Lightning prediction model generation method and device, equipment and medium

Abstract

The invention discloses a thunder and lightning prediction model generation method and device, equipment and a medium. The method comprises the following steps: acquiring historical thunder and lightning training data; determining a current stationary parameter according to historical thunder and lightning training data; the historical thunder training data and the current stationary parameters are input into an initial model, a prediction result is obtained, the initial model is used for predicting the occurrence time, the occurrence position and the thunder occurrence probability of a second time period according to the occurrence time, the occurrence position and the thunder occurrence intensity of the first time period, and the thunder occurrence probability is obtained; and in response to historical thunder training data and a prediction result, when it is determined that the prediction result meets an adjustment and optimization ending condition, determining the stationary parameter as an optimal stationary parameter, and training the initial model according to the optimal stationary parameter and the historical thunder training data. According to the embodiment of the invention, the lightning prediction accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method, apparatus, device and medium for generating a lightning prediction model. Background Art

[0002] With the rapid development of technology, the types of devices are gradually increasing. To ensure the normal operation of the devices, prevent the devices from being affected by lightning, and maintenance personnel need to monitor the lightning conditions corresponding to the devices, so as to perform corresponding lightning protection operations on each device in a timely manner.

[0003] Currently, by monitoring the lightning conditions corresponding to the devices, lightning rods or grounding devices are installed on the devices for safety protection.

[0004] However, the traditional method cannot predict lightning in advance, cannot achieve active protection before the lightning strikes, and is likely to cause device damage. Summary of the Invention

[0005] The present invention provides a method, apparatus, device and medium for generating a lightning prediction model to improve the accuracy of generating the lightning prediction model.

[0006] In a first aspect, an embodiment of the present invention provides a method for generating a lightning prediction model, the method including:

[0007] Obtaining historical lightning training data;

[0008] Determining a current stationary parameter according to the historical lightning training data;

[0009] Inputting the historical lightning training data and the current stationary parameter into an initial model to obtain a prediction result, where the initial model is used to predict the occurrence time, occurrence location and lightning occurrence probability in a second time period according to the occurrence time, occurrence location and lightning occurrence intensity in a first time period, and the first time period precedes the second time period;

[0010] When it is determined that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determining the stationary parameter as the optimal stationary parameter, and training the initial model according to the optimal stationary parameter and the historical lightning training data to obtain a target lightning prediction model;

[0011] When it is determined that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, updating the historical lightning training data, and updating the current stationary parameter and the prediction result.

[0012] In a second aspect, an embodiment of the present invention provides a lightning prediction method, the method including:

[0013] Obtaining real-time lightning monitoring data within a preset acquisition time period;

[0014] Input the historical lightning training data within a historical time period and the real-time lightning monitoring data within a preset acquisition time period into the target lightning prediction model to obtain predicted lightning data, where the predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability;

[0015] Obtain at least one lightning probability threshold and the lightning protection measures corresponding to each lightning probability threshold;

[0016] Compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine the target probability and the target lightning protection measure, and send the lightning protection measure to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measure.

[0017] In a third aspect, an embodiment of the present invention further provides a lightning prediction model generation device, and the device includes:

[0018] A data acquisition module, configured to acquire historical lightning training data;

[0019] A parameter determination module, configured to determine the current stationary parameter according to the historical lightning training data;

[0020] A result acquisition module, configured to input the historical lightning training data and the current stationary parameter into an initial model to obtain a prediction result, where the initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in a second time period according to the occurrence time, occurrence location, and lightning occurrence intensity in a first time period, and the first time period precedes the second time period;

[0021] A model training module, configured to, when it is determined that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determine the stationary parameter as the optimal stationary parameter, and train the initial model according to the optimal stationary parameter and the historical lightning training data to obtain the target lightning prediction model;

[0022] A data update module, configured to, when it is determined that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, update the historical lightning training data, and update the current stationary parameter and the prediction result.

[0023] In a fourth aspect, an embodiment of the present invention further provides a lightning prediction device, and the device includes:

[0024] A real-time data acquisition module, configured to acquire real-time lightning monitoring data within a preset acquisition time period;

[0025] A prediction data acquisition module, configured to input historical lightning training data within a historical time period and real-time lightning monitoring data within a preset acquisition time period into a target lightning prediction model to obtain predicted lightning data, where the predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability;

[0026] A measure acquisition module, configured to acquire at least one lightning probability threshold and lightning protection measures corresponding to each lightning probability threshold;

[0027] An information sending module, configured to compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine a target probability and a target lightning protection measure, and send the lightning protection measure to corresponding lightning protection devices and maintenance personnel, so that the lightning protection devices perform automatic lightning protection operations according to the lightning protection measures.

[0028] In a fifth aspect, an embodiment of the present invention further provides a lightning prediction model generation device, where the lightning prediction model generation device includes:

[0029] At least one processor; and

[0030] A memory communicatively connected to the at least one processor; wherein,

[0031] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the lightning prediction model generation method of any embodiment of the present invention.

[0032] According to another aspect of the present invention, there is provided a computer-readable storage medium storing computer instructions for causing a processor to implement the lightning prediction model generation method of any embodiment of the present invention when executed.

[0033] In the technical solution of the embodiment of the present invention, historical lightning training data is obtained; a current steady parameter is determined according to the historical lightning training data; the historical lightning training data and the current steady parameter are input into an initial model to obtain a prediction result. The initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in a second time period according to the occurrence time, occurrence location, and lightning occurrence intensity in a first time period, and the first time period precedes the second time period; in response to determining that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, the steady parameter is determined as the optimal steady parameter, and the initial model is trained according to the optimal steady parameter and the historical lightning training data to obtain a target lightning prediction model; in response to determining that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, the historical lightning training data is updated, and the current steady parameter and the prediction result are updated. By continuously optimizing the historical lightning training data, at least one steady parameter is obtained, the various steady parameters are tuned to obtain the current steady parameter, and the current optimal steady parameter is input into the model for training, which improves the accuracy of the generated lightning prediction model and can accurately predict the lightning occurrence situation.

