameter at the ith abnormal position; Cik is the current value of the k-th parameter at the ith abnormal position; Cnk is the mean value of the k-th parameter in the normal set; ;

[0227] According to the second difference and the influence coefficient of the corresponding parameter; determining a second comprehensive evaluation value of the parameter;

[0228] ;

[0229] A second sorting unit is used to select parameters corresponding to the second comprehensive evaluation values ​​ranked in the top N or parameters greater than a preset evaluation value threshold as preset target parameters for adjustment;

[0230] The priority selection unit is used to give priority to adjusting the same preset target parameters if multiple positions contain the same preset target parameters.

[0231] In some embodiments, the early warning module includes:

[0232] A first warning unit, used to issue a first warning if any parameter at any position is outside a preset threshold range;

[0233] The second warning unit is used to issue a second warning if the same parameter is outside the preset threshold range at multiple locations in the same nursery shed;

[0234] The third warning unit is used to issue a third warning if multiple parameters at multiple locations in the same seedling shed are outside the preset threshold range.

[0235] The working principle and effect of the above technical solution are the same as those in the embodiment of the method of the present application and will not be described in detail here.

[0236] This application is explained from the perspectives of purpose of use, effectiveness, progress and novelty, and has met the functional enhancement and usage requirements emphasized by the Patent Law. The above description and drawings of this application are only the preferred embodiments of this application, and are not intended to limit this application. Therefore, all structures, devices, features, etc. that are similar or identical to this application, that is, all equivalent replacements or modifications made in accordance with the scope of the patent application of this application, should fall within the scope of protection of the patent application of this application.

Claims

1. A tobacco seedling cultivation full-cycle monitoring and management method based on the Internet of Things, characterized in that: The method comprises: Collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices; Uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data; The pre-processed data is uploaded to the central processor, and the health scores of tobacco seedlings at different stages are determined based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the abnormal health type and abnormal location information are output; Automatically adjust the environmental conditions of the tobacco seedling shed and nutrient pool according to the tobacco seedling health score results; Establish an early warning mechanism to warn of abnormal results; The pre-processed data is uploaded to the central processor, and the tobacco seedling health score is determined based on the tobacco seedling evaluation model; including: A first classification is performed according to the types of tobacco leaves; under the first classification, each growth stage is divided by a preset growth cycle; the historical data is first divided according to the time series of each growth stage to obtain a first group; wherein, the start of each sowing preparation is regarded as the beginning of a new time series; Divide the data of the first group into the second group according to each day in the time series; Cluster the data of the second group according to natural light, temperature and humidity to obtain the third group; Counting the characteristics of the tobacco seedlings in each third group and the corresponding environmental characteristics, including the parameter mean, standard deviation and image characteristics of multiple samplings; the image characteristics include the height, color, number of leaf petals and size of leaf petals of the tobacco seedlings; Comparing the similarity of tobacco seedling characteristics of the same third group in a plurality of second groups adjacent to each other in time series; ； Among them, Sp is the total similarity of the two groups of tobacco seedling features, Sj is the similarity of the jth feature; xj is the weight of the jth feature; n is the total number of tobacco seedling features; If the similarity is greater than a preset similarity threshold, the adjacent multiple second groups are taken as a time series window, and the mean, maximum, minimum, and standard deviation values ​​of the tobacco seedling characteristics of the consecutive and identical third groups and the mean, maximum, minimum, and standard deviation values ​​of the corresponding environmental parameters are input as features into the tobacco seedling evaluation model; the environmental parameters include nutrient pool environmental information and seedling shed environmental information; According to the current data and the corresponding time series, based on the tobacco seedling evaluation model, the tobacco seedling health scores at different stages are determined.

2. The method according to claim 1, characterized in that The multi-dimensional data of each stage of the tobacco seedling raising cycle is collected by the Internet of Things sensing device; including: A plurality of positioning base stations are arranged in the seedling raising shed, and the seedling raising tray is positioned by using the positioning base stations and positioning tags; an intelligent system with positioning tags is used to trigger the track camera to locate and capture the tobacco seedling images in the seedling raising tray; The environmental parameters of the nursery shed and each nutrient pool are obtained through sensors; the environmental parameters include the water depth, water temperature, water quality parameters of the nutrient pool and the environmental temperature, humidity, soil parameters, water quality parameters and light conditions of the nursery shed.

