Method and device for determining wind speed random distribution parameters and storage medium

Abstract

The invention discloses a method and a device for determining a wind speed random distribution parameter, a storage medium, an electronic device and a computer program product. The method comprises the following steps: acquiring historical environment monitoring data of an observation point; and based on the historical environment monitoring data, estimating a first wind speed random distribution parameter of the observation point by adopting a pre-established Bayesian network model. Adopting the first wind speed random distribution parameter and historical environment monitoring data as training samples, training a long-short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model; and determining a second wind speed random distribution parameter of the observation point by adopting a wind speed random distribution parameter prediction model based on the current environment monitoring data of the observation point. The Bayesian network model and the long-short-term memory neural network model are combined, so that the wind speed random distribution parameters can be quickly and accurately predicted.

Description

Technical Field

[0001] The present application relates to the field of wind speed prediction, and more specifically, to a method and device for determining random distribution parameters of wind speed, a storage medium, an electronic device, and a computer program product. Background Art

[0002] With the rapid development of the new energy industry, wind power generation, as a renewable and clean energy source, is encouraged by national policies and environmental protection requirements. Wind is mainly generated by the thermal effect of air, and is affected by meteorological factors (such as atmospheric temperature, humidity, and air pressure) and geographical factors (such as terrain and ground obstacles), and has significant uncertainty. In order to improve the reliability and stability of wind power generation, it is necessary to predict wind speed.

[0003] In related technologies, physical models, statistical models, machine learning models, hybrid forecasting models, etc. are used to predict wind speed.

[0004] However, the prediction method in the related art cannot take into account both the accuracy and speed of prediction. Summary of the invention

[0005] Embodiments of the present application provide a method and device for determining wind speed random distribution parameters, a storage medium, an electronic device, and a computer program product.

[0006] According to one aspect of an embodiment of the present application, a method for determining a random distribution parameter of wind speed is provided, the method comprising: acquiring historical environmental monitoring data of an observation point; based on the historical environmental monitoring data, estimating a first random distribution parameter of wind speed at the observation point using a pre-established Bayesian network model, wherein the Bayesian network model is modeled based on the random distribution parameter of wind speed, and the random distribution parameter of wind speed is used to indicate characteristics of average wind speed and turbulent wind speed; using the first random distribution parameter of wind speed and the historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a prediction model for the random distribution parameter of wind speed; based on the current environmental monitoring data of the observation point, using the random distribution parameter of wind speed prediction model to determine a second random distribution parameter of wind speed at the observation point.

[0007] In an exemplary embodiment, before estimating the first wind speed random distribution parameter of the observation point based on the historical environmental monitoring data using a pre-established Bayesian network model, a method for determining the wind speed random distribution parameter includes: decomposing the wind speed into average wind speed and turbulent wind speed; using a first mathematical model to describe the average wind speed, wherein the first mathematical model includes a first parameter; using a second mathematical model to describe the turbulent wind speed, wherein the second mathematical model includes a second parameter, and the wind speed random distribution parameter includes the first parameter and the second parameter; constructing corresponding Bayesian network nodes based on the first parameter, the second parameter, and environmental influencing factors, wherein the historical environmental monitoring data includes data of multiple environmental influencing factors; adjusting the dependency between the nodes and determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged; and determining the current Bayesian network model as the pre-established Bayesian network model when the scoring function of the Bayesian network model remains unchanged.

[0008] In an exemplary embodiment, the step of adjusting the dependency relationship between nodes and determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged includes: converting the directed acyclic graph of the Bayesian network model into a moral graph, wherein the moral graph includes the Bayesian network nodes, and the Bayesian network nodes with dependency relationships are connected in the moral graph, and the moral graph is used to determine the joint probability distribution of the Bayesian network model; determining the scoring function value of the Bayesian network model based on the number of nodes and the likelihood of the nodes in the moral graph, wherein the number of nodes is used to indicate the complexity of the Bayesian network model, and the likelihood of the nodes is used to indicate the fit of the Bayesian network model; adjusting the dependency relationship between at least two nodes in the moral graph, and re-determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged.

[0009] In an exemplary embodiment, the step of adjusting the dependency relationship between at least two nodes in the moral graph and re-determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged includes: using multiple adjustment methods to adjust the dependency relationship between at least two nodes in the initial moral graph respectively to obtain multiple adjusted first moral graphs; using the first moral graph with the lowest scoring function value among the multiple adjusted first moral graphs as the second moral graph; using the second moral graph as the new initial moral graph, and returning to execute the step of using multiple adjustment methods to adjust the dependency relationship between at least two nodes in the initial moral graph respectively to obtain multiple adjusted first moral graphs until the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal.

[0010] In an exemplary embodiment, the method of using the first wind speed random distribution parameters and the historical environmental monitoring data as training samples to train a long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model includes: using the first wind speed random distribution parameters and the historical environmental monitoring data as training samples to train a long short-term memory neural network model; adjusting the parameters of the long short-term memory neural network model until the long short-term memory neural network model meets the optimization goal, wherein the parameters of the long short-term memory neural network model include a weight matrix of a forget gate, a weight matrix of an input gate, a weight matrix of a unit state, a weight matrix of an output gate, a weight matrix of an output layer, a bias vector of a forget gate, a bias vector of an input gate, a bias vector of a unit state, a bias vector of an output gate, and a bias vector of an output layer, and the optimization goal is that the cost function is less than a preset value; and using the long short-term memory neural network model that meets the optimization goal as the wind speed random distribution parameter prediction model.

