Wind speed prediction method, device, equipment, storage medium and product

Abstract

The invention discloses a wind speed prediction method and device, equipment, a storage medium and a product, and relates to the technical field of data processing, and the wind speed prediction method comprises the steps: obtaining wind speed measurement data, and inputting the wind speed measurement data to a preset long short-term memory model which comprises a forgetting gate and an input gate; based on the preset long-short-term memory model, extracting long-term memory data in the wind speed measurement data, and fusing the long-term memory data, an information flow in a forgetting gate and an information flow in an input gate to obtain hidden state data including long-term memory; and performing prediction processing on the hidden state data to obtain a wind speed prediction result. The accuracy of the wind speed prediction result is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a wind speed prediction method, device, equipment, storage medium and computer program product. Background Art

[0002] Due to the influence of temperature, humidity, clouds, topography, etc., wind speed is easily affected. The wind speed time series data collected by the wind speed sensor presents typical characteristics: fast changes, large fluctuations, and instability. At present, the wind speed is mainly predicted by the ARIMA model (Autoregressive Integrated Moving Average Model). However, the ARIMA model requires that the data input to the model has strong stability. When non-stationary time series data appears, the prediction effect is general and accurate wind speed prediction results cannot be output. Therefore, how to accurately predict the wind speed value at the next moment based on the historical data of the wind speed time series is an urgent problem to be solved.

[0003] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] The main purpose of the present application is to provide a wind speed prediction method, device, equipment, storage medium and computer program product, aiming to solve the technical problem in the related technology that when the wind speed is predicted by the ARIMA model, the prediction effect is general and accurate wind speed prediction results cannot be output when non-stationary time series data appears.

[0005] To achieve the above objectives, the present application proposes a wind speed prediction method, which includes: Acquire wind speed measurement data, and input the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; Based on the preset long-short term memory model, extracting long-term memory data from the wind speed measurement data, and fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; Prediction processing is performed on the hidden state data to obtain a wind speed prediction result.

[0006] In one embodiment, the step of extracting the long-term memory data from the wind speed measurement data based on the preset long-short term memory model comprises: Based on the preset long short-term memory model, performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data to obtain a first matrix; Based on the hidden state data at the previous moment, the first matrix is ​​dimensionally transformed to obtain long-term memory data.

[0007] In one embodiment, the step of performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data based on the preset long short-term memory model to obtain a first matrix includes: Based on the preset long short-term memory model, multiplying the augmented vector by a preset matrix coefficient to obtain a first vector, a second vector and a third vector; Calculating an outer product space of the first vector and the second vector, and normalizing the outer product space; Perform nonlinear transformation on the normalized outer product space to obtain linear vector data; The linear vector data is multiplied by the third vector to obtain a first matrix.

[0008] In one embodiment, the step of performing a nonlinear transformation on the normalized outer product space to obtain linear vector data includes: The normalized outer product space is nonlinearly transformed by a nonlinear activation function, and normalized to obtain linear vector data, wherein the calculation formula of the linear vector data is as follows:

[0009] Among them, B i，j For linear vector data, ReLu (A i，j ) is the outer product space after nonlinear transformation.

[0010] In one embodiment, the step of performing dimension transformation on the first matrix based on the hidden state data at the previous moment to obtain long-term memory data includes: Based on the data dimension of the hidden state data at the previous moment, a preset transformation matrix is ​​multiplied by the first matrix to obtain long-term memory data after dimension transformation, wherein the data dimension of the long-term memory data is the same as the data dimension of the hidden state data at the previous moment.

[0011] In one embodiment, the step of fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory includes: The augmented vector is respectively processed by a forget gate and an input gate to obtain a first information flow in the forget gate and a second information flow in the input gate; Multiplying the long-term memory data and the first information flow with the hidden state data at the previous moment respectively to obtain a long-term memory data product and a forget gate data product; The long-term memory data product, the forget gate data product and the second information flow are weightedly summed with preset weight values ​​to obtain hidden state data including long-term memory.

[0012] In addition, to achieve the above purpose, the present application also proposes a wind speed prediction device, which includes: An acquisition module, used for acquiring wind speed measurement data, and inputting the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; An extraction module, configured to extract long-term memory data from the wind speed measurement data based on the preset long-short term memory model, and fuse the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; The processing module is used to perform prediction processing on the hidden state data to obtain a wind speed prediction result.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes a wind speed prediction device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the wind speed prediction method described above.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the wind speed prediction method described above are implemented.