3. The method according to claim 2, characterized in that The uploading of the multi-dimensional data to the edge processing device for preprocessing to obtain preprocessed data includes: Clean the received data through edge processing devices to remove duplicate, abnormal or invalid data; Compress the image; The data is associated and integrated through positioning tags, including the tobacco seedling data of the seedling tray, the historical environmental parameters of the nutrient pool where the seedling tray is located, the historical environmental parameters of the seedling shed where the tobacco seedlings are located, and pictures of the growth process of the tobacco seedlings in the seedling tray and related information.

4. The method according to claim 3, characterized in that The methods for determining abnormal data include: Dividing a day into a first time period; dividing a day into a first time period, wherein the first time period includes early morning, daytime, first half of the night, and second half of the night; The plurality of data of the same sensor in the same first time period at the same stage of the seedling raising cycle and the variation of the adjacent data are taken as the first cluster; Taking multiple data of multiple adjacent sensors of the same type at the same time point as the second cluster; counting the difference between the sensors in the second cluster at the same time point as the second difference; taking multiple groups of second difference values ​​at multiple time points as the third cluster; According to the first cluster and the third cluster, an abnormality determination model for data of the same type of sensor in the first time period is established through machine learning; Determine abnormal data and the location of abnormal data through the data anomaly determination model; and send a reminder to the monitoring terminal; According to the feedback from the monitoring terminal, optimize the corresponding data anomaly judgment model.

5. The method according to claim 1, characterized in that The automatic adjustment of the environmental conditions of the tobacco seedling shed and the nutrient pool according to the tobacco seedling health score results includes: If the tobacco seedlings have abnormal health scores, the abnormal category is determined; Based on the correlation analysis between the environmental parameters and the abnormal categories; obtaining the correlation coefficient between the environmental parameters and the abnormal categories; If there are multiple abnormal positions in the same nursery shed and the abnormal categories are the same; then calculate the ratio of the total area of ​​the multiple abnormal positions to the total area of ​​the nursery shed as the first ratio, and obtain the distribution of the multiple abnormal positions; If the first ratio is less than or equal to a preset ratio threshold; or, multiple abnormal locations are concentrated; Then, the current parameters of the multiple first abnormal positions are taken as an abnormal set, and the current parameters of other positions are taken as a normal set; the difference between the current parameters in the abnormal set and the normal set is obtained as the first difference; wherein the first difference is: ； Where, Delk is the first difference of the k-th parameter; Cak is the mean of the k-th parameter in the abnormal set; Cnk is the mean of the k-th parameter in the normal set; ; Obtain a comprehensive parameter evaluation value based on the abnormal health conditions that affect the growth cycle or tobacco seedlings and the first difference; The comprehensive evaluation value of the parameters is: ; Where Wk is the correlation coefficient between the kth category and the abnormal health category; abs() is the absolute value; According to the comprehensive evaluation value of the parameters, the corresponding adjustment parameters and adjustment amounts are determined.

6. The method according to claim 5, characterized in that According to the comprehensive evaluation value of the parameters, the corresponding adjustment parameters and parameter adjustment amounts are determined, including: Select the parameters corresponding to the top N comprehensive evaluation values ​​or the parameters greater than the preset evaluation value threshold as the preset target parameters for adjustment; Use the mean of each parameter in the normal set as the corresponding parameter adjustment expected value; A parameter adjustment model is established through the expected values ​​of each preset target parameter, the mutual influence relationship of each preset target parameter, and the energy consumption; The adjustment amount of each target parameter is obtained through the parameter adjustment model; If the adjustment amount of a certain parameter is less than or equal to the absolute value of the corresponding adjustment threshold; Then the parameter at the position is adjusted directly by the adjustment amount; If the adjustment amount of a certain parameter is greater than the absolute value of the corresponding adjustment threshold; The absolute value of the parameter threshold is used as a single adjustment amount to adjust the parameter at the position.