[0011] In an exemplary embodiment, based on the historical environmental monitoring data, a pre-established Bayesian network model is used to estimate the first wind speed random distribution parameter of the observation point, including: using the historical environmental monitoring data as the observation data in the Bayesian network model; determining the posterior probability of the wind speed random distribution parameter of the observation point based on the dependency relationship between the nodes in the Bayesian network model and the observation data; determining the first wind speed random distribution parameter according to the posterior probability of the wind speed random distribution parameter of the observation point.

[0012] According to another aspect of an embodiment of the present application, a device for determining a wind speed random distribution parameter is also provided, the device comprising:

[0013] Data acquisition module, used to obtain historical environmental monitoring data of observation points;

[0014] A parameter estimation module, for estimating a first wind speed random distribution parameter of the observation point based on the historical environmental monitoring data using a pre-established Bayesian network model, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and the turbulent wind speed;

[0015] A model training module, used to use the first wind speed random distribution parameter and the historical environmental monitoring data as training samples to train a long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model;

[0016] The parameter determination module is used to determine the second wind speed random distribution parameter of the observation point based on the current environmental monitoring data of the observation point and using the wind speed random distribution parameter prediction model.

[0017] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the random distribution parameters of wind speed when running.

[0018] According to another aspect of an embodiment of the present application, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method for determining the random distribution parameters of wind speed through the computer program.

[0019] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the method described in each embodiment of the present application are implemented.

[0020] The above-mentioned method for determining the random distribution parameters of wind speed first obtains the historical environmental monitoring data of the observation point; then, based on the historical environmental monitoring data, a pre-established Bayesian network model is used to estimate the first random distribution parameter of wind speed at the observation point. Next, the first random distribution parameter of wind speed and the historical environmental monitoring data are used as training samples to train the long short-term memory neural network model to obtain a prediction model for the random distribution parameters of wind speed. Finally, based on the current environmental monitoring data of the observation point, the prediction model for the random distribution parameters of wind speed is used to determine the second random distribution parameter of wind speed at the observation point. Since the Bayesian network model can accurately determine the probability distribution of the random distribution parameters of wind speed and improve the prediction accuracy of the random distribution parameters of wind speed, and the long short-term memory neural network model can quickly and timely predict the random distribution parameters of wind speed according to the currently input environmental monitoring data, the combination of the Bayesian network model and the long short-term memory neural network model can quickly and accurately predict the random distribution parameters of wind speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0023] Figure 1 It is a hardware structure block diagram of the method for determining the wind speed random distribution parameters of the embodiment of the present application;

[0024] Figure 2 is a flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0025] Figure 3 is a second flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0026] Figure 4 is a structural diagram of a Bayesian network model according to an embodiment of the present application;

[0027] Figure 5 is a third flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0028] Figure 6 is a fourth flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0029] Figure 7 is a fifth flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0030] Figure 8 is a sixth flow chart of a method for determining wind speed random distribution parameters according to an embodiment of the present application;

[0031] Fig. 9 It is a structural block diagram of a device for determining wind speed random distribution parameters according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0034] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or a similar computing device. Taking running on a computer terminal as an example, Figure 1 1 is a hardware structure block diagram of a computer terminal for determining the random distribution parameters of wind speed according to an embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a microprocessor (Microprocessor Unit, referred to as MPU) or a programmable logic device (Programmable logic device, referred to as PLD)) and a memory 104 for storing data. In an exemplary embodiment, the computer terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0035] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the random distribution parameters of wind speed in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0036] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0037] In this embodiment, a method for determining wind speed random distribution parameters is provided. Figure 2 4 is a flow chart of an optional method for determining wind speed random distribution parameters according to an embodiment of the present application, the process comprising the following steps S200-S230:

[0038] Step S200, obtaining historical environmental monitoring data of the observation point.

[0039] Among them, wind speed is mainly generated by air thermal effect, which is mainly affected by meteorological factors (such as atmospheric temperature, humidity, air pressure) and geographical factors (such as terrain, ground obstacles). Environmental monitoring data can include monitoring data of the most important meteorological influencing factors, such as temperature and pressure.

[0040] Step S210, based on historical environmental monitoring data, a pre-established Bayesian network model is used to estimate a first wind speed random distribution parameter of an observation point.

[0041] Among them, the random distribution parameters of wind speed are of great significance for wind control and stochastic modeling. Using the random distribution parameters of wind speed, the dynamic characteristics of wind speed can be predicted, so that the state of the wind power generation system can be quickly adjusted to adapt to the dynamic changes of wind speed and increase power generation efficiency.

[0042] The Bayesian network model is based on the random distribution parameters of wind speed, which are used to indicate the characteristics of average wind speed and turbulent wind speed. The Bayesian network (BN) uses a directed acyclic graph to describe the dependency between attributes, in which the conditional probability table is used to describe the joint probability distribution of attributes, which can well express the complex dependencies between random variables. According to the joint probability distribution defined by BN, the posterior probability of the random distribution parameters of wind speed can be accurately calculated to achieve accurate prediction.

