Multi-algorithm fusion and multi-mode integration air temperature prediction method and related device

Abstract

The invention provides a multi-algorithm fusion and multi-mode integration air temperature prediction method and a related device. The method comprises the following steps: according to a mode forecast value of an observation station at a forecast moment, an observation live value of the observation station at the moment same as the forecast moment in the previous day, a mode forecast value of the observation station at the moment same as the forecast moment in the previous day, and a forecast deviation value of the observation station at the moment same as the forecast moment in the previous day; adopting a plurality of air temperature forecasting algorithms to construct a first air temperature forecasting model, and obtaining a first air temperature forecasting result of the observation station at the forecasting moment; according to the first air temperature forecasting result of the observation station at the forecasting moment, utilizing a dynamic weighted set average algorithm to construct a second air temperature forecasting model, and obtaining a second air temperature forecasting result of the observation station at the forecasting moment; and determining the result as an air temperature forecasting product of the observation station at the forecasting moment. The method can meet the requirement for precise forecasting of the air temperature in the complex terrain area, and the accuracy and stability of an air temperature forecasting product are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature prediction, and in particular to a temperature prediction method integrating multiple algorithms and multiple modes and a related device. Background Art

[0002] Temperature forecast is an important part of meteorological services and is of great significance to agricultural production, energy supply, forest fire prevention, traffic management, public health, daily life and other fields. Affected by global warming, accelerated urbanization and human activities, meteorological disasters directly related to temperature, such as high temperature drought, low temperature rain, snow and freezing, have occurred frequently, and society has put forward higher requirements for the accuracy and stability of temperature forecasts. Especially in complex terrain areas such as mountainous areas, the terrain has a significant impact on the temperature. In addition, the weather system is complex and it is difficult to accurately forecast the temperature. Therefore, improving the accuracy and stability of temperature forecasts in complex terrain areas has become an urgent problem to be solved. Summary of the invention

[0003] In view of the defects in the prior art, the present invention provides a temperature prediction method and related devices integrating multiple algorithms and multiple modes to improve the accuracy and stability of temperature forecast products in complex terrain areas.

[0004] A temperature prediction method integrating multiple algorithms and multiple models, including:

[0005] According to the model forecast value of the observation station at the forecast time, the actual observation value of the observation station at the same time as the forecast time the day before, the model forecast value of the observation station at the same time as the forecast time the day before, and the forecast deviation value of the observation station at the same time as the forecast time the day before, a plurality of temperature forecast algorithms are used to construct a first temperature forecast model, and obtain the first temperature forecast result of the observation station at the forecast time;

[0006] According to the first temperature forecast result of the observation station at the forecast time, a second temperature forecast model is constructed using a dynamic weighted ensemble average algorithm to obtain the second temperature forecast result of the observation station at the forecast time;

[0007] The second temperature forecast result is determined as the temperature forecast product of the observation station at the forecast time.

[0008] Furthermore, the method further comprises:

[0009] The bilinear interpolation method is used to interpolate the temperature forecast products of the target area in the numerical forecast model to the observation site at the forecast time to obtain the model forecast value of the observation site at the forecast time.

[0010] Furthermore, the numerical prediction models include the European Centre for Medium-Range Weather Forecasts global numerical prediction model, the China Meteorological Administration global numerical prediction model, the China Meteorological Administration regional numerical prediction model and the Chongqing mesoscale regional numerical prediction model.

[0011] Furthermore, the various temperature forecasting algorithms include a univariate linear regression equation, a bivariate linear regression equation, a ternary linear regression equation, a deviation linear regression equation, a model forecast and a deviation linear regression equation.

[0012] Furthermore, according to the model forecast value of the observation station at the forecast time, a plurality of temperature forecast algorithms are used to construct a first temperature forecast model, and the first temperature forecast result of the observation station at the forecast time is obtained, including:

[0013] According to the model forecast value of the observation station at the forecast time, the first temperature forecast model is constructed using the univariate linear regression equation, the bivariate linear regression equation, the ternary linear regression equation, the deviation linear regression equation, the model forecast and the deviation linear regression equation to obtain the first temperature forecast result of the observation station at the forecast time, where:

[0014] The linear regression equation is: F'(t+n)=a 1 F(t+n)+a 0

[0015] The binary linear regression equation is: F'(t+n)=a 1 F(t+n)+a 2 O(t 0 +n 0 )+a 0

[0016] The three-variable linear regression equation is: F'(t+n)=a 1 F(t+n)+a 2 O(t 0 +n 0 )+a 3 F(t 0 +n 0 )+a 0

[0017] The deviation linear regression equation is: F′(t+n)=F(t+n)-a 1 B(t 0 +n 0 )+a 0

[0018] The linear regression equation of model forecast and deviation is: F′(t+n)=a 1 F(t+n)+a 2 B(t 0 +n 0 )+a0

[0019] In the formula, F′(t+n) represents the first temperature forecast result of the observation station at the forecast time t+n, F(t+n) represents the model forecast value of the observation station at the forecast time t+n, and a 1 、a 2 、a 3 、a 0 represents the regression coefficient calculated based on the least squares method, O(t 0 +n 0 ) represents the actual value of the observation station at the same time as the forecast time t+n on the previous day, F(t 0 +n 0 ) represents the model forecast value of the observation station at the same time as the forecast time t+n on the previous day, B(t 0 +n 0 ) represents the forecast deviation value of the observation station at the same time as the forecast time t+n on the previous day.

[0020] Furthermore, according to the first temperature forecast result of the observation station at the forecast time, a second temperature forecast model is constructed using a dynamic weighted ensemble average algorithm, including:

[0021] According to the first temperature forecast result of the observation station at the forecast time, the second temperature forecast model is constructed using the method shown in the following formula:

[0022]

[0023] In the formula, MAE ij represents the average absolute error between the first temperature forecast results of all samples and the observed actual values ​​under the jth temperature forecast algorithm of the first temperature forecast model in the i-th numerical forecast mode within the sliding period; M represents the number of samples within the sliding period, k is the sample sequence number within the sliding period; F / ij,k is the first temperature forecast result of the kth sample under the jth temperature forecast algorithm of the first temperature forecast model by the i-th numerical forecast model; k is the observed actual value of the kth sample, w ij is the weight coefficient of the i-th numerical forecast model to the j-th temperature forecast algorithm of the first temperature forecast model, F / ij is the first temperature forecast result of the ith numerical forecast model under the jth temperature forecast algorithm in the first temperature forecast model, F WEM is the second temperature forecast result of the observation station at the forecast time output by the second temperature forecast model, and X is the product of the type of numerical forecast model participating in the ensemble average and the types of multiple temperature forecast algorithms in the first temperature forecast model.

[0024] A temperature prediction device integrating multiple algorithms and multiple modes, comprising:

[0025] The first forecasting unit is used to construct a first temperature forecast model using a plurality of temperature forecasting algorithms based on the model forecast value of the observation station at the forecast time, the observed actual value of the observation station at the same time as the forecast time the previous day, the model forecast value of the observation station at the same time as the forecast time the previous day, and the forecast deviation value of the observation station at the same time as the forecast time the previous day, and obtain the first temperature forecast result of the observation station at the forecast time;

[0026] The second forecasting unit is used to construct a second temperature forecasting model based on the first temperature forecasting result of the observation station at the forecasting time by using a dynamic weighted ensemble average algorithm to obtain the second temperature forecasting result of the observation station at the forecasting time;

[0027] The determination unit is used to determine the second temperature forecast result as the temperature forecast product of the observation site at the forecast time.

[0028] A device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of a temperature prediction method integrating multiple algorithms and multiple modes as described above are implemented.

[0029] A storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a temperature prediction method integrating multiple algorithms and multiple modes as described above are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the specific embodiments of the present invention, the following will briefly introduce the drawings required for the specific embodiments or the prior art description. In all the drawings, each element or part is not necessarily drawn according to the actual scale.

