Partitioned and graded meteorological data quality control method

Abstract

The invention discloses a zoning and grading meteorological data quality control method, and relates to the technical field of data processing. The method comprises the following steps: collecting historical data of a meteorological monitoring station, extracting corresponding historical time-space multi-dimensional features, introducing a typical rainfall mode, and dividing a monitoring area into a plurality of time-space classification quality control units; constructing a partition grading quality control parameter space and a multi-objective optimization function for a space-time classification quality control unit; based on the multi-objective optimization function, using a multi-objective swarm intelligence algorithm to perform iterative optimization on the partitioned and graded quality control parameter space to obtain a Pareto optimal solution set; and according to application scene requirements, selecting and loading an optimal partition grading quality control parameter vector from the Pareto optimal solution set, and carrying out quality control on meteorological monitoring station real-time data of the corresponding space-time classification quality control unit. The problems that in the prior art, quality control parameter adaptability is poor, abnormity recognition and data fidelity are difficult to consider at the same time, and quality control efficiency is low are solved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for regional and hierarchical meteorological data quality control. Background Technology

[0002] Meteorological data is a core foundation for fields such as hydrological forecasting, water resource management, and disaster prevention and mitigation. Meteorological monitoring stations are widely distributed across various geographical environments, from plains to mountains, and from urban areas to the countryside. Due to equipment malfunctions, transmission interference, harsh environments (such as strong winds and lightning), and untimely maintenance, the collected rainfall data often contains a large amount of noise, outliers, and even missing values. Directly using these uncontrolled raw data will seriously affect the accuracy of hydrological model analysis and the scientific validity of decision-making.

[0003] Existing data quality control methods for meteorological monitoring stations mostly employ uniform static thresholds or simple single rules, failing to fully consider the complex temporal and spatial variations of rainfall. For example, rainfall characteristics differ significantly across different geographical regions (mountainous areas, plains), and the rainfall patterns in the same region also vary drastically across different seasons (spring precipitation season, summer strong convection season, winter dry season, etc.). Traditional "one-size-fits-all" quality control methods struggle to maintain high data fidelity while ensuring accurate anomaly identification, and their computational efficiency often falls short of the demands of large-scale real-time monitoring.

[0004] Therefore, there is an urgent need for an intelligent, hierarchical data quality control method that can adapt to dynamic changes in time and space and take into account multiple objectives. Summary of the Invention

[0005] This invention provides a method for regional and hierarchical meteorological data quality control. This invention solves the problems of poor adaptability of quality control parameters, difficulty in balancing anomaly identification and data fidelity, and low quality control efficiency in existing technologies.

[0006] In a first aspect, embodiments of the present invention provide a method for regional and hierarchical meteorological data quality control, the method comprising: Historical data from meteorological monitoring stations were collected, corresponding historical spatiotemporal multidimensional features were extracted, and based on these features, typical rainfall patterns were introduced to divide the monitoring area into several spatiotemporal classification quality control units. For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function, as shown in the following formula:

[0007] In the formula, To load the partitioned and graded quality control parameter space A multi-objective optimization function for quality control; To load the partitioned and graded quality control parameter space Anomaly identification efficiency function for quality control; To load the partitioned and graded quality control parameter space Data fidelity function for quality control; To load the partitioned and graded quality control parameter space The quality control efficiency function for quality control; The formula for the anomaly detection performance function is:

[0008] In the formula, To load the partitioned and graded quality control parameter space The anomaly detection effectiveness value for quality control; To load the partitioned and graded quality control parameter space Precision on the validation set for quality control; To load the partitioned and graded quality control parameter space Recall rate on the validation set for quality control;

[0009] In the formula, To load the partitioned and graded quality control parameter space Data fidelity for quality control; To verify the first one that was misjudged as an anomaly in the set j Rainfall data; To After quality control, the replacement was the first j Rainfall data;

[0010] In the formula, To load the partitioned and graded quality control parameter space Quality control efficiency in the process of quality control; The average time taken to perform one quality control operation on the validation set; To find the minimum value, to prevent the denominator from being zero; Based on a multi-objective optimization function, a multi-objective swarm intelligence algorithm is used to iteratively optimize the parameter space of partitioned and graded quality control to obtain the Pareto optimal solution set. Based on the application scenario requirements of each spatiotemporal classification quality control unit, the optimal partitioned and graded quality control parameter vector is selected and loaded from the Pareto optimal solution set, and the real-time data of the meteorological monitoring station of the corresponding spatiotemporal classification quality control unit is subjected to quality control.

[0011] The technical solution provided in this application has at least the following beneficial effects: By incorporating historical spatiotemporal multidimensional features and typical rainfall modes, a [database / database] was constructed. K × M A spatiotemporal classification quality control unit was established, realizing the transformation of the quality control strategy from static and global to dynamic and regional, improving the parameter adaptability under different rainfall scenarios. A multi-objective optimization function was constructed, including anomaly identification efficiency, data fidelity, and quality control efficiency, avoiding the limitations of single-index optimization. Pareto optimal solution sets provide flexible choices for different application needs. The Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism were introduced into the multi-objective particle swarm optimization algorithm. The chaotic sequence was initialized using Logistic mapping, which effectively improved the algorithm's global search capability and ability to escape local optima, thereby obtaining a better combination of quality control parameters. A three-level quality control parameter (dynamic extrema, temporal mutation, and spatial consistency) was used to screen the data step by step. The main direction of the rainband was extracted by PCA for spatial consistency verification, which significantly improved the accuracy and robustness of anomaly identification.

[0012] In one optional implementation, historical data from meteorological monitoring stations is collected, corresponding historical spatiotemporal multidimensional features are extracted, and based on these features, typical rainfall modes are introduced to divide the monitoring area into several spatiotemporal classification quality control units, including: Collect historical data from meteorological monitoring stations, which includes historical rainfall data and corresponding historical geographic information data from all meteorological monitoring stations within the monitoring area. The historical data of meteorological monitoring stations are preprocessed to obtain preprocessed historical data of meteorological monitoring stations, and the preprocessed historical data of meteorological monitoring stations is stored in the historical database. Extract historical spatiotemporal multidimensional features from preprocessed historical meteorological monitoring station data. These features include at least rainfall statistical features, rainfall extreme value features, rainfall timeliness features, geographical features, and spatiotemporal distribution features. Based on the geographical features described in the historical spatiotemporal multidimensional characteristics, and combined with the preset terrain classification rules, all meteorological monitoring stations within the monitoring area are divided into... K A static geographic sub-region, of which K Total number of static geographic sub-regions; Based on static geographic sub-regions, a time dimension is introduced. According to historical spatiotemporal multi-dimensional rainfall statistics, extreme rainfall characteristics, and rainfall timeliness characteristics, a rule-based partitioning method combining calendar cycles and meteorological characteristic thresholds is used to divide the year into [number missing]. M A typical rainfall pattern includes at least the spring precipitation period, the summer strong convection period, the autumn drizzle period, and the winter dry period. M This represents the total number of typical rainfall modes; Combining each static geographic sub-region and its corresponding M A typical rainfall mode was obtained. K × M Spatiotemporal classification quality control unit.

