A tobacco yield prediction system and method based on precise meteorology

Abstract

The embodiment of this specification discloses a tobacco yield prediction system and method based on precision meteorology, which relates to the field of data analysis technology. The method includes: obtaining the yield influencing parameters of the target planting area in the current planting cycle; inputting the yield influencing parameters into the target yield prediction model for processing; obtaining the predicted tobacco yield according to the output result of the target yield prediction model; wherein the target yield prediction model includes a preprocessing layer and a neural network layer, the preprocessing layer is used to preprocess the yield influencing parameters, the neural network layer includes an input layer, a first processing layer, a second processing layer and an output layer, the first processing layer is used to perform feature extraction processing to obtain the time series feature fusion data of the influencing parameters; the second processing layer is used to extract the time series correlation between the yield influencing parameters to obtain the time series correlation feature data; the output layer is used to predict based on the time series correlation feature data and the time series feature fusion data of the influencing parameters to obtain the predicted tobacco yield.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular, to a tobacco yield prediction system and method based on precision meteorology. Background Art

[0002] In modern agricultural production, tobacco is an important cash crop, and its yield forecast is of great significance to growers, enterprises and governments. Traditional tobacco yield forecasting mainly relies on experience and intuition, lacks scientific basis, and leads to low forecast accuracy, which is difficult to meet the needs of modern agricultural development.

[0003] In recent years, with the development of precision meteorology and data analysis technology, tobacco yield prediction methods based on meteorological data and planting data have gradually become a research hotspot. These methods construct more accurate and reliable yield prediction models by collecting and analyzing meteorological data (such as temperature, rainfall, relative humidity, etc.), soil data (such as soil type, soil moisture, soil nutrient content, etc.) and planting data (such as planting area, transplanting time, planting density, fertilizer application amount, fertilizer type, etc.). However, these methods still face some challenges. For example, due to the low distribution density of meteorological monitoring points, there may be a certain deviation between the actual meteorological data of the target planting area and the forecast data of the meteorological station; for example, the same meteorology may have different degrees of influence on tobacco yield in different growth periods; for example, the combination of different meteorological data and different planting data may also have different effects on tobacco yield. Therefore, how to effectively analyze the relationship between meteorological data and tobacco yield and improve the accuracy and practicality of tobacco yield prediction is an urgent problem to be solved in current research. Summary of the invention

[0004] To solve at least one of the above technical problems, one aspect of the embodiments of this specification provides a tobacco yield prediction method based on precise meteorology, the method comprising:

[0005] Obtaining yield influencing parameters of a target planting area in a current planting cycle, wherein the yield influencing parameters include first time series data corresponding to meteorological data, second time series data corresponding to soil data, and third time series data corresponding to planting data, wherein the meteorological data include first meteorological data obtained from a meteorological forecasting platform and second meteorological data obtained through an agricultural meteorological monitor of the target planting area;

[0006] Inputting the yield influencing parameters into a target yield prediction model for processing;

[0007] The predicted tobacco yield of the target planting area in the current planting cycle is obtained according to the output result of the target yield prediction model; the target yield prediction model includes a preprocessing layer and a neural network layer, wherein the preprocessing layer is used to preprocess the yield influencing parameters, and the neural network layer is obtained by iterative training based on historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles; the neural network layer includes an input layer, a first processing layer, a second processing layer and an output layer connected in sequence, the first processing layer is used to perform feature extraction processing on multiple time series data contained in the yield influencing parameters to obtain influencing parameter time series feature fusion data; the second processing layer is used to extract the time series correlation between the influencing parameter time series feature fusion data corresponding to each yield influencing parameter to obtain time series correlation feature data; the output layer is used to predict based on the time series correlation feature data and the influencing parameter time series feature fusion data to obtain the predicted tobacco yield.

[0008] In some embodiments, the preprocessing of the yield influencing parameters includes:

[0009] The information contained in the first time series data, the second time series data and the third time series data is standardized and converted according to preset rules to obtain the converted first time series data, the converted second time series data and the converted third time series data; and the missing values ​​in the converted first time series data and the converted second time series data are filled by the mean filling method, and the missing values ​​in the converted third time series data are filled by the fixed value filling method to obtain the first target time series data, the second target time series data and the third target time series data.

[0010] In some embodiments, the first target time series data includes preprocessed first meteorological data and preprocessed second meteorological data, and the preprocessing of the yield influencing parameters further includes:

[0011] Calculating the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in a past time period;

[0012] Predicting the predicted meteorological parameters of the preprocessed second meteorological data in the future time period according to the Pearson correlation coefficient and the meteorological parameters of the preprocessed first meteorological data in the future time period;

[0013] Target meteorological data are obtained according to the preprocessed second meteorological data and the predicted meteorological parameters of the preprocessed second meteorological data in the future time period, and the time series data corresponding to the target meteorological data are used as the first target time series data.

[0014] In some embodiments, the first processing layer includes a feature extraction unit and a feature fusion unit; wherein,

[0015] The feature extraction unit is used to perform convolution calculation on each time series data included in the yield influencing parameter, so as to extract corresponding feature data from each time series data;

[0016] The feature fusion unit is used to fuse the feature data extracted from each time series data with the corresponding original time series data to obtain the influencing parameter time series feature fusion data.

[0017] In some embodiments, extracting the temporal correlation between the temporal feature fusion data of the influencing parameters corresponding to each yield influencing parameter to obtain the temporal correlation feature data includes:

[0018] The mutation points in the time series feature fusion data of the influencing parameters are obtained by using the BG sequence segmentation algorithm, and the mutation time series corresponding to each yield influencing parameter is determined based on the time corresponding to each mutation point;

[0019] The sum of DTW distances between mutation points with the same sequence number in the mutation time series corresponding to any two yield influencing parameters is calculated, and the time series correlation feature data between the any two yield influencing parameters is determined based on the sum of the DTW distances.

[0020] In some embodiments, calculating the sum of DTW distances between mutation points with the same sequence number in the mutation time series corresponding to any two yield influencing parameters includes:

[0021] When the number of mutation points in the mutation time series corresponding to any two yield influencing parameters is inconsistent, the mutation points are padded based on the time corresponding to the last mutation point in the one with fewer mutation points, and the sum of the DTW distances between the mutation points with the same sequence number is calculated based on the padded data.

[0022] In some embodiments, the first processing layer further includes a pooling processing unit, and the pooling processing unit is used to perform PCA dimensionality reduction processing on the influencing parameter time series feature fusion data;

[0023] The extracting of the temporal correlation between the temporal feature fusion data of the influencing parameters corresponding to each yield influencing parameter includes: extracting the temporal correlation of the temporal feature fusion data of the influencing parameters after dimensionality reduction processing corresponding to each yield influencing parameter to obtain the temporal correlation feature data.