[0043] Step S220, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model.

[0044] Among them, the Long Short-Term Memory (LSTM) neural network is a typical recursive neural network suitable for prediction based on time series data. The LSTM neural network can receive information from the previous state to the current state by maintaining a storage unit in the hidden layer. This architecture can maintain all currently available input information.

[0045] Step S230, based on the current environmental monitoring data of the observation point, a wind speed random distribution parameter prediction model is used to determine a second wind speed random distribution parameter of the observation point.

[0046] In this embodiment, the historical environmental monitoring data of the observation point is first obtained; then, based on the historical environmental monitoring data, a pre-established Bayesian network model is used to estimate the first wind speed random distribution parameter of the observation point. Next, the first wind speed random distribution parameter and the historical environmental monitoring data are used as training samples to train the long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model. Finally, based on the current environmental monitoring data of the observation point, the wind speed random distribution parameter prediction model is used to determine the second wind speed random distribution parameter of the observation point. Since the Bayesian network model can accurately determine the probability distribution of the wind speed random distribution parameter and improve the prediction accuracy of the wind speed random distribution parameter, and the long short-term memory neural network model can quickly and timely predict the wind speed random distribution parameter according to the currently input environmental monitoring data, the combination of the Bayesian network model and the long short-term memory neural network model can quickly and accurately predict the wind speed random distribution parameter.

[0047] In one embodiment, Figure 3 As shown, in step S210, based on historical environmental monitoring data, a pre-established Bayesian network model is used to estimate the first wind speed random distribution parameter of the observation point. Previously, the method for determining the wind speed random distribution parameter includes: steps S300-S350. Among them:

[0048] Step S300, decomposing the wind speed into average wind speed and turbulent wind speed.

[0049] Among them, through the statistical analysis of a large number of measured wind speeds, a random model of the actual wind speed can be obtained, which can be decomposed into two components, including the average wind speed (a strong potential seasonal component driven by the change in insolation caused by the tilt of the Earth's rotation axis) and the turbulent wind speed (a daily component driven by the change in insolation caused by the Earth's rotation). The average wind speed generally conforms to the Weibull distribution with a scale parameter and a shape parameter, and the turbulent wind speed generally conforms to the Gaussian distribution.

[0050] The wind speed is decomposed into average wind speed and turbulent wind speed by the following formula:

[0051] v=v m +v t

[0052] Where, v is the wind speed, v m is the average wind speed, v t is the turbulent wind speed. The average wind speed is the long-term trend component of the wind speed, reflecting the macroscopic characteristics of wind speed changes over time. The average wind speed is usually driven by the change in insolation caused by the tilt of the earth's axis of rotation, and has a significant seasonal component. The turbulent wind speed is the short-term fluctuation component of the wind speed, representing the randomness and local changes in the wind speed. Turbulent wind speed is usually affected by a variety of meteorological factors (such as atmospheric pressure, temperature, humidity) and terrain factors. The average wind speed reflects the long-term trend and regularity of the wind speed, which is of great significance for evaluating the power generation potential of wind farms and planning wind power projects. The turbulent wind speed reflects the short-term fluctuations and randomness of the wind speed, which has a direct impact on the control and operation of wind turbines. By establishing a random distribution model of wind speed, the change of wind speed can be predicted more accurately, thereby improving the efficiency and reliability of wind power generation systems.

[0053] Step S310: using a first mathematical model to describe the average wind speed.

[0054] The first mathematical model includes a first parameter. In the statistical model, the average wind speed generally conforms to a Weibull distribution with a scale parameter and a shape parameter. Therefore, the Weibull distribution is used to describe the average wind speed.

[0055] Exemplarily, the model for the average wind speed is as follows:

[0056]

[0057] Among them, v mis the average wind speed, s1 is the scale parameter, also known as the proportional parameter, which determines the scaling of the wind speed distribution, that is, the "width" of the distribution, and s2 is the shape parameter, which controls the shape of the wind speed distribution. A larger s2 value will make the tail of the distribution thicker, indicating the possibility of more extreme wind speeds. The s1 in the denominator adjusts the scale so that the distribution can adapt to different wind speed ranges. The s2 term in the numerator and Adjust the shape, affecting how steep or flat the distribution is. Exponential part It ensures that the total area of ​​the probability density function is 1, which meets the requirements of probability distribution. This form of Weibull distribution is particularly suitable for simulating wind speed because it can capture the extreme values ​​of wind speed well, and the parameters can be adjusted to adapt to different wind speed characteristics. The first parameter includes the scale parameter s1 and the shape parameter s2.

[0058] Step S320: using a second mathematical model to describe the turbulent wind speed.

[0059] The second mathematical model includes a second parameter, and the wind speed random distribution parameter includes a first parameter and a second parameter. In the statistical model, the turbulent wind speed generally conforms to a Gaussian distribution.

