Orchard frost protection system based on meteorological monitoring

Abstract

The invention discloses an orchard frost protection system based on meteorological monitoring, and the system comprises a sensor module, a main control module, a power supply module, a communication module, a monitoring module, a support, and a case. The power supply module provides power for the sensor module, the main control module, the communication module and the monitoring module, the main control module obtains data collected by the sensor module, abnormal values are removed through a DBSCAN algorithm, data continuity is guaranteed through a Lagrange interpolation algorithm, temperature change prediction results of different canopies of an orchard in a future preset time period are generated through a Bi-LSTM + Transformer algorithm, and the temperature change prediction results of the different canopies of the orchard in the future preset time period are obtained. And when the prediction result meets the frost judgment condition, early warning information is sent out through the communication module. Different frost judgment conditions are set for different fruit trees and growth stages, real-time monitoring and timely early warning are combined, fruit farmers are helped to take measures in advance, and yield reduction and quality decline caused by frost are remarkably reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural plant protection, and in particular to a frost protection system for orchards based on meteorological monitoring. BACKGROUND

[0002] In modern agricultural production, frost disasters have become one of the key factors restricting the economic benefits of orchards. Every year, frost leads to reduced yield and decreased fruit quality of fruit trees, causing huge economic losses to fruit farmers. Although the traditional meteorological monitoring system can provide meteorological data in a certain area, it still has significant shortcomings in the special scenario of orchards, especially in frost prediction at different heights of the orchard canopy.

[0003] The structure of the orchard canopy is complex, and the microclimate environment at different heights is significantly different. Due to the differences in heat exchange and air flow characteristics, the temperature variation laws of the upper, middle and lower parts of the canopy are different, and the traditional single-height meteorological data collection method cannot fully reflect the real temperature distribution inside the orchard. The data obtained based on conventional monitoring methods cannot accurately predict the frost risk at different canopy heights. The existing prediction model cannot effectively integrate multi-source heterogeneous data, and cannot solve the problem of spatial and temporal differences in frost prediction at different heights of the orchard canopy, resulting in large prediction result errors and failing to meet the actual needs of precise and fine frost protection in orchards. SUMMARY

[0004] The present application aims to solve the problem of insufficient frost prediction accuracy in existing technology and the inability to accurately reflect temperature changes at different canopy heights, and provides a frost protection system and method for orchards based on meteorological monitoring to achieve accurate prediction and timely protection of frost risk at different canopy heights in orchards.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions: In a first aspect, the present application provides a frost protection system for orchards based on meteorological monitoring, comprising a sensor module, a main control module, a power supply module, a communication module, a monitoring module, a bracket and a case; The main control module is connected to the sensor module, the power supply module and the communication module, the power supply module provides power for the sensor module, the main control module, the communication module and the monitoring module, the main control module obtains the data collected by the sensor module, removes outliers through the DBSCAN algorithm, ensures data continuity through the Lagrange interpolation algorithm, and generates temperature change prediction results for different canopy layers in the orchard within a preset time period through the Bi-LSTM+Transformer algorithm. When the prediction result meets the frost determination condition, an early warning message is sent through the communication module.

[0006] In an optional embodiment, the monitoring module comprises a monitoring camera holder, the monitoring camera holder is installed on the side surface of the support, and the surface of the support is further fixedly provided with a solar panel and a machine box; the sensor module, the main control module, the power supply module and the communication module are arranged in the interior of the machine box.

[0007] In an optional embodiment, the power supply module comprises a battery and a step-down unit, the step-down unit comprises a transformer and an interface component, the battery is electrically connected with the transformer, the interface component and the solar panel respectively, the sensor module and the communication module are connected with the power supply module through the interface component, and the step-down module is electrically connected with the main control module.

[0008] In an optional embodiment, the main control module comprises an edge computing MCU and a communication conversion unit, the edge computing MCU is connected with the communication conversion unit, the communication conversion unit is connected with the sensor module, the power supply module, the communication module and the monitoring module respectively, the sensor module comprises a temperature acquisition unit, a wind speed acquisition unit, a wind direction acquisition unit, a precipitation acquisition unit, a humidity acquisition unit, an altitude acquisition unit and a barometric pressure acquisition unit, the main control module is connected with a cloud server through a serial communication unit, the serial communication unit is connected with the edge computing MCU, and the sensor module is integrated in at least two sensor modules according to the monitoring parameter type.

