A fast calculation method for the optimal sowing date of winter wheat based on artificial intelligence

Abstract

The present invention discloses a method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence, which mainly consists of two parts: one is a data quality control part, which mainly includes a temperature calculation method considering wind speed and direction, a precipitation calculation method considering sky clouds, a solar radiation calculation method considering relative humidity, and a yield calibration method considering planting technology improvement and variety replacement; the other is an optimal sowing date calculation part, which mainly includes a Gaussian model construction method for suitable sowing period based on artificial intelligence, and a rapid sampling and iterative sowing date method based on mathematical expectation. Aiming at the problem that the current trial and error method simulates the problem of long calculation time for the optimum sowing date of winter wheat, the present invention realizes rapid calculation and prediction of the optimum sowing date of winter wheat with high resolution gridding through artificial intelligence technology, and meets the business needs of high resolution agricultural meteorological grid forecasting.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural production data prediction and calculation, and in particular to a method for quickly calculating the optimum sowing date of winter wheat based on artificial intelligence. Background Art

[0002] The sowing date of winter wheat is directly related to the growth and final yield of the crop, and the calculation of the sowing date is a core link in agricultural production management. However, accurately calculating the optimal sowing date for winter wheat faces many challenges.

[0003] Traditional methods mainly rely on farmers' experience and meteorological observation data. The empirical method has many limitations due to its lack of scientific basis and poor interpretability. The method based on meteorological observation data is difficult to fully reflect the dynamic changes of the optimal sowing date in different regions and years due to the uneven distribution of observation points, missing or discontinuous data, and the complexity of long-term changes in climate conditions.

[0004] In addition, differences in agricultural technology levels and planting habits in different regions further complicate the analysis of the optimal sowing date. Typically, field experiments conducted for scientific research purposes can be used to determine the optimal sowing date, but the results are not representative enough in large-scale areas, and the experimental cycle is long and costly.

[0005] In current agricultural meteorological forecasts, agricultural meteorological indicators are usually used to define the range of the most suitable sowing date, and the time scale is mostly about 10 days, which is difficult to meet the needs of refined agricultural meteorological services. In addition, winter wheat yield is affected by a variety of factors, and it is difficult to directly infer the optimal sowing date from historical data. This also makes the current agricultural meteorological forecast have great uncertainty in the prediction of the most suitable sowing date for winter wheat. Therefore, only the trial and error method is applicable in the current business. By exhaustively enumerating the winter wheat planting date and inputting the crop numerical model to calculate the yield data, the planting date corresponding to the highest yield is selected as the most suitable planting date. However, due to the large amount of calculation of the exhaustive method, it is only applicable to the simulation and calculation of the most suitable planting date for a small number of sites, and it is more difficult for large-scale simulations with high spatial resolution.

[0006] Against the backdrop of the rapid development of artificial intelligence (AI) technology, building a systematic and spatially consistent method for quickly calculating optimal sowing date data is of great significance to the development and scientific research of agricultural meteorological AI models. However, AI models require a large amount of high-precision training data to accurately predict the optimal sowing date in the future and guide agricultural production. Incomplete or inaccurate historical data will directly affect the generalization ability and prediction accuracy of the model. Summary of the invention

[0007] The present invention aims to solve one of the technical problems existing in the related art at least to a certain extent.

[0008] One object of the present invention is to provide a method for quickly calculating the optimal sowing date of winter wheat based on artificial intelligence. By integrating meteorological observation data, remote sensing images and agricultural experimental results, a comprehensive and accurate calculation model for the optimal sowing date of winter wheat is constructed, so as to realize high-resolution grid-based calculation and prediction of the optimal sowing date of winter wheat.

