Method and device for intelligent management of water and fertilizer for greenhouse tomatoes by integrating agricultural experience knowledge

Abstract

The application provides a facility tomato water and fertilizer intelligent management method and device fusing agricultural experience knowledge, and the method comprises the following steps: inputting a to-be-recognized target parameter into a crop coefficient prediction model to obtain a crop coefficient output by the crop coefficient prediction model; the crop coefficient prediction model is obtained by training an extreme learning machine model based on a training set and a knowledge function; the knowledge function is determined based on a loss function and a conflict function; the conflict function is determined based on an output result and domain experience knowledge; and water and fertilizer management is performed on the facility tomato based on the crop coefficient. The facility tomato water and fertilizer intelligent management method fusing agricultural experience knowledge provided by the application determines a knowledge function by combining a target parameter and domain experience knowledge, trains an extreme learning machine model, thereby obtaining a crop coefficient prediction model, and performs water and fertilizer management on the facility tomato according to the crop coefficient output by the crop coefficient prediction model, so that the irrigation precision is improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural irrigation technology, and in particular to a method and device for intelligent water and fertilizer management of greenhouse tomatoes that integrates agricultural experience and knowledge. Background Technology

[0002] Water management and fertilizer application are the two most important factors affecting tomato growth, development, yield, and quality. Excessive water deficit alters the chemical environment of the crop roots, thus affecting the photosynthetic rate. Conversely, excessive water supply weakens the crop's reactive oxygen metabolism, leading to a decrease in photosynthetic rate and crop yield.

[0003] Traditional water and fertilizer decision-making methods include mechanistic models and data-driven models. These models, based on soil data, crop characteristics, crop patterns, and climate change, consider crop requirements and model the interactions of all system elements to understand them. However, these models are complex in expression, their development is limited by specific observational data, their physical understanding of the system and irrigation process is narrow, and their performance is affected by the model structure and how the input data represents the irrigation process. Furthermore, not all interactions of system elements can be fully understood or modeled, and the calculated water usage may not provide acceptable actual water usage or meet the needs of agricultural practitioners.

[0004] Data-driven models directly explore the relationship between water and fertilizer application drivers (primarily facility environmental variables) and application rates without requiring any knowledge of the physical processes within the system. Data-driven modeling employs statistical methods such as artificial intelligence and multivariate time series analysis. While data-driven models can improve the predictive performance of water and fertilizer forecasting, they still have some limitations. The modeling process is implicit, and overfitting is a common problem. Data-driven models perform worse during periods of extremely high or low demand. This is primarily due to these models overgeneralizing during training.

[0005] Existing water and fertilizer decision-making methods have low irrigation precision, and excessive water and fertilizer can easily lead to problems such as soil moisture loss, reduced fertilizer utilization efficiency, soil compaction, and groundwater pollution. Summary of the Invention

[0006] To address the problems existing in the prior art, embodiments of the present invention provide a method and device for intelligent water and fertilizer management of greenhouse tomatoes that integrates agricultural experience and knowledge.

[0007] This invention provides a method for intelligent water and fertilizer management of greenhouse tomatoes that integrates agricultural experience and knowledge, comprising:

[0008] Input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model;

[0009] The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge.

[0010] Water and fertilizer management is carried out for greenhouse tomatoes based on the crop coefficient.

[0011] In some embodiments, the water and fertilizer management of greenhouse tomatoes based on the crop coefficient includes:

[0012] Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined;

[0013] Based on the theoretical daily water demand, the soil moisture deficit is determined;

[0014] If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined.

[0015] The greenhouse tomatoes are irrigated based on the stated amount of water used for irrigation.

[0016] In some embodiments, after irrigating the greenhouse tomatoes based on the irrigation amount, the method further includes:

[0017] The amount of fertilizer to be applied is determined based on the growth period of greenhouse tomatoes;

[0018] Based on the amount of fertilizer applied, fertilizer is applied to greenhouse tomatoes.

[0019] In some embodiments, the expression for the loss function is as follows:

[0020]

[0021] Where H(p,q) represents the loss function, x represents the dimension of the output vector, P(x) represents the value of the x-th element in the expected output vector, and q(x) represents the value of the x-th element in the actual output vector.

