Greenhouse temperature and humidity prediction method and system based on limited sensor and medium

Abstract

The invention relates to the technical field of greenhouse temperature prediction, in particular to a greenhouse temperature and humidity prediction method and system based on limited sensors and a medium. Collecting data of an indoor reference point, an indoor virtual point and an outdoor reference point of the greenhouse through a limited number of sensors; establishing a mapping relation between the two reference points and the virtual points based on the data, and obtaining three-dimensional information of indoor temperature and humidity; establishing a classification algorithm model, and clustering the three-dimensional information to obtain a classification result; constructing a corresponding neural network model according to a classification result, and training to obtain a trained model; and the trained model carries out prediction according to the input two datum point data, and outputs a temperature and humidity prediction result of the indoor virtual point of the greenhouse. On the basis of data of two reference points inside and outside a greenhouse, a mapping relation between the two reference points and a virtual point is obtained, so that a model and a method for multi-point temperature and humidity three-dimensional prediction based on data of a limited number of sensors are established, and the purposes of saving cost and improving efficiency are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of greenhouse temperature prediction, and particularly to a method, system and medium for predicting greenhouse temperature and humidity based on limited sensors. Background Art

[0002] A greenhouse is an important part of facility agriculture. The internal microclimate environmental factors include temperature, humidity, solar radiation, carbon dioxide, sunshine hours, cloud amount, etc., among which temperature is a dominant factor. Maintaining an appropriate temperature inside the greenhouse is very important for the healthy growth of crops. However, due to its own structural characteristics, the greenhouse has problems such as lower light intensity than natural light, changed light quality, and uneven light distribution, resulting in an uneven spatial distribution of temperature inside the greenhouse, large differences in temperature distribution at different positions, and a complex temperature distribution pattern that not only leads to relatively low utilization rate of thermal energy, but also causes differences in the growth of crops in different regions, and even reduces the comprehensive yield. Therefore, studying the distribution of the temperature field inside the greenhouse is an important topic facing greenhouse researchers.

[0003] Currently, the control of temperature changes inside the greenhouse is mainly to adjust the actuators inside the greenhouse through the current data measured by a single monitoring point sensor to change the greenhouse environment to meet the growth needs of crops. There are obvious one-sidedness in both the feedback of data collection and the effectiveness of regulation measures. In order to meet the demand for multi-point temperature monitoring, producers often need to configure corresponding sensors at different points, resulting in a sharp increase in the number of sensors in a single greenhouse, thus significantly increasing the regulation cost of the greenhouse environment. However, the greenhouse has the attribute of being sensitive to production costs and has relatively high requirements for sensor input costs and production management convenience. Considering economy and universality, the number of sensors used should be restricted. Therefore, reducing the redundancy of the number of sensors and using limited sensors to accurately evaluate the temperature condition inside the greenhouse has important theoretical value and practical significance.

[0004] With the continuous development of modern computing methods, identification models have received more attention and application from scholars (Tian Dong et al., 2020; Li Huan et al., 2020). A large number of scholars have constructed greenhouse environmental factor identification models based on methods such as statistical regression analysis, neural network prediction, and time series clustering by collecting environmental data, and studied the mathematical laws between a large amount of data, showing relatively high prediction accuracy. However, most of the existing studies are for predicting the greenhouse temperature at a single monitoring point or the average temperature at different points; few studies consider the differential characteristics of its environmental spatial distribution, and the few existing studies are also based on a relatively large number of sensors. Summary of the Invention

[0005] (I) Technical Problems to be Solved

[0006] In view of the above-mentioned disadvantages and deficiencies of the prior art, the present invention provides a method for predicting the temperature and humidity in a greenhouse based on limited sensors. Based on the data of two reference points inside and outside the greenhouse, the mapping relationship between the two reference points and the virtual point is obtained, so as to establish a model and method for three-dimensional prediction of temperature and humidity at multiple points based on data from a limited number of sensors, achieving the purpose of cost savings and efficiency improvement. In use, only two sensor points need to be arranged for a long time to predict the three-dimensional distribution of temperature and humidity at multiple points in the greenhouse, which can effectively reduce the number of sensors arranged and greatly reduce the equipment investment cost, and has high popularization value in production; the prediction model considers the differential characteristics of the environmental space distribution, and its prediction accuracy is higher.