[0015] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the wind speed prediction method described above are implemented.

[0016] The present application proposes a wind speed prediction method, device, equipment, storage medium and computer program product. The present application obtains wind speed measurement data and inputs the wind speed measurement data into a preset long-short term memory model. The long-term memory data in the wind speed measurement data is extracted through the preset long-short term memory model, and the long-term memory data, the information flow in the forgetting gate and the information flow in the input gate are merged to obtain hidden state data including long-term memory. When the hidden state data is predicted and processed to obtain the wind speed prediction result at the next moment, the influence of long-term memory can be taken into account more, thereby reducing the inaccuracy caused by considering only recent memory and increasing the accuracy of the wind speed prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] 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.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0019] Figure 1 A schematic diagram of a flow chart provided for the first embodiment of the wind speed prediction method of the present application; Figure 2 A schematic diagram of a flow chart provided for the second embodiment of the wind speed prediction method of the present application; Figure 3 A schematic diagram of the model architecture involved in the wind speed prediction method of this application; Figure 4 This is a schematic diagram of the module structure of the wind speed prediction device according to an embodiment of the present application; Figure 5 Schematic diagram of the equipment structure of the hardware operating environment involved in the wind speed prediction method in the embodiment of the present application.

[0020] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0022] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solutions of the embodiments of this application are: Acquire wind speed measurement data, and input the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; Based on the preset long-short term memory model, extracting long-term memory data from the wind speed measurement data, and fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; Prediction processing is performed on the hidden state data to obtain a wind speed prediction result.

[0024] In the prior art, the wind speed is predicted by the ARIMA model (Autoregressive Integrated Moving Average Model), but the ARIMA model requires that the data input to the model have strong stationarity. When non-stationary time series data appears, the prediction effect is general and it is impossible to output accurate wind speed prediction results.

[0025] In related technologies, there are also solutions for using recurrent neural networks and long short-term memory models to predict data, but both have their shortcomings: 1. RNN (recurrent neural network) does not have a hidden state update mechanism, and the hidden state has the characteristic of preserving the essential laws of time series. Therefore, RNN can usually only predict time series with relatively simple data changes. It cannot model time series that contain seasonality, periodicity, and oscillation well.

[0026] 2. LSTM overcomes the shortcomings of RNN to a certain extent. By introducing hidden states, it can model time series with more complex changes. However, since the calculation of hidden states is relatively simple, the time window of historical data used is small, and only historical data close to the next moment can be used to predict the future, and the accuracy is still low. In particular, when the size of the historical data time window is increased, the prediction accuracy is not significantly improved, and there is even the possibility of reduction. The reason is that the calculation of hidden states is relatively simple, and there is no effective modeling of data far from the next moment, so they cannot play their role in the prediction process.

[0027] The present application proposes a wind speed prediction method, device, equipment, storage medium and computer program product. The present application obtains wind speed measurement data and inputs the wind speed measurement data into a preset long-short term memory model. The long-term memory data in the wind speed measurement data is extracted through the preset long-short term memory model, and the long-term memory data, the information flow in the forgetting gate and the information flow in the input gate are merged to obtain hidden state data including long-term memory. When the hidden state data is predicted and processed to obtain the wind speed prediction result at the next moment, the influence of long-term memory can be taken into account more, thereby reducing the inaccuracy caused by considering only recent memory and increasing the accuracy of the wind speed prediction result.

[0028] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a wind speed prediction device, etc. The following takes the wind speed prediction device as an example to illustrate this embodiment and the following embodiments.

[0029] Based on this, the present application embodiment provides a wind speed prediction method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the wind speed prediction method of the present application.

[0030] In this embodiment, the wind speed prediction method includes steps S10 to S40: Step S10, obtaining wind speed measurement data, and inputting the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; It should be noted that the wind speed measurement data may be the wind speed data measured by the sensor at the current moment. The wind speed measurement data also includes noise data, which can be processed by inputting into the long short-term memory model to obtain the wind speed prediction result at the next moment.

[0031] It should be noted that the present application mainly involves the improvement of the long short-term memory model. Specifically, a processing module for extracting long-term memory data is added to the model, so that the long-term memory data can be added to the update process of the hidden state of the long short-term memory model. When predicting the data at the next moment, the influence of long-term memory can be taken into account to improve the accuracy of the prediction results.