[0022] In some embodiments, the target parameters for the greenhouse tomatoes include:

[0023] Tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weather forecast for the week.

[0024] This invention also provides an intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge, comprising:

[0025] The output module is used to input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model.

[0026] The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge.

[0027] The management module is used to manage water and fertilizer for greenhouse tomatoes based on the crop coefficient.

[0028] In some embodiments, the management module is specifically used for:

[0029] Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined;

[0030] Based on the theoretical daily water demand, the soil moisture deficit is determined;

[0031] If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined.

[0032] Irrigation of greenhouse tomatoes is carried out based on the aforementioned water volume.

[0033] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the program, it implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described above.

[0034] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described above.

[0035] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described above.

[0036] The present invention provides a method and device for intelligent water and fertilizer management of greenhouse tomatoes that integrates agricultural experience and knowledge. By combining target parameters and domain experience knowledge to determine a knowledge function, and training an extreme learning machine model, a crop coefficient prediction model is obtained. Water and fertilizer management of greenhouse tomatoes is carried out based on the crop coefficient output by the crop coefficient prediction model, thereby improving irrigation accuracy. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0038] Figure 1 This is one of the flowcharts of the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, provided in an embodiment of the present invention.

[0039] Figure 2 This is the second flowchart of the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, provided in this embodiment of the invention.

[0040] Figure 3 This is the third flowchart of the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, provided in this embodiment of the invention.

[0041] Figure 4 This is a schematic diagram of the structure of the intelligent water and fertilizer management system for greenhouse tomatoes that integrates agricultural experience and knowledge, provided in an embodiment of the present invention.

[0042] Figure 5 This is a schematic diagram of the structure of the intelligent water and fertilizer management device for facility tomatoes that integrates agricultural experience and knowledge, provided in an embodiment of the present invention.

[0043] Figure 6 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0045] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0046] Tomatoes are a typical greenhouse crop. The water and fertilizer absorbed by tomatoes mainly come from the soil, which determines their growth and yield. With the widespread availability of IoT online monitoring devices, time-series data on greenhouse climate and crop growth are easily accessible. Exploring new features based on time-series data to improve the accuracy of irrigation decisions in agricultural facilities is receiving increasing attention. Using time-series data of characteristic environments obtained from experiments under one or several conditions, along with target parameters, to measure the quantitative correlation between these parameters and vegetable yield and quality, and using this as a feature for irrigation decisions, can improve the intelligence level of facility water and fertilizer management. However, this method of decision-making based on monitoring data suffers from overly defined boundaries. This invention combines the irrigation experience of agricultural personnel with a data-driven intelligent irrigation decision-making model by constructing a knowledge function, enabling the water and fertilizer decision-making model to learn independently and adapt to different facility environments.

[0047] Figure 1 This is one of the flowcharts illustrating the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, as provided in this embodiment of the invention. Figure 1 As shown in the embodiment of the present invention, the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge includes:

[0048] Step 101: Input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model;

[0049] The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge.

[0050] Step 102: Based on the crop coefficient, perform water and fertilizer management on greenhouse tomatoes.

[0051] It should be noted that the implementing entity of the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge provided by this invention can be an electronic device, a component in an electronic device, an integrated circuit, or a chip. The electronic device can be a mobile electronic device or a non-mobile electronic device. For example, mobile electronic devices can be mobile phones, tablets, laptops, handheld computers, in-vehicle electronic devices, wearable devices, ultra-mobile personal computers (UMPCs), netbooks, or personal digital assistants (PDAs), etc., while non-mobile electronic devices can be servers, network attached storage (NAS), personal computers (PCs), televisions (TVs), ATMs, or self-service machines, etc. This invention does not impose specific limitations.

[0052] In step 101, the target parameters to be identified are input into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model.

[0053] The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge.

[0054] By inputting the target parameters to be identified into the trained crop coefficient prediction model, the crop coefficient predicted by the crop coefficient prediction model can be obtained.

[0055] The crop coefficient prediction model is trained using a training set and a knowledge function.

[0056] The training set is determined based on the target parameters of greenhouse tomatoes. The target parameters of greenhouse tomatoes are data on crop influencing factors and crop growth, such as tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weekly weather forecast.