[0034] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0036] Figure 1 is a flowchart of a method for generating a lightning prediction model according to Embodiment 1 of the present invention;

[0037] Figure 2 is a flowchart of a method for generating a lightning prediction model according to Embodiment 2 of the present invention;

[0038] Figure 3 is a flowchart of a lightning prediction method according to Embodiment 3 of the present invention;

[0039] Figure 4 is a structural diagram of a device for generating a lightning prediction model according to an embodiment of the present invention;

[0040] Figure 5 is a flowchart of a lightning prediction device according to an embodiment of the present invention;

[0041] Figure 6 It is a schematic structural diagram of a lightning prediction model generation device provided by an embodiment of the present invention. Specific embodiments

[0042] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0043] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0044] In the technical solution of the embodiment of the present invention, the acquisition, storage, application, etc. of historical lightning training data and the like all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0045] Embodiment 1

[0046] Figure 1 It is a flowchart of a lightning prediction model generation method provided by Embodiment 1 of the present invention. The embodiment of the present invention is applicable to the situation of generating a lightning prediction model. This method can be executed by a lightning prediction model generation device, and the lightning prediction model generation device can be implemented in the form of hardware and / or software.

[0047] See Figure 1 The lightning prediction model generation method shown includes:

[0048] S101. Obtain historical lightning training data.

[0049] Among them, the historical lightning training data can be used to describe the training set data to be used for model training.

[0050] Specifically, lightning monitoring devices such as lightning location detectors or electric field detectors can be deployed. These devices can monitor lightning activities in real time, record relevant parameters of lightning occurrences, and store data. After a period of monitoring, historical lightning data can be accumulated. Obtain a preset time period and filter the historical lightning data to select the historical lightning data within the preset time period as historical lightning training data for training the initial model. The filtering methods include, but are not limited to, sequential search algorithms, block search algorithms, tree table search algorithms, or hash search algorithms, etc. The embodiments of the present invention do not limit this.

[0051] S102. Determine the current stationary parameter according to the historical lightning training data.

[0052] Among them, the current stationary parameter can be used to describe the parameters in the model for controlling the lightning prediction result in the current time period.

[0053] Specifically, the initial model includes stationary parameters. The stationary parameters include at least one parameter item and the corresponding parameter values of each parameter item. The data values of the stationary parameters affect the prediction result of the model for the lightning situation. Different stationary parameters correspond to different lightning prediction results. By adjusting the parameter values of each parameter item of the stationary parameters, the stationary parameter with the highest prediction accuracy can be obtained as the current stationary parameter. The stationary parameters can be adjusted according to the historical lightning training data, and the stationary parameter with the highest prediction accuracy after adjustment is determined as the current stationary parameter. For example, the model can be an autoregressive integrated moving average (ARIMA) model, and the stationary parameters can be the autoregressive term (p), the differencing order (d), and the moving average term (q) of the ARIMA model, and the specific values corresponding to the autoregressive term (p), the differencing order (d), and the moving average term (q).

[0054] S103. Input the historical lightning training data and the current stationary parameter into the initial model to obtain a prediction result. The initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in the second time period based on the occurrence time, occurrence location, and lightning occurrence intensity in the first time period, and the first time period precedes the second time period.

[0055] Among them, the initial model can be used to describe the model for predicting the lightning situation to be performed. The prediction result can be used to describe the predicted data of the lightning information obtained by inputting the historical lightning training data and the current stationary parameter into the initial model. The first time period can be the time period to which the historical lightning training data belongs. The second time period can be a preset time period for obtaining lightning data to be obtained.

[0056] Specifically, the first time period precedes the second time period and the first time period is greater than the second time period. The historical lightning training data and the current steady-state parameters are input into the initial model, and the steady-state parameters in the initial model are updated by replacing them with the current steady-state parameters. The occurrence time, occurrence location, and occurrence lightning intensity data of the lightning in the first time period in the historical lightning training data are input into the initial model after the steady-state parameter update to obtain the prediction result of the lightning condition, and the prediction result is the occurrence time, occurrence location, and lightning occurrence probability of the lightning in the second time period.

[0057] S104. In response to determining that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determine the steady-state parameter as the optimal steady-state parameter, and train the initial model according to the optimal steady-state parameter and the historical lightning training data to obtain the target lightning prediction model.

[0058] Among them, the tuning end condition can be used to describe the preset condition for determining whether to end the training according to the prediction result. The optimal steady-state parameter can be used to describe the steady-state parameter when the steady-state parameter is adjusted to meet the tuning end condition. The target lightning prediction model can be used to describe the model obtained after training the initial model with the historical lightning training data.

[0059] Specifically, according to the initial model, the prediction result is obtained by predicting the historical lightning training data, and it is judged whether the prediction result meets the tuning end condition. If the tuning end condition is met, the steady-state parameter is determined as the optimal steady-state parameter, and the steady-state parameter in the initial model is replaced with the optimal steady-state parameter, and the historical lightning training data is input into the initial model to train the initial model to obtain the target lightning prediction model. Exemplarily, the tuning end condition can be that by presetting the thresholds of accuracy, recall, precision, and F1 score, and calculating whether the differences between the four values of accuracy, recall, precision, and F1 score corresponding to the prediction result of this prediction and the preset thresholds are within the preset range or whether these four values are greater than their respective thresholds, etc., and determining whether the prediction result meets the tuning end condition according to the judgment result. When it is determined that the prediction result meets the tuning end condition, the steady-state parameter is determined as the optimal steady-state parameter, and the initial model is trained according to the optimal steady-state parameter and the historical lightning training data to obtain the target lightning prediction model.