[0013] In one optional implementation, for the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly detection efficiency function, a data fidelity function, and a quality control efficiency function, including: For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space is constructed, which includes at least three levels of partitioned and hierarchical quality control parameters. For the partitioned and graded quality control parameter space, a multi-objective optimization function is constructed, which includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function.

[0014] In one optional implementation, the zoning and grading quality control parameters are used to perform grading quality control on meteorological monitoring station data, and include at least: Primary quality control parameters should include at least the dynamic extreme value threshold coefficient; Secondary quality control parameters should include at least the time-series difference window width and the mutation rate coefficient; The three-level quality control parameters include at least the spatial neighborhood search radius, the weight sensitivity parameter, and the relative error coefficient of rainfall.

[0015] In one alternative implementation, a multi-objective swarm intelligence algorithm is used based on a multi-objective optimization function to iteratively optimize the parameter space of the partitioned and graded quality control system, obtaining a Pareto optimal solution set, including: Encode the partitioned and graded quality control parameter vectors corresponding to the partitioned and graded quality control parameter space into individual vectors of the multi-objective particle swarm optimization algorithm, and initialize the multi-objective particle swarm optimization algorithm; The chaotic sequence is generated using the Logistic mapping, and then mapped to the solution space of the particles to obtain the initial particle swarm. Load the candidate partitioning and grading quality control parameter vectors corresponding to each particle in the initial particle swarm, perform quality control on several preprocessed historical meteorological monitoring station data in the historical database, and use a multi-objective optimization function to calculate the corresponding optimization target value. Based on the optimization objective value, the initial particle swarm is subjected to fast non-dominated sorting to obtain different frontier levels, and an external archive set is maintained to store all non-dominated solutions of the first frontier level. By introducing the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism, the initial particle swarm is updated to obtain an updated particle swarm. Using a multi-objective optimization function, the optimization objective value corresponding to each updated particle in the updated particle swarm is calculated, and the frontier level and external archive set are updated according to the optimization objective value. The iteration continues until the number of iterations reaches the threshold, at which point the iterative update of the particle swarm stops, and all non-dominated solutions in the external archive set are output as the Pareto optimal solution set.

[0016] In one alternative implementation, a Levy flight mechanism, a convergence factor, and a gray wolf cooperative mechanism are introduced to update the positions of the initial particle swarm, resulting in an updated particle swarm, including: In the external archive set, based on the degree of crowding, the top three local optima are identified and defined as Alpha particles, Beta particles, and Delta particles, respectively. The historical best position of each particle is then extended to Alpha particles, Beta particles, and Delta particles. Based on Alpha, Beta, and Delta particles, the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism are introduced to update the initial velocity of each initial particle in the initial particle swarm, resulting in the updated velocity, as shown in the formula:

[0017] In the formula, Number of iterations t+ 1 of i The rate of update of each particle; Number of iterations t The i The update rate of each particle, which is the initial rate during the first iteration; Number of iterations t The convergence factor improves the inertia weight; Number of iterations t The globally optimal particle; Number of iterations t The i A new particle is generated, which is the initial particle in the first iteration; t This represents the current iteration number; The cooperation coefficient; A random number between (0, 1); Number of iterations t Alpha particles, Beta particles, and Delta particles; t This is an indicator of the number of iterations. i For particle indication; for Levy Distribute random numbers; b for Levy Step length, and b ∈[1,2]; For flight coefficient;

[0018] In the formula, These represent the maximum and minimum values ​​of the inertia weight; This is the threshold for the number of iterations; , To adjust the parameters; It is the hyperbolic tangent function; Based on the update rate, the positions of the corresponding initial particles are updated to obtain updated particles. This process continues until the positions of all initial particles have been updated, resulting in an updated particle swarm. The formula is:

[0019] In the formula, Number of iterations t+ 1 of i An updated particle.

[0020] In one optional implementation, based on the application scenario requirements of each spatiotemporal classification quality control unit, the optimal partitioned and graded quality control parameter vector is selected and loaded from the Pareto optimal solution set, and quality control is performed on the real-time meteorological monitoring station data of the corresponding spatiotemporal classification quality control unit, including: In each spatiotemporal classification quality control unit, real-time data from the corresponding meteorological monitoring station is collected and preprocessed to obtain preprocessed real-time data from the meteorological monitoring station. Determine the typical rainfall mode corresponding to the current moment, use the typical rainfall mode as the application scenario requirement of the corresponding spatiotemporal classification quality control unit, and obtain the preset weighting coefficients corresponding to the application scenario requirement; Based on the preset weighting coefficients corresponding to the application scenario requirements, the optimization objective value of each solution in the Pareto optimal solution set is weighted to obtain the comprehensive optimization objective value; The partitioned and graded quality control parameter vector corresponding to the solution with the best overall optimization objective value in the Pareto optimal solution set is loaded as the optimal partitioned and graded quality control parameter vector; Based on the optimal partitioned and graded quality control parameter vector, quality control is performed on the preprocessed real-time data of meteorological monitoring stations for the corresponding spatiotemporal classification quality control units.

[0021] In one optional implementation, based on the optimal partitioned and graded quality control parameter vector, quality control is performed on the preprocessed real-time data of the meteorological monitoring station for the corresponding spatiotemporal classification quality control unit, including: Read the first l The preprocessed real-time rainfall data and corresponding preprocessed real-time geographic information data from the meteorological monitoring station's real-time data in the spatiotemporal classification quality control unit, among which,l This is the indicator quantity for the spatiotemporal classification quality control unit; If the preprocessed real-time rainfall data <0 or Then, the preprocessed real-time rainfall data will be marked as a Level 1 anomaly. This refers to the dynamic extreme value threshold coefficient of the primary quality control parameter. This represents the maximum historical rainfall value. Based on preprocessed real-time rainfall data Calculate the first l Spatiotemporal classification quality control unit in the past Real-time rainfall increments within the window ,like Then, the preprocessed real-time rainfall data will be marked as a level 2 anomaly. This represents the time-series difference window width for the secondary quality control parameters. The mutation rate coefficient is a secondary quality control parameter. This serves as a baseline value for the mutation rate. Based on the preprocessed real-time geographic information data, the search radius For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall at neighboring stations. ; like The preprocessed real-time rainfall data will then be marked as a level three anomaly. The relative error coefficient of rainfall is a secondary quality control parameter. If preprocessed real-time rainfall data is marked as Level 1, 2, and / or 3 anomalies, an anomaly handling algorithm is selected based on the source of the anomaly to replace the preprocessed real-time rainfall data, resulting in replaced preprocessed real-time rainfall data.