[0032] It should be noted that the preset long-term and short-term memory model mainly includes six parts: long-term memory extraction unit, long-term memory and current memory fusion unit, memory cell, input gate, output gate and forget gate. The functions of each module are as follows: The long-term memory extraction unit is mainly used to extract the long-term memory in the model output and wind speed measurement data of the previous moment. After the long-term memory data is extracted, the long-term memory is fused with the output of the input gate and the forget gate to synchronously update the memory cells in the model.

[0033] Memory cell: It is the core of LSTM and is responsible for storing important information and passing it to the subsequent network layers.

[0034] The role of memory cells is to maintain a persistent state. It is constantly updated with continuous input. Its characteristic is that it can save historical information and apply it to current predictions. Memory cells can be regarded as a simple number that is updated at each time step. Then the input gate, forget gate, and output gate control how to update and output this value at each time step.

[0035] Input gate: determines whether the current input information is written into the memory cell, that is, it can control the impact of the input information on the memory cell.

[0036] Forget gate: determines whether the information in the memory cell is forgotten, that is, it can control whether the information stored in the memory cell will disappear.

[0037] Output gate: determines whether the information in the memory cell is output, that is, it can control whether the information stored in the memory cell will affect the subsequent network layers.

[0038] These four parts calculate the dot product of the weight matrix and the input signal, calculate the output value of each gate through the activation function, and then multiply it by the value of the memory cell for the final calculation.

[0039] Step S20, based on the preset long-short term memory model, extracting long-term memory data from the wind speed measurement data, and fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; It should be noted that in the process of extracting long-term memory data, the wind speed measurement data and the output of the model at the previous moment are combined to form an augmented vector, which can be used Indicates that, h t-1 is the output of the model at the previous moment, x t The input wind speed measurement data is time series data.

[0040] It should be noted that the information flow in the forget gate and the information flow in the input gate are both through the augmented vector The data is obtained after processing, wherein the processing flow of the forget gate and the input gate for data is prior art and will not be described here.

[0041] It should be noted that the three information flows act on the hidden state vector C for real-time update; The first information flow is the augmented vector Acts on the long-term memory retrieval unit (hidden state vector at the previous moment); The second information flow is the augmented vector Acting on ; The third information flow is the augmented vector Through the input gate and Function acts on ; It should be noted that the process of fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate can be to multiply the long-term memory data and the information flow in the forget gate by the bitwise multiplication of the hidden state vector, and then perform a weighted summation of the information flow in the input gate and the former two, so as to obtain the fused hidden state vector.

[0042] It should be noted that when the hidden state that integrates the long-term memory data is updated at the next moment, the data that is far away from the next moment can be retained in the hidden state and participate in the prediction of the data at the next moment, thereby improving the accuracy of the prediction results.

[0043] In a feasible implementation manner, the step S20 of extracting the long-term memory data from the wind speed measurement data based on the preset long-short term memory model includes: Based on the preset long short-term memory model, performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data to obtain a first matrix; It should be noted that the augmented vector is To express, Represents a vector conduct The new vector obtained after the expansion, the step of extracting the long-term memory data in the wind speed measurement data is performed by the long-term memory extraction unit, and the long-term memory extraction unit mainly includes and Modules, The module is mainly used to perform matrix transformation on augmented vectors. The module is used to align the dimensions of the long-term memory vector with the dimensions of the hidden state vector.

[0044] Specifically, Module for augmenting vector The processing includes four steps. The first step is matrix transformation. , Matrix Transformation , Matrix Transformation , the second step: normalization, the third step: nonlinear transformation, and the fourth step: matrix transformation, are used together for the extraction of long-term memory.

[0045] Based on the hidden state data at the previous moment, the first matrix is ​​dimensionally transformed to obtain long-term memory data.

[0046] It should be noted that the hidden state data exists in the form of a vector, which is subsequently referred to as the hidden state vector. Indicates that it is updated every moment, for example, the current moment is , the previous moment is , the next moment is .

[0047] It should be noted that The module is used to store long-term memory vectors The dimension and vector of The dimensions of With vector For the same dimension, after aligning the dimensions, the required long-term memory data is output. By integrating the long-term memory data into the hidden state vector, the value of data far away from the next moment is brought into play in the process of predicting the data at the next moment, making the prediction result more accurate.

[0048] In a feasible implementation manner, the step of performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data based on the preset long short-term memory model to obtain a first matrix includes: Based on the preset long short-term memory model, multiplying the augmented vector by a preset matrix coefficient to obtain a first vector, a second vector and a third vector; It should be noted that the preset matrix coefficients may be matrix transformation coefficients , matrix transformation coefficients , matrix transformation coefficients , these three matrix change coefficients are randomly initialized in the model and automatically adjusted according to the gradient descent of the training process. The specific values ​​are not limited.