[0060] Exemplarily, the model for turbulent wind speed is as follows:

[0061]

[0062] Among them, v t is the turbulent wind speed, 0 is the mean (expected value), indicating that v t The average value of is 0. In many practical applications, especially when dealing with wind speed data, the long-term average of the turbulence component is usually close to zero. is the variance, indicating v t The square root of the variance is the standard deviation, which describes the intensity of the turbulent part of the wind speed. The normal distribution is a common continuous probability distribution in natural and social sciences, usually used to represent measurement errors, noise, etc. In wind speed modeling, the turbulent component is usually assumed to be an independent and identically distributed Gaussian random variable, which reflects the random fluctuation characteristics of wind speed on a short time scale.

[0063] Step S330: construct corresponding Bayesian network nodes based on the first parameter, the second parameter, and the environmental impact factor.

[0064] The historical environmental monitoring data includes data on multiple environmental impact factors. Environmental impact factors include time T, wind tower ambient temperature T e , atmospheric pressure P a , light H u , humidity σ, etc. The first parameter includes the scale parameter s1 and the shape parameter s2, and the second parameter includes s3, such as Figure 4Shown is a diagram of the Bayesian network model.

[0065] Step S340, adjusting the dependency relationship between the nodes and determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged.

[0066] Among them, one edge may be adjusted each time (such as adding, deleting or adjusting the direction), and then the scoring function value of the Bayesian network model is determined.

[0067] Exemplarily, the scoring function is selected as follows:

[0068] BIC(B|D)=f(λ)B|-LL(B|D)

[0069] Among them, BIC(B|D) is the Bayesian information criterion for a given model B and data D. BIC is often used for model selection, especially when comparing statistical models with different numbers of parameters. The BIC criterion combines the complexity of the model and the degree of fit to the data, providing a reference standard for model selection. The BIC criterion is to subtract the log-likelihood of the model from the complexity of the model. The smaller the BIC value, the better the model fits the data after considering the complexity. Indicates that each parameter is in bytes It is a measure of model complexity and is proportional to the number of parameters in the model. The more parameters and the more complex the model, the larger this value will be. B is the number of nodes in BN. is the log-likelihood of BN, which measures how well model B fits data D. The higher the log-likelihood, the better the model's ability to explain the data.

[0070] If the scoring function value of the Bayesian network model remains unchanged, it means that the structure of the Bayesian network model has stabilized. It can be that the scoring function value has reached the lowest and will not decrease any more. At this time, the structure of the Bayesian network model is the optimal structure.

[0071] Step S350, while the scoring function of the Bayesian network model remains unchanged, determining the current Bayesian network model as the pre-established Bayesian network model.

[0072] In this embodiment, the wind speed is decomposed into average wind speed and turbulent wind speed, and then described by corresponding mathematical models respectively, so that the prediction model can understand the composition of wind speed more carefully, thereby more accurately reflecting the dynamic characteristics of wind speed in wind speed prediction. Then, the basic structure of the Bayesian network model is first established using expert knowledge, and then the final Bayesian network model is obtained through an iterative learning process, thereby establishing a Bayesian network model that can accurately estimate the parameters of the random distribution of wind speed.

[0073] In one embodiment, Figure 5 As shown, step S340 adjusts the dependency between the nodes and determines the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged. It includes: steps S500-S520, wherein:

[0074] Step S500, converting the directed acyclic graph of the Bayesian network model into a moral graph.

[0075] The moral graph includes Bayesian network nodes, and the Bayesian network nodes with dependencies are connected in the moral graph. The moral graph is used to determine the joint probability distribution of the Bayesian network model. In the moral graph, by analyzing the connection relationship between nodes, the joint probability distribution of the entire network can be calculated.

[0076] Step S510, determining the scoring function value of the Bayesian network model according to the number of nodes in the moral graph and the likelihood of the nodes.

[0077] Among them, the number of nodes is used to indicate the complexity of the Bayesian network model, and the likelihood of the node is used to indicate the fit of the Bayesian network model.

[0078] Step S520, adjusting the dependency relationship between at least two nodes in the moral graph, and re-determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged.

[0079] Among them, the edges in the moral graph are added, deleted or adjusted in direction. These operations are aimed at finding a simpler model while maintaining a good description of the data. Through these adjustments, a network with the lowest score function value is selected as the initial network for the next iteration. The network structure is continuously adjusted and the score function value is recalculated until the adjustment can no longer reduce the score function value. At this time, the network structure is considered to be optimal. The resulting Bayesian network (RN) can effectively describe the dependencies between variables while simplifying the complexity of the model as much as possible.

[0080] In this embodiment, the construction of the Bayesian network model is achieved by converting the directed acyclic graph of the Bayesian network model into a moral graph, and then iterating and adjusting the network structure multiple times until the optimal structure of the Bayesian network model is found.

[0081] In one embodiment, Figure 6 As shown, step S520 adjusts the dependency between at least two nodes in the moral graph and re-determines the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged. It includes: steps S600-S640. Among them:

[0082] Step S600, using a plurality of adjustment methods to adjust the dependency relationship between at least two nodes in the initial moral graph respectively, to obtain a plurality of adjusted first moral graphs.

