A method for reconstructing typhoon missing data based on statistical interpolation

Through the statistical interpolation method and PSO algorithm, parameter adjustment of the typhoon model is performed, and the outliers are detected in combination with the Z-score method, the problems of typhoon data are solved, and the accuracy and reliability of typhoon path and intensity prediction are improved. enhances the stability of the meteorological model.

CN118484638BActive Publication Date: 2024-11-12NANJING UNIV OF INFORMATION SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410947160.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-12
Estimated Expiration
2044-07-16

AI Technical Summary

Technical Problem

The lack or incompleteness of key typhoon data affects the accurate prediction of typhoon paths and intensity, reduces the accuracy of establishing and predicting meteorological models, and has adverse effects on disaster prevention and emergency management.

Method used

Using statistical interpolation method, other complete typhoon data are used to construct typhoon models, and parameters are adjusted in combination with PSO algorithm. Outliers are detected using Z-score method, and data reconstruction and timing fine-tuning are carried out to achieve data completion and replication. structure.

Benefits of technology

It improves the accuracy and reliability of typhoon path and intensity prediction, reduces the situation of data reconstruction anomalies, and enhances the robustness of the model and the stability of the prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118484638B_ABST
    Figure CN118484638B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for reconstructing typhoon lost data based on statistical interpolation, comprising the following steps: S1, using complete typhoon elements as independent variables and missing key data as dependent variables to construct a typhoon model; S2, inputting typhoon key elements as independent variables into the typhoon model, and adjusting the independent variable parameters using a PSO algorithm; S3, using a Z-score method to detect data generated by a typhoon, and locating typhoon data anomalies; S4, using a trained typhoon model to reconstruct typhoon data anomalies; S5, performing time series anomaly detection on the reconstructed data, and fine-tuning based on the time series. The present invention can achieve small data reconstruction errors and stronger robustness.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of typhoon data processing, and in particular to a method for reconstructing typhoon lost data based on statistical interpolation. Background Art

[0002] With the intensification of climate change and the frequent occurrence of meteorological disasters, typhoon monitoring and prediction have become increasingly important. A typhoon is a strong tropical cyclone, and accurate prediction of its path and intensity is crucial to reducing casualties and property losses. In order to accurately predict the path, intensity and impact range of a typhoon, scientists and meteorologists need a large amount of typhoon observation data, including key data such as air pressure, wind speed, path, etc. However, due to factors such as the limitations of observation equipment, the imperfection of the observation network and the complexity of the natural environment, key typhoon data is often missing and incomplete, which poses a challenge to typhoon prediction and defense.

[0003] One of the main challenges is the missing or incomplete key data, which may affect the accurate prediction of typhoon path and intensity. For example, the typhoon center pressure data, as a key reference data, may be damaged by the typhoon, and the detection equipment may not be able to provide sufficiently accurate and timely pressure data. Missing typhoon data will affect the establishment and prediction accuracy of meteorological models, reduce the accuracy and reliability of typhoon predictions, and have an adverse impact on disaster prevention and emergency management. Therefore, a method and system are needed to complete the key data in typhoon monitoring to improve the accuracy and reliability of typhoon predictions.

[0004] At present, although there are some methods to collect and analyze typhoon-related data, and these methods rely on meteorological satellites and sensors to collect data, these data may be affected by factors such as sensor failure and cloud cover, resulting in incomplete data. There are also certain limitations, such as limited processing capabilities for the particularity and complexity of typhoon meteorological data, inability to fully utilize other meteorological data for completion, and the need to improve the accuracy and stability of prediction results. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a method for reconstructing typhoon lost data based on statistical interpolation, by utilizing other complete data of the typhoon and combining the typhoon model to interpolate and complete the key missing data, thereby improving the prediction ability of the typhoon path and intensity.

[0006] Technical solution: A method for reconstructing typhoon lost data based on statistical interpolation, comprising the following steps:

[0007] S1, using complete typhoon elements as independent variables and missing key data as dependent variables to construct a typhoon model;

[0008] S2, input the key elements of typhoon into the typhoon model as independent variables, and use the PSO algorithm to adjust the parameters of the independent variables;

[0009] S3, the Z-score method is used to detect the data generated by the typhoon and locate the abnormal values ​​of the typhoon data;

[0010] S4, using the trained typhoon model to reconstruct the data of typhoon data outliers;

[0011] S5, performs time series anomaly detection on the reconstructed data and performs fine-tuning based on the time series.

[0012] Further, the expression of the typhoon model is: Among them, P is the key data to be completed. , , …, It is a complete typhoon element. , , …, is the weight parameter of the key elements of the typhoon, is the error term.

