Aerosol optical thickness forecasting method and related equipment
By decomposing and modeling the measured values of the aerosol optical thickness, combined with time series forecasting and residual prediction models, the problem of lack of time dependence and physical driving factors for AOD forecasting in the prior art is solved, and a high-precision aerosol optical thickness forecasting is achieved.
Patent Information
- Application Number
- CN202411728039.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-03-07
AI Technical Summary
The lack of aerosol optical thickness (AOD) data forecasting methods that can effectively consider time dependence and physical drivers simultaneously, limiting accurate forecasting of AOD and its climate and environmental impacts.
By obtaining the time series of measured values of the optical thickness of the aerosol, decomposition is performed to obtain regular and irregular components. The timing prediction model is constructed using regular components, and the residual prediction model is constructed using the key influence factors of non-regular components, and the optical thickness of the aerosol is predicted by combining the two.
High-precision aerosol optical thickness prediction is achieved, time-dependent and physical drivers can be considered more accurately, and the prediction ability of AOD and its impact is improved.
Smart Images
Figure CN119578647A_ABST
Abstract
Description
[Technical field]
[0001] The invention relates to the technical field of meteorological data forecasting, and in particular to a forecasting method for aerosol optical thickness and related equipment. [Background technology]
[0002] Aerosol optical depth (AOD) is the integral of the medium's light elimination coefficient in the vertical direction, and is the most important representation of aerosol optical characteristics. AOD is often regarded as a physical quantity that represents the degree of atmospheric turbidity, and is also often used as a key factor in determining the climate effect of aerosols. Therefore, high-precision AOD observation and estimation are crucial to meteorological research and environmental monitoring.
[0003] Although there are a variety of methods for predicting AOD in the existing technology, most of them focus on statistical models or physical models. Therefore, there is still a lack of AOD data forecasting methods in the existing technology that can effectively consider both time dependence and physical driving factors at the same time, which limits the accurate forecast of AOD and its impact on climate and environment. [Summary of the invention]
[0004] In view of this, the present invention provides a method for predicting aerosol optical thickness and related equipment.
[0005] The specific technical scheme of the first embodiment of the present invention is: a method for predicting aerosol optical thickness, the method comprising: obtaining all aerosol optical thickness measured values within a preset time period; sorting all aerosol optical thickness measured values by time to obtain a time series of all aerosol optical thickness measured values; decomposing the time series to obtain multiple aerosol optical thickness components; sorting the multiple aerosol optical thickness components according to variance, and selecting the target aerosol optical thickness component whose cumulative variance contribution rate exceeds the preset value as the regularity component, and the unselected aerosol optical thickness component; The aerosol optical thickness component is used as a non-regular component; a time series forecast model is constructed using the regular component, and a first aerosol optical thickness forecast component is obtained based on the time series forecast model; a residual prediction model is constructed using the key influencing factors of the non-regular component, and a second aerosol optical thickness forecast component is obtained based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type; and an aerosol optical thickness forecast value is obtained according to the first aerosol optical thickness forecast component and the second aerosol optical thickness forecast component.
[0006] Preferably, the method of constructing a time series forecast model using the regularity component and obtaining a first aerosol optical thickness forecast component based on the time series forecast model includes: training a preset number of rolling time series forecast models using the regularity component; and obtaining the first aerosol optical thickness forecast component using the regularity component and all trained rolling time series forecast models.
[0007] Preferably, the method of obtaining the first aerosol optical thickness forecast component by using the regularity component and all the trained rolling time series forecast models includes: inputting the regularity component into the trained first rolling time series forecast model, and outputting the first aerosol optical thickness forecast value; merging the first aerosol optical thickness forecast value with the regularity component to obtain a merged regularity component; using the merged regularity component as the regularity component and the second rolling time series forecast model as the first rolling time series forecast model, and returning to the step of inputting the regularity component into the trained first rolling time series forecast model and outputting the first aerosol optical thickness forecast value, until all the trained rolling time series forecast models output the first aerosol optical thickness forecast value, and all the first aerosol optical thickness forecast values are the first aerosol optical thickness forecast component.