[0060] Accuracy is one of the commonly used indicators in model evaluation and is used to measure the accuracy of model prediction. The calculation formula is as follows:

[0061]

[0062] The recall rate measures the model's ability to identify positive samples and is also known as sensitivity or recall. The calculation formula is as follows:

[0063]

[0064] The precision rate measures the accuracy of the model's prediction of positive samples. The calculation formula is:

[0065]

[0066] The F1 score is an indicator that comprehensively considers the precision rate and the recall rate and is used to evaluate the performance of the model. It is the harmonic mean of the precision rate and the recall rate, and the calculation formula is as follows:

[0067]

[0068] In the above formula, T P is the number of true positive samples (the model predicts positive samples and they are actually positive samples), T N is the number of true negative samples (the model predicts negative samples and they are actually negative samples), F P is the number of false positive samples (the model predicts positive samples but they are actually negative samples), and FN is the number of false negative samples (the model predicts negative samples but they are actually positive samples).

[0069] S105. When it is determined that the prediction result does not meet the tuning end condition in response to the historical lightning training data and the prediction result, update the historical lightning training data, and update the current steady parameters and the prediction result.

[0070] Specifically, when it is determined according to the prediction result that the prediction result does not meet the optimization end condition, the historical lightning training data is updated. The update method can be differential processing. After differential processing, the change amount of the data at different time points or spatial points can be calculated, and the change trend and pattern of the data can be presented more clearly. For the prediction of this time series data of historical lightning training data, the stationary data obtained after differential processing is more conducive to establishing an accurate prediction model, can better capture the internal law of the data, and thus improve the accuracy and reliability of the prediction. At the same time, sensor data such as lightning monitoring equipment is often non-stationary, with trends (such as linear or quadratic trends) or seasonal components. Through differential processing, for example, the first-order difference can remove the linear trend, the second-order difference can remove the quadratic trend, and the seasonal difference can eliminate the seasonal fluctuation, converting the non-stationary sequence into a stationary sequence, meeting the requirements of most classical time series analysis models (such as the ARIMA model for example) for data stationarity, and improving the accuracy and reliability of the model. According to the updated historical lightning training data, its corresponding stationary parameters are obtained, and the current stationary parameters are updated through the obtained stationary parameters. The initial model corresponding to the updated current stationary parameters is used to predict the updated historical lightning training data to obtain a prediction result, and then it is judged whether the prediction result meets the optimization end condition, and it is judged whether it is necessary to update the historical lightning training data for the next time.

[0071] The technical solution of the embodiment of the present invention includes obtaining historical lightning training data; determining current stationary parameters according to the historical lightning training data; inputting the historical lightning training data and the current stationary parameters into an initial model to obtain a prediction result, where the initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in the second time period according to the occurrence time, occurrence location, and lightning occurrence intensity in the first time period, and the first time period precedes the second time period; in response to determining that the prediction result meets the optimization end condition according to the historical lightning training data and the prediction result, determining the stationary parameters as the optimal stationary parameters, and training the initial model according to the optimal stationary parameters and the historical lightning training data to obtain a target lightning prediction model; in response to determining that the prediction result does not meet the optimization end condition according to the historical lightning training data and the prediction result, updating the historical lightning training data, and updating the current stationary parameters and the prediction result. By continuously optimizing the historical lightning training data, at least one stationary parameter is obtained, the various stationary parameters are optimized to obtain the current stationary parameters, and the current optimal stationary parameters are input into the model to train the model, improving the accuracy of the generated lightning prediction model and enabling accurate prediction of lightning occurrence conditions.

[0072] Embodiment 2

[0073] Figure 2The flowchart of a lightning prediction model generation method provided in the second embodiment of the present invention. On the basis of the above embodiment, the present invention embodiment optimizes and improves the lightning prediction model generation operation.

[0074] Further, "determining the current stationary parameter according to the historical lightning training data" is refined to "taking the historical lightning training data for differencing to make the differenced sequence stationary as the goal, performing differencing update on the historical lightning training data, and obtaining the number of differencing times corresponding to the stationary historical lightning training data; calculating the autocorrelation parameter value and the partial autocorrelation parameter value according to the stationary historical lightning training data", so as to improve the operation of generating the lightning prediction model.

[0075] It should be noted that for the parts not detailed in the embodiments of the present invention, reference may be made to the descriptions of other embodiments.

[0076] See Figure 2 The lightning prediction model generation method shown includes:

[0077] S201. Obtain historical lightning training data.

[0078] S202. Taking the historical lightning training data for differencing to make the differenced sequence stationary as the goal, perform differencing update on the historical lightning training data, and obtain the number of differencing times corresponding to the stationary historical lightning training data.

[0079] Among them, the number of differencing times can be used to describe the number of times of differencing processing on the historical lightning training data.

[0080] Specifically, taking the historical lightning training data for differencing to make the differenced sequence stationary as the goal, perform differencing processing on the historical lightning training data to obtain the differenced historical lightning training data, and count the number of times of differencing processing on the historical lightning training data in the case of a stationary differenced sequence, so as to obtain the number of differencing times corresponding to the stationary historical lightning training data. For example, through the Augmented Dickey-Fuller (ADF) algorithm, it can be detected whether the time series data corresponding to the historical lightning training data has a unit root. A unit root is a characteristic of a non-stationary time series, which means that the mean and variance of the time series will change over time, making the long-term behavior of the time series unpredictable.

[0081] Based on a random walk (a special sequence of non-stationarity), perform regression on it. If it is detected that the parameter p = 1, it means that the sequence satisfies a random walk and is non-stationary. If the original sequence is non-stationary, then perform differencing processing to make the non-stationary sequence stationary. First-order differencing: Perform differencing processing on the original sequence Y(x) to obtain the differenced sequence D(x).

[0082] D(x) = Y (x) -Y (x-1)

[0083] If it is still non - stationary after the first - order difference, a second - order difference can be performed:

[0084] D (x) = (Y (x) -Y (x-1) ) - (Y (x-1) -Y (x-2) )

[0085] Until the sequence becomes stationary, obtain the number of differences, which can be denoted by d.

[0086] S203. Calculate the autocorrelation parameter value and the partial autocorrelation parameter value according to the stationary historical lightning training data.

[0087] Among them, the autocorrelation parameter value can be used to describe the value obtained by performing autocorrelation calculation on the stationary historical lightning training data. The partial autocorrelation parameter value can be used to describe the value obtained by performing partial autocorrelation calculation on the stationary historical lightning training data.