[0024] In some embodiments, the output layer includes a fully connected layer, the fully connected layer includes a plurality of neurons, each neuron is connected to each parameter in the time series associated feature data and the influencing parameter time series feature fusion data through a weight;

[0025] The iterative training based on the historical yield influencing parameters and historical tobacco leaf yield data of multiple historical planting cycles includes: optimizing the weights through a back propagation algorithm to minimize the error between the predicted tobacco leaf yield and the actual tobacco leaf yield;

[0026] The prediction based on the time series correlation feature data and the influencing parameter time series feature fusion data is performed to obtain the predicted tobacco leaf yield, including: predicting the standard yield and reduced yield based on the time series correlation feature data and the influencing parameter time series feature fusion data, and obtaining the predicted tobacco leaf yield based on the standard yield and reduced yield.

[0027] Another aspect of the embodiments of this specification also provides a tobacco yield prediction system based on precise meteorology, the system comprising:

[0028] A yield influencing parameter acquisition module, used to acquire the yield influencing parameters of the target planting area in the current planting cycle, wherein the yield influencing parameters include first time series data corresponding to meteorological data, second time series data corresponding to soil data, and third time series data corresponding to planting data, wherein the meteorological data include first meteorological data acquired from a meteorological forecasting platform and second meteorological data acquired through an agricultural meteorological monitor of the target planting area;

[0029] A tobacco yield prediction module is used to input the yield influencing parameters into a target yield prediction model for processing, and obtain the predicted tobacco yield of the target planting area in the current planting cycle according to the output result of the target yield prediction model; the target yield prediction model includes a preprocessing layer and a neural network layer, wherein the preprocessing layer is used to preprocess the yield influencing parameters, and the neural network layer is obtained by iterative training based on historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles; the neural network layer includes an input layer, a first processing layer, a second processing layer and an output layer connected in sequence, the first processing layer is used to perform feature extraction processing on multiple time series data contained in the yield influencing parameters to obtain influencing parameter time series feature fusion data; the second processing layer is used to extract the time series correlation between the influencing parameter time series feature fusion data corresponding to each yield influencing parameter to obtain time series correlation feature data; the output layer is used to predict based on the time series correlation feature data and the influencing parameter time series feature fusion data to obtain predicted tobacco yield.

[0030] In some embodiments, the pretreatment layer is specifically used for:

[0031] Performing standardized conversion on the information included in the first time series data, the second time series data, and the third time series data according to a preset rule to obtain converted first time series data, converted second time series data, and converted third time series data;

[0032] The missing values ​​in the converted first time series data and the converted second time series data are filled by using a mean filling method, and the missing values ​​in the converted third time series data are filled by using a fixed value filling method, so as to obtain first target time series data, second target time series data and third target time series data; wherein the first target time series data includes the preprocessed first meteorological data and the preprocessed second meteorological data, and the preprocessing layer is further used for:

[0033] Calculating the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in a past time period;

[0034] Predicting the predicted meteorological parameters of the preprocessed second meteorological data in the future time period according to the Pearson correlation coefficient and the meteorological parameters of the preprocessed first meteorological data in the future time period;

[0035] Target meteorological data are obtained according to the preprocessed second meteorological data and the predicted meteorological parameters of the preprocessed second meteorological data in the future time period, and the time series data corresponding to the target meteorological data are used as the first target time series data.

[0036] The tobacco yield prediction system and method based on precise meteorology provided in the embodiments of this specification may bring beneficial effects including at least:

[0037] (1) Tobacco yield is predicted by combining the first meteorological data obtained from the meteorological forecast platform, the second meteorological data obtained by the agricultural meteorological monitor in the target planting area, the soil data obtained by the soil sensor, and the planting data obtained by the planting management record. Not only can the accuracy of the meteorological data used in the yield prediction process be improved according to the correlation between the first meteorological data and the second meteorological data, but also the time series correlation between various yield influencing parameters can be extracted as the influencing factor of yield prediction, thereby improving the accuracy of tobacco yield prediction results to a certain extent;

[0038] (2) By predicting the predicted meteorological parameters of the second meteorological data in the future time period according to the correlation between the first meteorological data and the second meteorological data, and predicting the tobacco leaf yield based on this, the accuracy of the meteorological data used in the yield prediction process can be improved to a certain extent, thereby improving the accuracy of subsequent analysis and obtaining a more accurate and reliable tobacco leaf yield prediction result;

[0039] (3) By calculating the time series correlation characteristic data between various meteorological influencing parameters, the relationship characteristics of various factors affecting tobacco leaf yield at the time nodes are reflected, and the time series correlation characteristic data is used as a calculation basis for tobacco leaf yield prediction. The different effects of different influencing factors on tobacco leaf yield at different time nodes can be taken into account, thereby further improving the accuracy of tobacco yield prediction results.

[0040] Additional features will be described in part in the following description. For those skilled in the art, it will become apparent by reviewing the following and accompanying drawings, or it may be understood by the production or operation of the examples. The features of this specification may be implemented and obtained by practicing or using various aspects of the methods, tools, and combinations described in the following detailed examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0042] Figure 1 is an exemplary module diagram of a tobacco yield prediction system based on precise meteorology according to some embodiments of this specification;

[0043] Figure 2 is an exemplary structural diagram of a target yield prediction model according to some embodiments of this specification;

[0044] Figure 3 is an exemplary structural diagram of a first processing layer according to some embodiments of this specification;

[0045] Figure 4 It is an exemplary flow chart of the tobacco leaf yield prediction method based on precise meteorology shown in some embodiments of this specification. DETAILED DESCRIPTION

[0046] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0047] It should be understood that the "system", "device", "unit" and / or "module" used in this specification is a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0048] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0049] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0050] The tobacco yield prediction system and method based on precise meteorology provided in the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0051] Figure 1 Schematic diagram of a module of a tobacco yield prediction system based on precise meteorology according to some embodiments of this specification. Figure 1 The tobacco yield prediction system 100 based on precision meteorology shown can be implemented in software and / or hardware. For example, it can be configured in the form of software and / or hardware to a processing device and / or a terminal device to process the yield influencing parameters of the current planting cycle to obtain the corresponding predicted tobacco yield.

[0052] Reference Figure 1 In an embodiment of the present application, a tobacco yield prediction system 100 based on precise meteorology may include a yield influencing parameter acquisition module 110 and a tobacco yield prediction module 120 .