[0008] Preferably, the construction of the residual prediction model using the key influencing factors of the irregular components includes: training a preset regression model using the key influencing factors, and the trained regression model is the residual prediction model.
[0009] Preferably, the method of obtaining the second aerosol optical thickness forecast component based on the residual prediction model includes: taking the irregular component as the output layer parameter of the residual prediction model, and predicting the key influencing factor as the input layer parameter of the residual prediction model to obtain the second aerosol optical thickness forecast component.
[0010] Preferably, obtaining the aerosol optical thickness prediction value according to the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component includes: adding the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component to obtain the aerosol optical thickness prediction value.
[0011] Preferably, decomposing the time series to obtain a plurality of aerosol optical thickness components comprises:
[0012] The time series is decomposed according to time dependency using a multi-channel singular spectrum analysis method to obtain the multiple aerosol optical thickness components.
[0013] The specific technical scheme of the second embodiment of the present invention is: a forecasting system for aerosol optical thickness, the system comprising: a data acquisition module, a sorting module, a decomposition module, a component acquisition module, a first forecast component acquisition module, a second forecast component acquisition module and a forecast value output module; the data acquisition module is used to obtain all the measured values of aerosol optical thickness within a preset time period; the sorting module is used to sort all the measured values of aerosol optical thickness in units of time to obtain a time series of all the measured values of aerosol optical thickness; the decomposition module is used to decompose the time series to obtain multiple aerosol optical thickness components; the component acquisition module is used to sort the multiple aerosol optical thickness components according to variance, and select the components with cumulative variance contribution exceeding a preset number. The target aerosol optical thickness component of the target value is taken as the regular component, and the unselected aerosol optical thickness component is taken as the irregular component; the first forecast component acquisition module is used to construct a time series forecast model using the regular component, and obtain the first aerosol optical thickness forecast component based on the time series forecast model; the second forecast component acquisition module is used to construct a residual prediction model using the key influencing factors of the irregular component, and obtain the second aerosol optical thickness forecast component based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type; the forecast value output module is used to obtain the aerosol optical thickness forecast value according to the first aerosol optical thickness forecast component and the second aerosol optical thickness forecast component.
[0014] The specific technical solution of the third embodiment of the present invention is: an aerosol optical thickness forecasting device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method as described in any one of the first embodiments of the present application.
[0015] The specific technical solution of the fourth embodiment of the present invention is: a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to perform the steps of the method described in any one of the first embodiments of the present application.
[0016] Implementing the embodiments of the present invention will have the following beneficial effects:
[0017] The present invention obtains the corresponding time series by obtaining the actual measurement of aerosol optical thickness, and obtains multiple aerosol optical thickness components by decomposing the time series; the multiple aerosol optical thickness components are divided into regular components and irregular components; for the regular component, that is, the part with strong time dependence, a time series prediction model is used to predict to obtain the first aerosol optical thickness prediction component; for the irregular component, that is, the part with weak time dependence, a residual prediction model between the key influencing factors related to the weak time dependence part and the physical driving factor is constructed to predict to obtain the second aerosol optical thickness prediction component, and the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component are combined to realize high-precision aerosol optical thickness prediction.
Brief Description of the Drawings
[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 A flowchart of the steps of the prediction method for aerosol optical depth;
[0020] Figure 2 This is a schematic diagram of the structure of the prediction system for aerosol optical depth;
[0021] Figure 3 Flow chart of the steps for obtaining aerosol optical depth prediction;
[0022] Figure 4 A diagram of the internal structure of a computer device;
[0023] Among them, 201, data acquisition module; 202, sorting module; 203, decomposition module; 204, component acquisition module; 205, first forecast component acquisition module; 206, second forecast component acquisition module; 207, forecast value output module. [Specific implementation method]
[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0025] The terms "first", "second", etc. in the specification, claims and drawings of this application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally includes steps or modules that are not listed, or optionally includes other steps or modules that are inherent to these processes, methods, products or devices.