[0009] In order to achieve the above-mentioned purpose, the present invention provides a method for quickly calculating the optimum sowing date of winter wheat based on artificial intelligence, comprising:

[0010] S100, establishing an artificial intelligence training data set based on historical meteorological observation data, historical winter wheat yield data, and winter wheat sowing and variety change time data;

[0011] Design an interpolation algorithm that conforms to the changing laws of weather processes to perform quality control on missing or abnormal data of temperature, precipitation and solar radiation in historical meteorological observation data;

[0012] Conduct quality control on statistical errors in long-term series of historical winter wheat yield data to eliminate the impact of improved planting techniques and crop variety improvements on historical winter wheat yield data;

[0013] S200, constructing a multidimensional Gaussian process model for optimizing the calculation of the optimal sowing date using the quality-controlled artificial intelligence training data set; the multidimensional Gaussian process model uses yield data as the target variable and meteorological data and sowing time data as feature data;

[0014] S300, based on the constructed multidimensional Gaussian process model, establish a planting date sampling function based on mathematical expectation improvement, sample planting dates from the winter wheat planting date interval, perform numerical simulation of crop yields with the sampled planting dates, and then feed back the simulated yields of the sampled planting dates and the corresponding crop planting dates to the multidimensional Gaussian process model to update the multidimensional Gaussian process model; perform cyclic iterations, and when the number of cycles required and the accumulated temperature conditions for winter wheat planting are met, use the sampled planting date as the optimal sowing date for winter wheat;

[0015] If the condition is not met, step S300 is repeated until the accumulated temperature condition for winter wheat planting is met, and the most suitable sowing date for winter wheat is output.

[0016] A further preferred technical solution of the present invention is that, in step S100, a difference algorithm that conforms to the law of weather process change is designed to perform quality control on the missing or abnormal data of temperature, precipitation and solar radiation in the historical meteorological observation data; including:

[0017] Consider wind direction and speed to perform quality control on air temperature data;

[0018] Consider the changes in sky clouds and conduct quality control on precipitation data;

[0019] The solar radiation data were quality controlled by considering the changes in near-surface relative humidity.

[0020] Preferably, the wind direction and wind speed are considered to perform quality control on the temperature data, and the temperature difference function is expressed as:

[0021]

[0022] In the formula, it is assumed that the temperatures of the two time series are and , wind speed is , wind direction is ;in is the proportionality coefficient, ranging from 0.1 to 0.3;

[0023] The above-mentioned consideration of the cloud system changes in the sky and the quality control of precipitation data include:

[0024] The precipitation difference function is calculated by the precipitation difference and the specific humidity weighted adjustment factor, which is expressed as:

[0025] ;

[0026] Assume that the precipitation of the two time series is and , the specific humidity at 500 hPa and 850 hPa are and , is the weighted adjustment factor for specific humidity, expressed as:

[0027] ;

[0028] In the formula, and are the weights of the specific humidity at 500 hPa and 850 hPa, is the specific humidity value at 500 hPa or 850 hPa in the typical atmosphere of the station climatological state, used for function normalization;

[0029] The solar radiation data is quality controlled by considering the change of near-ground relative humidity. The solar radiation difference function is expressed as:

[0030] ;

[0031] Assume that the precipitation of the two time series is and , and are two relative humidities, is the reference relative humidity value, is the sensitivity coefficient of humidity to radiation attenuation.

[0032] Preferably, in step S100, quality control is performed on the statistical errors in the long-term series of historical winter wheat yield data to eliminate the impact of planting technology improvement and crop variety improvement on the historical winter wheat yield data; specifically,

[0033] Considering the impact of improved planting technology and crop variety on yield data, a yield function is established, which is expressed as:

[0034] ;

[0035] in, is a linear function of the time trend of output, used to represent the impact of technological improvement on output. , are the intercept and slope of the linear function of the production data time series, is the independent variable time; is a dummy variable for the variety category, Change the time segmentation function for local varieties, The total number of changes to the product. The date of a single product change. Used to eliminate the contribution of different varieties to production; It is a function of climate factors and is obtained by linear function fitting using temperature, precipitation and radiation meteorological data; is a random error;

[0036] After fitting the original yield according to the formula, the effects of technological improvements and varieties are eliminated, and the part related to climate factors is retained. This method is used to control the data quality of historical yield data.

[0037] Preferably, in step S200, the multidimensional Gaussian process model constructed is expressed as:

[0038] ;

[0039] In the formula, represents a multidimensional Gaussian function, is the mean of the Gaussian function, is the standard deviation of the Gaussian function, is the planting date composed of historical training data, as well as the matrix data composed of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date. The newly entered planting date and the matrix data consisting of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date are: The winter wheat yield matrix data corresponding to the planting date of historical training data.

[0040] is the covariance of the new input data and the training data, is the covariance of the training data and the new input data, is the covariance between the training data of each year, is the covariance of the new input data itself.