[0088] Specifically, according to the time series corresponding to the stationary historical lightning training data, respectively obtain its autocorrelation coefficient ACF and partial autocorrelation coefficient PACF. Through the analysis of the autocorrelation graph and the partial autocorrelation graph, obtain the optimal orders p and q, where the value of p is used as the autocorrelation parameter value and p is used as the partial autocorrelation parameter value.

[0089] The autocorrelation function measures the correlation between the values of a time series at different time lags. For a time series Y(x), the correlation coefficient between Y(x) and Y(x - k) is called the autocorrelation coefficient of Y(x) at lag k. The formula is as follows:

[0090]

[0091] The lag value of ACF represents the correlation between the sequence and its own lag value. The significant ACF cut - off point can be used to select q.

[0092] For a stationary model, when calculating the autocorrelation coefficient ρ(k) at lag k, what is actually obtained is not simply the correlation relationship between Y(x) and Y(x - k), but also affected by the intermediate k - 1 random variables Y(x - 1), Y(x - 2), … Y(x - k + 1). In order to simply measure the influence of Y(x - k) on Y(x), the concept of partial autocorrelation coefficient PACF is introduced. The value of PACF at lag k represents the correlation between the values at time points x and x - k under the condition that the values of the time series at x - 1, x - 2, … x - k + 1 are known. The formula is as follows:

[0093]

[0094] The lag values of the PACF represent the partial correlation of the sequence with its own lagged values. The significant PACF truncation points can be used to select p.

[0095] The autoregressive moving average model is a combination of the autoregressive model and the moving average model, and the formula is as follows:

[0096]

[0097] where μ is the constant term, γ i is the autoregressive coefficient, θ i is the moving average coefficient, and ε (x) is white noise.

[0098] S204. Input the historical lightning training data and the current stationary parameters into the initial model to obtain a prediction result. The initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in the second time period based on the occurrence time, occurrence location, and lightning occurrence intensity in the first time period, and the first time period precedes the second time period.

[0099] S205. In response to determining that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determine the stationary parameter as the optimal stationary parameter, and train the initial model according to the optimal stationary parameter and the historical lightning training data to obtain the target lightning prediction model.

[0100] S206. In response to determining that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, update the historical lightning training data, and update the current stationary parameter and the prediction result.

[0101] In the embodiment of the present invention, with the goal of making the difference sequence of the historical lightning training data stationary by taking the difference of the historical lightning training data, the historical lightning training data is differentially updated, and the number of differential times corresponding to the stationary historical lightning training data is obtained; according to the stationary historical lightning training data, the autocorrelation parameter value and the partial autocorrelation parameter value are calculated. For the prediction of the historical lightning training data, stationary data is more conducive to establishing an accurate prediction model. After the differential processing makes the data stationary, the prediction model established based on the stationary data can better capture the internal law of the data, thereby improving the accuracy and reliability of the prediction.

[0102] Optionally, perform differential update on historical lightning training data, including: obtaining the current change direction and current change speed corresponding to each lightning acquisition data in the historical lightning training data; obtaining the optimal lightning data for the next iteration round corresponding to each lightning acquisition data; for each lightning acquisition data, aiming at minimizing the difference between the lightning acquisition data and the optimal lightning data for the next iteration round, updating the current change direction and current change speed corresponding to the lightning acquisition data, and updating the optimal lightning data for the next iteration round corresponding to the lightning acquisition data to obtain the target lightning data of the lightning acquisition data; updating the historical lightning training data according to the target lightning data of each lightning acquisition data.

[0103] Among them, the current change direction can be used to describe the moving direction of the historical lightning training data relative to the optimal lightning data. The current change speed can be used to describe the moving speed of the historical lightning training data relative to the optimal lightning data. The optimal lightning data can be used to describe the historical optimal position corresponding to the historical lightning training data. The target lightning data can be used to describe the lightning data obtained after updating the current change direction and current change speed of the historical lightning training data.

[0104] Specifically, to obtain the current change direction and current change speed corresponding to each lightning acquisition data in the historical lightning training data, the historical lightning training data can be regarded as multiple particles. For example, in the Particle Swarm Optimization (PSO) algorithm, the flight process of the particle is the search process of the individual. The flight speed of the particle can be dynamically adjusted according to the optimal lightning data (the current change speed of the particle represents the speed of movement, and the current change direction represents the direction of movement), continuously updating the speed and position of the particle to obtain the target lightning data, updating the historical lightning training data, and finding the optimal steady-state parameters that meet the tuning termination conditions according to the updated historical lightning training data. The steps are as follows:

[0105] Initialize the parameters of the particle swarm algorithm: The population size (N) represents the number of different fault modes or initial guess values; the maximum number of iterations (T) represents the maximum number of loops for the algorithm to run; the inertia weight (ω) represents the influence of the current speed of the particle on the next speed; the acceleration coefficients (c1) and (c2) respectively represent the acceleration of the particle approaching the position corresponding to the optimal lightning data; the position boundary corresponds to the physical range of data points such as historical meteorology, lightning monitoring, and cloud layers; the speed boundary corresponds to the rate of parameter change.

[0106] Initialize the population: Randomly generate the positions and speeds of the population particles.

[0107] s i (0) = s min +(s max -smin )·rand()

[0108] v i (0) = v min +(v max - v min )·rand()

[0109] where s i (0), v i (0) are the position and velocity at the initial moment. s min , s max are the minimum and maximum values of the position respectively. v min and v max are the minimum and maximum values of the velocity respectively.

[0110] Update velocity and position: Update the velocity, position, current change direction, and current change speed of each particle according to the following formula.

[0111] v i (t + 1) = ω·v i (t)+c1·r1·(p best_i - s i (t))+c2·r2·(g best - s i (t))

[0112] s i (t + 1) = s i (t)+v i (t + 1)

[0113] where velocity = r1 and r2 are random numbers between [0, 1].

[0114] Introduce Levy flight: To expand the search range of particles, increase the diversity of the particle swarm, improve the search efficiency and particle vitality of particles in the fuzzy state, enable the solution to jump out of the local optimal solution, and better complete the search for the global optimal solution.