[0057] Optionally, wireless sensors can be networked to monitor air temperature and humidity, light radiation, atmospheric pressure, and soil temperature and humidity within the tomato greenhouse. Under expert guidance, different water and fertilizer strategies can be implemented for zoned irrigation experiments. Tomato yields under different water and fertilizer combinations can be collected, and tomato quality can be determined through professional evaluation. The zoned experiments first obtain crop coefficients using the dual-crop coefficient method, which serve as the sample labels for the training set.

[0058] Irrigation is carried out using the following formulas for calculating irrigation thresholds and irrigation volume, resulting in a dataset of facility water and fertilizer irrigation, expressed as follows:

[0059] T n =(x1,x2,…,x t ,…,x n )

[0060] x t ={a t ,b t ,c t ,d t}

[0061] Among them, T n Denotes the target parameter set, x t Denotes the target parameter, a t Indicates the tomato transplanting time, b t c represents the environmental parameters inside the greenhouse. t Denotes the evapotranspiration parameter, d t This indicates the local weather forecast for the week.

[0062] Among them, the greenhouse environmental parameter b t = {airt, airh, lr, atp, st, sh}, where airt represents the air temperature inside the greenhouse, airh represents the air humidity, lr represents the light radiation, atp represents the atmospheric pressure, st represents the soil temperature, and sh represents the soil moisture.

[0063] To ensure that the data-driven experiments were conducted in the same greenhouse environment as the above-mentioned experiments, different agricultural personnel in different zones determined the irrigation time and amount based on weather forecasts, tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weekly weather forecasts. Tomato yields under the relevant irrigation schemes were collected, and tomato quality was assessed.

[0064] Based on the target parameters of greenhouse tomatoes, a training set and a test set for the crop coefficient prediction model can be constructed as follows:

[0065] T tr =(x1,x2,…,x m )

[0066] T te =(x m+1 ,x m+2 ,…,x n )

[0067] 3≤m≤n-3, m, n∈N

[0068] Among them, T tr T represents the training set. tedenoted as the test set, n represents the number of original datasets, and m represents the number of training sets.

[0069] The data in the training and test sets are preprocessed to obtain the preprocessed training and test sets, as shown in the following expressions:

[0070] T t ′ r =(x1′,x2′,…,x′) m )

[0071] T t ′ e =(x′) m+1 ,x′ m+2 ,…,x′ n )

[0072]

[0073] 1≤t≤m,t∈N

[0074] Among them, T t ′ r T represents the preprocessed training set. t ′ e This represents the preprocessed test set; x t This represents the original value within the time series t.

[0075] The reference evapotranspiration can be estimated based on the collected data, as expressed below:

[0076]

[0077] Where ET0 represents the reference evapotranspiration, G represents the soil heat flux, which, according to existing research, is quite small and can be ignored. In this embodiment of the invention, G≈0 is taken, and R... n This represents the daily net radiation of the crop canopy, collected using a net radiation sensor. T represents the daily average temperature, RH represents the daily average relative humidity (collected via a wireless sensor), and γ represents the hygrometer constant, calculated as γ = 0.665 × 10⁻³ × P, where P represents atmospheric pressure. S Δ represents the saturated vapor pressure, and Δ represents the slope of the saturated vapor pressure curve when the greenhouse temperature is T.

[0078] e s The expression is as follows:

[0079]

[0080] Among them, T max T represents the highest temperature of the day. min This indicates the lowest temperature of the day.

[0081] The expression for Δ is as follows:

[0082]

[0083] The theoretical daily water requirement ET for tomato crops on day i i The expression is as follows:

[0084] ET i =ET0×(K c )

[0085] Among them, ET i This represents the theoretical daily water requirement for tomato crops on day i, ET0 represents the reference evapotranspiration, and K... c This represents the crop coefficient.

[0086] After preprocessing the training set and dataset, the extreme learning machine model can be trained based on the preprocessed training set. Under the condition of minimizing the knowledge function, a trained crop coefficient prediction model can be obtained, and the crop coefficient prediction model can be tested based on the preprocessed test set.

[0087] Extreme Learning Machine (ELM) is a machine learning algorithm applicable to single-hidden-layer feedforward neural networks. Among various data-driven models, it has outstanding advantages in terms of training speed and generalization performance. In this embodiment of the invention, it is selected as a data mining algorithm to construct a crop coefficient prediction model based on ELM.