[0007] (II) Technical Solution

[0008] To achieve the above object, the main technical solutions adopted by the present invention include:

[0009] In the first aspect, the present invention provides a method for predicting the temperature and humidity in a greenhouse based on limited sensors, including the steps of:

[0010] Collecting data of the indoor reference point, indoor virtual point and outdoor reference point in the greenhouse through a limited number of sensors;

[0011] Based on the data, establishing the mapping relationship between the two reference points and the virtual point to obtain the three-dimensional information of the indoor temperature and humidity;

[0012] Establishing a classification algorithm model, clustering the three-dimensional information according to the model to obtain a classification result;

[0013] Constructing a corresponding neural network model according to the classification result and training to obtain a trained model;

[0014] The trained model predicts based on the input data of the two reference points and outputs the temperature and humidity prediction results of the indoor virtual point in the greenhouse by collecting the data of the indoor reference point and outdoor reference point through a limited number of sensors.

[0015] Optionally, the data includes the regulation historical characteristics, crop growth characteristics and meteorological characteristics of the indoor reference point and virtual point, and the meteorological characteristics of the outdoor reference point.

[0016] Optionally, the regulation historical characteristics include one or any combination of two or more of the top window, sunshade net, rolling curtain film opening and irrigation historical data;

[0017] And / or the meteorological characteristics of the indoor reference point and virtual point or the meteorological characteristics of the outdoor reference point include one or any combination of two or more of temperature, humidity, light and wind speed and direction.

[0018] Optionally, it further includes preprocessing the collected data, and the preprocessing includes one or any combination of the following two or more:

[0019] (1) Analyzing and processing the spatio-temporal rules of temperature and humidity of the data;

[0020] (2) Conducting correlation analysis and processing on the data;

[0021] (3) Processing outliers of the data using the 3-sigma rule;

[0022] (4) Conducting data normalization processing on the data.

[0023] Optionally, a classification algorithm model is established, and the three-dimensional information is clustered according to the model, and the classification results include:

[0024] Classifying the three-dimensional information with the same spatial distance into one category through Euclidean clustering;

[0025] Applying the output result of Euclidean clustering to the cosine angle function for clustering again, and screening out the data with the same curve shape and classifying them into one category to obtain the classification result.

[0026] Optionally, the neural network model is an RBF neural network model; and / or during the training process, the mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE) are used to evaluate the model.

[0027] Optionally, the reference point in the greenhouse is the indoor center point, and all virtual points are arranged around the reference point.

[0028] In a second aspect, the present invention provides a prediction system based on limited sensors, and the system includes modules / units for executing the method of any possible design in the first aspect above. These modules / units can be implemented by hardware or by hardware executing corresponding software.

[0029] In a third aspect, the present invention provides an electronic device, including a memory and a processor, and a program is stored on the memory and can run on the processor. When the program is executed by the processor, the electronic device is enabled to execute the method of any possible design in any of the above aspects.

[0030] In a fourth aspect, the present invention provides a readable storage medium, and a program is stored in the readable storage medium. When the program is executed, the method of any possible design in any of the above aspects is implemented.

[0031] (III) Beneficial effects

[0032] The beneficial effects of the present invention are:

[0033] The prediction method of the present invention establishes an indoor multi-point temperature and humidity mapping through indoor reference points and the collected data, obtains rich three-dimensional temperature and humidity information in the greenhouse with only 2 limited sensors, realizes the balance of reducing the redundancy of the number of sensors and accurately predicting the spatial distribution of temperature and humidity. In use, only 2 sensor points need to be arranged for a long time to predict the three-dimensional distribution of multi-point temperature and humidity in the greenhouse, which can effectively reduce the number of arranged sensors and greatly reduce the equipment investment cost, and has high popularization value in production. It balances the relationship between the reduction of the number of sensors and the enhancement of the abundance of temperature information, solves the problem of sensitivity of the production cost of the greenhouse, improves the economy and universality of model application, and provides new ideas for the research on greenhouse temperature field prediction.