[0031] Figure 1 A flow chart of a temperature prediction method integrating multiple algorithms and multiple modes provided in one embodiment of the present invention;

[0032] Figure 2 A schematic diagram of temperature forecast of an observation station model in a bilinear interpolation process provided by an embodiment of the present invention;

[0033] Figure 3 A schematic diagram of the structure of a temperature prediction device integrating multiple algorithms and multiple modes provided by an embodiment of the present invention;

[0034] In the attached figure, (x, y) is the coordinate of the corresponding position of the observation station, (x 1 ,y 1 )、(x2 ,y 1 )、(x 1 ,y 2 )、(x 2 ,y 2 ) are the coordinates of the corresponding positions of the four grid points at the lower left, lower right, upper left, and upper right of the observation station near the forecast time, (x, y 1 )、(x,y 2 ) are the coordinates of the corresponding positions of the two middle grid points whose horizontal coordinates are the same as those of the observation site. DETAILED DESCRIPTION

[0035] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0036] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the invention belongs.

[0037] In one embodiment, if Figure 1 As shown, a temperature prediction method integrating multiple algorithms and multiple models is provided, including:

[0038] 1. Based on the model forecast value of the observation station at the forecast time, the actual observation value of the observation station at the same time as the forecast time the day before, the model forecast value of the observation station at the same time as the forecast time the day before, and the forecast deviation value of the observation station at the same time as the forecast time the day before, a plurality of temperature forecast algorithms are used to construct a first temperature forecast model and obtain the first temperature forecast result of the observation station at the forecast time;

[0039] Preferably, a bilinear interpolation method is used to interpolate the temperature forecast product of the numerical forecast model for the target area to the observation site to obtain the model forecast value of the observation site at the forecast time.

[0040] Preferably, the numerical forecast models include the European Centre for Medium-Range Weather Forecasts Global Numerical Forecast Model (EC for short), the China Meteorological Administration Global Numerical Forecast Model (CMA-GFS for short), the China Meteorological Administration Regional Numerical Forecast Model (CMA-MESO for short) and the Chongqing Mesoscale Regional Numerical Forecast Model (CQ-MESO for short).

[0041] Actual business applications show that both forecasters and forecast users usually refer to multiple numerical forecast models at the same time in their work, because each model has its own advantages, and different numerical forecast models have different forecast performance for different regions and different forecast time periods. For example, the global numerical forecast model of the European Center for Medium-Range Weather Forecasts is relatively stable in forecasting large-scale weather systems, but its forecasting capabilities for sudden or local weather events are often insufficient; while the regional high-resolution numerical model is relatively strong in forecasting small and medium-scale convective weather, but its forecast for large-scale weather systems is not as stable as the global numerical forecast model.

[0042] At present, in the weather forecast business, the station observation data with high credibility are mostly used to carry out the model statistical interpretation (MOS) on the station to improve the forecasting skills. The MOS method can effectively reduce the systematic deviation in the temperature forecast by establishing the statistical relationship between the numerical model output and the historical observation data. Based on the station MOS forecast, the grid temperature MOS temperature guidance product can also be generated through the CRESSMAN step-by-step correction method. At present, the difficulty of the MOS forecast method lies in the optimization of the model and the optimization of the selected forecast factors, and many studies are to use the numerical model forecast product to establish a statistical relationship with the site actual situation at the same time to carry out the site correction forecast, and the relationship between the site actual situation at the historical moment, the model forecast, the model deviation and the current forecast has not been considered. If the optimal forecast factor can be extracted, different MOS forecast algorithms are used, and their advantages are integrated, the temperature forecast performance will be significantly improved. Therefore, the present invention adopts the method of multi-mode integration and multi-algorithm fusion, by integrating the correction results of multiple numerical forecast models under different forecast algorithms, reducing the uncertainty in the single numerical model forecast, integrating the forecast advantages of different algorithms, to obtain a forecast result with higher accuracy and stability.

[0043] In numerical weather forecasting, temperature forecast products are usually generated on a two-dimensional or three-dimensional grid system that covers the forecast area (i.e., the target area). Each grid point represents a specific geographic location and contains the model forecast values ​​of meteorological elements (such as temperature, precipitation, wind speed, etc.) at that location.

[0044] The bilinear interpolation method is a commonly used numerical analysis method for interpolating data from grid points to non-grid points (such as observation sites). When a numerical forecast model predicts the temperature of a target area, it generates a series of temperature forecast values ​​on a grid system. Then, in order to obtain the model temperature forecast at the observation site (i.e., the model forecast value), the bilinear interpolation method is used to interpolate the temperature forecast values ​​on these grid points to the location of the observation site.

