Intelligent agriculture method and system based on Internet of Things and artificial intelligence

Abstract

The invention provides an intelligent agriculture method and system based on the Internet of Things and artificial intelligence, and the method comprises the steps: collecting and selecting environment parameters and crop growth state data in a target greenhouse in real time, obtaining high-precision environment parameters and crop growth state data, carrying out the scale conversion and integration of the high-precision data, and obtaining a high-precision environment parameter and crop growth state data; the method comprises the steps of obtaining first structured data, structuring historical experimental data into a sensor-like data format to obtain second structured data, analyzing and outputting the confidence coefficient of the first structured data, and performing crop growth prediction on the first structured data. And obtaining a crop growth state prediction value, a crop yield prediction result and a crop growth quality grade, and generating an optimal greenhouse environment regulation and control strategy and a farming management decision suggestion through an intelligent optimization algorithm. Based on an Internet of Things system and artificial intelligence, the accuracy of collected crop related data is ensured, and intelligent greenhouse environment regulation and control are carried out.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart agriculture, and in particular relates to a smart agriculture method and system based on the Internet of Things and artificial intelligence. Background Art

[0002] In the current agricultural crop growth management, there is a growing preference for automatic control and regulation of intelligent systems. The Internet of Things and artificial intelligence are also gradually being used in smart agricultural management systems to monitor the crop growth process. However, the existing solutions are still relatively simple management and monitoring, and there are mainly the following problems:

[0003] (1) Data collection is not comprehensive, there is a lack of systematic monitoring of crop physiological parameters, and the standardization and credibility assessment mechanisms for collected data are not perfect;

[0004] (2) The constructed crop growth prediction model is too simple and the dimensions considered are not comprehensive enough, making it difficult to accurately reflect the crop growth law;

[0005] (3) The environmental regulation of greenhouse crops mainly relies on manual experience and judgment, lacking a more accurate and comprehensive calculation basis, and the efficiency of adjusting to a better greenhouse crop environment is low.

[0006] Therefore, we need to develop a smart agricultural method and system based on the Internet of Things and artificial intelligence that can comprehensively consider factors related to crop growth, accurately predict crop growth data, and perform corresponding intelligent regulation. Summary of the invention

[0007] The purpose of the present invention is to provide a smart agriculture method and system based on the Internet of Things and artificial intelligence to solve the problems mentioned in the above background technology, such as insufficient data collection of existing greenhouse crop growth prediction schemes, poor accuracy of crop growth prediction models, and low efficiency of intelligent control strategies.

[0008] To achieve the above objectives, on the one hand, the present invention provides a smart agriculture method based on the Internet of Things and artificial intelligence, and the method is specifically as follows:

[0009] By deploying environmental sensors and physiological parameter sensors in the target greenhouse, the environmental parameters and crop growth status data in the target greenhouse are collected in real time;

[0010] Selecting the environmental parameters and the crop growth status data, detecting and removing the scattered values ​​in the sensor collected data, and obtaining high-precision environmental parameters and selected high-precision crop growth status data;

[0011] The high-precision environmental parameters and the high-precision crop growth status data are scaled and integrated to obtain first structured data, the historical experimental data is structured into a sensor data-like format to obtain second structured data, and the confidence of the first structured data is analyzed and output, where the confidence value is 0 or 1;

[0012] Recording the first structured data with a confidence level of 1 as credible structured data, performing crop growth prediction on the credible structured data, and obtaining prediction results including crop growth state prediction values, crop yield estimation results, and crop growth quality grades;

[0013] The analyzing and outputting the confidence of the first structured data is to calculate the deviation of the first structured data based on the second structured data;

[0014] Based on the prediction results and expert knowledge, a greenhouse environment control strategy is generated through an intelligent optimization algorithm, and further agricultural management decision recommendations are generated.

[0015] Based on the above solution, the step of calculating the deviation of the first structured data based on the second structured data is specifically:

[0016] The historical data of R monitoring periods are extracted from the second structured data, each monitoring period includes N monitoring points, and the data of each monitoring point includes d monitoring indicators, and a historical data reference matrix H is constructed. The dimension of the matrix H is ;

[0017] Calculate the mean vector of each monitoring indicator in the reference matrix H and the standard deviation vector , the dimensions are , for the first structured data newly collected , calculate its relative to the second structured data The normalized deviation .

[0018] Based on the above scheme, the crop growth prediction for the credible structured data includes:

[0019] Predicting crop growth trends on the credible structured data to obtain predicted values ​​of crop growth status in different time periods in the future;

[0020] Outputting a crop yield estimation result based on the credible structured data and the crop growth status prediction value;

[0021] The quality evaluation index is obtained through the second structured data, and a quality grade threshold range is set. The crop growth quality grade of the target greenhouse is predicted in real time through the quality evaluation index data in the trusted structured data.

[0022] Based on the above scheme, the crop growth prediction of the credible structured data includes constructing a crop growth trend prediction model to predict the crop growth status prediction value:

[0023] The input of the crop growth trend prediction model is the credible structured data, including plant height, stem diameter, leaf area index, chlorophyll content, light, temperature and humidity, and carbon dioxide concentration, and the output is the predicted values ​​of plant height, stem diameter, leaf area index, and chlorophyll content in the next week to one month;

[0024] Constructing samples for the time series data in the credible structured data by a data optimization method, wherein the data optimization method includes a sliding window method and a difference array method; each sample includes the credible structured data of the previous n days, i.e., model input, and a predicted value of crop growth status on the n+1th day, i.e., model output;

[0025] The mapping relationship from historical data to predicted values ​​is: [ x (t - n+1) , x (t - n+2) , ..., x (t) ] ≥ y (t+1) ,in arrive represent A sequence of historical observation data at consecutive time points, is the target value at the next time point to be predicted, and n is the time step, which is set according to the prediction period.