[0115] After introducing Levy flight, the update formula for the particle position is as follows:

[0116]

[0117] where s i '(t + 1) is the position of the i-th particle after introducing Levy flight at the t-th update; a is the step size control parameter, and L(u, v) is the random step size path.

[0118] Inertia weight adjustment: To further improve the convergence speed and global search ability of the algorithm, the inertia weight ω can be dynamically adjusted.

[0119]

[0120] Among them, ω max and ω min are the maximum and minimum values of the inertia weight respectively.

[0121] Iteratively update the current change direction and current change speed corresponding to the lightning collection data, and update the optimal lightning data for the next iteration round corresponding to the lightning collection data, to obtain the target lightning data of the lightning collection data. Update the historical lightning training data according to the target lightning data of each lightning collection data.

[0122] By obtaining the current change direction and current change speed corresponding to each lightning collection data in the historical lightning training data; obtaining the optimal lightning data for the next iteration round corresponding to each lightning collection data; aiming at minimizing the difference between each lightning collection data and the optimal lightning data for the corresponding next iteration round, update the current change direction and current change speed corresponding to the lightning collection data, and update the optimal lightning data for the next iteration round corresponding to the lightning collection data, to obtain the target lightning data of the lightning collection data; update the historical lightning training data according to the target lightning data of each lightning collection data. By iteratively updating the current change direction and current change speed multiple times, the influence of noise on the prediction result can be reduced, and the accuracy of lightning prediction can be improved.

[0123] Optionally, aiming at minimizing the difference between each lightning collection data and the optimal lightning data for the corresponding next iteration round, updating the current change direction and current change speed corresponding to the lightning collection data includes: obtaining the initial position and initial speed corresponding to the lightning collection data; calculating the fitness value corresponding to the lightning collection data according to the obtained fitness formula; aiming at minimizing the difference between each lightning collection data and the optimal lightning data for the corresponding next iteration round, determining the optimal position corresponding to the lightning collection data according to the fitness value; determining the current change direction and current change speed corresponding to the lightning collection data according to the fitness value and the optimal position.

[0124] Among them, the fitness value can be used to describe the index for evaluating the quality of lightning collection data.

[0125] Specifically, obtaining the initial position and initial speed corresponding to the lightning collection data, both the initial position and the initial speed are randomly generated; obtaining the fitness formula, which can be defined as the error between the actual meteorological, lightning monitoring, cloud layer and other data points x mea and the theoretical value x the :

[0126] f x =|x mea -x the |

[0127] Calculate the fitness value corresponding to the lightning collection data according to the obtained fitness formula; aiming at minimizing the difference between the lightning collection data and the optimal lightning data in the next iteration round, a difference degree value can be preset, compare the fitness value with the difference degree value, and calculate the optimal position corresponding to the lightning collection data according to the comparison result; determine the current change direction and current change speed corresponding to the lightning collection data according to the fitness value and the optimal position.

[0128] Obtain the initial position and initial speed corresponding to the lightning collection data; calculate the fitness value corresponding to the lightning collection data according to the obtained fitness formula; aiming at minimizing the difference between the lightning collection data and the optimal lightning data in the next iteration round, determine the optimal position corresponding to the lightning collection data according to the fitness value; determine the current change direction and current change speed corresponding to the lightning collection data according to the fitness value and the optimal position, and the current change direction and current change speed can be updated according to preset conditions to avoid infinite updates, saving data processing time and improving data processing efficiency.

[0129] Optionally, obtain historical lightning training data, including: obtain lightning data within a preset time period, and the lightning data includes: historical lightning monitoring data, historical meteorological data and satellite cloud images, and the historical lightning monitoring data includes: lightning occurrence time, lightning location and lightning intensity, and the historical meteorological data includes: historical temperature, historical humidity and electric field intensity; perform smoothing processing on the lightning data within the preset time period by the single exponential smoothing method to obtain smoothed lightning data; perform feature extraction on the smoothed lightning data to obtain historical lightning training data within the preset time period.

[0130] Specifically, with the development of meteorological observation technology and lightning monitoring technology, lightning warning systems have been gradually introduced. A typical lightning warning system relies on changes in meteorological conditions. For example: 1) Electric field intensity change: Before lightning occurs, the atmospheric electric field usually changes sharply. Monitoring the electric field intensity can help predict the approach of thunderstorm weather. 2) Temperature and humidity: The changes in temperature and humidity are closely related to thunderstorm weather. Under high temperature and high humidity conditions, the probability of lightning strikes usually increases. 3) Atmospheric pressure and wind speed: The fluctuations of these meteorological parameters also affect the formation of thunderstorms and are commonly used in meteorological monitoring of lightning activities. Obtain lightning data within a preset time period. The lightning data includes: historical lightning monitoring data, historical meteorological data, and satellite cloud images. The historical lightning monitoring data includes: lightning occurrence time, lightning location, and lightning intensity. The historical meteorological data includes: historical temperature, historical humidity, and electric field intensity. Smooth the lightning data within the preset time period through the first-order exponential smoothing method. By calculating the weighted average of the actual value of the current period and the exponential smoothing value of the previous period for data points such as meteorology, lightning monitoring, and clouds within a specified window, smooth lightning data is obtained. Extract features from the smooth lightning data, extract key features affecting lightning occurrence from historical data, such as sudden changes in electric field intensity and sharp changes in temperature, etc., to obtain historical lightning training data within the preset time period.

[0131] By obtaining lightning data within a preset time period, the lightning data includes: historical lightning monitoring data, historical meteorological data, and satellite cloud images. The historical lightning monitoring data includes: lightning occurrence time, lightning location, and lightning intensity. The historical meteorological data includes: historical temperature, historical humidity, and electric field intensity. Smooth the lightning data within the preset time period through the first-order exponential smoothing method to obtain smooth lightning data. Extract features from the smooth lightning data to obtain historical lightning training data within the preset time period. By screening multi-dimensional lightning-related data, the accuracy of lightning prediction can be improved.