[0045] Specifically, Figure 2As shown in the figure, the bilinear interpolation method is used to interpolate the temperature forecast product of the numerical forecast model for the target area to the observation site to obtain the model temperature forecast at the observation site in the target area, which is called the model forecast value (DMO).

[0046]

[0047] Where F(x,y) is the model forecast value of the observation station at the forecast time, (x,y) is the coordinates of the corresponding position of the observation station; F(x 1 ,y 1 )、F(x 2 ,y 1 )、F(x 1 ,y 2 )、F(x 2 ,y 2 ) are the model forecast values ​​of the four grid points at the lower left, lower right, upper left, and upper right near the observation station at the forecast time, (x 1 ,y 1 )、(x 2 ,y 1 )、(x 1 ,y 2 )、(x 2 ,y 2 ) are the coordinates of the four grid points; F(x,y 1 )、F(x,y 2 ) are the model forecast values ​​of the two middle grid points whose abscissas are the same as those of the observation sites.

[0048] The first temperature forecast model inputs the model forecast value of the observation site at the forecast time, the actual observation value of the observation site at the same time as the forecast time the day before, the model forecast value of the observation site at the same time as the forecast time the day before, and the forecast deviation value of the observation site at the same time as the forecast time the day before, and the output is the first temperature forecast result of the observation site at the forecast time.

[0049] The date of the forecast time is the forecast day, which includes multiple forecast times. The forecast day refers to the date of specific release or prediction in the weather forecast, and the forecast time refers to the specific release or prediction time point in the weather forecast. In order to provide more detailed and timely weather information, a forecast day usually sets multiple forecast times to release weather conditions in different time periods.

[0050] Preferably, it also includes adopting a quasi-symmetric mixed sliding training period method to construct a sample data set along with the sliding training period of the forecast start day to pre-train the first temperature forecast model.

[0051] The first day of the forecast period is called the forecast start day.

[0052] Preferably, the training period comprises 3n days in total.

[0053] Specifically, the training period includes:

[0054] n days before the forecast start date of the current year;

[0055] n days before the forecast start date in the previous year;

[0056] The day of the previous year that is the same as the forecast start date plus the n-1 days after the day of the previous year that is the same as the forecast start date.

[0057] The sample data set specifically includes:

[0058] The actual observation value, model forecast value, and forecast deviation value of the observation station at the same time as the forecast time n days before the forecast start date;

[0059] The actual observation value, model forecast value, and forecast deviation value of the observation station at the same time as the forecast time n days before the forecast start date in the previous year;

[0060] The actual observation value, model forecast value, and forecast deviation value of the observation station at the same time as the forecast time on the same day as the forecast start date in the previous year, and the actual observation value, model forecast value, and forecast deviation value at the same time as the forecast time on the n-1 days after the same day as the forecast start date in the previous year.

[0061] Preferably, the multiple temperature forecasting algorithms include a univariate linear regression equation, a bivariate linear regression equation, a ternary linear regression equation, a deviation linear regression equation, a model forecast and a deviation linear regression equation.

[0062] Preferably, according to the model forecast value of the observation site at the forecast time, a plurality of temperature forecast algorithms are used to construct a first temperature forecast model to obtain a first temperature forecast result of the observation site at the forecast time, including:

[0063] According to the model forecast value of the observation station at the forecast time, the first temperature forecast model is constructed using the univariate linear regression equation, the bivariate linear regression equation, the ternary linear regression equation, the deviation linear regression equation, the model forecast and the deviation linear regression equation to obtain the first temperature forecast result of the observation station at the forecast time, where:

[0064] The linear regression equation is: F'(t+n)=a 1 F(t+n)+a 0

[0065] The binary linear regression equation is: F'(t+n)=a 1 F(t+n)+a 2 O(t 0+n 0 )+a 0

[0066] The three-variable linear regression equation is: F'(t+n)=a 1 F(t+n)+a 2 O(t 0 +n 0 )+a 3 F(t 0 +n 0 )+a 0

[0067] The deviation linear regression equation is: F′(t+n)=F(t+n)-a 1 B(t 0 +n 0 )+a 0

[0068] The linear regression equation of model forecast and deviation is: F′(t+n)=a 1 F(t+n)+a 2 B(t 0 +n 0 )+a 0

[0069] In the formula, F′(t+n) represents the first temperature forecast result of the observation station at the forecast time t+n, F(t+n) represents the model forecast value of the observation station at the forecast time t+n, and a 1 、a 2 、a 3 、a 0 represents the regression coefficient calculated based on the least squares method, O(t 0 +n 0 ) represents the actual value of the observation station at the same time as the forecast time t+n on the previous day, F(t 0 +n 0 ) represents the model forecast value of the observation station at the same time as the forecast time t+n on the previous day, B(t 0 +n 0 ) represents the forecast deviation value of the observation station at the same time as the forecast time t+n on the previous day.