[0088] In this embodiment of the invention, the tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weather forecast for one week are used as decision feature parameters. Therefore, the number of neurons in the input layer is set to 15, the number of neurons in the output layer is set to 1, and the activation function of the hidden layer neurons is the sigmoid function.

[0089] The output vector is processed using the softmax function, so that the value of each output element in the vector is between 0 and 1, and its value represents the different crop coefficients corresponding to the input data.

[0090] Based on the crop coefficient prediction model, the predicted crop coefficient is obtained under data-driven conditions, and the irrigation time and irrigation amount are obtained according to the irrigation threshold calculation formula and the irrigation amount calculation formula.

[0091] To standardize the irrigation experience and knowledge of agricultural workers, three weather forecast type constraints can be set: whether the weather is consecutively sunny within a week before and after the current date, less than 2 rainy days, and more than or equal to 2 rainy days.

[0092] Under different weather conditions, agricultural personnel set irrigation event triggering conditions and irrigation amounts based on tomato planting time, greenhouse environmental parameters, and evapotranspiration parameters.

[0093] Taking greenhouse tomato irrigation under consecutive sunny weather as an example, the standardized irrigation knowledge is shown in Table 1.

[0094] Table 1 Standard Irrigation Knowledge

[0095]

[0096]

[0097] Tomato yield results are broadly categorized into three types: normal yield and quality, abnormal yield and quality, and abnormal yield and abnormal quality.

[0098] Principal component analysis is used to extract the mapping relationship between different output effects and features such as irrigation time and irrigation amount, and the classification threshold is determined. Finally, fuzzy rule knowledge between different features and output effects is extracted by combining fuzzy logic.

[0099] Table 2. Tomato Yield and Corresponding Vector Labels

[0100] Tomato yield Vector Labels Production and quality are normal (1,0,0) Abnormal production and quality (0,1,0) Abnormal output and abnormal quality (0,0,1)

[0101] In order to incorporate the influence of domain experience knowledge on irrigation decisions, embodiments of the present invention construct a conflict function based on a data-driven crop coefficient prediction model and domain knowledge.

[0102] The knowledge function consists of two parts: first, the loss function between the output of the pure data-driven model and the actual labels of the training set samples; and second, the conflict function between the output of the pure data-driven model and domain experience knowledge.

[0103] Cross-entropy is often used in data mining algorithms to characterize the distance between the actual output and the expected output. The smaller the value, the closer the probability distribution of the actual output and the expected output are. In this embodiment of the invention, cross-entropy is used as the loss function of the model.

[0104] In some embodiments, the expression for the loss function is as follows:

[0105]

[0106] Where H(p,q) represents the loss function, x represents the dimension of the output vector, P(x) represents the value of the x-th element in the expected output vector, and q(x) represents the value of the x-th element in the actual output vector.

[0107] Based on the experience and knowledge of agricultural workers, the irrigation amount and irrigation time were obtained. The correlation between irrigation time and irrigation amount and tomato yield was obtained using a fuzzy rule knowledge reasoning algorithm.

[0108] Meanwhile, based on the irrigation amount and irrigation time obtained from the data-driven model, the correlation between irrigation time and irrigation amount is obtained using the fuzzy rule knowledge reasoning algorithm to determine the tomato yield effect.

[0109] This invention defines two types of serious decision-making errors in purely data-driven models:

[0110] One issue is making incorrect decisions about irrigation timing, such as irrigating heavily before tomatoes bear fruit, causing the tomatoes to only grow in size but not produce fruit.

[0111] Second, the irrigation duration is incorrect.

[0112] Based on the above definition, the calculation rules for the conflict function in this embodiment of the invention are as follows:

[0113] Rule1:if((A≤1&&1<B<=3)||(A≤2&&2<B<=3)||(B≤1&&1<A<=3)||(B≤2&&2<A<=3))thenconflictfunction+1

[0114] Rule2: if((A≤1&&B≤1&&A≤B)||(1<A≤2&&1<B≤2&&A≤B)||(2<A≤3&&2<B≤3&&A≤B))thenconflictfunction+1

[0115] Where A represents the column number of the maximum value in the output vector of the Extreme Learning Machine model. A represents the probability that the pure data-driven model believes that the tomato yield and quality corresponding to that column are the most likely for a set of input data.