[0026] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0027] See also Figure 1 , is a flowchart of a method for predicting aerosol optical thickness in the first embodiment of the present application, which realizes high-precision prediction of aerosol optical thickness. The method includes:
[0028] Step 101, obtaining all measured values of aerosol optical thickness within a preset time period;
[0029] Step 102: sorting all the aerosol optical thickness measured values by time to obtain a time series of all the aerosol optical thickness measured values;
[0030] Step 103, decomposing the time series to obtain multiple aerosol optical thickness components;
[0031] Step 104: sort the multiple aerosol optical thickness components according to the variance, and select the target aerosol optical thickness components whose cumulative variance contribution rate exceeds a preset value as regular components, and the aerosol optical thickness components that are not selected as irregular components;
[0032] Step 105: constructing a time series prediction model using the regularity component, and obtaining a first aerosol optical thickness prediction component based on the time series prediction model;
[0033] Step 106: construct a residual prediction model using the key influencing factors of the irregular component, and obtain a second aerosol optical thickness forecast component based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type;
[0034] Step 107: Obtain an aerosol optical thickness prediction value according to the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component.
[0035] Specifically, we obtain n years of high-precision AOD measured data from the China Aerosol Remote Sensing Network. For example, we obtain 5 years of high-precision AOD measured data. The AOD measured data is named AOD. g The measured AOD was analyzed by Multichannel Singular Spectrum Analysis (MSSA). g The time series is decomposed into multiple components and sorted by variance. The first N components (the specific number is obtained from the decomposition structure) with cumulative variance contribution exceeding 85% are taken as regularity components. The sum is obtained to obtain the AOD sequence that is relatively consistent with the regular changes of the time series. This part is marked as the regularity component of AOD and named AOD. s From AOD g Subtract AOD from the time series s The irregular component of AOD is named AOD r . Using regularity components to construct AOD s Time series forecast model, using the regularity component AOD s And obtain the first aerosol optical thickness forecast component AOD based on the time series forecast model f1 ; Using the irregular component AOD r The residual prediction model is constructed based on the key influencing factors of the aerosol, and the second aerosol optical thickness prediction component AOD is obtained based on the residual prediction model. f2 ; Finally, according to AOD f1 and AOD f2 Get the final high-precision AOD forecast value AOD F .
[0036] The method in this embodiment obtains the corresponding time series by obtaining the actual measurement of aerosol optical thickness, and obtains multiple aerosol optical thickness components by decomposing the time series; the multiple aerosol optical thickness components are divided into regular components and irregular components; for the regular component, that is, the part with strong time dependence, a time series prediction model is used to predict the first aerosol optical thickness prediction component; for the irregular component, that is, the part with weak time dependence, a residual prediction model between the key influencing factors related to the weak time dependence part and the physical driving factors is constructed to predict the second aerosol optical thickness prediction component, and the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component are combined to achieve high-precision aerosol optical thickness prediction.
[0037] In a specific embodiment, the method of constructing a time series forecast model using the regularity component and obtaining a first aerosol optical thickness forecast component based on the time series forecast model includes: training a preset number of rolling time series forecast models using the regularity component; and obtaining the first aerosol optical thickness forecast component using the regularity component and all the trained rolling time series forecast models.
[0038] Specifically, considering AOD s The regular part has strong time correlation, long-term dependence and seasonality. Therefore, an LSTM rolling time series forecast model is constructed for AODs. This forecast model can use the data of the previous n1 days to forecast the data of the next n2 days. For example, the data of the previous 10 days can be used to forecast the data of the next 3 days. The data of the next 3 days is the first aerosol optical thickness forecast component.