[0041] The calculation formula of each covariance is expressed as:

[0042] ;

[0043] In the formula Corresponding to the multi-dimensional Gaussian process model construction function formula or , is the variance of all known production data, is the length scale constant.

[0044] Preferably, the specific method of step S300 is:

[0045] S310, according to the multidimensional Gaussian process model constructed in step S200, establishing a planting date sampling function based on mathematical expectation improvement, and retrieving the planting date;

[0046] S320, selecting a planting date with the largest mathematical expectation from the winter wheat planting date interval, and inputting it into a crop model to simulate yield;

[0047] S330, using the simulated yield and the selected planting date as training data again to update the multidimensional Gaussian process model of step S200;

[0048] S340, repeating steps S310 to S330, after the number of sequential times is met, verifying whether the simulation result meets the accumulated temperature planting condition, if so, outputting the planting date as the suitable sowing date;

[0049] S350: If the accumulated temperature condition for winter wheat planting is not met, repeat steps S320 to S340 until the accumulated temperature condition for winter wheat planting is met, and output the most suitable sowing date for winter wheat.

[0050] Preferably, the planting date sampling function based on mathematical expectation improvement established in step S310 is expressed as:

[0051] ;

[0052] ;

[0053] In the formula, is the mathematical expectation of the yield corresponding to the planting date, and are the standard deviation and mean of the yield corresponding to the planting date in the Gaussian function model, are the cumulative distribution function and probability density function of the yield corresponding to the planting date, is the maximum value of known yield. The last planting date sampling value is the planting date corresponding to the maximum mathematical expectation of yield.

[0054] On the other hand, the present invention provides a non-transitory computer-readable storage medium, characterized in that computer instructions are stored thereon, and the computer instructions enable the computer to execute the above-mentioned artificial intelligence-based method for quickly calculating the optimal sowing date of winter wheat.

[0055] Yet another aspect of the present invention provides an electronic device, characterized in that it comprises: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, and the processor calls the logic instructions in the memory to execute the above-mentioned artificial intelligence-based method for quickly calculating the optimal sowing date of winter wheat.

[0056] On the other hand, the present invention provides a computer program product, characterized in that the computer program product includes a computer program, the computer program is stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer executes the above-mentioned artificial intelligence-based method for quickly calculating the optimal sowing date of winter wheat.

[0057] Beneficial effects: In order to solve the problem of missing meteorological data, the present invention designs a temperature quality control method based on wind direction and wind speed, which solves the problem that the traditional interpolation method ignores the heating and cooling rates; based on the changes in sky clouds and ground relative humidity, the present invention designs precipitation and solar radiation data quality control methods respectively, which is conducive to solving the errors caused by missing meteorological data and further improves the accuracy of the subsequent reproduction of the optimal sowing date for winter wheat in history.

[0058] The present invention combines artificial intelligence with crop model algorithms, making the determination and calibration of the optimal sowing date for winter wheat in history more intelligent, reducing human errors and experience dependence in traditional methods. At the same time, through historical data quality control and artificial intelligence optimal sowing date calculation method, the number of trial and error simulation planting dates is significantly reduced, ensuring the reliability and availability of the results, and can quickly calculate the optimal planting date data with high-precision grid resolution.

[0059] The method designed by the present invention can not only be used to construct data on the optimal sowing date for winter wheat, but also has potential promotion value and can be applied to data analysis and calibration of other crops and agricultural fields. By constructing a standardized data set, the present invention provides a high-quality historical data foundation for the training of large artificial intelligence models, promoting AI research and application in the agricultural field. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of a method for quickly calculating the optimal sowing date for winter wheat based on artificial intelligence of the present invention;

[0061] Figure 2 This is a diagram showing the calculation results of the optimal sowing date in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0063] Combine the following Figure 1-Figure 2 The invention describes an artificial intelligence-based fast calculation method for the optimum sowing date of winter wheat, a non-transient storage medium, an electronic device and a program product.

[0064] Embodiment 1: This embodiment provides a method for quickly calculating the optimum sowing date of winter wheat based on artificial intelligence.