[0116] B represents the column number of the maximum value in the irrigation results obtained from domain knowledge, and represents the range of tomato yield and quality obtained through empirical knowledge.

[0117] Rule 1 means that when the output type of greenhouse tomatoes is divided into three categories: normal yield and quality, abnormal yield and quality, and mixed abnormal yield and quality, if the output result of the Extreme Learning Machine (ELM) model is not in the same category as the diagnostic result obtained from artificial irrigation, the conflict function value is increased to guide the retraining of the ELM model and reduce the probability of it making low-level decision-making errors.

[0118] Rule 2 means that if the output of the Extreme Learning Machine model and the irrigation result obtained by agricultural workers are in the same major category of decision, but their diagnoses of the output level are different, and the output level that the Extreme Learning Machine model believes is lower than the output level of the knowledge rule, then the conflict function value is increased to reduce the probability of diagnosing a serious abnormality in yield quality as a non-serious abnormality in yield quality, thus ensuring the accuracy of the output result.

[0119] Existing data-driven extreme learning machines rely too heavily on sample data, resulting in limited generalization performance. Models trained on a particular training set are difficult to apply to other environments.

[0120] In this embodiment of the invention, based on the actual irrigation experience of agricultural workers, the high-yield and high-quality correlation characteristics and irrigation strategy classifications under different irrigation experiences were determined. Knowledge of greenhouse tomato irrigation based on agricultural experience was extracted and introduced into the training of a pure data-driven model. This knowledge was used to guide and constrain the direction of data mining. The concept of a knowledge function was proposed by combining the loss function and the conflict function, and two calculation rules for the conflict function were defined to avoid serious decision-making errors. A data-knowledge jointly driven crop coefficient prediction model was established, thereby improving irrigation accuracy.

[0121] The knowledge function can be obtained by adding the calculated cross-entropy function and conflict function. The cross-entropy function reflects the error between the data mining algorithm output and the actual label, while the conflict function reflects the conflict between the data mining algorithm and domain experience knowledge.

[0122] With the goal of minimizing the aforementioned knowledge function, the initial weights and thresholds of the prediction model based on the Extreme Learning Machine are optimized by constructing a fitness function and using evolutionary learning algorithms such as particle swarm optimization and ant colony optimization, thereby completing the joint-driven model training.

[0123] This invention establishes a crop coefficient prediction model based on Extreme Learning Machines, which can solve the common problems of insufficient generalization performance and difficulty in guaranteeing the accuracy of output results in pure data-driven models. It analyzes the advantages and disadvantages of decision-making methods based on domain experience and intelligent decision-making methods based on data. Combining data-knowledge fusion methods, the concept of a conflict function and its two calculation rules are defined. A knowledge function is constructed by combining the loss function of the data-driven model with the conflict function between data and knowledge. Drawing on the greedy algorithm in reinforcement learning, a random factor and a trust function that varies with the number of iterations are introduced into the knowledge function. Adjusting the size of the random factor allows for adjustment of the participation of domain knowledge, and the introduction of the trust function can correct the adverse effects of domain knowledge errors on model training. Finally, a data-knowledge jointly driven crop coefficient prediction model is established.

[0124] Figure 2 This is the second flowchart illustrating the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, as provided in this embodiment of the invention. Figure 2 As shown, after determining the knowledge function, the extreme learning machine model is trained based on the data in the preprocessed training set. When the knowledge function is minimized, the trained crop coefficient prediction model is obtained, and the crop coefficient prediction model is validated based on the preprocessed test set.

[0125] By inputting the target parameters to be identified into the crop coefficient prediction model based on the trained crop coefficient prediction model, the crop coefficient output by the crop coefficient prediction model can be obtained.

[0126] In step 102, water and fertilizer management is carried out on greenhouse tomatoes based on the crop coefficient.

[0127] Based on the crop coefficient prediction model, water and fertilizer management can be implemented for greenhouse tomatoes. For example, the theoretical daily water requirement of greenhouse tomatoes can be determined based on the crop coefficient, and the soil moisture deficit can be calculated to determine whether watering is necessary. Fertilizer management can also be implemented according to the growth stage of the greenhouse tomatoes.