[0049] It should be noted that the first vector is expressed as: , the second vector is expressed as: , the third vector is expressed as: .

[0050] Calculating an outer product space of the first vector and the second vector, and normalizing the outer product space; It should be noted that after obtaining the three vectors, the first vectors v1 and v2 are calculated. and The outer product space , and normalized to get .

[0051] It should be noted that the outer product space is a specific subspace in the vector space obtained through the outer product operation. It is usually used for calculations between two vectors, and the result of the calculation is a vector rather than a scalar.

[0052] Perform nonlinear transformation on the normalized outer product space to obtain linear vector data; It should be noted that the normalized outer product space is subjected to nonlinear transformation to obtain the required vector data, and the specific process is calculated through the nonlinear activation function Relu.

[0053] The linear vector data is multiplied by the third vector to obtain a first matrix.

[0054] It should be noted that after the linear vector data is calculated, the linear vector data is multiplied by the third vector. The specific calculation formula is as follows: , and get the transformed vector

[0055] Among them, Bm,n represents linear vector data, v3 represents the third vector, and the transformed vector is the first matrix.

[0056] In a feasible implementation manner, the step of performing a nonlinear transformation on the normalized outer product space to obtain linear vector data includes: The normalized outer product space is nonlinearly transformed by a nonlinear activation function, and normalized to obtain linear vector data, wherein the calculation formula of the linear vector data is as follows:

[0057] Among them, Bi,j is the linear vector data, and ReLu(Ai,j) is the outer product space after nonlinear transformation.

[0058] Specifically, the process of performing nonlinear transformation can be: Implement ReLu nonlinear transformation according to the row: , , then B 1,1 = , where the molecule For the matrix The first row and first column element of , ReLu is a nonlinear activation function, and the denominator is a matrix Each element in the first row is activated nonlinearly and then added together, that is, the numerator is normalized. Similarly, the matrix The calculation method for any element of is as follows: , and we get the matrix .

[0059] In a feasible implementation manner, the step of performing dimension transformation on the first matrix based on the hidden state data at the previous moment to obtain long-term memory data includes: Based on the data dimension of the hidden state data at the previous moment, a preset transformation matrix is ​​multiplied by the first matrix to obtain long-term memory data after dimension transformation, wherein the data dimension of the long-term memory data is the same as the data dimension of the hidden state data at the previous moment.

[0060] It should be noted that after determining the data dimension of the hidden state data at the previous moment , the dimension of the first matrix is transformed through the β module. Specifically: Calculate the matrix transformation of the first matrix : , and the transformed vector is obtained, where [Shi 1]. This transformation matrix aligns the dimension of the vector with the dimension of the vector . After the dimension conversion, the data dimension of the long-term memory data is the same as the data dimension of the hidden state data at the previous moment. Among them, The calculation method is eigenvalue decomposition or SVD decomposition, which belongs to the common knowledge in this field and will not be elaborated here.

[0061] It should be noted that after the calculations of the α and β modules, the vector is obtained. This vector is subjected to a Hadamard product with the vector , that is, multiplied bit by bit [Shi 2]. After the bit-by-bit multiplication, is obtained, and this result is used as the input for the subsequent fusion link.

[0062] Step S30, perform prediction processing on the hidden state data to obtain a wind speed prediction result.

[0063] It should be noted that after fusing the information flows of the long-term memory data, the forgetting gate, and the input gate with the hidden state vector at the previous moment, the model updates C t-1 to obtain C t . After obtaining , will be updated. Specifically, the update process of h t is as follows: ,

[0064] Among them, ht is the hidden state at the t-th moment As the output, it is not the wind speed value f, is a vector, and the wind speed value is a scalar. An output matrix is required as a mapping transformation between them. And it is necessary to synthesize the at several moments to obtain the wind speed value at the t-th moment. For example: 5 moments (t-4, t-3, t-2, t-1, t, a total of 5 moments) jointly obtain the wind speed value f at the t-th moment. Then , and so on, , , , Among them, the matrix is the output matrix obtained through data training. Then, the wind speed value f at time t is: , where w is a scalar, representing the weight, which is also obtained through training, from which the required wind speed prediction result can be obtained.