[0026] Based on the above scheme, the crop growth prediction of the credible structured data further includes constructing a crop yield prediction model to predict the crop yield estimation result:

[0027] On the basis of the crop growth trend prediction model, yield component factors are added as inputs of the crop yield prediction model, wherein the yield component factors include fruit size, single fruit weight, and fruit number, and the output is the final single plant yield and unit area yield prediction values;

[0028] Specifically, the crop yield prediction model adopts a classical growth model, including a piecewise linear function, a logistic curve, a Richards equation, or a machine learning model, and the model parameter solution is aimed at minimizing the prediction error, and a gradient descent and least squares optimization algorithm are used for iterative training;

[0029] The trained crop yield prediction model is used in combination with the growth status prediction value to predict the crop yield per unit area in the next m days.

[0030] Based on the above scheme, the crop growth prediction of the credible structured data further includes constructing a crop growth quality prediction model to predict the crop growth quality grade and the quality grade threshold range:

[0031] Based on the second structured data, the correlation between product quality grade and environmental parameters and physiological indicators is statistically analyzed, leaf area index, fruit size, temperature and light intensity are extracted as quality evaluation indicators, a decision tree algorithm is used to construct a crop growth quality prediction model, and the quality grade threshold range is set; the credible structured data is divided into three quality grades: A, B, and C, among which grade A: leaf area index>6.5 and fruit size>220g; grade B: 3.5<leaf area index≤6.5 and 120g<fruit size≤220g; grade C: others.

[0032] Based on the above scheme, the greenhouse environment control strategy is the setting of environment control parameters, including the setting of temperature, humidity, light, and carbon dioxide concentration. The generation of the greenhouse environment control strategy specifically includes:

[0033] Taking the prediction results of the crop growth stage in period Z as the optimization target, sample data with leaf area index, plant height, yield and quality rating in the top 20% and grade A in the growth stage in period Z are selected from the prediction results, and the corresponding environmental parameters are extracted as reference values. Based on the crop growth trend prediction model and the influence of the environmental parameters on yield and quality, the optimal environmental parameter combination is explored through a genetic algorithm.

[0034] Based on the above scheme, the further generation of agricultural management decision suggestions is based on the crop growth state prediction value, the environmental control parameter setting and the expert knowledge base to obtain the agricultural management decision suggestions;

[0035] The agricultural management decision suggestions include agricultural management measures such as irrigation time and amount, fertilization time and amount, and pest control measures. A rule base is constructed for the agricultural management measures for management decision-making;

[0036] The irrigation time and amount are set according to the real-time soil moisture data fed back by the environmental sensor and the water demand estimated by the crop water consumption model, and the irrigation rules are as follows: when the soil moisture is less than the lower limit threshold or the estimated water demand is greater than the predetermined value, and the current time is the allowed irrigation time, the irrigation operation is performed, otherwise no irrigation is performed.

[0037] On the other hand, the present invention also provides a smart agricultural system based on the Internet of Things and artificial intelligence, comprising:

[0038] Perception layer: used to collect environmental parameters and crop growth status data in the target greenhouse in real time through environmental sensors and physiological parameter sensors deployed in the target greenhouse;

[0039] Control layer: As the core of the system, it is responsible for analyzing and processing the collected environmental parameters and crop growth status data, obtaining greenhouse environment control strategies and agricultural management decision-making suggestions, and controlling the application layer for specific implementation, including:

[0040] Data processing module: selects the environmental parameters and the crop growth status data, detects and removes the scattered values ​​in the sensor collected data, and obtains high-precision environmental parameters and selected high-precision crop growth status data;

[0041] Data integration module: scale conversion and integration of the high-precision environmental parameters and the high-precision crop growth status data to obtain first structured data, structure the historical experimental data into a sensor data-like format to obtain second structured data, analyze and output the confidence of the first structured data, where the confidence value is 0 or 1;

[0042] The analyzing and outputting the confidence of the first structured data is to calculate the deviation of the first structured data based on the second structured data;

[0043] Crop growth prediction module: record the first structured data with a confidence level of 1 as credible structured data, perform crop growth prediction on the credible structured data, and obtain prediction results including crop growth state prediction value, crop yield estimation result and crop growth quality grade;

[0044] Environmental control decision module: Based on the prediction results and expert knowledge, the greenhouse environmental control strategy is generated through intelligent optimization algorithm, and further agricultural management decision suggestions are generated;

[0045] Network layer: responsible for data transmission and communication between the perception layer and the control layer, and ensuring the reliability and real-time performance of the system;

[0046] Application layer: receives the implementation instructions of the control layer, executes the greenhouse environment control strategy and the agricultural management decision suggestions, and provides relevant information to the user operation and interaction interface.