[0013] Furthermore, the key elements of the typhoon include: wind speed, air pressure, humidity, and temperature; the PSO algorithm is used to adjust the parameters of the four independent variables, and the specific steps are as follows:

[0014] S21, taking each variable as an individual, initialize four particles to represent the candidate solution of each parameter, and randomly initialize the position and velocity of each particle;

[0015] S22, based on the current position and velocity of the particle, as well as its individual optimal solution and the global optimal solution, the position and velocity of the particle are updated; through iterative updates, the particle swarm gradually converges to the optimal solution of the objective function;

[0016] The speed update formula is as follows: in, is the inertia weight, and is the learning factor, and is a random number in the interval [0,1]; is the best individual history. is the individual's current optimal value;

[0017] Position update formula:

[0018] in, Represents the current particle location, Represents the current particle speed;

[0019] S23, use the root mean square error to evaluate the fitness of the typhoon model parameters, then continuously adjust the parameters and iterate the model until the root mean square error is less than the set threshold; the root mean square error formula is:

[0020] in, is the observed value of the data, is the predicted value of the typhoon model, and n is the total number of typhoon elements.

[0021] Further, in step S3, the expression of the Z-score method is as follows: Where X represents a single raw data value, represents the overall mean, represents the overall standard deviation;

[0022] When z is between -3 and 3, it is normal data; otherwise, it is abnormal data.

[0023] Further, in step S5, after the reconstruction is completed, the data is subjected to time series anomaly detection, such as when the sum of the difference between the reconstructed data value and the two previous and next moments is greater than twice the difference between the two previous and next moments:

[0024] Then reconstruct the data Fine-tune according to the following formula:

[0025] in, is the previous moment value of the data to be reconstructed, is the value of the next moment of the data to be reconstructed, is the predicted value of the typhoon model.

[0026] Compared with the prior art, the present invention has the following significant effects:

[0027] The present invention trains historical typhoon data using a linear model and adjusts the model parameters using a PSO algorithm to timely reconstruct the situation where typhoon-generated data is lost and thus causes data anomalies, so that the reconstruction of typhoon data by the trained typhoon model is more accurate and timely; by verifying and fine-tuning the reconstructed data, the situation of abnormal reconstructed data is also reduced, thereby achieving the purpose of small data reconstruction error and having stronger robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a flow chart of the present invention;

[0029] Figure 2 This is a schematic diagram of typhoon abnormal data;

[0030] Figure 3 Schematic diagram of particle velocity and position update in PSO;

[0031] Figure 4 Schematic diagram of reconstructed data. DETAILED DESCRIPTION

[0032] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0033] The framework flow chart of the typhoon data reconstruction method based on statistical interpolation of the present invention is as follows: Figure 1 As shown, the specific steps are:

[0034] Step 1: Establish a linear regression model for reconstructing typhoon key data

[0035] Using complete typhoon elements as independent variables and missing key data as dependent variables, a multivariate linear regression model, namely the typhoon model, is constructed; the formula of the typhoon model can be expressed as:

[0036] (1)

[0037] Among them, P is the key data to be completed. , , …, It is a complete typhoon element. , , …, is the weight parameter of the key elements of the typhoon, is the error term.

[0038] Step 2: Determine the weight parameters of the constructed model

[0039] The key elements of typhoon: wind speed, air pressure, humidity, and temperature are input into the typhoon model as independent variables, and the PSO algorithm is used to adjust the parameters of the four independent variables. Specifically:

[0040] Step 21, taking each variable as an individual and each individual as a particle; initializing four particles, representing the candidate solution of each parameter respectively, and randomly initializing the position and velocity of each particle;

[0041] Step 22: Update the particle's position and velocity according to the particle's current position and velocity, as well as its individual optimal solution and the global optimal solution. Through iterative updates, the particle swarm gradually converges to the optimal solution of the objective function, such as Figure 3 shown.

[0042] The speed update formula is as follows:

[0043] (2)

[0044] in, is the inertia weight, and is the acceleration coefficient or learning factor, and is a random number in the interval [0,1]; is the best individual history. is the current optimal for the individual.

[0045] Position update formula:

[0046] (3)

[0047] in, Represents the current particle location, Represents the current particle speed.

[0048] Step 23, use the root mean square error to evaluate the fitness of the typhoon model parameters, and then continuously adjust the parameters and iterate the model until the evaluation value of the model, that is, the root mean square error, is less than the set threshold. The root mean square error formula is:

[0049] (4)

[0050] in, is the observed value of the data, is the predicted value of the typhoon model, and n is the total number of typhoon elements.