[0065] This method mainly consists of two parts. The first part is the data quality control part, which aims to solve the problems of missing meteorological observation data and statistical errors in yield data. It mainly includes the temperature calculation method considering wind speed and direction, the precipitation calculation method considering the sky cloud system, the solar radiation calculation method considering relative humidity, and the yield calibration method considering planting technology improvement and variety replacement. The second part is the optimal sowing date calculation part, which mainly solves the problem that the calculation amount of the optimal sowing date of winter wheat in the current crop numerical simulation based on trial and error method is too large, resulting in the inability to calculate refined grid data. It mainly includes the construction method of the calculation model based on artificial intelligence and the method of fast sampling and iterative simulation and updating model of sowing date.

[0066] Specific steps are as follows Figure 1 As shown, including:

[0067] S100. Establish an artificial intelligence training data set based on historical meteorological observation data, historical winter wheat yield data, and winter wheat sowing time data.

[0068] S200. Design an interpolation algorithm that conforms to the changing laws of weather processes to perform quality control on missing or abnormal data on temperature, precipitation and solar radiation in historical meteorological observation data.

[0069] There are usually missing data in meteorological observation data. In order to ensure the continuity and availability of data, it is necessary to perform quality control on meteorological data. Generally, interpolation is used for missing data, but its accuracy will be greatly affected. This embodiment adopts an interpolation method that takes into account the changes in weather processes to perform quality control on temperature, precipitation and solar radiation respectively.

[0070] (1) Consider wind direction and wind speed to perform quality control on temperature data

[0071] The temperature difference function is expressed as:

[0072]

[0073] In the formula, it is assumed that the temperatures of the two time series are and , wind speed is , wind direction is ;in is the proportionality coefficient, ranging from 0.1 to 0.3;

[0074] when (For cooling scenario): North wind ( ) hour Large, the temperature drop is intensified; South wind ( ) hour The temperature drop is smaller. (For warming scenario): South wind ( ) hour Large, the temperature rise is intensified; north wind ( ) hour Smaller, the temperature rise is weaker.

[0075] (2) Consider the changes in sky clouds and perform quality control on precipitation data

[0076] The precipitation difference function is calculated by the precipitation difference and the specific humidity weighted adjustment factor, which is expressed as:

[0077] ;

[0078] Assume that the precipitation of the two time series is and , the specific humidity at 500 hPa and 850 hPa are and , is the weighted adjustment factor for specific humidity, expressed as:

[0079] ;

[0080] In the formula, and are the weights of the specific humidity at 500 hPa and 850 hPa, is the specific humidity value at 500 hPa or 850 hPa in the typical atmosphere of the station climatological state, used for function normalization;

[0081] (3) Considering the change of near-surface relative humidity, quality control of solar radiation data

[0082] The solar radiation difference function is expressed as:

[0083] ;

[0084] Assume that the precipitation of the two time series is and , and are two relative humidities, is the reference relative humidity value, is the sensitivity coefficient of humidity to radiation attenuation.

[0085] When humidity is low (e.g. < 50), , the radiation difference is not significantly affected; when the humidity is high (such as > 80), It is significantly less than 1, reflecting the effect of humidity on weakening radiation.

[0086] S300. Conduct quality control on long-term series of historical winter wheat yield data to eliminate the impact of improved planting technology and crop variety improvements on historical winter wheat yield data.

[0087] There are usually certain errors in historical yield statistics, and there are also interference problems such as changes in winter wheat varieties and improvements in planting techniques in observation. This leads to greater uncertainty in calculating the relationship between yield and meteorological factors. In order to solve this problem, this embodiment adopts a long-term series historical yield data quality control method, which takes into account the impact of improved planting techniques and crop variety improvements on yield data.

[0088] Considering the impact of improved planting technology and crop variety on yield data, a yield function is established, which is expressed as:

[0089] ;

[0090] in, is a linear function of the time trend of output, used to represent the impact of technological improvement on output. , are the intercept and slope of the linear function of the production data time series, is the independent variable time; is a dummy variable for the variety category, Change the time segmentation function for local varieties, The total number of changes to the product. The date of a single product change. Used to eliminate the contribution of different varieties to production; It is a function of climate factors and is obtained by linear function fitting using temperature, precipitation and radiation meteorological data; is a random error;

[0091] After fitting the original yield according to the formula, the effects of technological improvements and varieties are eliminated, and the part related to climate factors is retained. This method is used to control the data quality of historical yield data.