[0022] In one alternative implementation, the search radius is determined based on preprocessed real-time geographic information data. For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall from neighboring stations, including: Taking the current meteorological monitoring station corresponding to the preprocessed real-time geographic information data as the target meteorological monitoring station, and taking the coordinates of the target meteorological monitoring station as the center, and using... Draw a circle with a radius and retrieve all adjacent meteorological monitoring stations within that circle. If the number of adjacent meteorological monitoring stations is less than the threshold, skip the third-level quality control and rely only on the first and second-level quality control. Extract rainfall data from the target station and all neighboring stations at the current time and at several past time intervals to construct a rainfall matrix. Each row of the matrix represents a meteorological monitoring station, and each column represents the preprocessed real-time data of the meteorological monitoring station at the corresponding time. Using rainfall matrix Real-time rainfall data after preprocessing from all meteorological monitoring stations in China at the current moment. The spatial covariance matrix is ​​constructed using the geographic coordinates of the corresponding preprocessed real-time geographic information data. For the spatial covariance matrix Perform eigenvalue decomposition to obtain eigenvalues ​​and their corresponding eigenvectors; The direction of the eigenvector corresponding to the largest eigenvalue is defined as the direction of the maximum gradient of rainfall distribution in the region, and is thus defined as the main direction of the rain belt. For each adjacent meteorological monitoring station, calculate the geographic connection vector from the target meteorological monitoring station to the adjacent meteorological monitoring station; Calculate the cosine of the angle between the geographic connection vector and the main direction of the rain belt, and use it as the directional consistency coefficient; Using weight sensitivity parameters The consistency coefficient is nonlinearly amplified to obtain the final direction weighting factor; Calculate the traditional inverse distance weighting factor, multiply the direction weighting factor by the inverse distance weighting factor to obtain the final neighbor station coupling weight, normalize the coupling weights of all neighboring meteorological monitoring stations so that the sum of the weights is 1, and obtain the normalized coupling weight. Based on the normalized coupling weights, the preprocessed real-time rainfall data is weighted to obtain the weighted rainfall estimate from neighboring stations. .

[0023] A second aspect of this invention provides an electronic device, which includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by at least one processor, such that the at least one processor can perform the method proposed in the first aspect of the present invention.

[0024] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in the first aspect of the present invention. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the electronic device structure of the hardware operating environment involved in the embodiments of the present invention; Figure 2This is a flowchart of the steps of a regional and hierarchical meteorological data quality control method provided in an embodiment of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] The present invention will be further described below with reference to the accompanying drawings.

[0028] Reference Figure 1 , Figure 1 This is a schematic diagram of the electronic device structure of the hardware operating environment involved in the embodiments of the present invention.

[0029] like Figure 1 As shown, the electronic device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed random access memory (RAM) or a stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0030] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0031] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a data storage module, a network communication module, a user interface module, and an electronic program for a regional and hierarchical meteorological data quality control method.

[0032] exist Figure 1 In the electronic device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the electronic device of the present invention can be set in the electronic device. The electronic device calls the electronic program of the regional and graded meteorological data quality control method stored in the memory 1005 through the processor 1001 and executes the regional and graded meteorological data quality control method provided in the embodiment of the present invention.

[0033] Reference Figure 2 The present invention provides a method for regional and hierarchical meteorological data quality control, the method comprising: S201: Collect historical data from meteorological monitoring stations, extract corresponding historical spatiotemporal multidimensional features, and based on the historical spatiotemporal multidimensional features, introduce typical rainfall modes to divide the monitoring area into several spatiotemporal classification quality control units. S202: For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function, as shown in the following formula:

[0034] In the formula, To load the partitioned and graded quality control parameter space A multi-objective optimization function for quality control; To load the partitioned and graded quality control parameter space Anomaly identification efficiency function for quality control; To load the partitioned and graded quality control parameter space Data fidelity function for quality control; To load the partitioned and graded quality control parameter space The quality control efficiency function for quality control; The formula for the anomaly detection performance function is:

[0035] In the formula, To load the partitioned and graded quality control parameter space The anomaly detection effectiveness value for quality control; To load the partitioned and graded quality control parameter space Precision on the validation set for quality control; To load the partitioned and graded quality control parameter space Recall rate on the validation set for quality control;

[0036] In the formula, To load the partitioned and graded quality control parameter space Data fidelity for quality control; To verify the first one that was misjudged as an anomaly in the set j Rainfall data; To After quality control, the replacement was the first j Rainfall data;

[0037] In the formula, To load the partitioned and graded quality control parameter space Quality control efficiency in the process of quality control; The average time taken to perform one quality control operation on the validation set; To find the minimum value, to prevent the denominator from being zero; S203: Based on a multi-objective optimization function, a multi-objective swarm intelligence algorithm is used to iteratively optimize the parameter space of the partitioned and graded quality control to obtain the Pareto optimal solution set; S204: Based on the application scenario requirements of each spatiotemporal classification quality control unit, select and load the optimal partitioned and graded quality control parameter vector from the Pareto optimal solution set, and perform quality control on the real-time meteorological monitoring station data of the corresponding spatiotemporal classification quality control unit.

[0038] The technical solution provided in this application has at least the following beneficial effects: By incorporating historical spatiotemporal multidimensional features and typical rainfall modes, a [database / database] was constructed. K × M A spatiotemporal classification quality control unit was established, realizing the transformation of the quality control strategy from static and global to dynamic and regional, improving the parameter adaptability under different rainfall scenarios. A multi-objective optimization function was constructed, including anomaly identification efficiency, data fidelity, and quality control efficiency, avoiding the limitations of single-index optimization. Pareto optimal solution sets provide flexible choices for different application needs. The Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism were introduced into the multi-objective particle swarm optimization algorithm. The chaotic sequence was initialized using Logistic mapping, which effectively improved the algorithm's global search capability and ability to escape local optima, thereby obtaining a better combination of quality control parameters. A three-level quality control parameter (dynamic extrema, temporal mutation, and spatial consistency) was used to screen the data step by step. The principal component analysis (PCA) method was used to extract the main direction of the rainband for spatial consistency verification, which significantly improved the accuracy and robustness of anomaly identification.