[0132] Embodiment III

[0133] Figure 3 This is a flowchart of a lightning prediction method provided in Embodiment I of the present invention. The embodiments of the present invention are applicable to the situation of lightning prediction. This method can be executed by a lightning prediction device, and the lightning prediction device can be implemented in the form of hardware and / or software.

[0134] See Figure 3 The lightning prediction method shown includes:

[0135] S101. Obtain real-time lightning monitoring data within a preset acquisition time period.

[0136] Among them, the real-time lightning monitoring data can be used to describe the data that may affect lightning occurrence during real-time monitoring.

[0137] Specifically, real-time lightning monitoring data within a preset collection time period is obtained. The obtaining methods include, but are not limited to: using a professional lightning monitoring system, leveraging meteorological department resources, or using open-source components or related platforms, etc. The embodiments of the present invention do not limit this.

[0138] S102. Input historical lightning training data within a historical time period and real-time lightning monitoring data within a preset collection time period into a target lightning prediction model to obtain predicted lightning data. The predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability.

[0139] Among them, the target lightning prediction model can be used to describe a model that has been trained with historical lightning training data.

[0140] Specifically, input historical lightning training data within a historical time period and real-time lightning monitoring data within a preset collection time period into a target lightning prediction model to obtain predicted lightning data. The predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability. Historical lightning training data can reflect the occurrence rules, spatio-temporal distribution characteristics, etc. of lightning on a relatively long time scale, while real-time lightning monitoring data can present the latest information such as the current atmospheric conditions. The combination of the two enables the model to capture more comprehensive lightning-related features, avoiding information loss caused by relying on a single data source, thereby improving the accuracy of predicting the time, location, and probability of lightning occurrence.

[0141] S103. Obtain at least one lightning probability threshold and the lightning protection measures corresponding to each lightning probability threshold.

[0142] Among them, the lightning probability threshold can be used to describe a preset threshold of lightning occurrence probability.

[0143] Specifically, obtain at least one lightning probability threshold and the lightning protection measures corresponding to each lightning probability threshold. The obtaining methods include, but are not limited to: mouse selection or keyboard input, etc. The embodiments of the present invention do not limit this. For example, the lightning protection measures can be information warning for maintenance personnel, audible and visual warning, starting lightning protection equipment, or turning off sensitive devices, etc.

[0144] S104. Compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine the target probability and the target lightning protection measure, and send the lightning protection measure to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measure.

[0145] Among them, the target probability can be used to describe the largest lightning probability threshold that the predicted lightning probability is greater than. The target lightning protection measure can be used to describe the lightning protection measure corresponding to the target probability.

[0146] Specifically, compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine the target probability and the target lightning protection measures, and send the lightning protection measures to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automatic lightning protection operations according to the lightning protection measures. Compare the lightning occurrence probability output by the model with the preset threshold. When the probability exceeds the threshold, the system automatically triggers an early warning mechanism, sends an early warning message to relevant personnel, and activates corresponding protection measures. The system can automatically activate lightning protection equipment, cut off the power supply of key equipment, and notify the staff to take protection measures to reduce the damage caused by lightning. For example, yellow warning: when the lightning occurrence probability is relatively high but does not reach the critical value (such as 60% to 80%), the system issues a yellow warning to remind the operator to check and prepare to activate the lightning protection measures. Red warning: when the lightning occurrence probability exceeds 80%, the system issues a red warning, indicating that immediate emergency protection measures should be taken, such as shutting down sensitive equipment or evacuating important areas.

[0147] The technical solution of the embodiment of the present invention obtains real-time lightning monitoring data within a preset acquisition time period; inputs historical lightning training data within a historical time period and real-time lightning monitoring data within the preset acquisition time period into the target lightning prediction model to obtain predicted lightning data, where the predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability; obtains at least one lightning probability threshold and the corresponding lightning protection measures for each lightning probability threshold; compares the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine the target probability and the target lightning protection measures, and sends the lightning protection measures to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automatic lightning protection operations according to the lightning protection measures, can predict the lightning situation in advance, provide sufficient response time for maintenance personnel, and automatically activate protection measures to ensure equipment safety.

[0148] Embodiment 4

[0149] Figure 4 FIG. 10 is a structural schematic diagram of a lightning prediction model generation device provided in Embodiment 4 of the present invention. The embodiment of the present invention is applicable to the situation of generating a lightning prediction model. The device can execute the lightning prediction model generation method, and the device can be implemented in the form of hardware and / or software.

[0150] See Figure 4 The lightning prediction model generation device shown in FIG. 10 includes: a data acquisition module 401, a parameter determination module 402, a result acquisition module 403, a model training module 404, and a data update module 405, where

[0151] The data acquisition module 401 is used to acquire historical lightning training data;

[0152] The parameter determination module 402 is used to determine the current stationary parameter according to the historical lightning training data;

[0153] A result acquisition module 403 is configured to input historical lightning training data and current stationary parameters into an initial model to obtain a prediction result. The initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in a second time period based on the occurrence time, occurrence location, and lightning occurrence intensity in a first time period, where the first time period precedes the second time period.

[0154] A model training module 404 is configured to, when it is determined that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determine the stationary parameter as the optimal stationary parameter, and train the initial model according to the optimal stationary parameter and the historical lightning training data to obtain a target lightning prediction model.

[0155] A data update module 405 is configured to, when it is determined that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, update the historical lightning training data, and update the current stationary parameter and the prediction result.

[0156] The technical solution of the embodiment of the present invention obtains historical lightning training data; determines the current stationary parameter according to the historical lightning training data; inputs the historical lightning training data and the current stationary parameter into an initial model to obtain a prediction result. The initial model is used to predict the occurrence time, occurrence location, and lightning occurrence probability in a second time period based on the occurrence time, occurrence location, and lightning occurrence intensity in a first time period, where the first time period precedes the second time period; when it is determined that the prediction result meets the tuning end condition according to the historical lightning training data and the prediction result, determine the stationary parameter as the optimal stationary parameter, and train the initial model according to the optimal stationary parameter and the historical lightning training data to obtain a target lightning prediction model; when it is determined that the prediction result does not meet the tuning end condition according to the historical lightning training data and the prediction result, update the historical lightning training data, and update the current stationary parameter and the prediction result. By continuously optimizing the historical lightning training data, at least one stationary parameter is obtained, the stationary parameters are tuned to obtain the current stationary parameter, and the current optimal stationary parameter is input into the model for training, which improves the accuracy of the generated lightning prediction model and can accurately predict the lightning occurrence situation.