[0053] Among them, the yield influencing parameter acquisition module 110 can be used to obtain the yield influencing parameters of the target planting area in the current planting cycle, and the yield influencing parameters include first time series data corresponding to meteorological data, second time series data corresponding to soil data, and third time series data corresponding to planting data. The meteorological data includes first meteorological data obtained from a meteorological forecasting platform and second meteorological data obtained through an agricultural meteorological monitor in the target planting area.

[0054] Specifically, in an embodiment of the present application, the first meteorological data can be obtained in real time from a meteorological forecasting platform, and the second meteorological data can be collected by an agricultural meteorological monitor installed in the target planting area. These meteorological data may include, but are not limited to, one or more of the key meteorological indicators such as temperature, humidity, wind speed, wind direction, weather type (such as sunny, cloudy, rainy, snowy, hail, strong wind, etc.), precipitation, sunshine duration, etc. Soil data may include soil moisture, soil temperature, soil pH value, soil nutrient content, etc., and these soil data can usually be obtained through soil sensors. Planting data can include information such as planting area, planting density, planting time, fertilizer application amount, irrigation amount, etc., and these data can be collected through planting management records.

[0055] In the embodiment of the present application, the above-mentioned meteorological data, soil data and planting data can be combined with each corresponding data acquisition time to form corresponding time series data. For example, by integrating meteorological data acquired multiple times and their acquisition time, the first time series data corresponding to the meteorological data can be obtained, by integrating soil data acquired multiple times and their acquisition time, the second time series data corresponding to the soil data can be obtained, and similarly, by integrating planting data acquired multiple times and their acquisition time, the first time series data corresponding to the planting data can be obtained.

[0056] The tobacco yield prediction module 120 can be used to input the yield influencing parameters into the target yield prediction model for processing, and obtain the predicted tobacco yield of the target planting area in the current planting cycle based on the output result of the target yield prediction model.

[0057] Figure 2 is an exemplary structural diagram of a target yield prediction model according to some embodiments of this specification. Figure 2 In the embodiment of the present application, the target yield prediction model 200 includes a preprocessing layer 210 and a neural network layer 220, wherein the preprocessing layer 210 can be used to preprocess the yield influencing parameters, and the neural network layer 220 can be obtained by iterative training based on the historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles. The neural network layer 220 includes an input layer 221, a first processing layer 222, a second processing layer 223 and an output layer 224 connected in sequence, the first processing layer 222 can be used to perform feature extraction processing on the multiple time series data contained in the yield influencing parameters to obtain the time series feature fusion data of the influencing parameters; the second processing layer 223 can be used to extract the time series correlation between the time series feature fusion data of the influencing parameters corresponding to each yield influencing parameter to obtain the time series correlation feature data; the output layer 224 can be used to predict based on the time series correlation feature data and the time series feature fusion data of the influencing parameters to obtain the predicted tobacco yield.

[0058] Figure 3 is an exemplary structural diagram of the first processing layer according to some embodiments of this specification. Figure 3 In some embodiments of the present application, the first processing layer 222 may include a feature extraction unit 2221 and a feature fusion unit 2222. The feature extraction unit 2221 may be used to perform convolution calculation on each time series data included in the yield influencing parameter to extract corresponding feature data from each time series data; the feature fusion unit 2222 may be used to fuse the feature data extracted from each time series data with the corresponding original time series data to obtain the influencing parameter time series feature fusion data.

[0059] Continue to refer to Figure 3 In some embodiments of the present application, the first processing layer 222 may further include a pooling processing unit 2223, and the pooling processing unit 2223 may be used to perform PCA dimensionality reduction processing on the influencing parameter time series feature fusion data.

[0060] For more details about the above-mentioned target yield prediction model 200 and each module in the tobacco yield prediction system 100 based on precise meteorology, please refer to other places in this specification (for example, Figure 4 part and its related description), which will not be discussed in detail here.

[0061] It should be pointed out that in the embodiment of the present application, tobacco leaf yield is predicted by combining the first meteorological data obtained from the meteorological forecast platform, the second meteorological data obtained by the agricultural meteorological monitor in the target planting area, the soil data obtained by the soil sensor, and the planting data obtained by the planting management record. It can not only improve the accuracy of the meteorological data used in the yield prediction process based on the correlation between the first meteorological data and the second meteorological data, but also can extract the time series correlation between each yield influencing parameter as an influencing factor for yield prediction, thereby improving the accuracy of the tobacco leaf yield prediction results to a certain extent.

[0062] Figure 4 is an exemplary flow chart of a tobacco yield prediction method based on precision meteorology according to some embodiments of this specification. In some embodiments, the tobacco yield prediction method 300 based on precision meteorology can be executed by processing logic, which can include hardware (e.g., circuits, dedicated logic, programmable logic, microcode, etc.), software (instructions running on a processing device to perform hardware simulation), etc. or any combination thereof. In some embodiments, Figure 4One or more operations in the flowchart of the tobacco leaf yield prediction method 300 based on precise meteorology can be implemented by a processing device and / or a terminal device. For example, the tobacco leaf yield prediction method 300 based on precise meteorology can be stored in a storage device in the form of a computer program or instruction, and called and / or executed by a processing device and / or a terminal device.

[0063] Reference Figure 4 In some embodiments, the tobacco yield prediction method 300 based on precise meteorology may include:

[0064] Step S310 , obtaining the yield influencing parameters of the target planting area in the current planting cycle. In some embodiments, step S310 may be performed by the yield influencing parameter obtaining module 110 .

[0065] In this specification, the target planting area can be understood as the planting area where tobacco leaf yield prediction needs to be performed, and the planting area can be any specific geographical area, such as a piece of farmland, a village, a county, a province, etc. The yield influencing parameters can be understood as various factors that affect tobacco leaf yield. In the embodiment of the present application, the yield influencing parameters may include meteorological data, soil data, and planting data, wherein the meteorological data includes first meteorological data obtained from a meteorological forecasting platform and second meteorological data obtained through an agricultural meteorological monitor in the target planting area.

[0066] Specifically, in an embodiment of the present application, the first meteorological data can be obtained in real time from a meteorological forecasting platform, and the second meteorological data can be collected by an agricultural meteorological monitor installed in the target planting area. These meteorological data may include, but are not limited to, one or more of the key meteorological indicators such as temperature, humidity, wind speed, weather type (such as sunny, cloudy, rainy, snowy, hail, strong wind, etc.), precipitation, and sunshine duration. Soil data may include soil moisture, soil temperature, soil pH value, soil nutrient content, etc., and these soil data can usually be obtained through soil sensors. Planting data can include information such as planting area, planting density, planting time, fertilizer application amount, irrigation amount, etc., and these data can be collected through planting management records.