[0083] Among them, a variety of adjustment methods are used to adjust the dependency relationship between at least two nodes in the initial moral graph respectively, so as to obtain a series of first moral graphs, and then determine the scoring function value corresponding to each first moral graph respectively. The first moral graph with the lowest scoring function value is used as the most optimized first moral graph in this iteration for the next iteration.

[0084] Step S610, taking the first moral graph with the lowest corresponding scoring function value among the multiple adjusted first moral graphs as the second moral graph.

[0085] Step S620, determining whether the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal.

[0086] If the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal, it means that the scoring function value no longer decreases, that is, the structure of the Bayesian network model is optimal at this time. If the scoring function values ​​corresponding to the first moral graph and the second moral graph are not equal, step S630 is executed. If the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal, step S640 is executed.

[0087] Step S630, taking the second moral graph as a new initial moral graph, and returning to step S600, until the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal.

[0088] Step S640, taking the Bayesian network model structure corresponding to the second moral graph at this time as the optimal structure.

[0089] In this embodiment, the final Bayesian network model is obtained through an iterative learning process, thereby establishing a Bayesian network model that can accurately estimate the parameters of the random distribution of wind speed.

[0090] In one embodiment, Figure 7 As shown, step S220, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples, training the long short-term memory neural network model, and obtaining the wind speed random distribution parameter prediction model. It includes: steps S700-S720.

[0091] Step S700, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples to train the long short-term memory neural network model.

[0092] Step S710, adjusting the parameters of the long short-term memory neural network model until the long short-term memory neural network model meets the optimization goal.

[0093] Among them, the parameters of the long short-term memory neural network model include the weight matrix of the forget gate, the weight matrix of the input gate, the weight matrix of the unit state, the weight matrix of the output gate, the weight matrix of the output layer, the bias vector of the forget gate, the bias vector of the input gate, the bias vector of the unit state, the bias vector of the output gate, and the bias vector of the output layer. The optimization goal is that the cost function is less than the preset value.

[0094] Step S720, using the long short-term memory neural network model that meets the optimization goal as a wind speed random distribution parameter prediction model.

[0095] Specifically, the data is first normalized, and the weight matrix and bias vector are randomly initialized, including the weight matrix of the forget gate, the weight matrix of the input gate, the weight matrix of the unit state, the weight matrix of the output gate, the weight matrix of the output layer, the bias vector of the forget gate, the bias vector of the input gate, the bias vector of the unit state, the bias vector of the output gate, and the bias vector of the output layer. Then, the back propagation method based on gradient optimization is used to train the parameters to minimize the cost function. The corresponding LSTM units are trained for different observation points.

[0096] In this embodiment, the Bayesian network model determines the parameters through a posteriori estimation, and then the specific probability distribution can be determined. However, environmental data often include missing values, because problems in sampling may mean that some unique events or time points are missing. Due to the changes in the types of environmental data and the long time from monitoring to transmission, even if the Bayesian network model is very accurate, it is difficult to achieve timely estimation of parameters. Therefore, the data estimated by the Bayesian network model is used to establish and train a long short-term memory neural network model to describe the relationship between wind speed and wind speed random distribution parameters, thereby improving the real-time prediction of wind speed random distribution parameters.

[0097] In one embodiment, Figure 8 As shown, step S210, based on historical environmental monitoring data, uses a pre-established Bayesian network model to estimate the first wind speed random distribution parameter of the observation point. It includes: steps S800-S820. Among them:

[0098] Step S800, using historical environmental monitoring data as observation data in the Bayesian network model.

[0099] After obtaining the historical environmental monitoring data or the current environmental monitoring data, the continuous data is first discretized. We discretize the parameters involved in the random distribution modeling of wind speed into different numbers of categories (with the same difference) to determine whether finer discretization or coarser discretization is more suitable for improving the prediction accuracy. Define x = [TT e H u P as1 s2 s3]. ξ=[TT e H u P a ] is the evidence variable. θ=(s1 s2 s3) is the query variable. The sample number is n. ξ∈[ξ L ,ξ U ],θ∈[θ L ,θ U ], where superscript * L and* U represent the upper and lower bounds of the variable respectively. The discrete step length of each parameter is represented by:

[0100]

[0101] Among them, through N i The granularity of discretization is defined to determine how many equal-width intervals can be divided between the upper and lower bounds for each variable. In Bayesian networks or machine learning models, continuous variables usually need to be discretized for easy processing. Adding 1 to the calculation is to ensure that all possible intervals including the upper and lower bounds are covered. Through the above data preprocessing methods, a suitable discretization level can be defined for each variable, which is crucial for subsequent data analysis and model building.

[0102] Step S810, based on the dependency relationship between nodes in the Bayesian network model and the observation data, determine the posterior probability of the random distribution parameters of the wind speed at the observation point.

[0103] Among them, the BN structure effectively expresses the conditional independence between node variables. According to the BN structure, RNx is given i The parent node P i (i=1,2,...,6) set. BN assumes that each node is independent of its non-descendant nodes, so according to the BN structure, the posterior probability can be determined by the following formula:

[0104]

[0105] Among them, P B (x) is the B The joint probability distribution of the variable x in the network. Since the variables in the Bayesian network are conditionally independent, the joint probability distribution of the entire network can be decomposed into the product of each conditional probability. i represents the i-th variable in the Bayesian network. (P i ) represents the parent node set of the i-th variable, Represents the conditional probability, that is, given the parent node (P i) in the case of variable x i The probability of taking a specific value. The probability distribution of the entire system is represented by local dependencies (i.e., parent nodes to child nodes), making reasoning more efficient.