[0128] Optionally, Figure 3 This is the third flowchart of the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge, provided in this embodiment of the invention. Figure 3 As shown, after determining the theoretical daily water requirement of tomato crops, the soil moisture deficit threshold and the amount of soil moisture deficit can be calculated. When the amount of soil moisture deficit is greater than the threshold, the amount of irrigation water is calculated, as well as the amount of nitrogen, phosphorus and potassium applied according to the crop's growth stage. Based on the calculated amount of irrigation water and the amount of nitrogen, phosphorus and potassium applied, irrigation and fertilization can be carried out on greenhouse tomatoes.

[0129] The intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge provided in this invention determines a knowledge function by combining target parameters and domain experience and training an extreme learning machine model to obtain a crop coefficient prediction model. The method then manages water and fertilizer for greenhouse tomatoes based on the crop coefficients output by the crop coefficient prediction model, thereby improving irrigation accuracy.

[0130] Furthermore, this invention provides a crop coefficient prediction model based on a self-learning mechanism. Addressing the varying crop coefficients resulting from different regions, crop varieties, and production practices, a crop coefficient prediction model based on an Extreme Learning Machine (ELM) is established, incorporating leaf area index, tomato transplanting days, facility environmental parameters, actual irrigation volume, actual irrigation time, reference evapotranspiration, and tomato yield information. Leveraging the powerful feature extraction capabilities provided by the numerous nonlinear functions within the ELM, the model uncovers the hidden correlations between crop growth, greenhouse environment, and irrigation water usage.

[0131] In some embodiments, the water and fertilizer management of greenhouse tomatoes based on the crop coefficient includes:

[0132] Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined;

[0133] Based on the theoretical daily water demand, the soil moisture deficit is determined;

[0134] If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined.

[0135] The greenhouse tomatoes are irrigated based on the stated amount of water used for irrigation.

[0136] The crop coefficient K is output by the trained crop coefficient prediction model. c Based on the above formula for calculating theoretical daily water requirement, the theoretical daily water requirement ET for tomato crops on day i can be calculated. i .

[0137] Based on the theoretical daily water requirement ET of the tomato crop on day i. i The soil moisture deficit is calculated using the following expression:

[0138] D i =D i-1 +ET i

[0139] Among them, D i (cm) represents the soil moisture deficit on day i, D i-1 ET represents the soil moisture deficit value on day i-1. i This represents the theoretical daily water requirement for tomato crops on day i.

[0140] If the planting day is the first day, then D0 = 0.

[0141] When the soil moisture deficit exceeds the soil water and fertilizer deficit threshold D th (cm), i.e., the preset threshold, then irrigate the greenhouse tomatoes, D th The expression is as follows:

[0142]

[0143] MAD represents the allowable deficit percentage, which can be selected based on Huffman's research findings. This represents the soil field-holding moisture content, calculated as the weighted average of the soil field-holding moisture content at the root depth. The soil moisture content at the wilting point is represented by the weighted average of the soil moisture content at the root depth; RD (cm) represents the root depth, which can be estimated based on the crop growth stage and crop height.

[0144] Irrigation is triggered when the soil moisture deficit in the tomato greenhouse exceeds the amount of water available for crop use. The amount of water irrigated is equal to the total evapotranspiration since the last irrigation.

[0145] Based on the calculated irrigation amount, the greenhouse tomatoes are then irrigated.

[0146] The intelligent water and fertilizer management method for greenhouse tomatoes provided in this invention integrates agricultural experience and knowledge. It calculates the theoretical daily water requirement and soil moisture deficit of greenhouse tomatoes by using the crop coefficient output by the crop coefficient prediction model, thereby determining whether irrigation is needed and the amount of irrigation, thus achieving precise irrigation for greenhouse tomatoes.

[0147] In some embodiments, after irrigating the greenhouse tomatoes based on the irrigation amount, the method further includes:

[0148] The amount of fertilizer to be applied is determined based on the growth period of greenhouse tomatoes;

[0149] Based on the amount of fertilizer applied, fertilizer is applied to greenhouse tomatoes.