[0065] The present application proposes a wind speed prediction method, device, equipment, storage medium and computer program product. The present application obtains wind speed measurement data and inputs the wind speed measurement data into a preset long-short term memory model. The long-term memory data in the wind speed measurement data is extracted through the preset long-short term memory model, and the long-term memory data, the information flow in the forgetting gate and the information flow in the input gate are merged to obtain hidden state data including long-term memory. When the hidden state data is predicted and processed to obtain the wind speed prediction result at the next moment, the influence of long-term memory can be taken into account more, thereby reducing the inaccuracy caused by considering only recent memory and increasing the accuracy of the wind speed prediction result.

[0066] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 The step S20 of fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain the hidden state data including the long-term memory comprises: Step S21, the augmented vector is respectively processed by a forget gate and an input gate to obtain a first information flow in the forget gate and a second information flow in the input gate; It should be noted that the first information flow is the information flow obtained after the augmented vector is processed by the forget gate, which can be expressed as: .

[0067] It should be noted that the first information flow is the information flow obtained after the augmented vector is processed by the input gate, which can be expressed as:

[0068] in, It is through The output obtained is, is the output obtained by tanh, and the two outputs are multiplied bit by bit to get the content of the above formula.

[0069] Step S22, multiplying the long-term memory data and the first information flow with the hidden state data at the previous moment respectively to obtain a long-term memory data product and a forget gate data product; It should be noted that the long-term memory data product is expressed as: , which is composed of the output of the long-term memory retrieval unit and The bitwise product is obtained as the weight matrix Input

[0070] The product of the forget gate data is expressed as:

[0071] Obtained by the bitwise product of the output of the forget gate and as the input

[0072] Step S23: Perform weighted summation on the long-term memory data product, the forget gate data product, and the second information flow with preset weight values respectively to obtain hidden state data including long-term memory.

[0073] It should be noted that the preset weight values can be: , , , where , , are weight matrices [Shi 3], and the constraint condition is , , The matrices are all of the same scale and are non-negative matrices, that is, each element in the matrix is a value greater than or equal to 0, and for the i-th row of the three matrices respectively: , , The sum of is 1, that is . Preferably, to reduce the computational complexity and the implementation difficulty, in the embodiment, , , can be simplified to three scalars.

[0074] It should be noted that the calculation process of performing weighted summation on the long-term memory data product, the forget gate data product, and the second information flow with preset weight values respectively can be:

[0075] In this embodiment, by extracting the long-term memory data in the model input data and fusing it with the information flows of the forget gate and the input gate, the hidden state [Shi 4] including long-term memory is obtained, so that when predicting the data at the next moment, more consideration is given to the influence brought by the long-term memory, reducing the inaccuracy caused by only considering the recent memory and increasing the accuracy of the prediction.

[0076] Exemplarily, to help understand the implementation process of the wind speed prediction method obtained by combining this embodiment with the above-mentioned Embodiment 1, please refer to Figure 3 , Figure 3 which provides a schematic diagram of the brief process of a wind speed prediction method, specifically: The augmented vector corresponding to the wind speed measurement data is processed by the long-term memory extraction unit, the forget gate, and the input gate. After extracting the long-term memory data in the augmented vector, the hidden state vector C is updated by weighted summing the three information streams. t , then update h synchronously t , and then the calculation process of the model is used to perform calculation processing and output the wind speed prediction result. Since the subsequent calculation process is an existing technology, it will not be described in detail here.

[0077] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the wind speed prediction method of the present application. More simple transformations based on this technical concept are all within the protection scope of the present application.

[0078] This application also provides a wind speed prediction device, please refer to Figure 4 , the wind speed prediction device comprises: An acquisition module 10 is used to acquire wind speed measurement data, and input the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; An extraction module 20 is used to extract long-term memory data from the wind speed measurement data based on the preset long-short term memory model, and fuse the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; The processing module 30 is used to perform prediction processing on the hidden state data to obtain a wind speed prediction result.

[0079] Optionally, the extraction module includes: A processing unit, configured to perform matrix transformation processing on the augmented vector corresponding to the wind speed measurement data based on the preset long short-term memory model to obtain a first matrix; The dimension conversion unit is used to perform dimension conversion on the first matrix based on the hidden state data at the previous moment to obtain long-term memory data.

[0080] Optionally, the processing unit includes: A first calculation subunit is used to multiply the augmented vector by a preset matrix coefficient based on the preset long short-term memory model to obtain a first vector, a second vector and a third vector; A second calculation subunit, configured to calculate an outer product space of the first vector and the second vector, and normalize the outer product space; A third calculation subunit is used to perform nonlinear transformation on the normalized outer product space to obtain linear vector data; The fourth calculation subunit is used to multiply the linear vector data by the third vector to obtain a first matrix.