[0034] Among them, the present invention constructs a model for predicting the greenhouse temperature field. Based on the time series of environmental parameters of the reference points, it integrates the historical data characteristics of the outdoor macroclimate and the indoor microclimate, studies the mathematical laws among a large amount of data, solves the problems of difficult data acquisition and difficult real-time calculation existing in the mechanism model, and improves the universality of the prediction model in different greenhouses. Brief Description of the Drawings

[0035] Figure 1 It is a flowchart of the prediction method of greenhouse temperature and humidity based on limited sensors provided by an embodiment of the present invention;

[0036] Figure 2 It is a schematic diagram of the principle of the prediction method of greenhouse temperature and humidity based on limited sensors provided by an embodiment of the present invention;

[0037] Figure 3 It is the basic structure of the RBF neural network model provided by an embodiment of the present invention;

[0038] Figure 4 It is a schematic diagram of the overall structure of the prediction of greenhouse temperature and humidity based on limited sensors provided by an embodiment of the present invention. Detailed Embodiments

[0039] To illustrate in detail the possible application scenarios, technical principles, specific implementable solutions, achievable purposes and effects of the present application, the following is described in detail in combination with the listed specific embodiments and with reference to the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of the present application, so they are only examples and cannot be used to limit the protection scope of the present application.

[0040] References to "embodiments" in this application mean that the specific features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The term "embodiment" appearing in various positions in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or relevance to other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0041] Unless otherwise defined, the meanings of the technical terms used herein are the same as those commonly understood by those skilled in the technical field to which this application belongs; the use of the relevant terms herein is only for describing specific embodiments and is not intended to limit this application.

[0042] In the description of this application, the phrase "and / or" is an expression used to describe the logical relationship between objects, indicating that there can be three relationships. For example, A and / or B means: there is A, there is B, and there is both A and B at the same time. In addition, the character " / " in this application generally represents an "or" logical relationship between the associated objects before and after.

[0043] In this application, terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual quantitative, primary-secondary, or sequential relationships between these entities or operations.

[0044] Without further limitation, in this application, the use of "including", "comprising", "having" or other similar expressions in a statement is intended to cover non-exclusive inclusion. These expressions do not exclude the possibility that there may be additional elements in the process, method, or product that includes the said elements. Thus, a process, method, or product that includes a series of elements may include not only those defined elements, but also other elements not explicitly listed, or elements inherent to such a process, method, or product.

[0045] Similar to the understanding in the "Examination Guidelines", in this application, expressions such as "greater than", "less than", "exceeding" are understood not to include the recited number; expressions such as "above", "below", "within" are understood to include the recited number. In addition, in the description of the embodiments of this application, the meaning of "a plurality of" is two or more (including two). Similar expressions related to "many", such as "multiple groups", "multiple times", etc., are understood in the same way, unless otherwise specifically defined.

[0046] In the description of the embodiments of the present application, the spatially related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "perpendicular", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., the indicated orientation or positional relationship is based on the orientation or positional relationship shown in the specific embodiments or the drawings, and is only for the convenience of describing the specific embodiments of the present application or for the reader's understanding, rather than indicating or implying that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, it should not be construed as a limitation on the embodiments of the present application.