[0150] Based on the different growth stages of greenhouse tomatoes and soil fertility, the amount of fertilizer to be applied, namely the amount of nitrogen, phosphorus, and potassium nutrients, can be calculated using the following formula:

[0151] Nutrient application amount = nutrient absorption amount × correction coefficient × nutrient absorption coefficient at growth stage, where nutrient absorption amount = target yield × nutrient absorption per unit yield.

[0152] The correction coefficient and the nutrient absorption coefficient at different growth stages are affected by factors such as the crop's ability to absorb and utilize nutrients, soil fertility levels, and soil residues and losses. They can be determined through actual water and fertilizer production experiments.

[0153] Based on the calculated amount of fertilizer to be applied, fertilizer is then applied to greenhouse tomatoes.

[0154] The intelligent water and fertilizer management method for greenhouse tomatoes provided in this invention integrates agricultural experience and knowledge. By determining the growth stage of the crop, a simple and rapid recommendation model for crop fertilizer application is established based on the principle of soil nutrient balance, so as to achieve precise fertilization of greenhouse tomatoes and avoid environmental pollution and other problems caused by excessive fertilization.

[0155] Figure 4 This is a schematic diagram of the structure of the intelligent water and fertilizer management system for greenhouse tomatoes that integrates agricultural experience and knowledge, as provided in an embodiment of the present invention. Figure 4 As shown in the figure, this embodiment of the invention also provides a system for a precise water and fertilizer decision-making method suitable for greenhouses, including an environmental information acquisition unit, a decision training unit, and an intelligent irrigation and fertilization unit.

[0156] The decision training unit includes a control center computer, a database, and a model analysis module. The control center computer is used for data storage and computation. The database includes an expert experience database, an agricultural personnel experience database, a soil information database, and a crop information database. It is used to store and retrieve data such as nutrient uptake per unit yield of different crops, target yield data, nutrient uptake coefficients at different growth stages, and correction coefficients.

[0157] The model analysis module includes a nutrient application recommendation model, a crop growth period water requirement model, and a nutrient requirement model, which are used for crop water and fertilizer decision analysis.

[0158] The intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge, provided by this invention, is described below. The intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge described below can be referred to in correspondence with the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge described above.

[0159] Figure 5 This is a schematic diagram of the structure of the intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge, as provided in an embodiment of the present invention. Figure 5 As shown in the figure, the intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge provided in this embodiment of the invention includes:

[0160] Output module 510 is used to input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficient output by the crop coefficient prediction model;

[0161] The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge.

[0162] The management module 520 is used to manage water and fertilizer for greenhouse tomatoes based on the crop coefficient.

[0163] It should be noted that the intelligent water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge provided in this embodiment of the invention can realize all the method steps implemented in the above-mentioned intelligent water and fertilizer management method embodiment for greenhouse tomatoes that integrates agricultural experience and knowledge, and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.

[0164] Optionally, the management module 520 is specifically used for:

[0165] Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined;

[0166] Based on the theoretical daily water demand, the soil moisture deficit is determined;

[0167] If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined.

[0168] The greenhouse tomatoes are irrigated based on the stated amount of water used for irrigation.

[0169] Optionally, it also includes a fertilizer application module for:

[0170] The amount of fertilizer to be applied is determined based on the growth period of greenhouse tomatoes;

[0171] Based on the amount of fertilizer applied, fertilizer is applied to greenhouse tomatoes.

[0172] Optionally, the expression for the loss function is as follows:

[0173]

[0174] Where H(p,q) represents the loss function, x represents the dimension of the output vector, P(x) represents the value of the x-th element in the expected output vector, and q(x) represents the value of the x-th element in the actual output vector.

[0175] Optionally, the target parameters for the facility-grown tomatoes include:

[0176] Tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weather forecast for the week.

[0177] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a smart water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience knowledge. This method includes: inputting target parameters to be identified into a crop coefficient prediction model to obtain crop coefficients output by the crop coefficient prediction model; wherein the crop coefficient prediction model is obtained by training an extreme learning machine model based on a training set and a knowledge function; the training set is determined based on target parameters of greenhouse tomatoes; the knowledge function is determined based on a loss function and a conflict function; the loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set; the conflict function is determined based on the output and domain experience knowledge; and water and fertilizer management is performed on the greenhouse tomatoes based on the crop coefficients.