[0092] After the data is processed, the second part of this embodiment is to solve the problem of long calculation time and high consumption of computing resources caused by the trial and error method of simulating wheat yield by exhaustively enumerating planting dates. This embodiment adopts an artificial intelligence algorithm. First, the meteorological observation data and winter wheat yield data after quality control are used to build an artificial intelligence model. The historical planting date and meteorological data are used as feature variables, and the historical winter wheat yield is used as the target variable to establish a set of multidimensional Gaussian process models. Then, the mathematical expectation method of calculating the yield corresponding to the winter wheat planting date is adopted, and the optimal planting date is sampled from the winter wheat planting date interval, and the sampled planting date is selected for numerical simulation of crop yield. Through the cyclic iteration method, the yield simulated by the sampled planting date and the corresponding data are fed back to the artificial intelligence model, and the multidimensional Gaussian process model is updated. When the conditions of 5 cycles are met and the accumulated temperature conditions of winter wheat planting are met, the most suitable planting date is output. This method can reduce the number of numerical simulations of crops and improve the calculation efficiency of the optimal and most suitable sowing date. The specific method is as follows:

[0093] S400, artificial intelligence model construction method.

[0094] A multidimensional Gaussian process model for optimizing the calculation of the optimal sowing date was established for the quality-controlled artificial intelligence training data set. The model uses yield data as the target variable and meteorological data and sowing time data as feature data. The purpose is to establish a yield probability density distribution function corresponding to meteorological data and planting date to obtain the distribution of winter wheat yield under the corresponding planting date and meteorological conditions.

[0095] The constructed multidimensional Gaussian process model is expressed as:

[0096] ;

[0097] In the formula, represents a multidimensional Gaussian function, is the mean of the Gaussian function, is the standard deviation of the Gaussian function, is the planting date composed of historical training data, as well as the matrix data composed of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date. The newly entered planting date and the matrix data consisting of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date are: The winter wheat yield matrix data corresponding to the planting date of historical training data;

[0098] is the covariance of the new input data and the training data, is the covariance of the training data and the new input data, is the covariance between the training data of each year, is the covariance of the new input data itself;

[0099] The calculation formula of each covariance is expressed as:

[0100] ;

[0101] In the formula Corresponding to the multi-dimensional Gaussian process model construction function formula or , is the variance of all known production data, is the length scale constant.

[0102] S500, improved planting date sampling method based on mathematical expectation.

[0103] Based on the training yield data and planting date data, a planting date sampling function based on mathematical expectation improvement is established to quickly retrieve the planting date, thereby replacing the existing trial and error method and solving the problem of slow calculation speed caused by the trial and error method of large-scale calculation of simulated yield on each planting day. The trial and error method usually requires daily numerical simulation of crops from September 15th to November 15th, which requires a lot of computing resources in high-precision grid calculations. The mathematical expectation sampling method can reduce the amount of calculation exponentially and improve the calculation efficiency by simulating crops based on a small number of planting dates with the largest mathematical expectation of yield from September 15th to November 15th. The specific method is:

[0104] S510, according to the multidimensional Gaussian process model constructed in step S400, a planting date sampling function based on mathematical expectation improvement is established for rapid retrieval of planting dates; the established planting date sampling function based on mathematical expectation improvement is expressed as:

[0105] ;

[0106] ;

[0107] In the formula, is the mathematical expectation of the yield corresponding to the planting date, and are the standard deviation and mean of the yield corresponding to the planting date in the Gaussian function model, are the cumulative distribution function and probability density function of the yield corresponding to the planting date, is the maximum value of known yield. The last planting date sampling value is the planting date corresponding to the maximum mathematical expectation of yield. The planting date range is September 15th to November 15th.

[0108] S520, selecting a planting date with the largest mathematical expectation from the winter wheat planting date interval, and inputting it into a crop model to simulate yield;

[0109] S530, using the simulated yield and the selected planting date as training data again to update the multidimensional Gaussian process model of step S400;

[0110] S540, repeating steps S510 to S530, after the number of sequential times is met, verifying whether the simulation result meets the accumulated temperature planting condition, if so, outputting the planting date as the suitable sowing date;

[0111] S550: If the accumulated temperature condition for winter wheat planting is not met, repeat steps S520 to S540 until the accumulated temperature condition for winter wheat planting is met, and output the most suitable sowing date for winter wheat.