[0047] Compared with the prior art, the present invention has the following advantages and effects:

[0048] (1) A complete multi-parameter monitoring system was established to coordinate the collection of environmental parameters and physiological parameters of target greenhouse crops, and confidence assessment of the collected environmental parameters and physiological parameters was performed based on historical data to ensure the multi-source integrity and accuracy of the data;

[0049] (2) Construct crop growth trend prediction models, crop yield prediction models, and crop growth quality prediction models, process the collected environmental parameters and physiological parameters, and predict crop growth from multiple angles, effectively improving the accuracy of the future growth of target greenhouse crops and providing a good foundation for intelligent environmental control;

[0050] (3) Based on the crop growth status prediction value, crop yield estimation result, crop growth quality grade and other data predicted by the crop growth prediction model and expert knowledge, the environmental conditions of the target greenhouse crops are intelligently regulated in a highly automated and efficient manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0052] Figure 1 This is a flow chart of a smart agriculture method based on the Internet of Things and artificial intelligence provided by an embodiment of the present invention;

[0053] Figure 2 This is a structural diagram of a smart agriculture system based on the Internet of Things and artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to more clearly explain the purpose, technical solutions and advantages of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The example implementation methods can be implemented in various forms and should not be understood as being limited to the examples described herein. On the contrary, these implementation methods are provided to make the present invention more comprehensive and complete, and to fully convey the concepts of the example implementation methods to those skilled in the art.

[0055] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present invention. However, those skilled in the art will appreciate that the technical solution of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present invention.

[0056] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0057] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0058] The present invention is described in detail below in conjunction with specific embodiments:

[0059] Example 1

[0060] As attached Figure 1 As shown, Embodiment 1 of the present invention provides a smart agriculture method based on the Internet of Things and artificial intelligence, and the specific steps of the method are as follows:

[0061] Step S1, collecting environmental parameters and crop growth status data in the target greenhouse in real time through environmental sensors and physiological parameter sensors deployed in the target greenhouse;

[0062] Step S2, selecting the environmental parameters and the crop growth status data, detecting and removing the scattered values ​​in the sensor collected data, and obtaining high-precision environmental parameters and selected high-precision crop growth status data;

[0063] Step S3, scaling and integrating the high-precision environmental parameters and the high-precision crop growth status data to obtain first structured data, structuring the historical experimental data into a sensor data-like format to obtain second structured data, analyzing and outputting the confidence of the first structured data, where the confidence value is 0 or 1;

[0064] Step S4: record the first structured data with a confidence level of 1 as credible structured data, perform crop growth prediction on the credible structured data, and obtain prediction results including crop growth state prediction values, crop yield estimation results, and crop growth quality grades;

[0065] Step S5: Based on the prediction results and expert knowledge, the greenhouse environment control strategy is generated through the optimal intelligent optimization algorithm, including the setting of environmental control parameters such as temperature, humidity, light, CO2 concentration, etc., and further generates agricultural management decision suggestions including irrigation time and amount, fertilization time and amount, and pest and disease control measures.

[0066] Preferably, step S1 specifically includes:

[0067] The environmental sensor and the physiological parameter sensor collect the environmental parameters of the target greenhouse and the crop growth status data in real time according to a set frequency;

[0068] According to different crop types and different growth stages, several physiological indicators that have the greatest impact on crop yield and quality are screened from the crop growth status data. In this embodiment, the physiological indicators include leaf area index, chlorophyll content, plant height, stem diameter, and fruit size;

[0069] For example, the physiological indexes are screened by taking tomatoes as an example, wherein the leaf area index reflects the photosynthetic capacity of the tomato population, is closely related to the yield, and is selected in the range of 3-4; the chlorophyll content reflects the nitrogen nutritional status of tomato leaves, is positively correlated with the photosynthetic rate, and is selected in the range of 50-60; the plant height reflects the growth dominance of tomatoes, is related to the distribution of microclimates in the inner and outer layers of the plant, and is selected in the range of 1.5-2m; the fruit size reflects the weight of a single tomato fruit, is directly related to the yield quality, and generally the diameter of a mature tomato is 50-70mm, so it is selected in the range of 50-70mm;

[0070] For example, considering factors such as the scale of tomato planting and differences in microclimate distribution, no less than 5 monitoring points are set up per mu on average. Environmental sensors and physiological parameter sensors are deployed at the monitoring points. The environmental sensors are installed at a height of 1.5m from the ground to avoid direct sunlight. Data is collected every 5 minutes. The monitoring indicators include temperature, air humidity, light intensity, and carbon dioxide concentration. The monitoring of crop physiological parameters adopts a combination of automatic sensing and manual measurement. The parameters that can be automatically collected collect data every 30 minutes, and the indicators that need to be manually measured are set with different monitoring frequencies according to their nature and change characteristics. The leaf area index is measured once a week, the chlorophyll content is measured once every 2 weeks, and the plant height, stem thickness and fruit size are measured once a week.

[0071] Preferably, in step S2, the detecting and removing of the stray values ​​in the sensor collected data is specifically:

[0072] Data cleaning: By setting physical thresholds, outliers beyond the measurement range are eliminated, and random error data are detected and removed based on statistics such as the median and interquartile range;

[0073] Data verification: According to the growth law of crops, judge the rationality of data, use data mining methods to analyze the correlation between different monitoring indicators, and identify erroneous data;

[0074] Data assimilation: refer to the changes in environmental parameters to assimilate crop growth monitoring data in time and space to improve data timeliness and comparability;

[0075] Data compression: Use lossless or lossy compression algorithms to reduce data storage and transmission costs while ensuring information integrity.