[0039] In one optional implementation, historical data from meteorological monitoring stations is collected, corresponding historical spatiotemporal multidimensional features are extracted, and based on these features, typical rainfall modes are introduced to divide the monitoring area into several spatiotemporal classification quality control units, including: S2011: Collect historical data from meteorological monitoring stations. The historical data from meteorological monitoring stations includes historical rainfall data and corresponding historical geographic information data from all meteorological monitoring stations within the monitoring area. It is worth noting that the historical rainfall data includes hourly rainfall data and minute-level rainfall data from meteorological monitoring stations over the past 5-10 years; Historical geographic information data, including the latitude and longitude, elevation, slope, aspect, and wind direction of meteorological monitoring stations; S2012: Preprocess the historical data of the meteorological monitoring station to obtain the preprocessed historical data of the meteorological monitoring station, and store the preprocessed historical data of the meteorological monitoring station in the historical database; It is worth noting that the preprocessing includes: Missing value handling: For a small number of missing data, linear interpolation is used to complete them; for large missing segments, they are marked but not used in feature extraction. Outlier removal: Remove obvious physical errors (such as negative rainfall or noisy data that exceeds historical extremes by many times); Data standardization: Normalize rainfall and geographic data (e.g., Min-Max standardization) to eliminate the influence of units and facilitate subsequent calculations; S2013: Extract historical spatiotemporal multidimensional features from preprocessed historical meteorological monitoring station data. The historical spatiotemporal multidimensional features should include at least rainfall statistical features, rainfall extreme value features, rainfall timeliness features, geographical features, and spatiotemporal distribution features. It is worth noting that, for the preprocessed historical data of meteorological monitoring stations, the following feature vectors are extracted to form the feature fingerprint of each meteorological monitoring station or each time segment: Rainfall statistics include annual average rainfall, monthly average rainfall, rainfall standard deviation, rainfall skewness (reflecting the asymmetry of rainfall distribution), and kurtosis (reflecting the frequency of extreme rainfall events). Extreme rainfall characteristics: historical maximum hourly rainfall intensity, historical maximum daily rainfall, and threshold values ​​corresponding to design rainfall for different return periods (e.g., once in 50 years); Rainfall time-leading characteristics: average rainfall duration, rainfall interval, and annual rainfall concentration; Geographical features: longitude, latitude, elevation, slope, aspect; Spatiotemporal distribution characteristics: The local spatial autocorrelation coefficient calculated using Moran's I index or Geary's C index reflects the correlation of rainfall between this station and surrounding stations; S2014: Based on the elevation and slope of the geographical features described in the historical spatiotemporal multidimensional features, and combined with the preset terrain classification rules, all meteorological monitoring stations within the monitoring area are divided into... K A static geographic sub-region, of which K Total number of static geographic sub-regions; In this embodiment, the specific partitioning method is as follows: a deterministic rule based on geographic attribute thresholds is used for partitioning, avoiding the unreproducible problem caused by the randomness of clustering algorithms. For example, setting K=3, the partitioning rule is as follows: Rule 1: If the elevation of a meteorological monitoring station is greater than or equal to the first elevation threshold (e.g., 500 meters) and the slope is greater than or equal to the first slope threshold (e.g., 15 degrees), then it is classified as the 'first static geographic sub-region'. Rule 2: If the elevation of a meteorological monitoring station is less than the second elevation threshold (e.g., 50 meters) and the slope is less than the second slope threshold (e.g., 5 degrees), then it is classified as the 'second static geographic sub-region'. Rule 3: For other meteorological monitoring stations that do not meet the above rules, they shall be classified as 'Third Static Geographic Sub-region'; The first altitude threshold, the second altitude threshold, the first slope threshold, and the second slope threshold mentioned above can be adaptively adjusted according to the actual geographical and climatic characteristics of the monitoring area. This division method based on physical attribute thresholds ensures the determinism and reproducibility of the partitions and avoids the boundary drift problem caused by algorithm initialization. S2015: Based on static geographic sub-regions, a time dimension is introduced. According to historical spatiotemporal multi-dimensional rainfall statistics, extreme rainfall characteristics, and rainfall timeliness characteristics, a rule-based partitioning method combining calendar cycles and meteorological characteristic thresholds is used to divide the year into [number missing]. M A typical rainfall pattern includes at least the spring precipitation period, the summer strong convection period, the autumn drizzle period, and the winter dry period. M This represents the total number of typical rainfall modes; In this embodiment, the specific method of division is as follows: based on the historical meteorological patterns of the monitoring area and the key points of business focus, a fixed time window or dynamic meteorological threshold is set; Example of fixed time window division: Define the calendar period from March to May as the 'spring precipitation period', June to August as the 'summer strong convection period', September to November as the 'autumn continuous rain period', and December to February of the following year as the 'winter dry period'; Example of dynamic meteorological threshold division: When the average temperature over a consecutive 5-day period exceeds the first preset threshold and the rainfall exceeds the second preset threshold, it is determined that the 'summer severe convection period' has begun. This rule-based division method eliminates the seasonal drift caused by the randomness of algorithm initialization, ensures the accurate correspondence between quality control parameters and seasonal characteristics, and improves the determinism of quality control strategies. S2016: Combining each static geographic sub-region and its corresponding M A typical rainfall mode was obtained. K × M One spatiotemporal classification quality control unit; In this embodiment, the Cartesian product combination: K A static geographic sub-region and M Combining typical rainfall modes to generate K × M Each spatiotemporal classification quality control unit has its own independent parameter configuration.

[0040] In one optional implementation, for the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly detection efficiency function, a data fidelity function, and a quality control efficiency function, including: S2021: For the spatiotemporal classification quality control unit, construct a partitioned and hierarchical quality control parameter space, which includes at least three levels of partitioned and hierarchical quality control parameters. S2022: For the partitioned and graded quality control parameter space, construct a multi-objective optimization function, which includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function.

[0041] In one optional implementation, the zoning and grading quality control parameters are used to perform grading quality control on meteorological monitoring station data, and include at least: Primary quality control parameters should include at least the dynamic extreme value threshold coefficient. The highest historical rainfall data The coefficient may change with the increase of observation years. Make the threshold value , ,For example =1.2 indicates that a margin of 20% is allowed beyond the historical maximum value; Secondary quality control parameters should include at least the timing difference window width. The time window for calculating the rate of change of rainfall, such as 1 hour or 3 hours, and the abrupt change rate coefficient. This refers to the maximum permissible increase in rainfall per unit time, used to determine flypoints; Level 3 quality control parameters should include at least the spatial neighborhood search radius. Find the distance range of reference meteorological monitoring stations and the weighted sensitivity parameters. When calculating the weights of neighboring stations, the contribution coefficient of wind direction or main direction of rain belt to the weights, as well as the relative error coefficient of rainfall, are considered.

[0042] In one alternative implementation, a multi-objective swarm intelligence algorithm is used based on a multi-objective optimization function to iteratively optimize the parameter space of the partitioned and graded quality control system, obtaining a Pareto optimal solution set, including: S2031: Encode the partitioned and graded quality control parameter vector corresponding to the partitioned and graded quality control parameter space into an individual vector of the multi-objective particle swarm optimization algorithm, and initialize the multi-objective particle swarm optimization algorithm; S2032: Use Logistic mapping to generate chaotic sequences and map the chaotic sequences to the solution space of particles to obtain the initial particle swarm; The formula is:

[0043] In the formula, For the first n+ 1. n One chaotic variable; The stability coefficient is typically 4. This sequence is ergodic and random, ensuring that the initial particle swarm is uniformly distributed in the solution space, avoiding getting trapped in local optima, which is superior to traditional random initialization. n For chaotic variable indicators;

[0044] In the formula, For the initial particle swarm, the first i An initial particle; For the first i One chaotic variable; These are the upper and lower bounds of the search space; For particle indication;

[0045] In the formula, For the first i The initial velocity of the initial particle; A random number in the range (-1, 1); S2033: Load the candidate partitioning and grading quality control parameter vectors corresponding to each particle in the initial particle swarm, perform quality control on several preprocessed historical meteorological monitoring station data in the historical database, and use a multi-objective optimization function to calculate the corresponding optimization target value; S2034: Based on the optimization objective value, perform fast non-dominated sorting on the initial particle swarm to obtain different frontier levels, and maintain an external archive set to store all non-dominated solutions of the first frontier level; S2035: Introduces the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism to update the positions of the initial particle swarm and obtain an updated particle swarm. S2036: Using a multi-objective optimization function, calculate the optimization objective value for each updated particle in the updated particle swarm, and update the frontier level and external archive set based on the optimization objective value; S2037: Stop iterating the particle swarm until the number of iterations reaches the iteration threshold, and output all non-dominated solutions in the external archive as the Pareto optimal solution set.