[0047] Unless otherwise clearly specified or limited, in the description of the embodiments of the present application, the terms "installed", "connected", "joined", "fixed", "set", etc. shall be understood in a broad sense. For example, the "connection" may be a fixed connection, a detachable connection, or an integral setting; it may be a mechanical connection, an electrical connection, or a communication connection; it may be directly connected, or indirectly connected through an intermediate medium; it may be the communication inside two components or the interaction relationship between two components. For those skilled in the art to which the present application pertains, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0048] As Figure 1 shown, a prediction method for the temperature and humidity in a greenhouse based on limited sensors includes the steps:

[0049] S101. Collect data of the indoor reference points, indoor virtual points, and outdoor reference points in the greenhouse through a limited number of sensors respectively.

[0050] In some embodiments, the data includes the regulation historical characteristics, crop growth characteristics, and meteorological characteristics of the indoor reference points and indoor virtual points, as well as the meteorological characteristics of the outdoor reference points.

[0051] In some specific embodiments, the regulation historical characteristics include one or any combination of two or more of the top window opening, sunshade net, rolling curtain film opening degree, and irrigation historical data. In some other specific embodiments, the meteorological characteristics of the indoor reference points, indoor virtual points, or outdoor reference points include one or any combination of two or more of temperature, humidity, light, and wind speed and direction.

[0052] In some other embodiments, it further includes preprocessing the collected data, and the preprocessing includes one or any combination of two or more of the following:

[0053] (1) Perform spatio-temporal law analysis and processing on the data;

[0054] (2) Perform correlation analysis and processing on the said data;

[0055] (3) Use the Pauta criterion to process the outliers in the said data;

[0056] (4) Perform data normalization processing on the said data.

[0057] S102. Establish the mapping relationship between two reference points and virtual points based on the said data, and obtain the three-dimensional information of the indoor temperature and humidity.

[0058] In some other embodiments, the indoor reference point of the greenhouse is the indoor center point, and all virtual points are arranged around the reference point.

[0059] S103. Establish a classification algorithm model, and cluster the said three-dimensional information according to the said model to obtain a classification result.

[0060] In some embodiments, establishing a classification algorithm model and clustering the said three-dimensional information according to the said model to obtain a classification result includes: classifying those with the same spatial distance in the said three-dimensional information into one category through Euclidean clustering; applying the cosine angle function to the output result of Euclidean clustering again for clustering, and screening out the data with the same curve shape and classifying them into one category to obtain a classification result.

[0061] S104. Construct a corresponding neural network model according to the said classification result, and train to obtain a trained model.

[0062] In some embodiments, the neural network model is an RBF neural network model; and / or during the training process, the mean absolute error MAE, mean absolute percentage error MAPE, and root mean square error RMSE are used to evaluate the model.

[0063] S105. The trained model makes predictions based on the input data of two reference points, and outputs the temperature and humidity prediction results of the virtual points in the greenhouse.

[0064] The beneficial effect of this embodiment lies in that based on the data of two reference points inside and outside the greenhouse, the mapping relationship between the two reference points and all virtual points is obtained, so as to establish a model and method for multi-point temperature and humidity three-dimensional prediction based on the data of a limited number of sensors. In use, only the sensors at two points need to be arranged for a long time to predict the three-dimensional distribution of the temperature and humidity at multiple points in the greenhouse, which can effectively reduce the number of sensors arranged, greatly reduce the equipment investment cost, and has high promotion value in production.

[0065] For the convenience of understanding, in this embodiment, the specific implementation process of the above method is further elaborated in combination with a specific application scenario system, such as Figure 2 and Figure 3 shown, and specifically includes the following steps:

[0066] 1. Data acquisition and reference point setting

[0067] (1) Data acquisition method and preprocessing, data acquisition method and preprocessing