[0076] Specifically, the environmental parameters and the crop growth status data are selected to obtain complete, effective and high-precision environmental parameters, including temperature, humidity, light intensity, carbon dioxide concentration, and high-precision crop growth status data, including leaf area index, chlorophyll content, plant height and fruit size.

[0077] It should be noted that due to the complex greenhouse environment, it is inevitable that scattered values ​​will appear in the monitoring data of various sensors. In order to ensure the data quality, the collected raw data needs to be cleaned and verified.

[0078] Preferably, in step S3, obtaining the first structure data is specifically:

[0079] Performing scale conversion by using the maximum and minimum normalization method, unifying the scales of the high-precision environmental parameters and the high-precision crop growth status data, mapping them to the [0, 1] interval, and obtaining normalized parameters;

[0080] Integrate the normalized parameters and timestamps into the first structured data, where each data record contains the normalized values ​​and corresponding times of all monitoring indicators, facilitating subsequent data management and analysis;

[0081] It should be noted that, since the dimensions and numerical ranges of the monitoring indicator data vary greatly, in order to facilitate comprehensive analysis, dimensionless processing is required, that is, the high-precision environmental parameters and the high-precision crop growth status data are integrated and converted; historical experimental data related to the target greenhouse, including data obtained from various previous surveys, monitoring or research on the target greenhouse, are collected and converted into a structure similar to the sensor data, and the historical sensor experimental data of the target greenhouse is determined to be the second structured data.

[0082] Preferably, in step S3, the analyzing and outputting the confidence of the first structured data is to calculate the deviation of the first structured data based on the second structured data, specifically:

[0083] The historical data of R monitoring periods are extracted from the second structured data, each monitoring period includes N monitoring points, and the data of each monitoring point includes d monitoring indicators, and a historical data reference matrix H is constructed. The dimension of the matrix H is ;

[0084] Calculate the mean vector of each monitoring indicator in the reference matrix H and the standard deviation vector , the dimensions are For the first structured data newly collected , calculate its relative to the second structured data The normalized deviation :

[0085]

[0086] in, For the The weight coefficient of each monitoring indicator is For the first structured data Middle The value of the monitoring indicator, and The second structured data Middle The mean and standard deviation of the monitoring indicators; weight coefficient The calculation is based on the partial correlation coefficients of each indicator with crop yield and quality ,as follows:

[0087]

[0088] The calculated deviation With preset threshold Compare, when When the confidence level of the first structured data is 1, otherwise it is marked as 0; wherein the preset threshold Dynamically adjusted with the crop growth stage t, the calculation formula is as follows:

[0089]

[0090] in is the baseline threshold, is the fluctuation coefficient, For the complete growth cycle of crops, this dynamic threshold mechanism takes into account the differences in data quality requirements at different growth stages.

[0091] It should be noted that in order to ensure data quality, it is necessary to evaluate the confidence of the first structured data and the second structured data. In this embodiment, the Euclidean distance between each data and the mean of the historical samples is calculated to determine whether it is a suspicious outlier; a batch of the second structured data is randomly selected from the historical data set, and the mean vector of all monitoring indicators is calculated. As a reference sample;

[0092] Exemplarily, the preset deviation distance is set , in this embodiment, preferably 3; if , then the data is judged as a suspicious outlier and the confidence level is marked as 0; otherwise, the confidence level is marked as 1.

[0093] Preferably, in step S4, performing crop growth prediction on the credible structured data includes:

[0094] Predicting crop growth trends based on the credible structured data to obtain predicted values ​​of crop growth status at different time periods in the future, including leaf area index and plant height;

[0095] Outputting a crop yield estimation result based on the credible structured data and the crop growth status prediction value;

[0096] The quality evaluation index is obtained through the second structured data, and a quality grade threshold range is set. The crop growth quality grade of the target greenhouse is predicted in real time through the quality evaluation index data in the trusted structured data.

[0097] Specifically, a crop growth trend prediction model, a crop yield prediction model and a crop growth quality prediction model are constructed based on the trusted structured data. The following describes in detail the model input and output, feature selection, sample collection, training and verification processes.

[0098] For example, taking tomatoes as an example, data from 30 consecutive days of planting process are selected for modeling analysis:

[0099] Step S401: The input of the crop growth trend prediction model is the credible structured data, including plant height, stem diameter, leaf area index, chlorophyll content, light, temperature and humidity, and carbon dioxide concentration, and the output is the predicted values ​​of plant height, stem diameter, leaf area index, and chlorophyll content in the next week to one month;

[0100] A sample is constructed for the time series data in the credible structured data by a data optimization method, wherein the data optimization method includes a sliding window method and a difference array method; each sample includes the credible structured data of the previous n days, i.e., the model input, and the predicted value of the crop growth status on the n+1th day, i.e., the model output. [ x (t - n+1) , x (t - n+2) , ..., x (t) ] ≥ y (t+1) , which describes the mapping relationship from historical data to predicted values, where arrive represent A sequence of historical observation data at consecutive time points, is the target value at the next moment to be predicted, n is the time step, which is set according to the prediction period. In this embodiment, it is preferred that: n=7 is selected for short-term prediction, and n=30 is selected for long-term prediction;

[0101] The sample data set is divided into a training set, a validation set and a test set, with the division ratios of 70%, 20% and 10% respectively. The training set is used to train the crop growth trend prediction model, the validation set is used to adjust the parameters, and the test set is used to evaluate the performance of the crop growth trend prediction model. The main parameter settings of the crop growth trend prediction model are preferably: the number of hidden layer nodes is 64, the learning rate is 0.01, the batch size is 32, and the number of training iterations is 1000; after the crop growth trend prediction model is trained, the predicted value of the crop growth status in the next m days can be predicted, including the change trend of the crop leaf area index, chlorophyll content and plant height.