[0067] In the embodiment of the present application, the above-mentioned meteorological data, soil data and planting data can be combined with each corresponding data acquisition time to form corresponding time series data. For example, by integrating meteorological data acquired multiple times and their acquisition time, the first time series data corresponding to the meteorological data can be obtained, by integrating soil data acquired multiple times and their acquisition time, the second time series data corresponding to the soil data can be obtained, and similarly, by integrating planting data acquired multiple times and their acquisition time, the first time series data corresponding to the planting data can be obtained.

[0068] The current planting cycle can be understood as the entire time period from the beginning of planting to the current time. It should be noted that in the embodiment of the present application, the current time can be any time point after the tobacco leaves enter the harvest period, or any time point before the tobacco leaves enter the harvest period. If the current time is a time point after the tobacco leaves enter the harvest period, it can be used to guide growers to choose a suitable date for tobacco leaf harvesting, thereby avoiding tobacco leaf production reduction caused by bad weather; if the current time is a time point before the tobacco leaves enter the harvest period, it can be used to analyze the yield of tobacco leaves planted this season.

[0069] It should be noted that in the embodiment of the present application, the accuracy of the aforementioned yield-influencing parameters may vary to a certain extent according to the size of the target planting area. For example, when the target planting area is a piece of farmland, the corresponding yield-influencing parameters include the first meteorological data obtained from the meteorological forecasting platform, the second meteorological data obtained by the agricultural meteorological monitor installed near the farmland, the soil data obtained by monitoring the farmland through the soil sensor, and the planting data for tobacco leaf planting management of the farmland; when the target planting area is a village or a county, the meteorological data, soil data and planting data may refer to the average level or average data of the corresponding area (the average level or average data can be calculated by averaging the data that can be obtained).

[0070] For example, in some embodiments of the present application, the time series data corresponding to the first meteorological data can be expressed as:

[0071] ;

[0072] in, arrive For the The time of obtaining weather data, For the 1st~ x Meteorological parameters, the meteorological data may include but are not limited to the above-mentioned temperature, humidity, wind speed, weather type, precipitation, sunshine duration, etc.

[0073] The time series data corresponding to the second meteorological data can be expressed as:

[0074] ;

[0075] in, arrive For the The time of obtaining weather data, For the 1st~ x It should be noted that, in the embodiment of the present application, the acquisition time granularity of the first meteorological data and the second meteorological data can be the same. In other words, the first meteorological data and the second meteorological data have the same acquisition time granularity. arrive is the same value, the first meteorological data The second meteorological data They are the same type of data. However, due to the different ways of obtaining them (different data accuracy), there may be some deviations.

[0076] The second time series data corresponding to the above soil data can be expressed as:

[0077] ;

[0078] in, arrive For the 1st~ m Soil data acquisition time, For the 1st~ y soil parameters, which may include but are not limited to the above-mentioned soil moisture, soil temperature, soil pH value, soil nutrient content, etc. It should be noted that in the embodiment of the present application, arrive The time granularity can be arrive The time granularity is the same or different. Similarly, the number of data types of meteorological parameters x Number of data types related to soil parameters y They may also be the same or different.

[0079] The third time series data corresponding to the above planting data can be expressed as:

[0080] ;

[0081] in, arrive For the 1st~ n The planting data acquisition time, The 1st to zth planting data may include planting management behaviors and corresponding quantitative data. For example, the planting behaviors may include transplanting, fertilizing, irrigation, etc., and the corresponding quantitative data may include planting area, planting density, fertilizer application amount, irrigation amount, etc. These data can be recorded by entering into a related system, and these data can be exported through the system to obtain the third time series data corresponding to the planting data. It should be noted that in the embodiment of the present application, arrive The time intervals between two adjacent data can be the same or different (that is, the implementation intervals between two adjacent planting management behaviors can be the same or different).

[0082] For example only, in some embodiments, They represent transplanting, fertilization and irrigation respectively. Represents the planting area, planting density, fertilizer application amount, and irrigation amount respectively. Transplanting is carried out at a certain time point, the planting area is 2 mu, the irrigation volume is 100 cubic meters, and the planting density is 1,000 plants per mu. The corresponding time series data is: , where NULL means the corresponding data position is empty (i.e. no relevant data). Fertilization was carried out at a time point, and the amount of fertilizer applied was 50 kg per mu (a total of 100 kg for an area of ​​2 mu), then the corresponding time series data is: It should be noted that, in some embodiments, the time series data corresponding to the above planting data may also include other data, such as fertilizer type and the like.

[0083] Step S320, inputting the yield influencing parameters into the target yield prediction model for processing. In some embodiments, step S320 can be performed by the tobacco yield prediction module 120.

[0084] After obtaining the time series data corresponding to the above-mentioned yield influencing parameters, they can be input into the target yield prediction model for processing, so as to predict the tobacco leaf yield of the target planting area in the current planting cycle.

[0085] Reference Figure 2 In the embodiment of the present application, the target yield prediction model 200 includes a preprocessing layer 210 and a neural network layer 220, wherein the preprocessing layer 210 can be used to preprocess the yield influencing parameters. Specifically, in the embodiment of the present application, the time series data corresponding to each yield influencing parameter obtained in the aforementioned process may contain non-quantitative data (such as text data, etc.), which may not be directly recognized in the subsequent process, so it is necessary to perform standardized conversion first. Exemplarily, in some embodiments, the wind direction and weather type (such as sunny, cloudy, rainy, snowy, hail, strong wind, etc.) in the first time series data corresponding to the meteorological data can be converted into corresponding numerical values ​​(for example, wind direction can be represented by 1~8, and weather type can be represented by 1~10), wherein the conversion rule can be preconfigured. For example, in some embodiments, the data mapping table corresponding to the conversion rule can be preconfigured, and then the information contained in the aforementioned first time series data is converted according to the data mapping table to obtain the converted first time series data.

[0086] Similarly, the non-quantized data contained in the second time series data and the third time series data can be converted to obtain the converted second time series data and the converted third time series data. Exemplarily, in some embodiments, the planting behavior (such as transplanting, fertilizing, and irrigation) in the third time series data can be in text form. For this situation, when any planting behavior is implemented, the corresponding data position can be set to 1, otherwise the corresponding data position can be set to 0.