[0046] In one alternative implementation, a Levy flight mechanism, a convergence factor, and a gray wolf cooperative mechanism are introduced to update the positions of the initial particle swarm, resulting in an updated particle swarm, including: S20351: In the external archive set, based on the crowding degree, the top three local optima are identified and defined as Alpha particles, Beta particles, and Delta particles, respectively. The historical best position of each particle is expanded to Alpha particles, Beta particles, and Delta particles. The information of the top three local optima is used to guide the particle flight direction, which enhances the convergence speed and accuracy. S20352: Based on Alpha, Beta, and Delta particles, the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism are introduced to update the initial velocity of each initial particle in the initial particle swarm, obtaining the updated velocity, as shown in the formula:

[0047] In the formula, Number of iterations t+ 1 of i The rate of update of each particle; Number of iterations t The i The update rate of each particle, which is the initial rate during the first iteration; Number of iterations t The convergence factor improves the inertia weight; Number of iterations t The globally optimal particle; Number of iterations t The i A new particle is generated, which is the initial particle in the first iteration; t This represents the current iteration number; The cooperation coefficient; A random number between (0, 1); Number of iterations t Alpha particles, Beta particles, and Delta particles; t This is an indicator of the number of iterations. i For particle indication; for Levy Distribute random numbers; b for Levy Step length, and b ∈[1,2]; The flight coefficient typically ranges from [0.1, 0.3] and is used to control the magnitude of randomness introduced by the Levy flight mechanism, balancing the algorithm's global exploration capability and local exploitation capability. The introduction of Levy random walk (long-tailed distribution) allows particles to make long-distance jumps, effectively escaping local extremum traps.

[0048] In the formula, These represent the maximum and minimum values ​​of the inertia weight; This is the threshold for the number of iterations; , To adjust the parameters; It is a hyperbolic tangent function; this design results in a larger weight in the early stage, which is beneficial for global search; and a smaller weight with a gradual change in the later stage, which is beneficial for fine-grained local mining. S20353: Based on the update rate, update the positions of the corresponding initial particles to obtain updated particles, until the positions of all initial particles are updated, resulting in an updated particle swarm. The formula is:

[0049] In the formula, Number of iterations t+ 1 of i An updated particle.

[0050] In one optional implementation, based on the application scenario requirements of each spatiotemporal classification quality control unit, the optimal partitioned and graded quality control parameter vector is selected and loaded from the Pareto optimal solution set, and quality control is performed on the real-time meteorological monitoring station data of the corresponding spatiotemporal classification quality control unit, including: S2041: In each spatiotemporal classification quality control unit, collect the real-time data of the corresponding meteorological monitoring station, and preprocess the real-time data of the meteorological monitoring station to obtain the preprocessed real-time data of the meteorological monitoring station. S2042: Determine the typical rainfall mode corresponding to the current moment, use the typical rainfall mode as the application scenario requirement of the corresponding spatiotemporal classification quality control unit, and obtain the preset weighting coefficient corresponding to the application scenario requirement; S2043: Based on the preset weighting coefficients corresponding to the application scenario requirements, the optimization objective value of each solution in the Pareto optimal solution set is weighted to obtain the comprehensive optimization objective value; In this embodiment, different rainfall modes correspond to different meteorological operational concerns, and a set of weighting coefficients are predefined for each typical rainfall mode. ,in,m These are modal indicators, corresponding to anomaly detection performance. Data fidelity and quality control efficiency ; Example of coefficient allocation strategy: Summer severe convective weather / Spring precipitation period (high-risk scenario): Strategy Setup: During periods of strong convection in summer and heavy rainfall in spring, rainfall events are concentrated, sudden, and intense, easily triggering floods or secondary disasters. Therefore, the strategy is set as "preferring false alarms (lowering fidelity requirements) to missed alarms (improving identification efficiency requirements), and requiring extremely fast response (improving efficiency requirements)." In other words, it is better to capture more suspected outliers to prevent missing real extreme rainfall events, while ensuring the real-time nature of data processing to meet flood control emergency needs. Example of coefficient setting: =[0.6,0.1,0.3], at this time Highest weight; Winter dry season (low-risk scenario): Strategy: During the dry winter season, rainfall is scarce and relatively stable, with an extremely low probability of disastrous rainstorms. At this time, the focus should be on the accuracy and long-term stability of the data, avoiding the accidental deletion or correction of weak, valid rainfall signals due to overly sensitive quality control thresholds, while maintaining a moderate level of real-time performance. Example of coefficient setting: =[0.2,0.6,0.2], at this time Highest weight; Autumn drizzle season: Strategy Setup: During the autumn drizzle season, rainfall is characterized by its long duration and moderate total amount, but low sunshine. Therefore, a "balanced" strategy is adopted, focusing on overall performance. This involves ensuring the accuracy of long-term rainy data (fidelity) while also considering the accuracy of identifying outliers and processing efficiency. Example of coefficient setting: =[0.4,0.4,0.2]; Due to the different objective function values ​​of solutions in the Pareto optimal solution set ( , , Because the dimensions and orders of magnitude are different, direct weighting will lead to the large numerical objective dominating the result. Therefore, normalization is required first. From the current Pareto optimal solution set { In the context of}, find the minimum value of each objective function. and maximum value For any solution in the solution set Its normalized target value The formula is:

[0051] In the formula, For the first The solution k The normalized target value of the objective function. ; For the first The solution k The objective value of the objective function; For the first The solution k The maximum and minimum values ​​of the objective function; To solve the indicator; k The objective function indicator; The formula for the overall optimization objective value is:

[0052] In the formula, For the first The comprehensive optimization objective value of the solution; For the first k The weighting coefficients of the objective function; S2044: Load the partitioned and graded quality control parameter vector corresponding to the solution with the best comprehensive optimization objective value in the Pareto optimal solution set as the optimal partitioned and graded quality control parameter vector; S2045: Based on the optimal partitioned and graded quality control parameter vector, perform quality control on the preprocessed real-time data of meteorological monitoring stations for the corresponding spatiotemporal classification quality control units.