[0009] In an optional embodiment, the monitoring module is used for monitoring the occurrence state of orchard frost in real time, and the installation height of the temperature acquisition unit 7 in the sensor module is determined through statistical analysis. The installation height determination method of the temperature acquisition unit 7 is as follows: the height parameters of fruit trees of the same age in the orchard are counted, the height distribution characteristics of the fruit trees of the same age in the orchard are calculated through the discrete degree, and at least three height intervals are divided to collect temperature parameters of different crown layers.

[0010] In a second aspect, the application provides an orchard frost protection method based on weather monitoring, which comprises the following steps: S1, performing DBSCAN calculation on the data values collected by the sensor module to identify and eliminate abnormal values; S2, performing Lagrange interpolation on the data after the abnormal values are eliminated to complete the eliminated abnormal values and ensure data continuity; S3, using a Bi-LSTM+Transformer algorithm to predict the future temperature change trend of different crown layers in the orchard; S4, when the predicted temperature meets the preset frost determination condition, the main control module sends an early warning information to a monitoring terminal.

[0011] In an optional embodiment, the specific steps of identifying and eliminating abnormal values in S1 comprise: S11, defining the neighborhood range of data points and calculating the spatial distance between each point and other points in the data set; S12, determine the core point based on the number of data points in the neighborhood, form data clusters by core point expansion, and determine data that is not included in any data cluster and is not a core point as an outlier.

[0012] In an optional embodiment, the data completion processing in step S2 adopts a polynomial interpolation algorithm, an interpolation function is constructed through known data points, and supplementary data at the outlier elimination position is calculated.

[0013] In an optional embodiment, step S3 comprises: S31, input historical environment data, the historical environment data comprising solar radiation, airflow state, precipitation condition, cloud cover, air humidity, temperature at different heights, geographic elevation and air pressure parameters; S32, standardize the historical environment data to unify the data scale; S33, use the Pearson correlation coefficient to sort the importance of all features to obtain the optimal feature set order; S34, construct a time series input window to generate model training samples; S35, train the sample data using a BiLSTM network to output temperature prediction results in the future preset time period.

[0014] In an optional embodiment, in step S4, the frost determination condition is set according to the low-temperature tolerance threshold of the fruit tree variety and the growth stage, and different fruit trees correspond to different temperature determination ranges in the bud stage, the flowering stage and the young fruit stage.

[0015] Compared with the prior art, the beneficial effects of the present application are: 1. The present application determines the installation height of the temperature sensor through statistical analysis, combines the meteorological data of different crown layers in the orchard collected by multi-dimensional sensors, comprehensively reflects the differences in crown layer microclimate, solves the problem that the traditional single height monitoring cannot capture the three-dimensional temperature distribution, and lays a data foundation for accurate prediction.

[0016] 2. The present application adopts the DBSCAN algorithm to eliminate outliers and Lagrange interpolation to complete data, effectively improves the data quality, and the Bi-LSTM+Transformer algorithm can effectively process time series data, accurately predict the temperature of different crown layers in the future preset time period, greatly reduces the prediction error, and overcomes the defects that the traditional model is difficult to handle the spatiotemporal differences.

[0017] 3. Different frost determination conditions are set for different fruit trees and growth stages, combined with real-time monitoring and timely warning, to realize fine protection, help fruit farmers to take measures in advance, significantly reduce the yield reduction and quality decline caused by frost, protect the economic benefits of the orchard, and have strong practical value and promotional significance. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a structural schematic diagram of the present application; Figure 2 is a structural diagram of the master control module proposed by the present application; Figure 3 is a flow chart of a fruit orchard frost protection method based on meteorological monitoring proposed by the present application.

[0019] The figure mark: 1, the first sensor module; 2, the second sensor module; 3, monitoring camera holder; 4, the case; 5, support; 6, solar panel; 7, temperature acquisition unit; 8, edge computing MCU; 9, communication conversion unit; 10, interface components; 11, serial communication unit. DETAILED DESCRIPTION

[0020] In order to more clearly understand the purpose, technical scheme and advantages of the present application, the present application is described and explained below in conjunction with the drawings and examples.