[0087] Furthermore, considering that some data may be missing in the above-mentioned production influencing parameters due to various reasons (such as data acquisition failure or failure to record relevant data), in an embodiment of the present application, after obtaining the above-mentioned converted first time series data, the converted second time series data and the converted third time series data, the missing values ​​can be filled.

[0088] Specifically, in some embodiments of the present application, the missing values ​​in the converted first time series data and the converted second time series data can be filled using a mean filling method. Moment The value of the parameter is a, Moment The value of the parameter is b, Moment If the parameter is missing, you can Moment Parameters are filled in as .

[0089] For the converted third time series data, in the embodiment of the present application, a fixed value filling method can be used to fill in the missing values ​​(for example, Specifically, in the embodiment of the present application, different data bits in the converted third time series data can be filled with different data. For example, for the planting area and planting density, the data obtained at the most recent time point before the current moment (for example The initial value obtained at the moment, which is regarded as a fixed value) is used to fill in the missing values, and other missing values ​​can be filled with the value 0. It should be pointed out that in the embodiment of the present application, after the planting area and planting density in the planting data are filled with the data obtained at the most recent time point, the characteristics of the planting data can be more accurately reflected.

[0090] In some embodiments of the present application, data filling can be performed so that the time series data corresponding to the meteorological data, soil data, and planting data have the same time granularity, thereby ensuring that these data can be analyzed synchronously when predicting tobacco leaf yields, thereby improving the accuracy of the prediction. For example, if the time granularity of the meteorological data is hours, and the time granularity of the soil data and planting data is days, the time granularity of the soil data and planting data can be adjusted to hours through interpolation methods (such as the aforementioned mean filling method and fixed value filling method) to match the time granularity of other data. In this way, it can be ensured that when predicting tobacco leaf yields, all relevant data are analyzed based on the same time interval, thereby avoiding prediction errors caused by inconsistent time granularity.

[0091] After filling missing values ​​for the first time series data after the above conversion, the first target time series data can be obtained, after filling missing values ​​for the second time series data after the above conversion, the second target time series data can be obtained, and after filling missing values ​​for the third time series data after the above conversion, the third target time series data can be obtained.

[0092] In the embodiment of the present application, the first target time series data includes the preprocessed first meteorological data and the preprocessed second meteorological data. Further, the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in the past time period can be calculated, and then the predicted meteorological parameters of the preprocessed second meteorological data in the future time period can be predicted based on the Pearson correlation coefficient and the meteorological parameters of the preprocessed first meteorological data in the future time period.

[0093] It can be understood that the first meteorological data is meteorological data obtained from a meteorological forecasting platform, which includes meteorological forecast data corresponding to a past time period and meteorological forecast data for a certain time period in the future. However, due to the problem of prediction accuracy, the first meteorological data usually has a certain deviation from the actual meteorology of the target planting area. The second meteorological data is obtained by an agricultural meteorological detector set in the target planting area, which can obtain the meteorological data of the target planting area more accurately, but does not have the meteorological forecast function for the future time period. Based on this, in an embodiment of the present application, the Pearson correlation coefficient between the aforementioned pre-processed first meteorological data and the pre-processed second meteorological data in the past time period is calculated, and then the predicted meteorological parameters of the target planting area in the future time period are predicted based on the Pearson correlation coefficient and the meteorological parameters of the pre-processed first meteorological data in the future time period, so as to more accurately obtain the meteorological data of the target planting area in the future time period.

[0094] Specifically, in the embodiment of the present application, for each meteorological parameter (for example, The Pearson correlation coefficient between the two is then calculated x The mean of the Pearson correlation coefficients corresponding to the meteorological parameters in the above formula is used as the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in the past time period. It can be understood that the Pearson correlation coefficient can represent the correlation between the preprocessed first meteorological data and the preprocessed second meteorological data in the past time period. In this way, the matching degree between the two can be effectively evaluated.

[0095] In some embodiments of the present application, the Pearson correlation coefficient may range from 0 to 1. When the Pearson correlation coefficient is close to 1, it indicates that there is a strong positive correlation between the two meteorological data series; and when the Pearson correlation coefficient is close to 0, it indicates that there is no obvious linear correlation between the two meteorological data series.

[0096] In the embodiment of the present application, the meteorological parameters of the first meteorological data in the future time period can be used as the basic value, and then the Pearson correlation coefficient is used as the coefficient of the error mean between the first meteorological data and the second meteorological data, so as to calculate the predicted meteorological parameters of the second meteorological data in the future time period. The process can be expressed as follows:

[0097] ;

[0098] in, It indicates the prediction result corresponding to the nth meteorological parameter of the second meteorological data at the i-th moment in the future time period. Indicates the nth meteorological parameter corresponding to the first meteorological data at the i-th moment in the future time period, represents the Pearson correlation coefficient between the first meteorological data and the second meteorological data, It represents the mean error between the nth meteorological parameter in the first meteorological data and the second meteorological data in the past time period.

[0099] Furthermore, the preprocessed second meteorological data can be spliced ​​with the predicted meteorological parameters of the target planting area in the future time period obtained in the aforementioned process to obtain target meteorological data, and the time series data corresponding to the target meteorological data can be used as the first target time series data for subsequent analysis.

[0100] It should be pointed out that in the embodiment of the present application, by predicting the predicted meteorological parameters of the second meteorological data in a future time period based on the correlation between the first meteorological data and the second meteorological data, and making tobacco leaf yield predictions based on this, the accuracy of the meteorological data used in the yield prediction process can be improved to a certain extent, thereby improving the accuracy of subsequent analysis and obtaining more accurate and reliable tobacco leaf yield prediction results.

[0101] In the embodiment of the present application, the neural network layer 220 can be obtained by iterative training based on historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles. The specific process of training can be referred to later, and will not be discussed in detail here.

[0102] Continue to refer to Figure 2 In the embodiment of the present application, the neural network layer 220 may include an input layer 221, a first processing layer 222, a second processing layer 223 and an output layer 224 connected in sequence. The input layer 221 may be used to receive the yield influencing parameters (i.e., the aforementioned first target time series data, second target time series data and third target time series data) obtained through preprocessing by the preprocessing layer 210. The first processing layer 222 may be used to perform feature extraction processing on the multiple time series data (first target time series data, second target time series data and third target time series data) contained in the yield influencing parameters to obtain the influencing parameter time series feature fusion data.