[0053] In one optional implementation, based on the optimal partitioned and graded quality control parameter vector, quality control is performed on the preprocessed real-time data of the meteorological monitoring station for the corresponding spatiotemporal classification quality control unit, including: S20451: Read the... l The preprocessed real-time rainfall data and corresponding preprocessed real-time geographic information data from the meteorological monitoring station's real-time data in the spatiotemporal classification quality control unit, among which, l This is the indicator quantity for the spatiotemporal classification quality control unit; S20452: If the preprocessed real-time rainfall data <0 or Then, the preprocessed real-time rainfall data will be marked as a Level 1 anomaly. This refers to the dynamic extreme value threshold coefficient of the primary quality control parameter. This represents the maximum historical rainfall value. S20453: Based on preprocessed real-time rainfall data Calculate the first l Spatiotemporal classification quality control unit in the past Real-time rainfall increments within the window ,like Then, the preprocessed real-time rainfall data will be marked as a level 2 anomaly. This represents the time-series difference window width for the secondary quality control parameters. The mutation rate coefficient is a secondary quality control parameter. This serves as a baseline value for the mutation rate. S20454: Based on the preprocessed real-time geographic information data, the search radius... For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall at neighboring stations. ; It is worth noting that this step analyzes the spatial structure and directionality of rainfall to determine whether the data of the target station conforms to the surrounding rainfall distribution pattern. Its core innovation lies in using Principal Component Analysis (PCA) to dynamically identify the direction of the rainband and adjust the weights of neighboring stations accordingly, rather than simply using physical distance. S20455: If The preprocessed real-time rainfall data will then be marked as a level three anomaly. The relative error coefficient of rainfall is a secondary quality control parameter. S20456: If there are preprocessed real-time rainfall data marked as level one, two and / or three anomalies, select an anomaly handling algorithm according to the source of the anomaly to replace the preprocessed real-time rainfall data, and obtain the replaced preprocessed real-time rainfall data; It is worth noting that Level 1 anomaly (physical extreme value / instrument malfunction): if <0 or This is a serious error, usually caused by sensor malfunction, transmission interference, or exceeding the physical range, and the data is completely unreliable. Level 2 anomaly (temporal mutation): If the Level 1 anomaly is not triggered, but if This is a suspected flying point, which may be caused by instantaneous signal jitter or local strong convection (it needs to be combined with the three-level judgment). If the three-level anomaly is not triggered, it is an isolated timing noise point. Level 3 Exception (Space Inconsistency): If Level 1 and Level 2 exceptions are not triggered, but This is a site-specific error (such as single-site congestion or calibration deviation), and its value is unreasonable within a local range. For handling Level 1 anomalies, a spatial neighborhood-based weighted estimation algorithm is used to replace the preprocessed real-time rainfall data. For handling secondary anomalies, time-series smoothing and interpolation algorithms are used to replace the preprocessed real-time rainfall data; For handling level 3 anomalies, a multi-source fusion correction algorithm is used to replace the preprocessed real-time rainfall data; Handling compound exceptions: Priority coverage strategy: If both Level 1 and Level 2 anomalies are triggered simultaneously, they should be treated as Level 1 anomalies and replaced with a weighted estimation algorithm based on spatial neighborhood, because Level 1 anomalies indicate that the data has exceeded physical boundaries and the data is completely unreliable. If both Level 2 and Level 3 anomalies are triggered simultaneously, Level 3 anomalies should be treated first (spatial estimation), because Level 2 anomalies detect temporal abrupt changes, while Level 3 anomalies are verified using spatial correlation. Spatial estimation results are usually closer to the actual rainfall distribution than temporal interpolation. If only a level 2 exception is triggered and no level 3 exception is triggered, it is considered an isolated timing noise point and replaced using timing smoothing and interpolation algorithms.

[0054] In one alternative implementation, the search radius is determined based on preprocessed real-time geographic information data. For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall at neighboring stations. ,include: S204541: Taking the current meteorological monitoring station corresponding to the preprocessed real-time geographic information data as the target meteorological monitoring station, and using the coordinates of the target meteorological monitoring station as the center, ... Draw a circle with a radius of 1, and retrieve the set of all adjacent meteorological monitoring stations located within that circle. If the number of adjacent meteorological monitoring stations is... If the number of neighboring stations is less than 3, the third-level quality control will be automatically skipped, and only the first and second-level quality control will be relied upon. S204542: Extract the target station and all neighboring stations at the current time and in the past. Rainfall data at time intervals (e.g., the past hour, with six 10-minute intervals) are used to construct a... Rainfall matrix Each row of the matrix represents a meteorological monitoring station, and each column represents the preprocessed real-time data of the meteorological monitoring station at a given moment. S204543: Utilizing the rainfall matrix Real-time rainfall data after preprocessing from all meteorological monitoring stations in China at the current moment. The spatial covariance matrix is ​​constructed using the geographic coordinates of the corresponding preprocessed real-time geographic information data. For the spatial covariance matrix Perform eigenvalue decomposition to obtain eigenvalues. and its corresponding eigenvectors ; S204544: with the largest eigenvalue Corresponding feature vector The direction of the rain belt is the direction of the maximum rainfall gradient in the region, and is defined as the main direction of the rain belt. The direction vector is represented as ,in, The angle between the main axis of the rain belt and true north; S204545: For each adjacent meteorological monitoring station, calculate the geographic vector connecting the target meteorological monitoring station to the adjacent meteorological monitoring station, using the following formula:

[0055] In the formula, Geographical connection vector; For the first g Geographic coordinates of adjacent meteorological monitoring stations; The geographic coordinates of the target meteorological monitoring station; g The readings are from adjacent meteorological monitoring stations; S204546: Calculate geographic connection vectors With the main direction of the rainband The cosine of the included angle is used as the directional consistency coefficient, and the formula is:

[0056] In the formula, For the target meteorological monitoring station and the first g The directional consistency coefficient of adjacent meteorological monitoring stations; Geographic connection vector With the main direction of the rainband The cosine of the included angle; S204547: Using weight sensitivity parameters For the consistency coefficient After nonlinear amplification, the final direction weighting factor is obtained, as shown in the formula:

[0057] In the formula, For the target meteorological monitoring station and the first g Directional weighting factor for adjacent meteorological monitoring stations; S204548: Calculate the traditional inverse distance weighting factor, multiply the direction weighting factor by the inverse distance weighting factor to obtain the final neighbor station coupling weight. Normalize the coupling weights of all neighboring meteorological monitoring stations so that the sum of the weights is 1, and obtain the normalized coupling weight. The formula is:

[0058] In the formula, In order to be with the first g Coupling weights of adjacent meteorological monitoring stations; In order to be with the first g Normalized coupling weights of adjacent meteorological monitoring stations; In order to be with the first g Traditional inverse distance weighting factor for adjacent meteorological monitoring stations; In order to be with the first g Distance between adjacent meteorological monitoring stations; With the r Coupling weights of adjacent meteorological monitoring stations; r The readings are from adjacent meteorological monitoring stations; S204549: Based on the normalized coupling weights, the preprocessed real-time rainfall data is weighted to obtain the weighted rainfall estimate from neighboring stations. The formula is:

[0059] In the formula, For the first g Preprocessed real-time rainfall data from adjacent meteorological monitoring stations; This represents the total number of adjacent meteorological monitoring stations.