[0021] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the general meaning understood by a person with ordinary skill in the art to which the present application belongs. In the present application, "one", "a", "an", "the", "these" and similar words do not represent a quantitative limitation, and they can be singular or plural. In the present application, the terms "include", "contain", "have" and any variants thereof are intended to cover non-exclusive inclusion; for example, a process, method and system, product or device containing a series of steps or modules (units) are not limited to the listed steps or modules (units), but can include steps or modules (units) not listed, or can include other steps or modules (units) inherent to the process, method, product or device. In the present application, the words "connected", "connected", "coupled" and similar words do not limit to physical or mechanical connection, but can include electrical connection, whether direct or indirect. In the present application, "multiple" means two or more. The association between the associated objects is described by "and / or", which means that there can be three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. In general, the character " / " represents an "or" relationship between the associated objects. In the present application, the terms "first", "second", "third" and the like are only used to distinguish similar objects, and do not represent a specific order of the objects.

[0022] As shown in Figure 1 and Figure 2 A fruit orchard frost protection system based on meteorological monitoring includes a sensor module, a master control module, a power supply module, a communication module, a monitoring module, a support 5 and a case 4. The main control module is connected with the sensor module, the power supply module and the communication module respectively, the power supply module provides power for the sensor module, the main control module, the communication module and the monitoring module, the main control module obtains the data collected by the sensor module, removes the abnormal values through the DBSCAN algorithm, ensures the data continuity through the Lagrange interpolation algorithm, and then generates the temperature change prediction result of different canopy layers in the orchard within a preset time period through the Bi-LSTM+Transformer algorithm, when the prediction result meets the frost determination condition, the warning information is sent through the communication module.

[0023] Specifically, the monitoring module includes a monitoring camera holder 3, which is installed on the side surface of the support 5, and the surface of the support 5 is also fixedly provided with a solar panel 6 and a case 4; the sensor module, the main control module, the power supply module and the communication module are all arranged in the interior of the case 4.

[0024] Specifically, the power supply module includes a battery 13 and a voltage reduction unit, the voltage reduction unit includes a transformer 12 and an interface component 10, the battery 13 is electrically connected with the transformer 12, the interface component 10 and the solar panel 6 respectively, the sensor module and the communication module are connected with the power supply module through the interface component 10, and the voltage reduction module is electrically connected with the main control module.

[0025] Specifically, the main control module includes an edge computing MCU 8 and a communication conversion unit 9, the edge computing MCU 8 is connected with the communication conversion unit 9, the communication conversion unit 9 is connected with the sensor module, the power supply module, the communication module and the monitoring module respectively, the sensor module includes a temperature acquisition unit 7, a wind speed acquisition unit, a wind direction acquisition unit, a precipitation acquisition unit, a humidity acquisition unit, an altitude acquisition unit and a pressure acquisition unit, the main control module is connected with a cloud server through a serial communication unit, the serial communication unit is connected with the edge computing MCU 8, and the sensor module is integrated in at least two sensor modules according to the monitoring parameter type.

[0026] And in the embodiment, as shown in Figure 1 the sensor module is divided into two, including a first sensor module 1 and a second sensor module 2, and the two are installed at different parts of the support 5.

[0027] In addition, it should be noted that the embodiment also provides a method for determining the installation height of the temperature acquisition unit 7; Specifically, the monitoring module is used for monitoring the frost occurrence state of the orchard in real time, the installation height of the temperature acquisition unit 7 in the sensor module is determined through a statistical analysis method, and the installation height determination method of the temperature acquisition unit 7 is that: the height parameters of the same age fruit trees in the orchard are counted, the height distribution characteristics of the same age fruit trees in the orchard are calculated through the discrete degree, and at least three height intervals are divided to collect the temperature parameters of different canopy layers.