[0068] The indoor environmental parameters are collected by the sensor spatial distribution method, with the collection frequency of once every 15 minutes and the monitoring time of 1 - 2 years. Horizontally, the length and width of the greenhouse are equally divided into 4 parts, and one node is arranged at each intersection point in the room; vertically, 3 monitoring heights are set, which are 0.6m, 1.2m, and 1.8m high from the ground respectively; there are a total of 27 monitoring nodes, all of which collect the meteorological characteristics of indoor temperature, humidity, light, wind speed and direction; the monitoring node located at the central position is set as the indoor reference point SN0, and the remaining 26 monitoring nodes are virtual points (SN1,..., SN26). At the same time, the control historical characteristics such as the opening degrees of the top window, sunshade net, rolling curtain film, irrigation history, and the growth characteristic data such as the crop growth period are recorded. The outdoor collected parameters are meteorological characteristics such as temperature, humidity, light, wind speed and direction. The outdoor monitoring point is located in an open and unobstructed area beside the test greenhouse, with a height of 1.5m from the ground; the outdoor monitoring node is set as the outdoor reference point SW0.

[0069] (2) Temperature and humidity spatio-temporal distribution law and correlation analysis

[0070] The same environmental factors at different heights and different orientations are systematically described to understand the spatio-temporal distribution law of temperature and humidity. Taking the analysis of the temperature environmental factor as an example, a detailed analysis is carried out on its daily average temperature, daily maximum temperature, daily minimum temperature, night-time daily average temperature, indoor and outdoor temperature difference at night, and the temperature changes under the conditions of typical months such as the hottest month and the coldest month, etc., to explore the temperature change law of the greenhouse; the analysis method of humidity is similar. Correlation analysis is used to reveal the correlation between the temperature and humidity of the virtual points and the historical environmental parameters, control historical characteristics and crop growth characteristics of the indoor reference point SN0 and the outdoor reference point SW0.

[0071] (3) Outlier processing and data normalization

[0072] The method of Pauta criterion is adopted, that is, when the absolute value of the difference between the measured value and the average value is greater than 3 standard deviations, the measured value is considered as an outlier, and this value is replaced by the average value of the data on both sides of the outlier point. To avoid the training error caused by too large value of a certain dimension of data and improve the prediction accuracy, the greenhouse environmental data is normalized, and then restored through inverse normalization after the prediction analysis is completed.

[0073] The normalization processing of the greenhouse environmental data satisfies Equation 1:

[0074]

[0075] Among them, x represents the characteristic data in the greenhouse environmental data;

[0076] x max represents the maximum value in the greenhouse characteristic data;

[0077] x min represents the minimum value in the greenhouse characteristic data;

[0078] x* represents the normalized value.

[0079] 2. Multi-point Temperature and Humidity Three-dimensional Prediction in Greenhouse Based on Sensor Reduction

[0080] (1) The multi-point temperature and humidity three-dimensional prediction in the greenhouse based on sensor reduction refers to establishing an indoor multi-point temperature and humidity mapping through the data collected by the indoor reference points SN0 and SW0, obtaining rich three-dimensional temperature and humidity information in the greenhouse with limited 2 sensors, and achieving a balance between reducing the redundancy of sensor quantity and accurately predicting the spatial distribution of temperature and humidity. Specifically, based on the correlation between the virtual point temperature and humidity and the historical meteorological characteristics, regulation historical characteristics, and crop growth characteristics of the reference points, characteristic data is selected as the representation of indoor temperature, light, and humidity, and data normalization is performed to remove the influence of dimension differences; a classification algorithm model is established, clustering is performed with the representation data of the virtual point temperature and humidity as the input, and the optimal number of classifications and classification scheme are obtained based on the selected evaluation indicators; for different classifications, a multi-point three-dimensional prediction model of temperature and humidity based on 2 reference points is established respectively.

[0081] (2) Establishing a classification algorithm model

[0082] The K-means clustering algorithm is a classic algorithm for solving clustering analysis. The traditional K-means clustering algorithm mainly uses the Euclidean distance as the discriminant function, but the Euclidean distance only reflects the size of the spatial distance between 2 samples and cannot reflect the similarity in shape between 2 sample data. Therefore, when selecting the metric function in this project, both the reflection of the Euclidean distance on the sample spatial distance and the reflection of the cosine angle function on the similarity of sample data shapes are considered. The mixed metric function considering 2 factors of sample values and sample changes can improve the similarity judgment ability of multi-dimensional data to a certain extent.