[0103] Figure 3 is an exemplary structural diagram of the first processing layer according to some embodiments of this specification. Figure 3 In some embodiments of the present application, the first processing layer 222 may include a feature extraction unit 2221 and a feature fusion unit 2222. The feature extraction unit 2221 may be used to perform convolution calculations on each time series data included in the yield influencing parameter to extract corresponding feature data from each time series data; the feature fusion unit 2222 may be used to fuse the feature data extracted from each time series data with the corresponding original time series data to obtain the influencing parameter time series feature fusion data. The original time series data refers to the first target time series data, the second target time series data, and the third target time series data obtained by the aforementioned preprocessing. As an example only, in some embodiments of the present application, the feature fusion unit 2222 may splice the feature data extracted from each time series data with the corresponding original time series data to form a feature matrix with a higher data dimension, namely, the aforementioned influencing parameter time series feature fusion data. In the embodiment of the present application, the aforementioned first target time series data, the second target time series data, and the third target time series data may be processed to obtain a corresponding influencing parameter time series feature fusion data.

[0104] It can be understood that the influencing parameter time series feature fusion data not only includes the original yield influencing parameters, but also includes the characteristic data corresponding to each yield influencing parameter, which can enable the data to provide richer and more accurate information in the subsequent tobacco leaf yield prediction process, thereby better capturing the key factors affecting tobacco leaf yield and improving the accuracy and reliability of the prediction.

[0105] Continue to refer to Figure 3 In some embodiments of the present application, considering that the aforementioned influencing parameter time series feature fusion data may cause subsequent calculations to be too complicated due to the high data dimension, therefore, in order to reduce the complexity of subsequent calculations to a certain extent, in some embodiments of the present application, the first processing layer 222 may also include a pooling processing unit 2223, which can be used to perform PCA dimensionality reduction (i.e., principal component analysis dimensionality reduction) processing on the influencing parameter time series feature fusion data.

[0106] It can be understood that through PCA dimensionality reduction processing, the dimension of the data can be effectively reduced while retaining important information in the data as much as possible. This dimensionality reduction processing helps to simplify the subsequent calculation process, improve the operation efficiency, and reduce the demand for storage space. Specifically, the pooling processing unit 2223 can effectively extract the features that have the greatest impact on tobacco leaf yield prediction by selecting the main components in the fusion data of the time series features of the influencing parameters, so as to make the subsequent prediction more accurate and efficient.

[0107] The second processing layer 223 can be used to extract the temporal correlation between the time series feature fusion data of the influencing parameters corresponding to each yield influencing parameter to obtain the time series correlation feature data. Specifically, in the embodiment of the present application, the BG (Bernaola-Galvan) sequence segmentation algorithm can be used to obtain the mutation points in the time series feature fusion data of the influencing parameters (the process of determining the mutation points using the BG sequence segmentation algorithm can be regarded as the prior art), and the mutation time series corresponding to each yield influencing parameter is determined based on the time corresponding to each mutation point (that is, the time series obtained by arranging the time corresponding to each mutation point in chronological order). Then, the sum of the DTW (Dynamic Time Warping) distances between the mutation points with the same sequence number in the mutation time series corresponding to any two yield influencing parameters is calculated, and the time series correlation feature data between the any two yield influencing parameters is determined based on the sum of the DTW distances. Among them, the specific calculation method of the DTW distance can be regarded as the prior art, and it will not be discussed in detail in this specification.

[0108] It should be noted that the above mutation points can be regarded as key points that affect tobacco leaf yield. For example, the mutation points in meteorological data can be regarded as the occurrence of extreme weather (such as strong winds, heavy rains, hail, etc.), the mutation points in soil data can be regarded as the growth environment of tobacco plants has undergone major changes, and the mutation points in planting data can be understood as corresponding planting management behaviors (such as fertilization, irrigation, etc.). In the embodiment of the present application, by calculating the sum of the DTW distances between the mutation points with the same sequence number in the mutation time series corresponding to the two yield influencing parameters, the relationship characteristics of each factor affecting tobacco leaf yield at the time node can be reflected. Therefore, by taking the sum of the DTW distances as the time series correlation feature data between the two yield influencing parameters and performing subsequent processing based on this, the accuracy of the subsequent tobacco yield prediction results is further improved.

[0109] In some embodiments, the number of mutation points in the mutation time series corresponding to the two yield influencing parameters may be inconsistent. In this case, the mutation time series corresponding to the last mutation point in the one with fewer mutation points can be padded, and the sum of the DTW distances between the mutation points with the same sequence number can be calculated based on the padded data. For example, the mutation time series corresponding to the meteorological data is , the mutation time series corresponding to the planting data is When , the mutation time series corresponding to the planting data can be filled in as Furthermore, the temporal correlation feature data Y between the meteorological data and the planting data can be expressed as:

[0110] ;

[0111] Among them, Y represents the time series correlation feature data, and D represents the DTW distance.

[0112] It should be noted that in the embodiment of the present application, the value of the time series correlation characteristic data has no specific meaning, and is only used to represent the time series relationship characteristics of the key points in the two yield influencing parameters. In the subsequent process, by using the time series correlation characteristic data as a calculation basis, the accuracy of the tobacco yield prediction result can be improved to a certain extent.

[0113] The output layer 224 can be used to make predictions based on the time series associated feature data and the influencing parameter time series feature fusion data to obtain predicted tobacco leaf yield. Specifically, in an embodiment of the present application, the output layer may include a fully connected layer, which includes a plurality of neurons, each of which is connected to each parameter in the time series associated feature data and the influencing parameter time series feature fusion data through a weight. In the process of iterative training based on historical yield influencing parameters and historical tobacco leaf yield data of multiple historical planting cycles, the weights can be optimized by a back propagation algorithm to minimize the error between the predicted tobacco leaf yield and the actual tobacco leaf yield.

[0114] Specifically, in an embodiment of the present application, the historical meteorological data, historical soil data, and historical planting data contained in the historical tobacco yield data can be processed by the above-mentioned preprocessing method to obtain the corresponding first target time series data, second target time series data, and third time series data. Further, the first target time series data, second target time series data, and third time series data corresponding to the historical meteorological data can be input into the neural network layer 220 for training, and the corresponding historical tobacco yield data can be used as the label corresponding to the first target time series data, the second target time series data, and the third time series data, and then the performance of the model is evaluated by calculating the difference between the label and the predicted output result of the neural network layer. If the difference is large, it means that the prediction accuracy of the model needs to be improved, and it is necessary to further adjust the parameters of the neural network layer or optimize the training algorithm. If the difference is small, it means that the model already has a high prediction accuracy and can be used for actual tobacco yield prediction.