[0039] In a specific embodiment, the method of obtaining the first aerosol optical thickness forecast component by using the regularity component and all the trained rolling time series forecast models includes: inputting the regularity component into the trained first rolling time series forecast model, and outputting the first aerosol optical thickness forecast value; merging the first aerosol optical thickness forecast value with the regularity component to obtain a merged regularity component; using the merged regularity component as the regularity component and the second rolling time series forecast model as the first rolling time series forecast model, and returning to the step of inputting the regularity component into the trained first rolling time series forecast model and outputting the first aerosol optical thickness forecast value, until all the trained rolling time series forecast models output the first aerosol optical thickness forecast value, and all the first aerosol optical thickness forecast values are the first aerosol optical thickness forecast component.
[0040] Specifically, the first embodiment of obtaining the first aerosol optical thickness forecast component is:
[0041] It is necessary to train three LSTM rolling time series forecast models in the manner of multi-model single-step forecasting. Then, model 1 predicts the data of day
[11] based on the data of day [1…10]. Model 2 adds the data of day
[11] predicted by model 1, that is, the data of day [2…11] is used to predict the data of day
[12] . Finally, model 3 uses the data of day [3…12] to predict the data of day
[13] . Finally, based on the time series forecast model, the forecast data of the AOD regular part of days 11-13 can be obtained, which is the first aerosol optical thickness forecast component AOD. f1 .
[0042] Specifically, the first embodiment of obtaining the second aerosol optical thickness forecast component is:
[0043] It is necessary to train three LSTM rolling time series forecast models in the manner of multi-model single-step forecasting. Then, model 1 predicts the data of day
[11] based on the data of day [1…10]. Model 2 adds the data of day 11 predicted by model 1, that is, the data of day [1…11] is used to predict the data of day
[12] . Finally, model 3 uses the data of day [1…12] to predict the data of day
[13] . Finally, based on the time series forecast model, the forecast data of the AOD regular part of days 11-13 can be obtained, which is the first aerosol optical thickness forecast component AOD. f1 .
[0044] In a specific embodiment, the construction of the residual prediction model using the key influencing factors of the irregular components includes: using the key influencing factors to train a preset regression model, and the trained regression model is the residual prediction model. Specifically, the residual prediction model can reveal the relationship between the key influencing factors and the dependent variables that are not fully considered in the original regression model, thereby enhancing the model's ability to explain the data.
[0045] In a specific embodiment, the method of obtaining the second aerosol optical thickness forecast component based on the residual prediction model includes: taking the irregular component as the output layer parameter of the residual prediction model, and predicting the key influencing factor as the input layer parameter of the residual prediction model to obtain the second aerosol optical thickness forecast component.
[0046] Specifically, considering the irregular component AOD r The correlation with time is poor and there is no long-term dependence, so some AOD components that can affect the irregularity are used. r The specific steps are as follows: first, the key influencing factors (i.e., air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitation, land cover type, etc.) are screened out, and then an XGBoost (eXtreme GradientBoosting) regression model is constructed based on the key influencing factors to predict the irregular component AOD. r There are seven key influencing factors here, namely N1-N7. N1-N7 are used as input layer parameters, and the irregular component AOD rAs the output layer parameters, after the input layer parameters are input, the data will be processed through a combination of decision trees, and the error of the previous tree will be gradually corrected through the output of each tree, and finally the prediction results of the model will be generated. In the process of training the prediction model, the grid search algorithm is used to determine the optimal parameters of the model (such as the number of trees, learning rate, tree depth, etc.), and the model is verified using the ten-fold cross-validation method to evaluate the generalization ability of the model. After completing the model training, the residual prediction model is obtained, and the optimal parameters and feature importance of the residual prediction model are saved. At the same time, the input parameters corresponding to the model are saved for subsequent prediction and application. Finally, the irregular component AOD can be obtained based on the residual prediction model. r The forecast data is the second aerosol optical thickness forecast component AOD f2 .
[0047] In a specific embodiment, obtaining the aerosol optical thickness prediction value according to the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component includes: adding the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component to obtain the aerosol optical thickness prediction value. Specifically, the AOD f1 and AOD f2 Add them together to get the final AOD forecast value, which is the aerosol optical thickness forecast value AOD F .