[0083] ① First, cluster the sample data with the same spatial distance into one category through Euclidean clustering:

[0084]

[0085] Among them, x fi * represents the i-th eigenvalue corresponding to the f-th variable in the greenhouse environmental data sample;

[0086] x gi * represents the i-th eigenvalue corresponding to the g-th variable in the greenhouse environmental data sample;

[0087] deuc(x f *,x g *) represents the value after Euclidean clustering;

[0088] ② Apply the cosine angle function to the output result of Euclidean clustering for clustering again, and screen out the sample data with the same curve shape and classify them into one category:

[0089]

[0090] Among them, d cos (x f *,x g *) represents the value after clustering by the cosine angle function.

[0091] After clustering, the monitoring data of the first year is used as the training set, and the data of the second year is used as the test set.

[0092] (3) RBF prediction of multi-point temperature and humidity in the greenhouse

[0093] For different classifications, establish an RBF prediction model for multi-point temperature and humidity in the greenhouse. The RBF neural network belongs to the type of 3-layer forward network, and has the characteristics of simple structure, fast calculation convergence speed, stable results and high approximation accuracy. Its basic idea is to use RBF as the "basis" of the hidden unit to form the hidden layer space. The hidden layer transforms the input vector, transforms the low-dimensional pattern input data into a high-dimensional space, so that the linearly inseparable problem in the low-dimensional space becomes linearly separable in the high-dimensional space.

[0094] In the RBF prediction of multi-point temperature and humidity, at the input nodes, to reduce parameter redundancy, according to the correlation analysis results in data preprocessing, select the meteorological historical features, regulation historical features and crop growth features of the reference points SN0 and SW0 with relatively large correlations with the virtual point temperature and humidity as input variables; the output layer is the temperature and humidity values of 47 virtual points, with a total of 94 output nodes.

[0095] Use the monitoring data of the first year to train the model, and use the data of the second year to test the training effect. The model is evaluated using the mean absolute error MAE, mean absolute percentage error MAPE, and root mean square error RMSE.

[0096] RBF neural network expression:

[0097] Y m (k) = w1h1 + … + w j h j + … + w m h m Equation Four;

[0098] Among them,

[0099] Y m (k) is the output of the neural network at the k-th iteration, m is the number of nodes in the hidden layer; [w1,..., w m is the weight vector; [h1,..., h m is the Gaussian basis function; X = [x1,..., x n is the input vector, n is the number of nodes in the network input layer; C j = [c j1 ,... c ji ,..., c jn is the center vector of the j-th node; b j is the node base width parameter of the radial basis function.

[0100] The learning metric function of the RBF neural network is:

[0101]

[0102] Y(k) is the actual value at time k.

[0103] Using the gradient descent method to obtain w j , b j , c ji , satisfying the requirements of Formula Seven, Formula Eight and Formula Nine;

[0104] w j (k) = w j (k - 1) + ρ(Y(k) - Y m (k))h j + α(w j (k - 1) - w j (k - 2)) Formula Seven;

[0105]

[0106] In the formula, ρ is the learning rate; α is the momentum factor.

[0107] Based on the above prediction method, such as Figure 4As shown in the figure, the present invention also provides a prediction system for greenhouse temperature and humidity based on limited sensors, including: a collection unit 404 for respectively collecting data of indoor reference points and outdoor reference points in the greenhouse through a limited number of sensors; a mapping unit 402 for establishing a multi-point temperature and humidity mapping in the greenhouse based on the data to obtain three-dimensional information of indoor temperature and humidity; a classification unit 403 for establishing a classification algorithm model and clustering the three-dimensional information to obtain a classification result; a training unit 404 for constructing a corresponding neural network model according to the classification result and training to obtain a trained model; and a prediction unit 405 for predicting according to the input reference point data by the trained model and outputting the temperature and humidity prediction result of virtual points in the greenhouse.