[0115] In the embodiment of the present application, the weights and the parameters in the first processing layer 221 and the second processing layer 222 can be optimized by a back propagation algorithm to minimize the error between the predicted tobacco leaf yield output by the neural network layer 220 and the actual tobacco leaf yield. More details about the optimization using the back propagation algorithm can be regarded as prior art and will not be discussed in detail in this specification.

[0116] In some embodiments of the present application, in the process of iterative training based on historical yield influencing parameters and historical tobacco leaf yield data of multiple historical planting cycles, by setting the label information as the standard yield and the reduced yield (the standard yield can be estimated by the planting area and the maximum yield per unit area, and the reduced yield can be obtained by the difference between the actual yield and the standard yield, and the reduced yield can be positive or negative), then based on the predicted standard yield and the predicted reduced yield corresponding to the historical yield influencing parameters, the comprehensive loss value is calculated (in the embodiment of the present application, the first difference between the predicted standard yield and the standard yield in the label information, and the sum of the predicted reduced yield and the second difference between the reduced yield in the label information can be used as the comprehensive loss), and then optimized by the back propagation algorithm to minimize the comprehensive loss value. In the embodiment of the present application, the training can be terminated when the comprehensive loss value reaches the preset threshold value to obtain the target yield prediction model. Among them, the preset threshold value can be set according to actual needs, and it is not specifically limited in this specification.

[0117] After training in this way, the neural network layer 220 can predict the standard yield and reduced yield based on the yield influencing parameters of the target planting area in the current planting cycle, and then obtain the predicted tobacco leaf yield of the target planting area in the current planting cycle based on the difference between the predicted standard yield and the predicted reduced yield.

[0118] Step S330, obtaining the predicted tobacco yield of the target planting area in the current planting cycle according to the output result of the target yield prediction model. In some embodiments, step S330 can be performed by the tobacco yield prediction module 120.

[0119] In summary, the beneficial effects that may be brought about by the embodiments of the present specification include but are not limited to: (1) In the tobacco yield prediction system and method based on precision meteorology provided in some embodiments of the present specification, tobacco yield is predicted by combining the first meteorological data obtained from the meteorological prediction platform, the second meteorological data obtained by the agricultural meteorological monitor in the target planting area, the soil data obtained by the soil sensor, and the planting data obtained by the planting management record. Not only can the accuracy of the meteorological data used in the yield prediction process be improved according to the correlation between the first meteorological data and the second meteorological data, but also the time series correlation between each yield influencing parameter can be extracted as an influencing factor for yield prediction, thereby improving the accuracy of tobacco yield prediction results to a certain extent; (2) In the tobacco yield prediction system based on precision meteorology provided in some embodiments of the present specification, In the method, by predicting the predicted meteorological parameters of the second meteorological data in a future time period according to the correlation between the first meteorological data and the second meteorological data, and based on this, tobacco yield prediction is performed, which can improve the accuracy of the meteorological data used in the yield prediction process to a certain extent, thereby improving the accuracy of subsequent analysis and obtaining more accurate and reliable tobacco yield prediction results; (3) In the tobacco yield prediction system and method based on precision meteorology provided in some embodiments of the present specification, by calculating the time series correlation feature data between various meteorological influencing parameters to reflect the relationship characteristics of various factors affecting tobacco yield at time nodes, and using the time series correlation feature data as a calculation basis for tobacco yield prediction, the different effects of different influencing factors on tobacco yield at different time nodes can be taken into account, thereby further improving the accuracy of tobacco yield prediction results.

[0120] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or may be any other beneficial effects that may be obtained.

[0121] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0122] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0123] In addition, it will be understood by those skilled in the art that various aspects of this specification may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of this specification may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of this specification may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0124] A computer storage medium may include a propagated data signal containing computer program code, for example, in baseband or as part of a carrier wave. The propagated signal may be in a variety of forms, including electromagnetic, optical, etc., or a suitable combination. A computer storage medium may be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, device or apparatus to communicate, propagate or transmit the program for use. The program code on the computer storage medium may be transmitted via any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of the above media.

[0125] The computer program codes required for the operation of each part of this specification can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C language, Visual Basic, Fortran2003, Perl, COBOL2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages, etc. The program code can be run entirely on the user's computer, or run on the user's computer as an independent software package, or run partly on the user's computer and partly on a remote computer, or run entirely on a remote computer or processing device. In the latter case, the remote computer can be connected to the user's computer through any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as software as a service (SaaS).

[0126] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing processing device or mobile device.

[0127] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0128] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining the digits. Although the numerical domains and parameters used to confirm the breadth of the range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0129] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification is hereby incorporated by reference in its entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0130] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A tobacco leaf yield prediction method based on precise meteorology, characterized in that: include: Obtaining yield influencing parameters of a target planting area in a current planting cycle, wherein the yield influencing parameters include first time series data corresponding to meteorological data, second time series data corresponding to soil data, and third time series data corresponding to planting data, wherein the meteorological data include first meteorological data obtained from a meteorological forecasting platform and second meteorological data obtained through an agricultural meteorological monitor of the target planting area; Inputting the yield influencing parameters into a target yield prediction model for processing; The predicted tobacco yield of the target planting area in the current planting cycle is obtained according to the output result of the target yield prediction model; the target yield prediction model includes a preprocessing layer and a neural network layer, wherein the preprocessing layer is used to preprocess the yield influencing parameters, and the neural network layer is obtained by iterative training based on historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles; the neural network layer includes an input layer, a first processing layer, a second processing layer and an output layer connected in sequence, the first processing layer is used to perform feature extraction processing on multiple time series data contained in the yield influencing parameters to obtain influencing parameter time series feature fusion data; the second processing layer is used to extract the time series correlation between the influencing parameter time series feature fusion data corresponding to each yield influencing parameter to obtain time series correlation feature data; the output layer is used to predict based on the time series correlation feature data and the influencing parameter time series feature fusion data to obtain the predicted tobacco yield; The step of extracting the temporal correlation between the temporal feature fusion data of the influencing parameters corresponding to each output influencing parameter to obtain the temporal correlation feature data includes: The mutation points in the time series feature fusion data of the influencing parameters are obtained by using the BG sequence segmentation algorithm, and the mutation time series corresponding to each yield influencing parameter is determined based on the time corresponding to each mutation point; The sum of the DTW distances between the mutation points with the same serial number in the mutation time series corresponding to any two yield-influencing parameters is calculated, and the time series correlation feature data between the any two yield-influencing parameters is determined based on the sum of the DTW distances; the sum of the DTW distances between the mutation points with the same serial number in the mutation time series corresponding to any two yield-influencing parameters is calculated, including: When the number of mutation points in the mutation time series corresponding to any two yield influencing parameters is inconsistent, the mutation points are padded based on the time corresponding to the last mutation point in the one with fewer mutation points, and the sum of the DTW distances between the mutation points with the same sequence number is calculated based on the padded data.