[0028] As Figure 3 shown, based on the basis of example one, the application also provides a weather monitoring based orchard frost protection method, comprising the following steps: S1, DBSCAN calculation is performed on the data values collected by the sensor module to identify and eliminate outliers; S2, Lagrange interpolation is performed on the data after the outliers are eliminated to complete the eliminated outliers and ensure data continuity; S3, using Bi-LSTM+Transformer algorithm to predict the future temperature change trend of different canopy layers of the orchard; S4, when the predicted temperature meets the preset frost determination condition, the main control module sends the warning information to the monitoring terminal.

[0029] Specifically, the specific steps of outlier identification and elimination in S1 include: S11, define the neighborhood range of data points, calculate the spatial distance of each point in the data set and other points; In this step, the formula for defining the neighborhood range of the input data is as follows: ; Among them, x i is the current data point to be judged, D is the entire data set (including all data points), x j is any other point in the data set, dist ( x i , x j ) is the set of all points in the neighborhood of point x i and x j is the Euclidean distance between them, epsilon is the neighborhood radius parameter, which controls the determination of "adjacent", N ε ( x i ) is the set of all points in the neighborhood of point x i . epsilon

[0030] And it should be noted that the Euclidean distance formula represents the Euclidean distance of two points in a multi-dimensional space: ; Among them, is the dimension of the data (such as two-dimensional or multi-dimensional feature space), is the two ​Data points in 3D space, and Points and In the The values ​​of each dimension.

[0031] S12. Determine the core point based on the number of data points in the neighborhood, and form a data cluster by expanding the core point. Data that is not included in any data cluster and is not a core point is determined to be an outlier.

[0032] Specifically, in this step, the core point is determined as follows: if the number of points in the neighborhood is greater than or equal to MinPts, then the point is considered a core point. ; Among them, | N ε ( x i | represents the number of data points in the neighborhood of point xi. MinPts The threshold is the minimum number of neighborhood points; Specifically, starting from a core point, points in its neighborhood are added to the current cluster; this process is repeated for all core points in the neighborhood until it can no longer be expanded. If a data point is neither a core point nor in the neighborhood of any core point, it is considered an outlier.

[0033] In step S2, the data completion process uses a polynomial interpolation algorithm to construct an interpolation function based on known data points and calculate the supplementary data for the outlier removal positions.

[0034] Specifically, step S2 is characterized by: constructing the Lagrange interpolation formula to solve for the unknowns, i.e., the values ​​to be interpolated. x， The specific calculation method is as follows: ; ; in, P n ( x )exist x The interpolation polynomial value at that point, x The unknown x-coordinate to be interpolated. x 0, x 1, ..., x n Given the x-coordinates of the data points. y 0, y 1, ..., y n The x-coordinate of the known data point corresponds to the y-coordinate value.

[0035] In addition, step S3 includes: S31, input historical environment data, the historical environment data including solar radiation, airflow state, precipitation condition, cloud cover, air humidity, temperature at different heights, geographical elevation and air pressure parameters; S32, standardize the historical environment data, unify the data scale; Specifically, normalize the input historical data of each element, transform the data into a certain interval or a certain standard value range, ensure that all input variables are in the same scale, and facilitate network training; For example, if the optimal interval is assumed to be [ c, d ], then the specific range is as follows: ;

[0036] Among them, is the lower limit that cannot be tolerated, is the upper limit that cannot be tolerated; Among them, ; Among them, is the vector standard sample transformation matrix, is a sample number, is the average value of the total sample number, is the standard deviation of the total sample number.

[0037] S33, use the Pearson correlation coefficient to sort the importance of all features, and obtain the optimal feature set order; Here, the specific calculation method is as follows: ;

[0038] Among them, x i is the specific value of the meteorological factor such as total solar radiation, wind speed and precipitation, y i is the specific value of the temperature at different heights, , are the mean values of variables X and Y , respectively, n is the sample number.

[0039] S34, construct a time sequence input window and generate model training samples; Specifically, a sliding window mechanism of input sequence is constructed, and for each time point t , the generated input sample s t : ;

[0040] Among them, For t the input feature vector at the moment, including total solar radiation, wind speed and direction, precipitation, cloud amount, relative humidity, and in this embodiment, three different height temperatures, altitude and pressure, etc., a total of D =11 features; In addition, t denotes the total length of the time series, L denotes the length of the sliding window.