[0106] Step S820, determining the first wind speed random distribution parameter according to the posterior probability of the wind speed random distribution parameter at the observation point.

[0107] Among them, Bayesian network (BN) and Gibbs sampling can be used for posterior probability estimation. Gibbs sampling is a special Markov chain Monte Carlo method for generating samples from complex probability distributions. This method is particularly suitable for Bayesian networks with multivariate dependencies. The specific process can be input: the total number of sampling times, i.e. the total number of iterations, and the Bayesian network model. Then perform the following steps:

[0108] Step 1: Initialize the loop: for each sampling, sample from 1 to N, where N is the total number of samples.

[0109] Step 2: Get evidence: In each iteration, get the value of the evidence variable.

[0110] Step 3: Loop through each variable: Operate on each query variable. Operate on each possible value. That is, adjusting different step lengths

[0111] Step 4: Initialize variable values: randomly assign an initial value to the variable.

[0112] Step 5: Iterative update: For each state, the conditional probability distribution under the current evidence is calculated based on the Bayesian network, that is, P B (θ i |ξ=e (t) ).

[0113] Then from this distribution according to P B (θ i |ξ=e (t) ) to get the variable θ i Value

[0114] Update the variable q s-1 In Replace with Set the updated q s-1 for q s

[0115] Step 6. Count Match: If the updated value q s With the currently considered value q ij If a match occurs, the counter is incremented by one.

[0116] Step 7: Calculate the posterior probability:

[0117] Step 8: After completing the iteration of all variables and all values, define h = (h ijt ) is the posterior probability distribution matrix.

[0118] The posterior probability distribution represents the probability of each query variable taking each possible value given the evidence.

[0119] This process samples every possible value of each variable in the Bayesian network through Gibbs sampling, thereby estimating the posterior probability under given evidence. This method is particularly suitable for processing high-dimensional data and complex dependencies, making the estimation of the posterior probability more accurate and effective.

[0120] On this basis, accurate Bayesian estimates can also be made for the three parameters in the wind speed random distribution. In theory, each component of the Bayesian risk B(d) = [B1(d1)B2(d2)B3(d3)] should be minimized to calculate the estimated Different Bayesian risks usually produce different results. In Bayesian estimation, the quadratic form is widely used as a loss function because of its simplicity and efficiency:

[0121] Bayesian risk B(d) = [B1(d1) B2(d2) B3(d3)] is set

[0122]

[0123] Among them, L(θ i ,d i )=(θ i -d i (ξ (t) )) 2 is the loss function. The Bayesian estimator of parameter θ is Essentially, it is a vector containing s1, s2, and s3, which is the decision function that minimizes each Bayesian risk.

[0124]

[0125] In this embodiment, after the Bayesian network model is established, the Bayesian network model can be used to accurately estimate the wind speed random distribution parameters, thereby improving the accuracy of wind speed prediction. Combined with the long short-term memory neural network model, the response speed of wind speed prediction can be improved.

[0126] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0127] In the present embodiment, a device for determining the random distribution parameters of wind speed is also provided, and the device for determining the random distribution parameters of wind speed is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions thereof will not be repeated. As used below, the term "module" may implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and contemplated.

[0128] Fig. 9 is a structural block diagram of an optional device for determining wind speed random distribution parameters according to an embodiment of the present application; Fig. 9 As shown, including:

[0129] The data acquisition module 901 is used to acquire the historical environmental monitoring data of the observation point.

[0130] The parameter estimation module 902 is used to estimate the first wind speed random distribution parameter of the observation point based on historical environmental monitoring data using a pre-established Bayesian network model, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and turbulent wind speed.

[0131] The model training module 903 is used to use the first wind speed random distribution parameter and historical environmental monitoring data as training samples to train the long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model.

[0132] The parameter determination module 904 is used to determine the second wind speed random distribution parameter of the observation point based on the current environmental monitoring data of the observation point and using the wind speed random distribution parameter prediction model.

[0133] Through the above device, firstly, the historical environmental monitoring data of the observation point are obtained. Then, based on the historical environmental monitoring data, the first wind speed random distribution parameter of the observation point is estimated by using a pre-established Bayesian network model. Next, the first wind speed random distribution parameter and the historical environmental monitoring data are used as training samples to train the long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model. Finally, based on the current environmental monitoring data of the observation point, the wind speed random distribution parameter prediction model is used to determine the second wind speed random distribution parameter of the observation point. Since the Bayesian network model can accurately determine the probability distribution of the wind speed random distribution parameter, the prediction accuracy of the wind speed random distribution parameter is improved, and the long short-term memory neural network model can quickly and timely predict the wind speed random distribution parameter according to the environmental monitoring data currently input, the Bayesian network model and the long short-term memory neural network model are combined to quickly and accurately predict the wind speed random distribution parameter.