[0102] Step S402: Based on the crop growth trend prediction model, yield component factors are added as inputs of the crop yield prediction model, wherein the parameter component factors include fruit size, single fruit weight, and fruit number, and the output is the final single plant yield and unit area yield prediction values;

[0103] It should be noted that in order to improve the prediction accuracy of the model, it is necessary to screen features with strong correlation with the target value and large amount of information. In this embodiment, the Pearson correlation coefficient between each monitoring indicator and the yield is calculated through dimensionality reduction algorithms such as correlation analysis and principal component analysis. The monitoring indicators with a coefficient greater than 0.6 are taken as the candidate feature set, and the candidate feature set is further optimized through algorithms including stepwise regression and Lasso to balance prediction accuracy and computational complexity.

[0104] For example, taking tomatoes as an example, by analyzing the second structured data, the selection results of the candidate feature set for predicting the yield per unit area are: plant height 40 days, 60 days, and 80 days after planting, leaf area index 50 days and 70 days after planting, chlorophyll content 30 days and 50 days after planting, average sunshine hours, average temperature, cumulative nitrogen application, etc.

[0105] It should be noted that in order to ensure that the model has good generalization performance, sample collection should be random and representative. The modeling sample and the validation sample should be divided into 8:2 ratios. The sample size should be no less than 10 times the number of features. The samples should be preprocessed, including missing value interpolation, outlier removal, and data normalization. Missing values ​​are filled using multiple interpolation and KNN algorithms, outliers are identified and removed using boxplots, and data normalization uses Min-Max and Z-score dimensionless processing.

[0106] Specifically, the crop yield prediction model adopts a classical growth model, including a piecewise linear function, a logistic curve, a Richards equation, or a machine learning model, including a neural network, a support vector machine, and the model parameter solution is aimed at minimizing the prediction error, and a gradient descent and least squares optimization algorithm are used for iterative training;

[0107] Specifically, the trained crop yield prediction model is used in combination with the growth status prediction value to predict the crop yield per unit area in the next m days.

[0108] Step S403: Based on the second structured data, the correlation between product quality grade and environmental conditions and physiological indicators is statistically analyzed, leaf area index, fruit size, temperature and light intensity are extracted as quality evaluation indicators, a decision tree algorithm is used to build a crop growth quality prediction model, and a quality grade threshold range is set; the credible structured data training set is divided into three quality grades: A, B, and C, where grade A: leaf area index>6.5 and fruit size>220g; grade B: 3.5<leaf area index≤6.5 and 120g<fruit size≤220g; grade C: others.

[0109] Specifically, a decision tree is generated by the CART algorithm, the feature with the smallest Gini index is selected by splitting the node, and the tree structure is optimized by pruning. The quality evaluation index in the credible structured data of the target greenhouse is input into the decision tree model to predict the crop growth quality grade under the current environmental conditions in real time.

[0110] Preferably, in step S5, the greenhouse environment control strategy is the setting of environment control parameters, including the setting of temperature, humidity, light, and carbon dioxide concentration. The generation of the greenhouse environment control strategy is specifically as follows:

[0111] Taking the prediction results of the crop growth stage in period Z as the optimization target, sample data with leaf area index, plant height, yield and quality rating in the top 20% and grade A in the growth stage in period Z are selected from the prediction results, and the corresponding environmental parameters such as temperature, humidity, light, CO2 concentration, etc. are extracted as reference values. Based on the crop growth trend prediction model and the influence of the environmental parameters on yield and quality, the optimal environmental parameter combination is explored through genetic algorithm, as follows:

[0112] The environmental parameters are used as the initial population of the genetic algorithm, and the individuals are encoded as environmental parameter vectors. The parameter combination is iteratively optimized through genetic operations such as selection, crossover, and mutation. The fitness function of the genetic algorithm is designed as a weighted combination of the crop yield estimation result and the crop growth quality grade. After optimization, the parameter value corresponding to the optimal individual searched by the genetic algorithm is used as the environmental control parameter setting for the Z period growth stage and is sent to the hardware control module for execution.

[0113] Preferably, in step S5, the further generating of the agricultural management decision suggestion is based on the crop growth state prediction value, the environmental control parameter setting and the expert knowledge base to obtain the agricultural management decision suggestion;

[0114] Specifically, the agricultural management decision recommendations include agricultural management measures such as irrigation time and amount, fertilization time and amount, and pest control measures. A rule base is constructed for the agricultural management measures for management decisions. Taking the agricultural management of irrigation time and amount as an example, based on the real-time soil moisture data fed back by the environmental sensor and the water demand estimated by the crop water consumption model, the irrigation rules are set as follows: when the soil moisture is less than the lower limit threshold or the estimated water demand is greater than the predetermined value, and the current time is the allowed irrigation time, the irrigation operation is performed, otherwise no irrigation is performed.