[0070] Preferably, B(t 0 +n 0 ) is calculated using the following formula:

[0071] B(t 0 +n 0 )=F(t 0 +n 0 )-O(t 0 +n 0 )

[0072] In the formula, t represents the start time of the report, t+n represents the forecast time, and t 0 = t–24h is the starting time of the previous day (the day before), t 0 +n 0 Indicates the time on the previous day that is the same as the forecast time t+n.

[0073] Preferably, it also includes constructing a historical sample data set according to the optimal training period n, solving the linear regression equations in multiple temperature forecast algorithms, and obtaining the optimal first temperature forecast model (including 5 linear regression algorithms).

[0074] The steps for selecting the optimal training period n are as follows:

[0075] The value range of n is set to 5 to 60 days with an interval of 5 days. A 12-category training period historical sample data set is constructed with the sliding of the forecast start date. The five linear regression equations in the first temperature forecast model (the single-variable linear regression equation, the binary linear regression equation, the ternary linear regression equation, the deviation linear regression equation, the pattern forecast and the deviation linear regression equation) are trained and solved respectively to obtain the regression coefficients in the first temperature forecast model.

[0076] The first temperature forecast model is then used to revise the forecast value of the model temperature forecast of the previous year to obtain the first temperature forecast result of the previous year, and then it is tested and evaluated. When the first temperature forecast result of the previous year is closest to the actual temperature of the previous year, it is the optimal training period. The specific test and evaluation indicator is the mean absolute error, which is calculated as follows:

[0077]

[0078] In the formula, F / i,j , O i,j are the first temperature forecast value under the jth temperature forecast algorithm on the i-th day of the previous year and the observed actual value at the same time, N is the number of days in the previous year, MAE j is the mean absolute error of the first temperature forecast value of the jth temperature forecast algorithm in the previous year. i When , is the smallest, the training period at this time is the optimal training period, thus 5 optimal training periods can be obtained, thereby 5 optimal linear regression equations, that is, 5 optimal temperature forecasting algorithms, thereby constructing the optimal first temperature forecast model.

[0079] Then the model forecast value of the observation site at the forecast time, the actual observation value at the same time as the forecast time the previous day, the model forecast value at the same time as the forecast time the previous day, and the forecast deviation value at the same time as the forecast time the previous day are input into the first temperature forecast model, and the first temperature forecast result of the observation site at the forecast time is output.

[0080] 2. According to the first temperature forecast result of the observation station at the forecast time, a second temperature forecast model is constructed using a dynamic weighted ensemble average algorithm to obtain the second temperature forecast result of the observation station at the forecast time.

[0081] According to the first temperature forecast result of the observation station at the forecast time, the second temperature forecast model is constructed using the method shown in the following formula:

[0082]

[0083] In the formula, MAE ij represents the average absolute error between the first temperature forecast results of all samples and the observed actual values ​​under the jth temperature forecast algorithm of the first temperature forecast model in the i-th numerical forecast mode within the sliding period; M represents the number of samples within the sliding period, k is the sample sequence number within the sliding period; F / ij,k is the first temperature forecast result of the kth sample under the jth temperature forecast algorithm of the first temperature forecast model under the ith numerical forecast model (each numerical forecast model has 5 first temperature forecast results, and each linear regression equation algorithm corresponds to 1 result); k is the observed actual value of the kth sample, w ij is the weight coefficient of the i-th numerical forecast model to the j-th temperature forecast algorithm of the first temperature forecast model, F / ij is the first temperature forecast result of the ith numerical forecast model under the jth temperature forecast algorithm in the first temperature forecast model, F WEM is the second temperature forecast result of the observation station at the forecast time output by the second temperature forecast model, and X is the product of the types of numerical forecast models participating in the ensemble average and the types of multiple temperature forecast algorithms in the first temperature forecast model (in fact, i=4, because there are four numerical forecast models; j=5, because there are 5 algorithms in the first temperature forecast model, and the final second temperature forecast result is a fusion integration of 20=4*5 first temperature forecast results).