[0134] In an exemplary embodiment, the above device further includes:

[0135] Decomposition module, used to decompose wind speed into average wind speed and turbulent wind speed.

[0136] The first description model module is used to describe the average wind speed by using a first mathematical model, wherein the first mathematical model includes a first parameter.

[0137] The second description model module is used to describe the turbulent wind speed by using a second mathematical model, wherein the second mathematical model includes a second parameter, and the wind speed random distribution parameter includes a first parameter and a second parameter.

[0138] The model building module is used to construct corresponding Bayesian network nodes based on the first parameter, the second parameter, and the environmental impact factor, wherein the historical environmental monitoring data includes data on multiple environmental impact factors.

[0139] The model adjustment module is used to adjust the dependency relationship between the nodes and determine the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged.

[0140] The model determination module is used to determine the current Bayesian network model as the pre-established Bayesian network model while maintaining the scoring function of the Bayesian network model unchanged.

[0141] In an exemplary embodiment, the model adjustment module further includes:

[0142] The conversion unit is used to convert the directed acyclic graph of the Bayesian network model into a moral graph, wherein the moral graph includes Bayesian network nodes, and the Bayesian network nodes with dependencies are connected in the moral graph, and the moral graph is used to determine the joint probability distribution of the Bayesian network model.

[0143] The scoring unit is used to determine the scoring function value of the Bayesian network model according to the number of nodes and the likelihood of the nodes in the moral graph, wherein the number of nodes is used to indicate the complexity of the Bayesian network model, and the likelihood of the nodes is used to indicate the fit of the Bayesian network model.

[0144] The adjustment unit is used to adjust the dependency relationship between at least two nodes in the moral graph and re-determine the score function value of the Bayesian network model until the score function value of the Bayesian network model remains unchanged.

[0145] In an exemplary embodiment, the adjustment unit is further configured to use a plurality of adjustment methods to respectively adjust the dependency relationship between at least two nodes in the initial moral graph to obtain a plurality of adjusted first moral graphs.

[0146] The first moral graph with the lowest score function value among the multiple adjusted first moral graphs is used as the second moral graph.

[0147] The second moral graph is used as a new initial moral graph, and the steps of using multiple adjustment methods to adjust the dependency between at least two nodes in the initial moral graph respectively to obtain multiple adjusted first moral graphs are returned until the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal.

[0148] In an exemplary embodiment, the model training module is also used to train the long short-term memory neural network model using the first wind speed random distribution parameter and historical environmental monitoring data as training samples.

[0149] Adjust the parameters of the long short-term memory neural network model until the long short-term memory neural network model meets the optimization goal, wherein the parameters of the long short-term memory neural network model include the weight matrix of the forget gate, the weight matrix of the input gate, the weight matrix of the unit state, the weight matrix of the output gate, the weight matrix of the output layer, the bias vector of the forget gate, the bias vector of the input gate, the bias vector of the unit state, the bias vector of the output gate, and the bias vector of the output layer, and the optimization goal is that the cost function is less than a preset value.

[0150] The long short-term memory neural network model that meets the optimization objectives is used as the wind speed random distribution parameter prediction model.

[0151] In an exemplary embodiment, the parameter estimation module is also used to use historical environmental monitoring data as observation data in the Bayesian network model.

[0152] Based on the dependency relationship between nodes in the Bayesian network model and the observed data, the posterior probability of the random distribution parameters of wind speed at the observation point is determined.

[0153] The first wind speed random distribution parameter is determined according to the posterior probability of the wind speed random distribution parameter at the observation point.

[0154] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0155] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0156] S1, obtain the historical environmental monitoring data of the observation point.

[0157] S2, based on historical environmental monitoring data, uses a pre-established Bayesian network model to estimate the first wind speed random distribution parameter of the observation point, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and turbulent wind speed.

[0158] S3, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model.

[0159] S4, based on the current environmental monitoring data of the observation point, using the wind speed random distribution parameter prediction model to determine the second wind speed random distribution parameter of the observation point.

[0160] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0161] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0162] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0163] S1, obtain the historical environmental monitoring data of the observation point.

[0164] S2, based on historical environmental monitoring data, uses a pre-established Bayesian network model to estimate the first wind speed random distribution parameter of the observation point, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and turbulent wind speed.

[0165] S3, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model.

[0166] S4, based on the current environmental monitoring data of the observation point, using the wind speed random distribution parameter prediction model to determine the second wind speed random distribution parameter of the observation point.

[0167] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0168] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores the computer program product, and when the computer program is executed by a processor, the steps of the method in each embodiment of the present application are implemented.

[0169] Optionally, in this embodiment, the above computer program may be configured to implement the following steps when executed by a processor:

[0170] S1, obtain the historical environmental monitoring data of the observation point.

[0171] S2, based on historical environmental monitoring data, uses a pre-established Bayesian network model to estimate the first wind speed random distribution parameter of the observation point, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and turbulent wind speed.

[0172] S3, using the first wind speed random distribution parameter and historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model.

[0173] S4, based on the current environmental monitoring data of the observation point, using the wind speed random distribution parameter prediction model to determine the second wind speed random distribution parameter of the observation point.