[0157] Optionally, the parameter determination module 402 includes:

[0158] An occurrence times acquisition unit is configured to take the historical lightning training data for differencing with the goal of making the differenced sequence stationary, perform differencing update on the historical lightning training data, and obtain the differencing times corresponding to the stationary historical lightning training data.

[0159] A parameter value calculation unit is configured to calculate the autocorrelation parameter value and the partial autocorrelation parameter value according to the stationary historical lightning training data.

[0160] Optionally, the number acquisition unit includes:

[0161] A current data acquisition subunit, configured to acquire the current change direction and the current change speed corresponding to each lightning collection data in the historical lightning training data;

[0162] An iterative data acquisition subunit, configured to acquire the optimal lightning data for the next iteration round corresponding to each lightning collection data;

[0163] A target data acquisition subunit, configured to, for each lightning collection data, with the goal of minimizing the difference between the lightning collection data and the optimal lightning data for the next iteration round, update the current change direction and the current change speed corresponding to the lightning collection data, and update the optimal lightning data for the next iteration round corresponding to the lightning collection data, to obtain the target lightning data of the lightning collection data;

[0164] A training data update subunit, configured to update the historical lightning training data according to the target lightning data of each lightning collection data.

[0165] Optionally, the current data acquisition subunit is specifically configured to:

[0166] Acquire the initial position and the initial speed corresponding to the lightning collection data;

[0167] Calculate the fitness value corresponding to the lightning collection data according to the obtained fitness formula;

[0168] With the goal of minimizing the difference between the lightning collection data and the optimal lightning data for the next iteration round, determine the optimal position corresponding to the lightning collection data according to the fitness value;

[0169] Determine the current change direction and the current change speed corresponding to the lightning collection data according to the fitness value and the optimal position.

[0170] Optionally, the data acquisition module 401 is specifically configured to:

[0171] Acquire lightning data within a preset time period, where the lightning data includes: historical lightning monitoring data, historical meteorological data, and satellite cloud images, the historical lightning monitoring data includes: lightning occurrence time, lightning position, and lightning intensity, and the historical meteorological data includes: historical temperature, historical humidity, and electric field intensity;

[0172] Perform smoothing processing on the lightning data within the preset time period by using the single exponential smoothing method to obtain smoothed lightning data;

[0173] Perform feature extraction on the smoothed lightning data to obtain historical lightning training data within the preset time period.

[0174] The lightning prediction model generation device provided by the embodiments of the present invention can execute the lightning prediction model generation method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the lightning prediction model generation method.

[0175] Embodiment 5

[0176] Figure 5 It is a schematic structural diagram of a lightning prediction device provided by Embodiment 5 of the present invention. The embodiments of the present invention are applicable to the situation of lightning prediction. This device can execute the lightning prediction method, and this device can be implemented in the form of hardware and / or software.

[0177] Refer to Figure 5 the lightning prediction device shown in the figure, including: a real-time data acquisition module 501, a prediction data acquisition module 502, a measure acquisition module 503, and an information sending module 504, where

[0178] The real-time data acquisition module 501 is used to acquire real-time lightning monitoring data within a preset acquisition time period;

[0179] The prediction data acquisition module 502 is used to input historical lightning training data within a historical time period and real-time lightning monitoring data within a preset acquisition time period into a target lightning prediction model to obtain predicted lightning data, where the predicted lightning data includes: prediction time, prediction location, and predicted lightning occurrence probability;

[0180] The measure acquisition module 503 is used to acquire at least one lightning probability threshold and lightning protection measures corresponding to each lightning probability threshold;

[0181] The information sending module 504 is used to compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine a target probability and a target lightning protection measure, and send the lightning protection measure to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measure.

[0182] The technical solution of the embodiment of the present invention is to acquire real-time lightning monitoring data within a preset acquisition time period; input historical lightning training data within a historical time period and real-time lightning monitoring data within a preset acquisition time period into a target lightning prediction model to obtain predicted lightning data, acquire at least one lightning probability threshold and lightning protection measures corresponding to each lightning probability threshold; compare the predicted lightning probability in the predicted lightning data with each lightning probability threshold to determine a target probability and a target lightning protection measure, and send the lightning protection measure to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measure to ensure the safety of the equipment.

[0183] Embodiment 6

[0184] Figure 6The structural schematic diagram of a lightning prediction model generation device 600 that can be used to implement the embodiments of the present invention is shown.

[0185] As Figure 6 shown, the lightning prediction model generation device 600 includes at least one processor 601 and a memory communicatively connected to the at least one processor 601, such as a read-only memory (ROM) 602, a random access memory (RAM) 603, etc. Among them, the memory stores a computer program executable by the at least one processor. The processor 601 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 602 or the computer program loaded from the storage unit 608 into the random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the lightning prediction model generation device 600 can also be stored. The processor 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. The input / output (I / O) interface 605 is also connected to the bus 604.

[0186] Multiple components in the lightning prediction model generation device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, an optical disc, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the lightning prediction model generation device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0187] The processor 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 601 executes the various methods and processes described above, such as the lightning prediction model generation method.

[0188] In some embodiments, the lightning prediction model generation method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed onto the lightning prediction model generation device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the processor 601, one or more steps of the lightning prediction model generation method described above may be performed. Alternatively, in other embodiments, the processor 601 may be configured to execute the lightning prediction model generation method by any other suitable means (e.g., by means of firmware).

[0189] Various embodiments of the systems and techniques described above in this document may be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0190] The computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The computer programs may be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on the remote machine or server.