[0048] In a specific embodiment, the decomposing the time series to obtain multiple aerosol optical thickness components includes: using a multi-channel singular spectrum analysis method to decompose the time series according to time dependence to obtain the multiple aerosol optical thickness components. Specifically, multi-channel singular spectrum analysis MSSA can decompose the time series into multiple components, such as long-term trends, periodic changes, noise, etc. These components reflect the variation characteristics of AOD on different time scales, which helps to more accurately understand the temporal and spatial distribution of AOD. By decomposing the time series, MSSA can identify and remove noise components, thereby reducing the interference of noise on AOD inversion. This helps to improve the accuracy and stability of AOD inversion.
[0049] In the specific embodiments, see Figure 2, is a structural schematic diagram of an aerosol optical thickness prediction system in the second embodiment of the present application, the system comprising: a data acquisition module 201, a sorting module 202, a decomposition module 203, a component acquisition module 204, a first prediction component acquisition module 205, a second prediction component acquisition module 206 and a prediction value output module 207; the data acquisition module 201 is used to obtain all aerosol optical thickness measured values within a preset time period; the sorting module 202 is used to sort all aerosol optical thickness measured values in units of time to obtain a time series of all aerosol optical thickness measured values; the decomposition module 203 is used to decompose the time series to obtain multiple aerosol optical thickness components; the component acquisition module 204 is used to sort the multiple aerosol optical thickness components according to variance, and select The target aerosol optical thickness component whose cumulative variance contribution rate exceeds the preset value is taken as the regular component, and the aerosol optical thickness component that is not selected is taken as the irregular component; the first forecast component acquisition module 205 is used to construct a time series forecast model using the regular component, and obtain the first aerosol optical thickness forecast component based on the time series forecast model; the second forecast component acquisition module 206 is used to construct a residual prediction model using the key influencing factors of the irregular component, and obtain the second aerosol optical thickness forecast component based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type; the forecast value output module 207 is used to obtain the aerosol optical thickness forecast value according to the first aerosol optical thickness forecast component and the second aerosol optical thickness forecast component.
[0050] Specifically, the steps for obtaining the predicted value of aerosol optical thickness by the system in this embodiment can be found in Figure 3 The system in this embodiment obtains the corresponding time series by obtaining the actual measurement of aerosol optical thickness, and obtains multiple aerosol optical thickness components by decomposing the time series; the multiple aerosol optical thickness components are divided into regular components and irregular components; for the regular component, that is, the part with strong time dependence, a time series prediction model is used to predict and obtain the first aerosol optical thickness prediction component; for the irregular component, that is, the part with weak time dependence, a residual prediction model between the key influencing factors related to the weak time dependence part and the physical driving factors is constructed to predict and obtain the second aerosol optical thickness prediction component, and the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component are combined to achieve high-precision aerosol optical thickness prediction.
[0051] In a specific embodiment, the third embodiment of the present application provides an aerosol optical thickness forecasting device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method as described in any one of the first embodiments of the present application.
[0052] In a specific embodiment, the fourth embodiment of the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the method as described in any one of the first embodiments of the present application.
[0053] Figure 4 The internal structure of a computer device in one embodiment is shown. The computer device can be a terminal or a server. Figure 4 The computer device includes a processor, a memory, etc. connected via a system bus. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement the method in this embodiment. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can implement the method in this embodiment. Those skilled in the art can understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0054] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
[0055] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.
Claims
1. A method for predicting aerosol optical thickness, characterized in that: The method comprises: Obtain all measured values of aerosol optical thickness within a preset time period; Sorting all the aerosol optical thickness measured values in units of time to obtain a time series of all the aerosol optical thickness measured values; Decomposing the time series to obtain multiple aerosol optical thickness components; The multiple aerosol optical thickness components are sorted according to the variance, and the target aerosol optical thickness components whose cumulative variance contribution rate exceeds a preset value are selected as regular components, and the aerosol optical thickness components that are not selected are selected as irregular components; constructing a time series prediction model using the regularity component, and obtaining a first aerosol optical thickness prediction component based on the time series prediction model; A residual prediction model is constructed using the key influencing factors of the irregular component, and a second aerosol optical thickness forecast component is obtained based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type; An aerosol optical thickness prediction value is obtained according to the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component.