[0178] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0179] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience knowledge provided by the above methods. The method includes: inputting the target parameters to be identified into a crop coefficient prediction model to obtain the crop coefficient output by the crop coefficient prediction model; wherein, the crop coefficient prediction model is obtained by training an extreme learning machine model based on a training set and a knowledge function; the training set is determined based on the target parameters of greenhouse tomatoes; the knowledge function is determined based on a loss function and a conflict function; the loss function is determined based on the output result of the extreme learning machine model and the sample labels corresponding to the training set; the conflict function is determined based on the output result and domain experience knowledge; and water and fertilizer management is performed on greenhouse tomatoes based on the crop coefficient.

[0180] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience knowledge, as provided by the methods described above. This method includes: inputting target parameters to be identified into a crop coefficient prediction model to obtain crop coefficients output by the crop coefficient prediction model; wherein the crop coefficient prediction model is obtained by training an extreme learning machine model based on a training set and a knowledge function; the training set is determined based on target parameters of the greenhouse tomatoes; the knowledge function is determined based on a loss function and a conflict function; the loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set; the conflict function is determined based on the output and domain experience knowledge; and water and fertilizer management is performed on the greenhouse tomatoes based on the crop coefficients.

[0181] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0182] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent water and fertilizer management of greenhouse tomatoes that integrates agricultural experience and knowledge, characterized in that, include: Input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model; The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge. Water and fertilizer management is carried out for greenhouse tomatoes based on the crop coefficient. The water and fertilizer management for greenhouse tomatoes based on the crop coefficient includes: Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined; Based on the theoretical daily water demand, the soil moisture deficit is determined; If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined. Irrigate the greenhouse tomatoes based on the aforementioned water volume; After irrigating the greenhouse tomatoes based on the stated water volume, the process further includes: The amount of fertilizer to be applied is determined based on the growth period of greenhouse tomatoes; Based on the amount of fertilizer applied, fertilizer is applied to greenhouse tomatoes.

2. The intelligent water and fertilizer management method for greenhouse tomatoes integrating agricultural experience and knowledge as described in claim 1, characterized in that, The expression for the loss function is as follows: in, Let x represent the loss function, and let x represent the dimension of the output vector. This represents the value of the x-th element in the expected output vector. This represents the value of the x-th element in the actual output vector.

3. The intelligent water and fertilizer management method for greenhouse tomatoes integrating agricultural experience and knowledge as described in claim 1 or 2, characterized in that, The target parameters for the facility-grown tomatoes include: Tomato planting time, greenhouse environmental parameters, evapotranspiration parameters, and local weather forecast for the week.

4. A smart water and fertilizer management device for greenhouse tomatoes that integrates agricultural experience and knowledge, characterized in that, include: The output module is used to input the target parameters to be identified into the crop coefficient prediction model to obtain the crop coefficients output by the crop coefficient prediction model. The crop coefficient prediction model is obtained by training an extreme learning machine model using a training set and a knowledge function. The training set is determined based on the target parameters of greenhouse tomatoes. The knowledge function is determined based on a loss function and a conflict function. The loss function is determined based on the output of the extreme learning machine model and the sample labels corresponding to the training set. The conflict function is determined based on the output and domain experience knowledge. The management module is used to manage water and fertilizer for greenhouse tomatoes based on the crop coefficient. The management module is specifically used for: Based on the crop coefficient, the theoretical daily water requirement of greenhouse tomatoes is determined; Based on the theoretical daily water demand, the soil moisture deficit is determined; If the soil moisture deficit is determined to be greater than a preset threshold, the irrigation amount for greenhouse tomatoes is determined. Irrigate the greenhouse tomatoes based on the aforementioned water volume; The device also includes a fertilizer application module for: The amount of fertilizer to be applied is determined based on the growth period of greenhouse tomatoes; Based on the amount of fertilizer applied, fertilizer is applied to greenhouse tomatoes.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described in any one of claims 1 to 3.

6. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described in any one of claims 1 to 3.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent water and fertilizer management method for greenhouse tomatoes that integrates agricultural experience and knowledge as described in any one of claims 1 to 3.

Images