[0112] This example uses the optimum sowing period of winter wheat in Jiangsu Province from 1995 to 1996 as an example and calculates the optimum sowing period using the above method with a grid resolution of 10 km. The results are as follows: Figure 2 As shown, the calculation time of the result is only one tenth of that of the trial and error method. It can be seen that the calculation method of the present invention is particularly suitable for high-precision calculation of suitable sowing period distribution, which is beneficial to improving the quality of agriculture-related services.

[0113] Embodiment 2: This embodiment provides a non-transitory computer-readable storage medium on which computer instructions are stored. The computer instructions enable a computer to execute the artificial intelligence-based method for quickly calculating the optimal sowing date for winter wheat of Embodiment 1.

[0114] Embodiment 3: This embodiment provides an electronic device, which may include: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus. The processor may call the logic instructions in the memory to execute the method for quickly calculating the optimal sowing date of winter wheat based on artificial intelligence in Embodiment 1.

[0115] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0116] Example 4: This example provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the artificial intelligence-based method for quickly calculating the optimal sowing date for winter wheat in Example 1.

[0117] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0118] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quickly calculating the optimal sowing date of winter wheat based on artificial intelligence, characterized in that: include: S100, establishing an artificial intelligence training data set based on historical meteorological observation data, historical winter wheat yield data, and winter wheat sowing time data; Design an interpolation algorithm that conforms to the changing laws of weather processes to perform quality control on missing or abnormal data of temperature, precipitation and solar radiation in historical meteorological observation data; Conduct quality control on statistical errors in long-term series of historical winter wheat yield data to eliminate the impact of improved planting techniques and crop variety improvements on historical winter wheat yield data; S200, using the quality-controlled artificial intelligence training data set, constructing a multidimensional Gaussian process model for optimizing the calculation of the optimal sowing date; the multidimensional Gaussian process model uses yield data as the target variable, and meteorological data and sowing time data as feature data; the constructed multidimensional Gaussian process model is expressed as: Where H(x) represents a multidimensional Gaussian function, μ is the mean of the Gaussian function, σ is the standard deviation of the Gaussian function, X is the planting date composed of historical training data, and the matrix data composed of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date, x is the newly input planting date, and the matrix data composed of the maximum temperature, minimum temperature, precipitation and solar radiation on the planting date, and Y is the winter wheat yield matrix data corresponding to the planting date of historical training data; K(x,X) is the covariance between the new input data and the training data, K(X,x) is the covariance between the training data and the new input data, K(X,X) is the covariance between the training data of each year, and K(x,x) is the covariance of the new input data itself; The calculation formula of each covariance is expressed as: Where a and b correspond to x or X in the function formula of the multidimensional Gaussian process model, is the variance of all known production data, l is the length scale constant; S300, based on the constructed multidimensional Gaussian process model, establish a planting date sampling function based on mathematical expectation improvement, sample planting dates from the winter wheat planting date interval, perform numerical simulation of crop yields with the sampled planting dates, and then feed back the simulated yields of the sampled planting dates and the corresponding crop planting dates to the multidimensional Gaussian process model to update the multidimensional Gaussian process model; perform cyclic iterations, and when the number of cycles required and the accumulated temperature conditions for winter wheat planting are met, use the sampled planting date as the optimal sowing date for winter wheat; If the condition is not met, step S300 is repeated until the accumulated temperature condition for winter wheat planting is met, and the most suitable sowing date for winter wheat is output.

2. The method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence according to claim 1, characterized in that: In step S100, a difference algorithm that conforms to the changing law of weather processes is designed to perform quality control on the missing or abnormal data of temperature, precipitation and solar radiation in the historical meteorological observation data; including: Consider wind direction and speed to perform quality control on air temperature data; Consider the changes in sky clouds and conduct quality control on precipitation data; The solar radiation data were quality controlled by considering the changes in near-surface relative humidity.