7. The method according to claim 5, characterized in that The method further comprises: If the first ratio is greater than a preset ratio threshold and the first abnormal positions are dispersedly distributed; Then send an early warning to remind the terminal to conduct analysis, and recommend improvement measures based on the terminal analysis results; If the abnormality categories of multiple abnormal locations are different; Then, the influencing parameters of the abnormal health categories of multiple abnormal positions are obtained respectively; and respectively calculate the difference between each parameter and the corresponding mean in the normal set as the second difference; ； Where Dik is the second difference of the k-th parameter at the ith abnormal position; Cik is the current value of the k-th parameter at the ith abnormal position; Cnk is the mean value of the k-th parameter in the normal set; ; According to the second difference and the influence coefficient of the corresponding parameter; determining a second comprehensive evaluation value of the parameter; ； Select the parameter corresponding to the second comprehensive evaluation value ranked in the top N or the parameter greater than the preset evaluation value threshold as the preset target parameter for adjustment; If multiple locations contain the same preset target parameter, the same preset target parameter is adjusted first.

8. The method according to claim 1, characterized in that The early warning mechanism includes: If any parameter at any location is outside the preset threshold range, a first warning is issued; If the same parameter is outside the preset threshold range at multiple locations in the same nursery shed, a second warning will be issued; If multiple parameters are outside the preset threshold range at multiple locations in the same nursery shed, a third warning will be issued.

9. The tobacco seedling cultivation full cycle monitoring and management system based on the Internet of Things is characterized by: The system comprises: The data collection module is used to collect multi-dimensional data at each stage of the tobacco seedling cycle through IoT sensing devices; A preprocessing module, used for uploading the multi-dimensional data to an edge processing device for preprocessing to obtain preprocessed data; The evaluation module is used to upload the pre-processed data to the central processor and determine the health scores of tobacco seedlings at different stages based on the tobacco seedling evaluation model; if the health score is less than the score threshold of the corresponding stage, the type of abnormal health and abnormal location information are output; A control module is used to automatically adjust the environmental conditions of the tobacco seedling shed and nutrient pool according to the tobacco seedling health score results; The early warning module is used to establish an early warning mechanism and issue early warnings for abnormal results; The pre-processed data is uploaded to the central processor, and the tobacco seedling health score is determined based on the tobacco seedling evaluation model; including: A first classification is performed according to the types of tobacco leaves; under the first classification, each growth stage is divided by a preset growth cycle; the historical data is first divided according to the time series of each growth stage to obtain a first group; wherein, the start of each sowing preparation is regarded as the beginning of a new time series; Divide the data of the first group into the second group according to each day in the time series; Cluster the data of the second group according to natural light, temperature and humidity to obtain the third group; Counting the characteristics of the tobacco seedlings in each third group and the corresponding environmental characteristics, including the parameter mean, standard deviation and image characteristics of multiple samplings; the image characteristics include the height, color, number of leaf petals and size of leaf petals of the tobacco seedlings; Comparing the similarity of tobacco seedling characteristics of the same third group in a plurality of second groups adjacent to each other in time series; ； Among them, Sp is the total similarity of the two groups of tobacco seedling features, Sj is the similarity of the jth feature; xj is the weight of the jth feature; n is the total number of tobacco seedling features; If the similarity is greater than a preset similarity threshold, the adjacent multiple second groups are taken as a time series window, and the mean, maximum, minimum, and standard deviation values ​​of the tobacco seedling characteristics of the consecutive and identical third groups and the mean, maximum, minimum, and standard deviation values ​​of the corresponding environmental parameters are input as features into the tobacco seedling evaluation model; the environmental parameters include nutrient pool environmental information and seedling shed environmental information; According to the current data and the corresponding time series, based on the tobacco seedling evaluation model, the tobacco seedling health scores at different stages are determined.

Images