[0174] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0175] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0176] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining wind speed random distribution parameters, characterized in that: The method comprises: Obtain historical environmental monitoring data of observation points; Based on the historical environmental monitoring data, a pre-established Bayesian network model is used to estimate a first wind speed random distribution parameter of the observation point, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and the turbulent wind speed; Using the first wind speed random distribution parameter and the historical environmental monitoring data as training samples, training a long short-term memory neural network model, and obtaining a wind speed random distribution parameter prediction model; Based on the current environmental monitoring data of the observation point, the wind speed random distribution parameter prediction model is adopted to determine the second wind speed random distribution parameter of the observation point.

2. The method according to claim 1, characterized in that Before estimating the first wind speed random distribution parameter of the observation point based on the historical environmental monitoring data using a pre-established Bayesian network model, the method includes: Decompose wind speed into average wind speed and turbulent wind speed; Using a first mathematical model to describe the average wind speed, wherein the first mathematical model includes a first parameter; Using a second mathematical model to describe the turbulent wind speed, wherein the second mathematical model includes a second parameter, and the wind speed random distribution parameter includes the first parameter and the second parameter; Based on the first parameter, the second parameter, and the environmental impact factor, construct a corresponding Bayesian network node, wherein the historical environmental monitoring data includes data of multiple environmental impact factors; Adjusting the dependency between the nodes and determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged; When the scoring function of the Bayesian network model remains unchanged, the current Bayesian network model is determined as the pre-established Bayesian network model.

3. The method according to claim 2, characterized in that The step of adjusting the dependency relationship between the nodes and determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged includes: Converting the directed acyclic graph of the Bayesian network model into a moral graph, wherein the moral graph includes the Bayesian network nodes, and the Bayesian network nodes with dependency relationships are connected in the moral graph, and the moral graph is used to determine the joint probability distribution of the Bayesian network model; Determining a scoring function value of the Bayesian network model according to the number of nodes and the likelihood of the nodes in the moral graph, wherein the number of nodes is used to indicate the complexity of the Bayesian network model, and the likelihood of the nodes is used to indicate the fit of the Bayesian network model; The dependency relationship between at least two nodes in the moral graph is adjusted, and the scoring function value of the Bayesian network model is re-determined until the scoring function value of the Bayesian network model remains unchanged.

4. The method according to claim 3, characterized in that: The step of adjusting the dependency relationship between at least two nodes in the moral graph and re-determining the scoring function value of the Bayesian network model until the scoring function value of the Bayesian network model remains unchanged comprises: Using multiple adjustment methods, respectively adjusting the dependency relationship between at least two nodes in the initial moral graph, to obtain multiple adjusted first moral graphs; The first moral graph with the lowest score function value corresponding to the multiple adjusted first moral graphs is used as the second moral graph; The second moral graph is used as a new initial moral graph, and the step of using multiple adjustment methods to adjust the dependency relationship between at least two nodes in the initial moral graph respectively to obtain multiple adjusted first moral graphs is returned until the scoring function values ​​corresponding to the first moral graph and the second moral graph are equal.

5. The method according to claim 1, characterized in that The method of using the first wind speed random distribution parameter and the historical environmental monitoring data as training samples to train a long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model includes: Using the first wind speed random distribution parameter and the historical environmental monitoring data as training samples to train a long short-term memory neural network model; Adjusting the parameters of the long short-term memory neural network model until the long short-term memory neural network model meets the optimization goal, wherein the parameters of the long short-term memory neural network model include a weight matrix of a forget gate, a weight matrix of an input gate, a weight matrix of a unit state, a weight matrix of an output gate, a weight matrix of an output layer, a bias vector of a forget gate, a bias vector of an input gate, a bias vector of a unit state, a bias vector of an output gate, and a bias vector of an output layer, and the optimization goal is that the cost function is less than a preset value; A long short-term memory neural network model that meets the optimization goal is used as the wind speed random distribution parameter prediction model.

6. The method according to claim 1, characterized in that The method of estimating the first wind speed random distribution parameter of the observation point based on the historical environmental monitoring data by using a pre-established Bayesian network model includes: Using the historical environmental monitoring data as observation data in the Bayesian network model; Determine the posterior probability of the random distribution parameter of the wind speed at the observation point based on the dependency relationship between the nodes in the Bayesian network model and the observation data; The first wind speed random distribution parameter is determined according to the posterior probability of the wind speed random distribution parameter of the observation point.

7. A device for determining wind speed random distribution parameters, characterized in that: The device comprises: Data acquisition module, used to obtain historical environmental monitoring data of observation points; A parameter estimation module, for estimating a first wind speed random distribution parameter of the observation point based on the historical environmental monitoring data using a pre-established Bayesian network model, wherein the Bayesian network model is modeled based on the wind speed random distribution parameter, and the wind speed random distribution parameter is used to indicate the characteristics of the average wind speed and the turbulent wind speed; A model training module, used to use the first wind speed random distribution parameter and the historical environmental monitoring data as training samples to train a long short-term memory neural network model to obtain a wind speed random distribution parameter prediction model; The parameter determination module is used to determine the second wind speed random distribution parameter of the observation point based on the current environmental monitoring data of the observation point and using the wind speed random distribution parameter prediction model.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images