[0084] Preferably, the sliding period of the second temperature forecast model is 15 days, and MAE ij It is the average absolute error between the first temperature forecast results of all samples and the observed actual values ​​under the jth linear regression equation of the first temperature forecast model of the observation station within 15 days before the forecast time under the i-th numerical forecast model at the same time as the forecast time.

[0085] The input of the second temperature forecast model is the first temperature forecast result of the observation site at the forecast time, and the output is the second temperature forecast result of the observation site at the forecast time.

[0086] 3. The second temperature forecast result is determined as the temperature forecast product of the observation station at the forecast time.

[0087] The present invention effectively integrates the temperature forecast results from different modes and different algorithms, and adopts dynamic weight allocation to obtain the best temperature prediction effect. It also comprehensively considers the relationship between the temperature forecast products of the numerical forecast model, the observed actual values ​​and the forecast deviation values, thereby improving the practicality and accuracy of the temperature forecast products.

[0088] In one embodiment, if Figure 3 As shown, a temperature prediction device integrating multiple algorithms and multiple modes is provided, comprising:

[0089] The first forecasting unit is used to construct a first temperature forecast model using a plurality of temperature forecasting algorithms according to the model forecast value of the observation station at the forecast time, the observed actual value of the observation station at the same time as the forecast time on the previous day, the model forecast value of the observation station at the same time as the forecast time on the previous day, and the forecast deviation value of the observation station at the same time as the forecast time on the previous day, so as to obtain the first temperature forecast result of the observation station at the forecast time;

[0090] The second forecasting unit is used to construct a second temperature forecasting model based on the first temperature forecasting result of the observation station at the forecasting time by using a dynamic weighted ensemble average algorithm to obtain the second temperature forecasting result of the observation station at the forecasting time;

[0091] The determination unit is used to determine the second temperature forecast result as the temperature forecast product of the observation site at the forecast time.

[0092] A device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of a temperature prediction method integrating multiple algorithms and multiple modes as described above are implemented.

[0093] A storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a temperature prediction method integrating multiple algorithms and multiple modes as described above are implemented.

[0094] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0095] 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 replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A temperature prediction method integrating multiple algorithms and multiple models, characterized in that: include: According to the model forecast value of the observation station at the forecast time, the actual observation value of the observation station at the same time as the forecast time the day before, the model forecast value of the observation station at the same time as the forecast time the day before, and the forecast deviation value of the observation station at the same time as the forecast time the day before, a plurality of temperature forecast algorithms are used to construct a first temperature forecast model, and obtain the first temperature forecast result of the observation station at the forecast time; According to the first temperature forecast result of the observation site at the forecast time, a second temperature forecast model is constructed using a dynamic weighted ensemble average algorithm to obtain the second temperature forecast result of the observation site at the forecast time; The second temperature forecast result is determined as the temperature forecast product of the observation site at the forecast time.

2. A temperature prediction method integrating multiple algorithms and multiple modes as claimed in claim 1, characterized in that: The method further comprises: The bilinear interpolation method is used to interpolate the temperature forecast product of the target area in the numerical forecast model at the forecast time to the observation site to obtain the model forecast value of the observation site at the forecast time.

3. A temperature prediction method integrating multiple algorithms and multiple modes as claimed in claim 2, characterized in that: The numerical prediction models include the European Centre for Medium-Range Weather Forecasts global numerical prediction model, the China Meteorological Administration global numerical prediction model, the China Meteorological Administration regional numerical prediction model and the Chongqing mesoscale regional numerical prediction model.

4. The temperature prediction method integrating multiple algorithms and multiple modes as claimed in claim 1, characterized in that: The multiple temperature forecasting algorithms include a univariate linear regression equation, a bivariate linear regression equation, a ternary linear regression equation, a deviation linear regression equation, a model forecast and a deviation linear regression equation.