3. The method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence according to claim 2, characterized in that: The wind direction and wind speed are considered to control the quality of the temperature data, specifically: The temperature difference function is expressed as: In the formula, it is assumed that the temperatures of the two time series are T1 and T2, the wind speed is v, and the wind direction is θ; Where k is the proportionality coefficient, ranging from 0.1 to 0.3; The above-mentioned consideration of the cloud system changes in the sky and the quality control of precipitation data include: The precipitation difference function is calculated by the precipitation difference and the specific humidity weighted adjustment factor, which is expressed as: ΔP=(P2-P1)·f(q 500 ,q 850 ); Assume that the precipitation of the two time series is P1 and P2, and the relative humidity at 500hPa and 850hPa is q 500 and q 850 , f(q 500 ,q 850 ) is the weighted adjustment factor for specific humidity, expressed as: In the formula, w 500 and w 850 are the weights of the relative humidity at 500hPa and 850hPa, q ref It is the relative humidity value of 500hPa or 850hPa in the typical atmosphere of the station climate state, which is used for function normalization; The solar radiation data is quality controlled by considering the change of near-ground relative humidity. The solar radiation difference function is expressed as: Assume that the precipitation of the two time series is R1 and R2, RH1 and RH2 are the two relative humidity, RH ref is the reference relative humidity value, and α is the sensitivity coefficient of humidity to radiation attenuation.

4. The method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence according to claim 1, characterized in that: In step S100, the statistical errors in the long-term series of historical winter wheat yield data are quality controlled to eliminate the impact of planting technology improvement and crop variety improvement on the historical winter wheat yield data; specifically: Considering the impact of improved planting technology and crop variety on yield data, a yield function is established, which is expressed as: Among them, β0+β1·T t is a linear function of the time-varying trend of output, which is used to represent the impact of technological improvement on output. β0 and β1 are the intercept and slope of the linear function of the output data time series, respectively. t is the independent variable time; C i,t is a dummy variable for the variety category, is the time segmentation function of local variety change, n is the total number of variety changes, C i,t The date of a single product change. Used to eliminate the contribution of different varieties to production; f(X clin ) is a function of climate factors, obtained by linear function fitting using temperature, precipitation and radiation meteorological data; ∈ t is a random error; After fitting the original yield according to the formula, the effects of technological improvements and varieties are eliminated, and the part related to climate factors is retained. This method is used to control the data quality of historical yield data.

5. The method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence according to claim 1, characterized in that: The specific method of step S300 is: S310, according to the multidimensional Gaussian process model constructed in step S200, establishing a planting date sampling function based on mathematical expectation improvement, and retrieving the planting date; S320, selecting a planting date with the largest mathematical expectation from the winter wheat planting date interval, and inputting it into a crop model to simulate yield; S330, using the simulated yield and the selected planting date as training data again to update the multidimensional Gaussian process model of step S200; S340, repeating steps S310 to S330, after the number of sequential times is met, verifying whether the simulation result meets the accumulated temperature planting condition, if so, outputting the planting date as the suitable sowing date; S350: If the accumulated temperature condition for winter wheat planting is not met, repeat steps S320 to S340 until the accumulated temperature condition for winter wheat planting is met, and output the most suitable sowing date for winter wheat.

6. The method for rapidly calculating the optimum sowing date of winter wheat based on artificial intelligence according to claim 5, characterized in that: The improved planting date sampling function based on mathematical expectation established in step S310 is expressed as: EI(D)=σ(D)·[z(D)·Φ(z(D))+φ(z(D))]; Where EI(D) is the mathematical expectation of the yield corresponding to the planting date, σ(D) and μ(D) are the standard deviation and mean of the yield corresponding to the planting date in the Gaussian function model, Φ and φ are the cumulative distribution function and probability density function of the yield corresponding to the planting date, max(Y) is the maximum known yield; the last planting date sampling value is the planting date corresponding to the maximum mathematical expectation of yield.

7. A non-transitory computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and the computer instructions enable the computer to execute the method for quickly calculating the optimal sowing date of winter wheat based on artificial intelligence as described in any one of claims 1 to 6.

8. An electronic device, characterized in that: include: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, and the processor calls the logic instructions in the memory to execute the artificial intelligence-based method for quickly calculating the optimal sowing date of winter wheat as described in any one of claims 1 to 6.

9. A computer program product, characterized in that The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer executes the artificial intelligence-based method for quickly calculating the optimal sowing date for winter wheat as described in any one of claims 1 to 6.

Images