[0115] Specifically, the lower limit threshold is set according to the crop type and growth stage, and is preferably 10% in the present embodiment, and the allowed irrigation time is preferably 8:00-10:00; similarly, fertilization decision rules are constructed based on soil nutrient content feedback and fertilization models, and pest control rules are constructed based on pest monitoring information and a control knowledge base to form a comprehensive decision-making plan for agricultural management, i.e., the agricultural management decision recommendation.

[0116] Preferably, based on the environmental control parameter settings, the agricultural management measures, the crop growth state prediction value, and the yield estimation result, the decision effect of the agricultural management decision suggestion is simulated and evaluated, specifically:

[0117] Based on the crop yield prediction model and the environmental control parameter settings, the optimized greenhouse environmental control strategy and the agricultural management measures are simulated and executed to estimate the crop yield estimation results and the crop growth quality level. If the decision effect reaches the preset effect, the agricultural management measures are directly sent to the hardware control module for execution. Otherwise, the agricultural management decision suggestions are adjusted and the simulation is iterated again until the decision effect reaches the preset effect.

[0118] In this embodiment, the environmental parameters and crop growth status data in the target greenhouse are collected in real time and selected to obtain high-precision environmental parameters and crop growth status data, the high-precision data is scaled and integrated to obtain the first structured data, the historical experimental data is structured into a sensor data-like format to obtain the second structured data, the confidence of the first structured data is analyzed and output, the first structured data is predicted for crop growth, the crop growth status prediction value, the crop yield estimation result and the crop growth quality grade are obtained, and the optimal greenhouse environment control strategy and agricultural management decision-making suggestions are generated through an intelligent optimization algorithm. The present invention is based on an Internet of Things system and artificial intelligence, and through historical related data, the accuracy of the collected crop-related data is ensured, the growth trend, growth yield and growth quality of crops in different periods can be predicted, and greenhouse intelligent environment control can be performed to achieve highly automated greenhouse environment control, providing a certain reference for the implementation of smart agriculture.

[0119] Example 2

[0120] As attached Figure 2 As shown, Embodiment 2 of the present invention provides a smart agricultural system S200 based on the Internet of Things and artificial intelligence, the system comprising:

[0121] The perception layer S201 is used to collect environmental parameters and crop growth status data in the target greenhouse in real time through environmental sensors and physiological parameter sensors deployed in the target greenhouse;

[0122] The network layer S202 is responsible for data transmission and communication between the perception layer and the control layer, and ensures the reliability and real-time performance of the system;

[0123] The control layer S203, as the core of the system, is responsible for analyzing and processing the collected environmental parameters and crop growth status data, obtaining greenhouse environment control strategies and agricultural management decision-making suggestions and intelligent control strategies, and controlling the application layer to implement them;

[0124] The application layer S204 receives the implementation instructions from the control layer, executes the greenhouse environment control strategy and the agricultural management decision suggestions, and provides relevant information to the user operation and interaction interface.

[0125] Preferably, the perception layer S201 includes:

[0126] Environmental sensors are used to monitor environmental parameters in the greenhouse in real time, such as temperature, humidity, light intensity, carbon dioxide concentration, soil moisture, air quality, etc.

[0127] Physiological parameter sensors are used to monitor the growth status of crops, such as leaf temperature, photosynthesis efficiency, crop height, leaf area, etc.

[0128] The data acquisition module is responsible for obtaining raw data from the sensor and performing preliminary filtering on the data through certain preprocessing to remove noise and ensure the accuracy of the data.

[0129] For example, the system deploys multiple types of sensors to comprehensively monitor the greenhouse environment, soil conditions, crop growth, etc. Among them, temperature and humidity sensors, light intensity sensors, carbon dioxide concentration sensors, etc. are used to collect environmental parameters, soil moisture sensors, nutrient sensors, etc. are used to monitor soil conditions, and crop growth parameters are obtained through special equipment such as leaf area index meters and chlorophyll meters. The layout of various sensors fully considers the temporal and spatial representativeness of sampling and the convenience of farmland operations, and at the same time, the sensor accuracy calibration and stability maintenance are done well.

[0130] Preferably, the network layer S202 includes:

[0131] The data transmission network uses wired or wireless methods, including Wi-Fi, LoRa, and Zigbee, to transmit data, ensuring that data can be transmitted from the sensor to the cloud or local data storage in real time and stably.

[0132] Data transmission protocol: Use reliable communication protocols, such as MQTT or HTTP, to ensure secure and accurate data transmission;

[0133] For example, a variety of communication technologies are used to build an agricultural sensor network to achieve real-time feedback of monitoring data and reliable issuance of control instructions; in view of the openness and complexity of agricultural scenarios, the system integrates short-range wireless communication modules such as ZigBee, LoRa, and NB-IoT, and wide-area communication modules such as 4G / 5G, to form a network topology with multi-network integration and multi-level interconnection; at the same time, the system deploys security authentication, data encryption and other mechanisms at the network layer to prevent network intrusion and data tampering risks, and provide a reliable data channel for agricultural production control.