[0108] In some other embodiments of the present invention, embodiments of the present invention disclose an electronic device, which may include: one or more processors; a memory; a display; one or more applications; and one or more computer programs. The above-mentioned devices may be connected through one or more communication buses. Wherein the one or more computer programs are stored in the above-mentioned memory and configured to be executed by the one or more processors. The one or more computer programs include instructions, and the above-mentioned instructions may be used to execute as Figure 1 the respective steps in the corresponding embodiments.

[0109] Based on the above embodiments, the present invention also discloses a computer-readable storage medium, on which at least one computer program is stored. When the computer program is executed by a processor, it implements a prediction method for greenhouse temperature and humidity based on limited sensors in the foregoing embodiments.

[0110] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or equivalently replace some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting greenhouse temperature and humidity based on limited sensors, characterized in that: Includes steps: The data of indoor reference points, indoor virtual points and outdoor reference points of the greenhouse are collected respectively by a limited number of sensors; Based on the data, a mapping relationship between two reference points and a virtual point is established to obtain three-dimensional information of indoor temperature and humidity; Establishing a classification algorithm model, clustering the stereo information according to the above, and obtaining a classification result; Constructing a corresponding neural network model according to the classification results, and training to obtain a trained model; The trained model makes predictions based on the two reference point data input, and outputs the temperature and humidity prediction results of the indoor virtual point of the greenhouse.

2. The method according to claim 1, characterized in that The data include control history characteristics, crop growth characteristics and meteorological characteristics of indoor reference points and indoor virtual points, and meteorological characteristics of outdoor reference points.

3. The method according to claim 1, characterized in that: The control history features include one or a combination of any two or more of the top opening window, sunshade net, roller blind film opening and irrigation history data; And / or the meteorological characteristics of the indoor reference point, indoor virtual point or outdoor reference point include one or a combination of any two or more of temperature, humidity, light and wind speed and direction.

4. The method according to claim 1, characterized in that The method further includes preprocessing the collected data, wherein the preprocessing includes one or a combination of any two or more of the following: (1) analyzing and processing the temporal and spatial regularities of temperature and humidity on the data; (2) performing correlation analysis on the data; (3) using the Laida rule to process the data for outliers; (4) Performing data normalization on the data.

5. The method according to claim 1, characterized in that Establishing a classification algorithm model, clustering the stereo information, and obtaining classification results including: Classifying the three-dimensional information with the same spatial distance into one category through Euclidean clustering; The output results of Euclidean clustering are clustered again using the cosine angle function, from which data with the same curve shape are selected and classified into one category to obtain the classification results.

6. The method according to claim 1, characterized in that The neural network model is an RBF neural network model; and / or the mean absolute error MAE, mean absolute percentage error MAPE, and root mean square error RMSE are used to evaluate the model during the training process.

7. The method according to any one of claims 1 to 6, characterized in that: The indoor reference point of the greenhouse is the indoor center point, and all virtual points are set around the reference point.

8. A greenhouse temperature and humidity prediction system based on limited sensors, used in the method according to any one of claims 1 to 7, characterized in that: include: A collection unit, used to collect data of indoor reference points, indoor virtual points and outdoor reference points of the greenhouse respectively through a limited number of sensors; A mapping unit, used to establish a mapping relationship between two reference points and a virtual point based on the data to obtain three-dimensional information of indoor temperature and humidity; A classification unit, used to establish a classification algorithm model, cluster the three-dimensional information according to the clustering, and obtain a classification result; A training unit, used to construct a corresponding neural network model according to the classification result, and train to obtain a trained model; The prediction unit is used for the trained model to make predictions based on the two reference point data input, and output the temperature and humidity prediction results of the indoor virtual point of the greenhouse.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores a program executable on the processor, and when the program is executed by the processor, the electronic device implements the method according to any one of claims 1 to 7.

10. A readable storage medium, wherein a program is stored in the readable storage medium, characterized in that: When the program is executed, the method according to any one of claims 1 to 7 is implemented.

Images