2. The method for predicting aerosol optical depth according to claim 1, characterized in that: The method of constructing a time series prediction model by using the regularity component and obtaining a first aerosol optical thickness prediction component based on the time series prediction model includes: Using the regularity component to train a preset number of rolling time series forecast models; The first aerosol optical thickness forecast component is obtained by using the regularity component and all trained rolling time series forecast models.
3. The method for predicting aerosol optical depth according to claim 2, characterized in that: The step of obtaining the first aerosol optical thickness forecast component by using the regularity component and all trained rolling time series forecast models includes: Inputting the regularity component into the trained first rolling time series forecast model, and outputting the first aerosol optical thickness forecast value; Combining the first aerosol optical thickness prediction value with the regularity component to obtain a combined regularity component; The merged regularity component is used as the regularity component, and the second rolling time series forecast model is used as the first rolling time series forecast model. The step of inputting the regularity component into the trained first rolling time series forecast model and outputting the first aerosol optical thickness forecast value is returned to, until all the trained rolling time series forecast models output the first aerosol optical thickness forecast value, and all the first aerosol optical thickness forecast values are the first aerosol optical thickness forecast component.
4. The method for predicting aerosol optical depth according to claim 1, characterized in that: The method of constructing a residual prediction model using the key influencing factors of the irregular components includes: The key influencing factors are used to train a preset regression model, and the trained regression model is the residual prediction model.
5. The method for predicting aerosol optical depth according to claim 1, characterized in that: The obtaining a second aerosol optical thickness forecast component based on the residual prediction model comprises: The irregular component is used as the output layer parameter of the residual prediction model, and the key influencing factor is used as the input layer parameter of the residual prediction model for prediction, so as to obtain the second aerosol optical thickness prediction component.
6. The method for predicting aerosol optical depth according to claim 1, characterized in that: The step of obtaining an aerosol optical thickness prediction value according to the first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component comprises: The first aerosol optical thickness prediction component and the second aerosol optical thickness prediction component are added to obtain an aerosol optical thickness prediction value.
7. The method for predicting aerosol optical depth according to claim 1, characterized in that: Decomposing the time series to obtain multiple aerosol optical thickness components includes: The time series is decomposed according to time dependency using a multi-channel singular spectrum analysis method to obtain the multiple aerosol optical thickness components.
8. A prediction system for aerosol optical depth, characterized in that: The system comprises: a data acquisition module, a sorting module, a decomposition module, a component acquisition module, a first forecast component acquisition module, a second forecast component acquisition module and a forecast value output module; The data acquisition module is used to obtain all measured values of aerosol optical thickness within a preset time period; The sorting module is used to sort all the aerosol optical thickness measured values by time unit to obtain a time series of all the aerosol optical thickness measured values; The decomposition module is used to decompose the time series to obtain multiple aerosol optical thickness components; The component acquisition module is used to sort the multiple aerosol optical thickness components according to the variance, and select the target aerosol optical thickness components whose cumulative variance contribution rate exceeds a preset value as regular components, and the aerosol optical thickness components that are not selected as irregular components; The first forecast component acquisition module is used to construct a time series forecast model using the regularity component, and acquire a first aerosol optical thickness forecast component based on the time series forecast model; The second forecast component acquisition module is used to construct a residual prediction model using the key influencing factors of the irregular component, and acquire the second aerosol optical thickness forecast component based on the residual prediction model; the key influencing factors include air pressure, temperature, humidity, wind speed, cloud cover, atmospheric precipitable water and surface cover type; The forecast value output module is used to obtain an aerosol optical thickness forecast value according to the first aerosol optical thickness forecast component and the second aerosol optical thickness forecast component.
9. A forecasting device for aerosol optical thickness, comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.