2. The method according to claim 1, characterized in that The preprocessing of the yield influencing parameters includes: Performing standardized conversion on the information included in the first time series data, the second time series data, and the third time series data according to a preset rule to obtain converted first time series data, converted second time series data, and converted third time series data; and The missing values ​​in the converted first time series data and the converted second time series data are filled using a mean filling method, and the missing values ​​in the converted third time series data are filled using a fixed value filling method to obtain first target time series data, second target time series data, and third target time series data.

3. The method according to claim 2, characterized in that The first target time series data includes preprocessed first meteorological data and preprocessed second meteorological data, and the preprocessing of the yield influencing parameters further includes: Calculating the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in a past time period; Predicting the predicted meteorological parameters of the preprocessed second meteorological data in the future time period according to the Pearson correlation coefficient and the meteorological parameters of the preprocessed first meteorological data in the future time period; Target meteorological data are obtained according to the preprocessed second meteorological data and the predicted meteorological parameters of the preprocessed second meteorological data in the future time period, and the time series data corresponding to the target meteorological data are used as the first target time series data.

4. The method according to claim 1, characterized in that The first processing layer includes a feature extraction unit and a feature fusion unit; wherein, The feature extraction unit is used to perform convolution calculation on each time series data included in the yield influencing parameter, so as to extract corresponding feature data from each time series data; The feature fusion unit is used to fuse the feature data extracted from each time series data with the corresponding original time series data to obtain the influencing parameter time series feature fusion data.

5. The method according to claim 4, characterized in that The first processing layer further includes a pooling processing unit, which is used to perform PCA dimensionality reduction processing on the influencing parameter time series feature fusion data; The extracting of the temporal correlation between the temporal feature fusion data of the influencing parameters corresponding to each yield influencing parameter includes: extracting the temporal correlation of the temporal feature fusion data of the influencing parameters after dimensionality reduction processing corresponding to each yield influencing parameter to obtain the temporal correlation feature data.

6. The method according to claim 1, characterized in that The output layer includes a fully connected layer, and the fully connected layer includes a plurality of neurons, each of which is connected to each parameter in the time series associated feature data and the influencing parameter time series feature fusion data through a weight; The iterative training based on the historical yield influencing parameters and historical tobacco leaf yield data of multiple historical planting cycles includes: optimizing the weights through a back propagation algorithm to minimize the error between the predicted tobacco leaf yield and the actual tobacco leaf yield; The prediction based on the time series correlation feature data and the influencing parameter time series feature fusion data is performed to obtain the predicted tobacco leaf yield, including: predicting the standard yield and reduced yield based on the time series correlation feature data and the influencing parameter time series feature fusion data, and obtaining the predicted tobacco leaf yield based on the standard yield and reduced yield.

7. A tobacco yield prediction system based on precise meteorology, characterized in that: include: A yield influencing parameter acquisition module, used to acquire the yield influencing parameters of the target planting area in the current planting cycle, wherein the yield influencing parameters include first time series data corresponding to meteorological data, second time series data corresponding to soil data, and third time series data corresponding to planting data, wherein the meteorological data include first meteorological data acquired from a meteorological forecasting platform and second meteorological data acquired through an agricultural meteorological monitor of the target planting area; A tobacco yield prediction module is used to input the yield influencing parameters into a target yield prediction model for processing, and obtain the predicted tobacco yield of the target planting area in the current planting cycle according to the output result of the target yield prediction model; the target yield prediction model includes a preprocessing layer and a neural network layer, wherein the preprocessing layer is used to preprocess the yield influencing parameters, and the neural network layer is obtained by iterative training based on historical yield influencing parameters and historical tobacco yield data of multiple historical planting cycles; the neural network layer includes an input layer, a first processing layer, a second processing layer and an output layer connected in sequence, the first processing layer is used to perform feature extraction processing on multiple time series data contained in the yield influencing parameters to obtain influencing parameter time series feature fusion data; the second processing layer is used to extract the time series correlation between the influencing parameter time series feature fusion data corresponding to each yield influencing parameter to obtain time series correlation feature data; the output layer is used to predict based on the time series correlation feature data and the influencing parameter time series feature fusion data to obtain predicted tobacco yield; The step of extracting the temporal correlation between the temporal feature fusion data of the influencing parameters corresponding to each output influencing parameter to obtain the temporal correlation feature data includes: The mutation points in the time series feature fusion data of the influencing parameters are obtained by using the BG sequence segmentation algorithm, and the mutation time series corresponding to each yield influencing parameter is determined based on the time corresponding to each mutation point; The sum of the DTW distances between the mutation points with the same serial number in the mutation time series corresponding to any two yield-influencing parameters is calculated, and the time series correlation feature data between the any two yield-influencing parameters is determined based on the sum of the DTW distances; the sum of the DTW distances between the mutation points with the same serial number in the mutation time series corresponding to any two yield-influencing parameters is calculated, including: When the number of mutation points in the mutation time series corresponding to any two yield influencing parameters is inconsistent, the mutation points are padded based on the time corresponding to the last mutation point in the one with fewer mutation points, and the sum of the DTW distances between the mutation points with the same sequence number is calculated based on the padded data.

8. The system according to claim 7, characterized in that The pretreatment layer is specifically used for: Performing standardized conversion on the information included in the first time series data, the second time series data, and the third time series data according to a preset rule to obtain converted first time series data, converted second time series data, and converted third time series data; The missing values ​​in the converted first time series data and the converted second time series data are filled by using a mean filling method, and the missing values ​​in the converted third time series data are filled by using a fixed value filling method, so as to obtain first target time series data, second target time series data and third target time series data; wherein the first target time series data includes the preprocessed first meteorological data and the preprocessed second meteorological data, and the preprocessing layer is further used for: Calculating the Pearson correlation coefficient between the preprocessed first meteorological data and the preprocessed second meteorological data in a past time period; Predicting the predicted meteorological parameters of the preprocessed second meteorological data in the future time period according to the Pearson correlation coefficient and the meteorological parameters of the preprocessed first meteorological data in the future time period; Target meteorological data are obtained according to the preprocessed second meteorological data and the predicted meteorological parameters of the preprocessed second meteorological data in the future time period, and the time series data corresponding to the target meteorological data are used as the first target time series data.

Images