[0060] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EI) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus. The communication interface is used for communication between the aforementioned terminal and other devices. The memory can include Random Access Memory (RAM), or non-volatile memory, such as at least one disk storage device. Optionally, the memory can also be at least one storage device located remotely from the aforementioned processor.

[0061] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0062] Furthermore, to achieve the above objectives, embodiments of the present invention also propose a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the zoning and grading meteorological data quality control method of the present invention.

[0063] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable hardware devices (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0064] The embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (apparatus), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. "And / or" indicates that either one or both can be chosen. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the element.

[0068] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for quality control of meteorological data by region and level, characterized in that, The method includes: Historical data from meteorological monitoring stations were collected, corresponding historical spatiotemporal multidimensional features were extracted, and based on these features, typical rainfall patterns were introduced to divide the monitoring area into several spatiotemporal classification quality control units. For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function, as shown in the following formula: In the formula, To load the partitioned and graded quality control parameter space A multi-objective optimization function for quality control; To load the partitioned and graded quality control parameter space Anomaly identification efficiency function for quality control; To load the partitioned and graded quality control parameter space Data fidelity function for quality control; To load the partitioned and graded quality control parameter space The quality control efficiency function for quality control; The formula for the anomaly detection performance function is: In the formula, To load the partitioned and graded quality control parameter space The anomaly detection effectiveness value for quality control; To load the partitioned and graded quality control parameter space Precision on the validation set for quality control; To load the partitioned and graded quality control parameter space Recall rate on the validation set for quality control; In the formula, To load the partitioned and graded quality control parameter space Data fidelity for quality control; To verify the first one that was misjudged as an anomaly in the set j Rainfall data; To After quality control, the replacement was the first j Rainfall data; In the formula, To load the partitioned and graded quality control parameter space Quality control efficiency in the process of quality control; The average time taken to perform one quality control operation on the validation set; To find the minimum value, to prevent the denominator from being zero; Based on a multi-objective optimization function, a multi-objective swarm intelligence algorithm is used to iteratively optimize the parameter space of partitioned and graded quality control to obtain the Pareto optimal solution set. Based on the application scenario requirements of each spatiotemporal classification quality control unit, the optimal partitioned and graded quality control parameter vector is selected and loaded from the Pareto optimal solution set, and the real-time data of the meteorological monitoring station of the corresponding spatiotemporal classification quality control unit is subjected to quality control.

2. The method for regional and hierarchical meteorological data quality control according to claim 1, characterized in that, Historical data from meteorological monitoring stations were collected, and corresponding historical spatiotemporal multidimensional features were extracted. Based on these features, typical rainfall modes were introduced, and the monitoring area was divided into several spatiotemporal classification quality control units, including: Collect historical data from meteorological monitoring stations, including historical rainfall data and corresponding historical geographic information data from all meteorological monitoring stations within the monitoring area; The historical data of meteorological monitoring stations are preprocessed to obtain preprocessed historical data of meteorological monitoring stations, and the preprocessed historical data of meteorological monitoring stations is stored in the historical database. Extract historical spatiotemporal multidimensional features from preprocessed historical meteorological monitoring station data. These features include at least rainfall statistical features, rainfall extreme value features, rainfall timeliness features, geographical features, and spatiotemporal distribution features. Based on the geographical features described in the historical spatiotemporal multidimensional characteristics, and combined with the preset terrain classification rules, all meteorological monitoring stations within the monitoring area are divided into... K A static geographic sub-region, of which K This represents the total number of static geographic sub-regions. Based on static geographic sub-regions, a time dimension is introduced. According to historical spatiotemporal multi-dimensional rainfall statistics, extreme rainfall characteristics, and rainfall timeliness characteristics, a rule-based partitioning method combining calendar cycles and meteorological characteristic thresholds is used to divide the year into [number missing]. M A typical rainfall pattern, wherein the typical rainfall pattern includes at least the spring precipitation period, the summer strong convection period, the autumn continuous rain period, and the winter dry period, wherein, M This represents the total number of typical rainfall modes; Combining each static geographic sub-region and its corresponding M A typical rainfall mode was obtained. K × M Spatiotemporal classification quality control unit.

3. The method for regional and hierarchical meteorological data quality control according to claim 2, characterized in that, For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space and a multi-objective optimization function are constructed. The multi-objective optimization function includes an anomaly identification performance function, a data fidelity function, and a quality control efficiency function, including: For the spatiotemporal classification quality control unit, a partitioned and hierarchical quality control parameter space is constructed, which includes at least three levels of partitioned and hierarchical quality control parameters; For the partitioned and graded quality control parameter space, a multi-objective optimization function is constructed, which includes an anomaly identification efficiency function, a data fidelity function, and a quality control efficiency function.

4. The method for regional and hierarchical meteorological data quality control according to claim 3, characterized in that, The zoning and grading quality control parameters are used for grading and controlling the quality of meteorological monitoring station data, and include at least: Primary quality control parameters should include at least the dynamic extreme value threshold coefficient; Secondary quality control parameters should include at least the time-series difference window width and the mutation rate coefficient; The three-level quality control parameters include at least the spatial neighborhood search radius, the weight sensitivity parameter, and the relative error coefficient of rainfall.

5. The method for regional and hierarchical meteorological data quality control according to claim 4, characterized in that, Based on a multi-objective optimization function, a multi-objective swarm intelligence algorithm is used to iteratively optimize the parameter space of the partitioned and graded quality control system, obtaining a Pareto optimal solution set, including: Encode the partitioned and graded quality control parameter vectors corresponding to the partitioned and graded quality control parameter space into individual vectors of the multi-objective particle swarm optimization algorithm, and initialize the multi-objective particle swarm optimization algorithm; The chaotic sequence is generated using the Logistic mapping, and then mapped to the solution space of the particles to obtain the initial particle swarm. Load the candidate partitioning and grading quality control parameter vectors corresponding to each particle in the initial particle swarm, perform quality control on several preprocessed historical meteorological monitoring station data in the historical database, and use a multi-objective optimization function to calculate the corresponding optimization target value. Based on the optimization objective value, the initial particle swarm is subjected to fast non-dominated sorting to obtain different frontier levels, and an external archive set is maintained to store all non-dominated solutions of the first frontier level. By introducing the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism, the initial particle swarm is updated to obtain an updated particle swarm. Using a multi-objective optimization function, the optimization objective value corresponding to each updated particle in the updated particle swarm is calculated, and the frontier level and external archive set are updated according to the optimization objective value. The iteration continues until the number of iterations reaches the threshold, at which point the iterative update of the particle swarm stops, and all non-dominated solutions in the external archive set are output as the Pareto optimal solution set.