[0051] Step 3: Locating outliers in typhoon data

[0052] For data outliers in the typhoon monitoring process, the Z-score method is used for detection, and the formula is as follows:

[0053] (5)

[0054] Where X represents a single raw data value, represents the overall mean, represents the population standard deviation. If it is less than -3 or greater than 3, it means that an abnormal value has been detected and the original data needs to be processed. Figure 2 shown.

[0055] Step 4: Data reconstruction

[0056] The typhoon model adjusted in step 2 is used to reconstruct the abnormal data detected in step 3 to ensure the integrity of the data. The reconstructed data is as follows: Figure 4 shown.

[0057] Step 5: Verification and fine-tuning

[0058] To ensure the accuracy and reliability of typhoon reconstruction data, the data is tested for time series anomalies after reconstruction. The previous and next moments are used as references for simple difference detection. If the difference deviates too much, that is, the sum of the difference between the reconstructed data value and the previous and next moments is greater than twice the difference between the previous and next moments, the formula is as follows:

[0059] (6)

[0060] Then reconstruct the data Fine-tune according to the following formula:

[0061] (7)

[0062] in, is the previous moment value of the data to be reconstructed, is the value of the next moment of the data to be reconstructed, is the predicted value of the typhoon model.

[0063] The method of the present invention solves the problem of reconstructing abnormal typhoon data, can effectively and timely apply the remaining complete data of the typhoon to reconstruct the abnormal data or the missing data, utilizes the characteristic that PSO has good adaptability to model parameter adjustment, constructs and adjusts the parameters of the typhoon regression model, and reconstructs in time after the abnormal data is detected, so that the typhoon data has integrity, effectively improves the accuracy and reliability of typhoon prediction by the meteorological model, and provides timely and accurate information for disaster prevention and emergency management.

Claims

1. A method for reconstructing typhoon lost data based on statistical interpolation, which is characterized in that it includes the following steps: S1, using complete typhoon elements as independent variables and missing key data as dependent variables to construct a typhoon model; S2, input the key elements of typhoon into the typhoon model as independent variables, and use the PSO algorithm to adjust the parameters of the independent variables; S3, the Z-score method is used to detect the data generated by the typhoon and locate the abnormal values ​​of the typhoon data; S4, using the trained typhoon model to reconstruct the data of typhoon data outliers; S5, perform time series anomaly detection on the reconstructed data and make fine adjustments based on the time series; if the sum of the difference between the reconstructed data value and the two moments before and after is greater than twice the difference between the two moments before and after: Then reconstruct the data x t Fine-tune according to the following formula: in, x t-1 is the previous moment value of the data to be reconstructed, x t+1 is the value of the next moment of the data to be reconstructed, x model is the predicted value of the typhoon model; The expression of the typhoon model is: P=β0+β1X1+β2X2+…+β n X n +e Among them, P is the key data to be completed, X1, X2, ..., X n is a complete typhoon element, β0, β1, ..., β n is the weight parameter of the key elements of the typhoon, and ε is the error term.

2. The method for reconstructing typhoon lost data based on statistical interpolation according to claim 1, wherein the typhoon key elements include: Wind speed, air pressure, humidity, temperature; The PSO algorithm is used to adjust the parameters of the four independent variables. The specific steps are as follows: S21, taking each variable as an individual, initialize four particles to represent the candidate solution of each parameter, and randomly initialize the position and velocity of each particle; S22, based on the current position and velocity of the particle, as well as its individual optimal solution and the global optimal solution, the position and velocity of the particle are updated; through iterative updates, the particle swarm gradually converges to the optimal solution of the objective function; The speed update formula is as follows: Where ω is the inertia weight, c1 and c2 are learning factors, and is a random number in the interval [0,1]; gBest is the individual's historical best, and pBest is the individual's current best; Position update formula: Among them, x d i represents the current position of particle i, v d i Represents the current velocity of particle i; S23, use the root mean square error to evaluate the fitness of the typhoon model parameters, then continuously adjust the parameters and iterate the model until the root mean square error is less than the set threshold; the root mean square error formula is: Among them, X obs,i is the observed value of the data, X model,i is the predicted value of the typhoon model, and n is the total number of typhoon elements.

3. The method for reconstructing typhoon lost data based on statistical interpolation according to claim 1 is characterized in that, in step S3, the expression of the Z-score method is as follows: in, X represents a single raw data value, μ represents the population mean, and σ represents the population standard deviation; When z is between -3 and 3, it is normal data; otherwise, it is abnormal data.