[0134] Preferably, the control layer S203 specifically includes:

[0135] A data processing module selects the environmental parameters and the crop growth status data, detects and removes the scattered values ​​in the sensor collected data, and obtains high-precision environmental parameters and selected high-precision crop growth status data;

[0136] A data integration module performs scale conversion and integration on the high-precision environmental parameters and the high-precision crop growth status data to form structured data for further analysis and processing;

[0137] Specifically, the high-precision environmental parameters and the high-precision crop growth status data are scaled and integrated to obtain first structured data, the historical experimental data is structured into a sensor data-like format to obtain second structured data, and the confidence of the first structured data is analyzed and output, where the value of the confidence is 0 or 1;

[0138] A crop growth prediction module, based on the structured data, uses machine learning algorithms such as regression models, deep learning models, etc. to predict crop growth status, crop yield, quality grade, etc.;

[0139] Specifically, the first structured data with a confidence level of 1 is recorded as credible structured data, and crop growth prediction is performed on the credible structured data to obtain prediction results including crop growth state prediction values, crop yield estimation results, and crop growth quality grades;

[0140] The environmental control decision module generates greenhouse environmental control strategies based on crop growth prediction results and expert knowledge through intelligent optimization algorithms such as genetic algorithms and particle swarm algorithms, and automatically adjusts the environmental parameters in the greenhouse;

[0141] For example, the IoT control layer is responsible for parsing and executing agricultural production management decisions, accurately controlling environmental conditioning equipment, and supervising agricultural operation processes. For greenhouses, the system digitally transforms ventilation, shading, irrigation, fertilization and other equipment, automatically adjusts the equipment operation status according to environmental optimization parameters, and ensures that crop growth conditions meet the model presets. For agricultural activities, the system formulates standard operating procedures for key links such as planting, fertilization, and harvesting, uses IoT technologies such as video surveillance and RFID to digitally manage the operation process, and provides real-time guidance to operators through the human-computer interaction interface. The execution status of each control command is transmitted back through the intelligent gateway, which facilitates the system to evaluate the execution effect of the decision.

[0142] Preferably, the application layer S204 specifically includes:

[0143] The user interface module provides a friendly operation interface for farm managers, who can view the environmental data in the greenhouse, crop growth prediction results and regulation strategies in real time, and make manual interventions and adjustments;

[0144] The decision support module generates agricultural management decision suggestions based on greenhouse environmental data and crop growth prediction results, such as specific operation instructions such as fertilization, irrigation, and light adjustment;

[0145] Intelligent reporting module, the system regularly generates reports to show the predicted results of crop growth status, the effects of environmental control strategies and the implementation of agricultural management suggestions, helping farm managers make scientific decisions;

[0146] Exemplarily, the application layer service of the Internet of Things control system provides information support for agricultural managers through visualization, data analysis and other technologies. In this embodiment, the system designs a visual management interface based on the Web and mobile terminals to centrally display information such as environmental status, equipment operation, and agricultural management, so that farm managers can timely grasp the dynamics of agricultural production. The system also provides a series of data analysis services, including sensor monitoring data mining, equipment control log analysis, and agricultural process execution efficiency evaluation, to provide auxiliary decision-making support for improving the quality and efficiency of agricultural production. The application layer also connects to agricultural product traceability services to provide consumers with full-chain supervision information from field to table, and enhance the credibility of agricultural product quality and safety.

[0147] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present invention, which follows the general principles of the present invention and includes common knowledge or customary techniques in the art that are not disclosed by the present invention. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present invention are indicated by the claims. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A smart agricultural method based on the Internet of Things and artificial intelligence, characterized in that: include: By deploying environmental sensors and physiological parameter sensors in the target greenhouse, the environmental parameters and crop growth status data in the target greenhouse are collected in real time; Selecting the environmental parameters and the crop growth status data, detecting and removing the scattered values ​​in the sensor collected data, and obtaining high-precision environmental parameters and selected high-precision crop growth status data; The high-precision environmental parameters and the high-precision crop growth status data are scaled and integrated to obtain first structured data, the historical experimental data is structured into a sensor data-like format to obtain second structured data, and the confidence of the first structured data is analyzed and output, where the confidence value is 0 or 1; Recording the first structured data with a confidence level of 1 as credible structured data, performing crop growth prediction on the credible structured data, and obtaining prediction results including crop growth state prediction values, crop yield estimation results, and crop growth quality grades; The analyzing and outputting the confidence of the first structured data is to calculate the deviation of the first structured data based on the second structured data; Based on the prediction results and expert knowledge, a greenhouse environment control strategy is generated through an intelligent optimization algorithm, and further agricultural management decision recommendations are generated.

2. According to claim 1, a smart agricultural method based on the Internet of Things and artificial intelligence is characterized in that: The calculating the deviation of the first structured data based on the second structured data is specifically: The historical data of R monitoring periods are extracted from the second structured data, each monitoring period includes N monitoring points, and the data of each monitoring point includes d monitoring indicators, and a historical data reference matrix H is constructed. The dimension of the matrix H is ; Calculate the mean vector of each monitoring indicator in the reference matrix H and the standard deviation vector , the dimensions are , for the first structured data newly collected , calculate its relative to the second structured data The normalized deviation .

3. According to claim 1, a smart agricultural method based on the Internet of Things and artificial intelligence is characterized in that: The performing crop growth prediction on the credible structured data comprises: Predicting crop growth trends on the credible structured data to obtain predicted values ​​of crop growth status in different time periods in the future; Outputting a crop yield estimation result based on the credible structured data and the crop growth status prediction value; The quality evaluation index is obtained through the second structured data, and a quality grade threshold range is set. The crop growth quality grade of the target greenhouse is predicted in real time through the quality evaluation index data in the trusted structured data.