6. The method for regional and hierarchical meteorological data quality control according to claim 5, characterized in that, By introducing the Levy flight mechanism, convergence factor, and gray wolf cooperative mechanism, the initial particle swarm is updated with positions to obtain an updated particle swarm, including: In the external archive set, based on the degree of crowding, the top three local optima are identified and defined as Alpha particles, Beta particles, and Delta particles, respectively. The historical best position of each particle is then extended to Alpha particles, Beta particles, and Delta particles. Based on Alpha, Beta, and Delta particles, the Levy flight mechanism, convergence factor, and gray wolf cooperation mechanism are introduced to update the initial velocity of each initial particle in the initial particle swarm, resulting in the updated velocity, as shown in the formula: In the formula, Number of iterations t+ 1 of i The rate of update of each particle; Number of iterations t The i The update rate of each particle, which is the initial rate during the first iteration; Number of iterations t The convergence factor improves the inertia weight; Number of iterations t The globally optimal particle; Number of iterations t The i A new particle is generated, which is the initial particle in the first iteration; t This represents the current iteration number; The cooperation coefficient; A random number between (0, 1); Number of iterations t Alpha particles, Beta particles, and Delta particles; t This is an indicator of the number of iterations. i For particle indication; for Levy Distribute random numbers; b for Levy Step length, and b ∈[1,2]; For flight coefficient; In the formula, These represent the maximum and minimum values ​​of the inertia weight; This is the threshold for the number of iterations; , To adjust the parameters; It is the hyperbolic tangent function; Based on the update rate, the positions of the corresponding initial particles are updated to obtain updated particles. This process continues until the positions of all initial particles have been updated, resulting in an updated particle swarm. The formula is: In the formula, Number of iterations t+ 1 of i An updated particle.

7. The method for regional and hierarchical meteorological data quality control according to claim 6, characterized in that, Based on the application scenario requirements of each spatiotemporal classification quality control unit, the optimal partitioned and graded quality control parameter vector is selected and loaded from the Pareto optimal solution set, and quality control is performed on the real-time meteorological monitoring station data of the corresponding spatiotemporal classification quality control unit, including: In each spatiotemporal classification quality control unit, real-time data from the corresponding meteorological monitoring station is collected and preprocessed to obtain preprocessed real-time data from the meteorological monitoring station. Determine the typical rainfall mode corresponding to the current moment, use the typical rainfall mode as the application scenario requirement of the corresponding spatiotemporal classification quality control unit, and obtain the preset weighting coefficients corresponding to the application scenario requirement; Based on the preset weighting coefficients corresponding to the application scenario requirements, the optimization objective value of each solution in the Pareto optimal solution set is weighted to obtain the comprehensive optimization objective value; The partitioned and graded quality control parameter vector corresponding to the solution with the best overall optimization objective value in the Pareto optimal solution set is loaded as the optimal partitioned and graded quality control parameter vector; Based on the optimal partitioned and graded quality control parameter vector, quality control is performed on the preprocessed real-time data of meteorological monitoring stations for the corresponding spatiotemporal classification quality control units.

8. The method for quality control of regional and hierarchical meteorological data according to claim 7, characterized in that, Based on the optimal partitioned and graded quality control parameter vector, quality control is performed on the preprocessed real-time data of meteorological monitoring stations for the corresponding spatiotemporal classification quality control units, including: Read the first l The preprocessed real-time rainfall data and corresponding preprocessed real-time geographic information data from the meteorological monitoring station's real-time data in the spatiotemporal classification quality control unit, among which, l This is the indicator quantity for the spatiotemporal classification quality control unit; If the preprocessed real-time rainfall data <0 or Then, the preprocessed real-time rainfall data will be marked as a Level 1 anomaly. This refers to the dynamic extreme value threshold coefficient of the primary quality control parameter. This represents the maximum historical rainfall value. Based on preprocessed real-time rainfall data Calculate the first l Spatiotemporal classification quality control unit in the past Real-time rainfall increments in the window ,like Then, the preprocessed real-time rainfall data will be marked as a level 2 anomaly. This represents the time-series difference window width for the secondary quality control parameters. The mutation rate coefficient is a secondary quality control parameter. This serves as a baseline value for the mutation rate. Based on the preprocessed real-time geographic information data, the search radius For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall at neighboring stations. ; like The preprocessed real-time rainfall data will then be marked as a level three anomaly. The relative error coefficient of rainfall is a secondary quality control parameter. If preprocessed real-time rainfall data is marked as Level 1, 2, and / or 3 anomalies, an anomaly handling algorithm is selected based on the source of the anomaly to replace the preprocessed real-time rainfall data, resulting in replaced preprocessed real-time rainfall data.

9. The method for regional and hierarchical meteorological data quality control according to claim 8, characterized in that, Based on the preprocessed real-time geographic information data, the search radius For all adjacent meteorological monitoring stations within the area, the PCA algorithm is used to extract the main direction of the regional rainband, the angular consistency between the target station and its neighbors is calculated, and a weighted sensitivity parameter is applied. Calculate the weighted estimate of rainfall from neighboring stations, including: Taking the current meteorological monitoring station corresponding to the preprocessed real-time geographic information data as the target meteorological monitoring station, and taking the coordinates of the target meteorological monitoring station as the center, and using... Draw a circle with a radius and retrieve all adjacent meteorological monitoring stations within that circle. If the number of adjacent meteorological monitoring stations is less than the threshold, skip the third-level quality control and rely only on the first and second-level quality control. Extract rainfall data from the target station and all neighboring stations at the current time and at several past time intervals to construct a rainfall matrix. Each row of the matrix represents a meteorological monitoring station, and each column represents the preprocessed real-time data of the meteorological monitoring station at the corresponding time. Using rainfall matrix Real-time rainfall data after preprocessing from all meteorological monitoring stations in China at the current moment. The spatial covariance matrix is ​​constructed using the geographic coordinates of the corresponding preprocessed real-time geographic information data. For the spatial covariance matrix Perform eigenvalue decomposition to obtain eigenvalues ​​and their corresponding eigenvectors; The direction of the eigenvector corresponding to the largest eigenvalue is defined as the direction of the maximum gradient of rainfall distribution in the region, and is thus defined as the main direction of the rain belt. For each adjacent meteorological monitoring station, calculate the geographic connection vector from the target meteorological monitoring station to the adjacent meteorological monitoring station; Calculate the cosine of the angle between the geographic connection vector and the main direction of the rain belt, and use it as the directional consistency coefficient; Using weight sensitivity parameters The consistency coefficient is nonlinearly amplified to obtain the final direction weighting factor; Calculate the traditional inverse distance weighting factor, multiply the direction weighting factor by the inverse distance weighting factor to obtain the final neighbor station coupling weight, normalize the coupling weights of all neighboring meteorological monitoring stations so that the sum of the weights is 1, and obtain the normalized coupling weight. Based on the normalized coupling weights, the preprocessed real-time rainfall data is weighted to obtain the weighted rainfall estimate from neighboring stations. .

Images