[0041] S35, training the sample data using the BiLSTM network, outputting the temperature prediction result in the future preset time length.

[0042] The following steps are also included in step S35: S351, divide the training sample into training set and validation machine according to 7:3, import BiLSTM network training, the calculation method is: ; ; ; ; ; ; ; ; ; ; ; wherein, , is the forward / reverse hidden state of the previous time step, is the forward / reverse input gate, is the forward / reverse forget gate, is the forward / reverse output gate, is the memory cell state, is the sigmoid activation function, is the hyperbolic tangent activation function, is the weight matrix of each gate, is the bias term of each gate, h t is the result of splicing the forward and backward hidden states.

[0043] S352, encode the data after BiLSTM, specifically, after BiLSTM network training processing, get the output , d h the hidden layer dimension of LSTM, the output of BiLSTM H i position encoding, let the position encoding function be PosEnc, the encoded data : ; S353, self-attention calculation is performed to obtain an output; Specifically, self-attention calculation is performed to calculate query Q , key K , and value V , specifically: ; wherein, is a learnable weight matrix, d k is the dimension of the key and the query; In addition, the output is obtained through self-attention calculation, and the attention weight formula is: ; S353, the self-attention output is input into multiple heads for calculation, and after splicing, a linear transformation is performed; Specifically, the self-attention output O is input into m different heads for calculation, and the outputs of the m heads are spliced together, and the dimension is : ; ; Finally, a linear transformation is performed to obtain , and ; wherein, W O is an output weight matrix.

[0044] S354, the multi-head attention output is input into a feedforward neural network, the output feature vector at the last time step is taken, and three predicted height temperatures are obtained through a regression layer; Specifically, the multi-head attention output is input into a feedforward neural network, and the output is obtained through FFN: ; wherein, is a feedforward network weight matrix, b 1,b 2 is a bias vector, d ff is the dimension of the intermediate layer of the FFN.

[0045] In addition, taking the feature vector FFFN of the FFN output F at the last time step F L , a prediction is made by a regression layer, specifically: ; wherein, W reg is a regression layer weight matrix, b reg is a regression layer bias vector, is the predicted temperature value of the three heights.

[0046] It should be further pointed out that in step S4, the frost determination condition is set according to the low-temperature tolerance threshold of the fruit tree variety and the growth stage, and different fruit trees correspond to different temperature determination ranges in the bud stage, the flowering stage and the young fruit stage.

[0047] In addition, when the predicted temperature meets the preset frost determination condition, the MCU8 sends the result to the detection platform through edge computing.

[0048] For example, in this embodiment, the judgment range of part of the fruit trees is illustrated with reference to the following table:

[0049] In the table: Apple tree: -2.8~ -3.8℃ in the bud stage (late March~ early April), -1.6~ -2.2℃ in the flowering stage (early April~ mid-May), and -1.1~ -2.2℃ in the young fruit stage (early May~ early June); Peach tree: -2.8~ -3.9℃ in the bud stage (mid-March~ early April), -2.0~ -1.5℃ in the flowering stage (late March~ mid-April), and -1.8~ -1.2℃ in the young fruit stage (late April~ early May); Pear tree: -3.9~ -8℃ in the bud stage (early March~ early April), -2.5~ -3.4℃ in the flowering stage (mid-March~ mid-April), and -0.7~ -2.5℃ in the young fruit stage (late April~ early June).

[0050] Pomegranate tree: 0℃ in the bud stage (late March~ early April), -2~ -5℃ in the flowering stage (mid-late May~ early-mid June), and 0℃ in the young fruit stage (late June~ early July).

[0051] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods of the various embodiments or some parts of the embodiments.

[0052] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An orchard frost protection system based on weather monitoring, characterized in that, The system comprises a sensor module, a main control module, a power supply module, a communication module, a monitoring module, a support and a case; The main control module is connected with the sensor module, the power supply module and the communication module, the power supply module provides power for the sensor module, the main control module, the communication module and the monitoring module, the main control module acquires data collected by the sensor module, removes abnormal values through a DBSCAN algorithm, ensures data continuity through a Lagrange interpolation algorithm, and generates a prediction result of temperature change of different canopy layers in the orchard in a preset time period in the future through a Bi-LSTM+Transformer algorithm, and sends a warning information to the monitoring terminal when the prediction result meets a frost determination condition.