[0191] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0192] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a lightning prediction model generation device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the lightning prediction model generation device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0193] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by any form or medium of digital data communication (e.g., a communication network). Examples of the communication network include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0194] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The relationship between the client and the server is created by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS (Virtual Private Server) services.

[0195] It should be understood that various forms of the processes shown above can be used, steps can be reordered, added or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0196] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A lightning prediction model generation method, characterized in that: The method comprises: Obtain historical lightning training data; Determining current stable parameters according to the historical lightning training data; Inputting the historical lightning training data and the current stable parameters into an initial model to obtain a prediction result, wherein the initial model is used to predict the occurrence time, occurrence location and probability of lightning occurrence in a second time period according to the occurrence time, occurrence location and lightning intensity in a first time period, wherein the first time period precedes the second time period; In response to determining, based on the historical lightning training data and the prediction result, that the prediction result meets the tuning end condition, determining the stationary parameter as the optimal stationary parameter, and training the initial model based on the optimal stationary parameter and the historical lightning training data to obtain a target lightning prediction model; In response to determining, based on the historical lightning training data and the prediction result, that the prediction result does not meet the tuning end condition, the historical lightning training data is updated, and the current stable parameter and the prediction result are updated.

2. The method according to claim 1, characterized in that: The determining of the current stable parameter according to the historical lightning training data comprises: The historical lightning training data is differentiated to make the differential sequence stable, the historical lightning training data is differentially updated, and the number of differentials corresponding to the stable historical lightning training data is obtained; According to the stable historical lightning training data, autocorrelation parameter values ​​and partial autocorrelation parameter values ​​are calculated.

3. The method according to claim 2, characterized in that The differential updating of the historical lightning training data includes: Obtaining the current change direction and current change speed corresponding to each lightning collection data in the historical lightning training data; Obtaining optimal lightning data for the next iteration round corresponding to each of the lightning collection data; For each of the lightning collection data, with the goal of minimizing the difference between the lightning collection data and the corresponding optimal lightning data of the next iteration round, updating the current change direction and current change speed corresponding to the lightning collection data, and updating the optimal lightning data of the next iteration round corresponding to the lightning collection data, to obtain target lightning data of the lightning collection data; The historical lightning training data is updated according to the target lightning data of each lightning collection data.

4. The method according to claim 3, characterized in that The updating of the current change direction and the current change speed corresponding to the lightning collection data with the goal of minimizing the difference between the lightning collection data and the corresponding optimal lightning data of the next iteration round includes: Obtaining an initial position and an initial speed corresponding to the lightning collection data; The fitness value corresponding to the lightning collection data is calculated according to the obtained fitness formula; With the goal of minimizing the difference between the lightning collection data and the corresponding optimal lightning data of the next iteration round, determining the optimal position corresponding to the lightning collection data according to the fitness value; According to the fitness value and the optimal position, a current change direction and a current change speed corresponding to the lightning collection data are determined.

5. The method according to claim 1, characterized in that The obtaining of historical lightning training data includes: Acquire lightning data within a preset time period, the lightning data including: historical lightning monitoring data, historical meteorological data and satellite cloud images, the historical lightning monitoring data including: lightning occurrence time, lightning location and lightning intensity, the historical meteorological data including: historical temperature, historical humidity and electric field intensity; Smoothing the lightning data within the preset time period by a primary exponential smoothing method to obtain smoothed lightning data; Feature extraction is performed on the smoothed lightning data to obtain historical lightning training data within a preset time period.

6. A lightning prediction method, characterized in that: include: Obtain real-time lightning monitoring data within a preset collection time period; Inputting the historical lightning training data within the historical time period and the real-time lightning monitoring data within the preset collection time period into the target lightning prediction model to obtain predicted lightning data, wherein the predicted lightning data includes: predicted time, predicted location and predicted probability of lightning occurrence; Obtaining at least one lightning probability threshold and lightning protection measures corresponding to each lightning probability threshold; The predicted lightning probability in the predicted lightning data is compared with each of the lightning probability thresholds to determine the target probability and target lightning protection measures, and the lightning protection measures are sent to the corresponding lightning protection equipment and maintenance personnel so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measures.

7. A lightning prediction model generation device, characterized in that: The device comprises: A data acquisition module, used to acquire historical lightning training data; A parameter determination module, used to determine the current stable parameters according to the historical lightning training data; a result acquisition module, for inputting the historical lightning training data and the current stable parameters into an initial model to obtain a prediction result, wherein the initial model is used to predict the occurrence time, occurrence location and lightning occurrence probability of a second time period according to the occurrence time, occurrence location and lightning intensity of a first time period, wherein the first time period precedes the second time period; A model training module, configured to, in response to determining, based on the historical lightning training data and the prediction result, that the prediction result satisfies the tuning end condition, determine the stationary parameter as the optimal stationary parameter, and train the initial model based on the optimal stationary parameter and the historical lightning training data to obtain a target lightning prediction model; A data updating module is used to update the historical lightning training data and the prediction result in response to determining that the prediction result does not meet the tuning end condition based on the historical lightning training data and the prediction result, and to update the current stable parameters and the prediction result.

8. A lightning prediction device, characterized in that: The device comprises: A real-time data acquisition module is used to obtain real-time lightning monitoring data within a preset collection time period; A prediction data acquisition module is used to input the historical lightning training data within the historical time period and the real-time lightning monitoring data within the preset collection time period into the target lightning prediction model to obtain predicted lightning data, wherein the predicted lightning data includes: predicted time, predicted location and predicted probability of lightning occurrence; A measure acquisition module, used to acquire at least one lightning probability threshold and the lightning protection measures corresponding to each lightning probability threshold; An information sending module is used to compare the predicted lightning probability in the predicted lightning data with each of the lightning probability thresholds, determine the target probability and target lightning protection measures, and send the lightning protection measures to the corresponding lightning protection equipment and maintenance personnel, so that the lightning protection equipment performs automated lightning protection operations according to the lightning protection measures.

9. A lightning prediction model generation device, characterized in that: The lightning prediction model generating device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the lightning prediction model generation method according to any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the lightning prediction model generation method according to any one of claims 1 to 5 when executed.

Images