5. A temperature prediction method integrating multiple algorithms and multiple modes as claimed in claim 4, characterized in that: The method uses a plurality of temperature forecast algorithms to construct a first temperature forecast model based on the model forecast value of the observation station at the forecast time, the observed actual value of the observation station at the same time as the forecast time the day before, the model forecast value of the observation station at the same time as the forecast time the day before, and the forecast deviation value of the observation station at the same time as the forecast time the day before, and obtains the first temperature forecast result of the observation station at the forecast time, including: According to the model forecast value of the observation station at the forecast time, the actual observation value of the observation station at the same time as the forecast time the day before, the model forecast value of the observation station at the same time as the forecast time the day before, and the forecast deviation value of the observation station at the same time as the forecast time the day before, the first temperature forecast model is constructed using a univariate linear regression equation, a binary linear regression equation, a ternary linear regression equation, a deviation linear regression equation, a model forecast and a deviation linear regression equation to obtain the first temperature forecast result of the observation station at the forecast time, where: The linear regression equation is: F′(t+n)=a1F(t+n)+a0 The binary linear regression equation is: F′(t+n)=a1F(t+n)+a2O(t0+n0)+a0 The three-variable linear regression equation is: F′(t+n)=a1F(t+n)+a2O(t0+n0)+a3F(t0+n0)+a0 The deviation linear regression equation is: F′(t+n)=F(t+n)-a1B(t0+n0)+a0 The linear regression equation of model forecast and deviation is: F′(t+n)=a1F(t+n)+a2B(t0+n0)+a0 Where F′(t+n) represents the first temperature forecast result of the observation station at the forecast time t+n, F(t+n) represents the model forecast value of the observation station at the forecast time t+n, a1, a2, a3, a0 represent the regression coefficients calculated based on the least squares method, O(t0+n0) represents the observed actual value of the observation station at the same time as the forecast time t+n on the previous day, F(t0+n0) represents the model forecast value of the observation station at the same time as the forecast time t+n on the previous day, and B(t0+n0) represents the forecast deviation value of the observation station at the same time as the forecast time t+n on the previous day.

6. A temperature prediction method integrating multiple algorithms and multiple modes as described in any one of claims 1 to 5, characterized in that: The second temperature forecast model is constructed based on the first temperature forecast result of the observation station at the forecast time by using a dynamic weighted ensemble average algorithm, including: According to the first temperature forecast result of the observation station at the forecast time, the second temperature forecast model is constructed using the method shown in the following formula: In the formula, MAE ij represents the average absolute error between the first temperature forecast results of all samples and the observed actual values ​​under the jth temperature forecast algorithm of the first temperature forecast model in the i-th numerical forecast mode within the sliding period; M represents the number of samples within the sliding period, k is the sample sequence number within the sliding period; F / ij,k is the first temperature forecast result of the kth sample under the jth temperature forecast algorithm of the first temperature forecast model by the i-th numerical forecast model; k is the observed actual value of the kth sample, w ij is the weight coefficient of the i-th numerical forecast model to the j-th temperature forecast algorithm of the first temperature forecast model, F / ij is the first temperature forecast result of the ith numerical forecast model under the jth temperature forecast algorithm in the first temperature forecast model, F WEM is the second temperature forecast result of the observation station at the forecast time output by the second temperature forecast model, and X is the product of the type of numerical forecast model participating in the ensemble average and the types of multiple temperature forecast algorithms in the first temperature forecast model.

7. A temperature prediction device integrating multiple algorithms and multiple modes, characterized in that: It includes a first forecasting unit, a second forecasting unit and a determining unit, wherein: The first forecasting unit is used to construct a first temperature forecast model using multiple temperature forecasting algorithms according to the model forecast value of the observation station at the forecast time, the observed actual value of the observation station at the same time as the forecast time the previous day, the model forecast value of the observation station at the same time as the forecast time the previous day, and the forecast deviation value of the observation station at the same time as the forecast time the previous day, so as to obtain the first temperature forecast result of the observation station at the forecast time; The second forecasting unit is used to construct a second temperature forecast model based on the first temperature forecast result of the observation site at the forecast time by using a dynamic weighted ensemble average algorithm to obtain the second temperature forecast result of the observation site at the forecast time; The determining unit is used to determine the second temperature forecast result as the temperature forecast product of the observation site at the forecast time.

8. A device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 6 when executing the computer program.

9. A storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images