2. An orchard frost protection system based on weather monitoring according to claim 1, characterized in that, The monitoring module comprises a monitoring camera holder, the monitoring camera holder is installed on the side surface of the support, and the surface of the support is further fixedly provided with a solar panel and a case; the sensor module, the main control module, the power supply module and the communication module are arranged in the case.

3. An orchard frost protection system based on weather monitoring according to claim 2, characterized in that, The power supply module comprises a battery and a voltage reduction unit, the voltage reduction unit comprises a transformer and an interface component, the battery is electrically connected with the transformer, the interface component and the solar panel, the sensor module and the communication module are connected with the power supply module through the interface component, and the voltage reduction module is electrically connected with the main control module.

4. An orchard frost protection system based on weather monitoring according to claim 3, characterized in that, The main control module comprises an edge computing MCU and a communication conversion unit, the edge computing MCU is connected with the communication conversion unit, the communication conversion unit is connected with the sensor module, the power supply module, the communication module and the monitoring module, the sensor module comprises a temperature acquisition unit, a wind speed acquisition unit, a wind direction acquisition unit, a precipitation acquisition unit, a humidity acquisition unit, an altitude acquisition unit and an air pressure acquisition unit, the main control module is connected with a cloud server through a serial communication unit, the serial communication unit is connected with the edge computing MCU, and the sensor module is integrated in at least two sensor modules according to monitoring parameter types.

5. An orchard frost protection system based on weather monitoring according to claim 4, characterized in that, The monitoring module is used for monitoring the frost occurrence state of the orchard in real time, and the installation height of the temperature acquisition unit 7 in the sensor module is determined through a statistical analysis method. The installation height determination method of the temperature acquisition unit 7 is as follows: the height parameters of fruit trees of the same age in the orchard are counted, the height distribution characteristics of the fruit trees of the same age in the orchard are calculated according to the discrete degree, and at least three height intervals are divided to collect temperature parameters of different canopy layers.

6. A weather monitoring-based orchard frost protection method, characterized by, The method is applied to the system of any one of claims 1-5, and comprises the following steps: S1, performing DBSCAN calculation on the data values collected by the sensor module to identify and remove abnormal values; S2, performing Lagrange interpolation on the data after removing the abnormal values, complementing the removed abnormal values, and ensuring data continuity; S3, using a Bi-LSTM+Transformer algorithm to predict the future temperature change trend of different canopy layers in the orchard; S4, when the predicted temperature meets a preset frost determination condition, the main control module sends a warning information to a monitoring terminal.

7. A method of orchard frost protection based on weather monitoring according to claim 6, characterized in that, The specific steps of identifying and removing abnormal values in S1 comprise: S11, defining the neighborhood range of data points, and calculating the spatial distance between each point and other points in the data set; S12, determining the core point based on the number of data points in the neighborhood, forming a data cluster through the core point expansion, and determining the data that is not included in any data cluster and is not a core point as an abnormal value.

8. A method of orchard frost protection based on weather monitoring according to claim 6, characterized in that, The data completion processing in step S2 adopts a polynomial interpolation algorithm, and the known data points are used to construct an interpolation function to calculate the supplementary data of the abnormal value elimination position.

9. A method of orchard frost protection based on weather monitoring according to claim 6, characterized in that, Step S3 includes: S31, input historical environment data, the historical environment data includes solar radiation, airflow state, precipitation condition, cloud cover, air humidity, temperature at different heights, geographical elevation and air pressure parameters; S32, standardizing the historical environment data to unify the data scale; S33, using the Pearson correlation coefficient to sort the importance of all features to obtain the optimal feature set order; S34, constructing a time series input window to generate model training samples; S35, using a BiLSTM network to train the sample data to output the temperature prediction result in the future preset time length.

10. A method for orchard frost protection based on weather monitoring according to claim 6, characterized in that, In step S4, the frost determination condition is set according to the low-temperature tolerance threshold of the fruit tree variety and the growth stage. Different fruit trees correspond to different temperature determination ranges in the bud stage, flowering stage and young fruit stage.

Images