4. According to claim 3, a smart agricultural method based on the Internet of Things and artificial intelligence is characterized in that: The performing crop growth prediction on the credible structured data includes constructing a crop growth trend prediction model to predict the crop growth status prediction value: The input of the crop growth trend prediction model is the credible structured data, including plant height, stem diameter, leaf area index, chlorophyll content, light, temperature and humidity, and carbon dioxide concentration, and the output is the predicted values ​​of plant height, stem diameter, leaf area index, and chlorophyll content in the next week to one month; Constructing samples for the time series data in the credible structured data by a data optimization method, wherein the data optimization method includes a sliding window method and a difference array method; each sample includes the credible structured data of the previous n days, i.e., model input, and a predicted value of crop growth status on the n+1th day, i.e., model output; The mapping relationship from historical data to predicted values ​​is: ,in arrive represent A sequence of historical observation data at consecutive time points, is the target value at the next time point to be predicted, and n is the time step, which is set according to the prediction period.

5. A smart agricultural method based on the Internet of Things and artificial intelligence according to claim 4, characterized in that: The performing crop growth prediction on the credible structured data further includes constructing a crop yield prediction model to predict the crop yield estimation result: On the basis of the crop growth trend prediction model, yield component factors are added as inputs of the crop yield prediction model, wherein the yield component factors include fruit size, single fruit weight, and fruit number, and the output is the final single plant yield and unit area yield prediction values; Specifically, the crop yield prediction model adopts a classical growth model, including a piecewise linear function, a logistic curve, a Richards equation, or a machine learning model, and the model parameter solution is aimed at minimizing the prediction error, and gradient descent and least squares optimization algorithms are used for iterative training; The trained crop yield prediction model is used in combination with the growth status prediction value to predict the crop yield per unit area in the next m days.

6. The smart agriculture method based on the Internet of Things and artificial intelligence according to claim 4, characterized in that: The performing crop growth prediction on the credible structured data further includes constructing a crop growth quality prediction model to predict the crop growth quality grade and the quality grade threshold range: Based on the second structured data, the correlation between product quality grade and environmental parameters and physiological indicators is statistically analyzed, leaf area index, fruit size, temperature and light intensity are extracted as quality evaluation indicators, a decision tree algorithm is used to construct a crop growth quality prediction model, and a quality grade threshold range is set; the credible structured data is divided into three quality grades: A, B, and C.

7. The smart agriculture method based on the Internet of Things and artificial intelligence according to claim 1, characterized in that: The greenhouse environment control strategy is the setting of environment control parameters, including the setting of temperature, humidity, light, and carbon dioxide concentration. The generation of the greenhouse environment control strategy specifically includes: Taking the prediction results of the crop growth stage in period Z as the optimization target, sample data with leaf area index, plant height, yield and quality rating in the top 20% and grade A in the growth stage in period Z are selected from the prediction results, and the corresponding environmental parameters are extracted as reference values. Based on the crop growth trend prediction model and the influence of the environmental parameters on yield and quality, the optimal environmental parameter combination is explored through a genetic algorithm.

8. The smart agriculture method based on the Internet of Things and artificial intelligence according to claim 7, characterized in that: The further generating of the agricultural management decision suggestion is based on the crop growth state prediction value, the environmental control parameter setting and the expert knowledge base to obtain the agricultural management decision suggestion; The agricultural management decision suggestion includes agricultural management measures, which include irrigation time and amount, fertilization time and amount, and pest control measures. A rule base is constructed for the agricultural management measures for management decision-making; The irrigation time and amount are set according to the real-time soil moisture data fed back by the environmental sensor and the water demand estimated by the crop water consumption model, and the irrigation rules are as follows: when the soil moisture is less than the lower limit threshold or the estimated water demand is greater than the predetermined value, and the current time is the allowed irrigation time, the irrigation operation is performed, otherwise no irrigation is performed.

9. A smart agricultural system based on the Internet of Things and artificial intelligence, characterized in that: include: Perception layer: used to collect environmental parameters and crop growth status data in the target greenhouse in real time through environmental sensors and physiological parameter sensors deployed in the target greenhouse; Control layer: As the core of the system, it is responsible for analyzing and processing the collected environmental parameters and crop growth status data, obtaining greenhouse environment control strategies and agricultural management decision-making suggestions, and controlling the application layer for specific implementation, including: Data processing module: selects the environmental parameters and the crop growth status data, detects and removes the scattered values ​​in the sensor collected data, and obtains high-precision environmental parameters and selected high-precision crop growth status data; Data integration module: scale conversion and integration of the high-precision environmental parameters and the high-precision crop growth status data to obtain first structured data, structure the historical experimental data into a sensor data-like format to obtain second structured data, analyze and output the confidence of the first structured data, where the confidence value is 0 or 1; The analyzing and outputting the confidence of the first structured data is to calculate the deviation of the first structured data based on the second structured data; Crop growth prediction module: record the first structured data with a confidence level of 1 as credible structured data, perform crop growth prediction on the credible structured data, and obtain prediction results including crop growth state prediction value, crop yield estimation result and crop growth quality grade; Environmental control decision module: Based on the prediction results and expert knowledge, the greenhouse environmental control strategy is generated through intelligent optimization algorithm, and further agricultural management decision suggestions are generated; Network layer: responsible for data transmission and communication between the perception layer and the control layer, and ensuring the reliability and real-time performance of the system; Application layer: receives the implementation instructions of the control layer, executes the greenhouse environment control strategy and the agricultural management decision suggestions, and provides relevant information to the user operation and interaction interface.

Images