[0081] Optionally, the third computing subunit includes: The calculation component is used to perform nonlinear transformation on the normalized outer product space through a nonlinear activation function, and perform normalization calculation to obtain linear vector data, wherein the calculation formula of the linear vector data is as follows:

[0082] Among them, Bi,j is the linear vector data, and ReLu(Ai,j) is the outer product space after nonlinear transformation.

[0083] Optionally, the dimension conversion unit includes: The fifth computing subunit is used to multiply the preset transformation matrix by the first matrix based on the data dimension of the hidden state data at the previous moment to obtain the long-term memory data after the transformation dimension, wherein the data dimension of the long-term memory data is the same as the data dimension of the hidden state data at the previous moment.

[0084] Optionally, the extraction module further includes: A first calculation unit, configured to calculate and process the augmented vector through a forget gate and an input gate respectively, to obtain a first information flow in the forget gate and a second information flow in the input gate; a second calculation unit, configured to multiply the long-term memory data and the first information stream by the hidden state data at the previous moment respectively, to obtain a long-term memory data product and a forget gate data product; The third calculation unit is used to perform weighted summation on the product of the long-term memory data, the product of the forget gate data and the second information flow with preset weight values ​​respectively to obtain hidden state data including long-term memory.

[0085] The wind speed prediction device provided by the present application adopts the wind speed prediction method in the above embodiment to solve the technical problem of wind speed prediction. Compared with the prior art, the beneficial effects of the wind speed prediction device provided by the present application are the same as the beneficial effects of the wind speed prediction method provided by the above embodiment, and other technical features in the wind speed prediction device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0086] The present application provides a wind speed prediction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the wind speed prediction method in the above-mentioned embodiment one.

[0087] Reference below Figure 5, which shows a schematic diagram of the structure of a wind speed prediction device suitable for implementing the embodiment of the present application. The wind speed prediction device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The wind speed prediction device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0088] like Figure 5 As shown, the wind speed prediction device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the wind speed prediction device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the wind speed prediction device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a wind speed prediction device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0089] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0090] The wind speed prediction device provided by the present application adopts the wind speed prediction method in the above embodiment to solve the technical problem of wind speed prediction. Compared with the prior art, the beneficial effects of the wind speed prediction device provided by the present application are the same as the beneficial effects of the wind speed prediction method provided by the above embodiment, and other technical features in the wind speed prediction device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0091] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0092] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0093] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the wind speed prediction method in the above-mentioned embodiment.

[0094] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0095] The computer-readable storage medium may be included in the wind speed prediction device; or may exist independently without being assembled into the wind speed prediction device.

[0096] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the wind speed prediction device, the wind speed prediction device: Acquire wind speed measurement data, and input the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; Based on the preset long-short term memory model, extracting long-term memory data from the wind speed measurement data, and fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; Prediction processing is performed on the hidden state data to obtain a wind speed prediction result.

[0097] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0098] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0099] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0100] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned wind speed prediction method, and can solve the technical problem of wind speed prediction. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the wind speed prediction method provided in the above-mentioned embodiment, and will not be described in detail here.

[0101] The present application also provides a computer program product, including a computer program, which implements the steps of the wind speed prediction method as described above when executed by a processor.

[0102] The computer program product provided in this application can solve the technical problem of wind speed prediction. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the wind speed prediction method provided in the above embodiment, which will not be repeated here.

[0103] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A wind speed prediction method, characterized in that: The method comprises: Acquire wind speed measurement data, and input the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; Based on the preset long-short term memory model, extracting long-term memory data from the wind speed measurement data, and fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; Prediction processing is performed on the hidden state data to obtain a wind speed prediction result.

2. The method according to claim 1, characterized in that The step of extracting the long-term memory data from the wind speed measurement data based on the preset long-short term memory model comprises: Based on the preset long short-term memory model, performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data to obtain a first matrix; Based on the hidden state data at the previous moment, the first matrix is ​​dimensionally transformed to obtain long-term memory data.

3. The method according to claim 2, characterized in that The step of performing matrix transformation processing on the augmented vector corresponding to the wind speed measurement data based on the preset long short-term memory model to obtain a first matrix includes: Based on the preset long short-term memory model, multiplying the augmented vector by a preset matrix coefficient to obtain a first vector, a second vector and a third vector; Calculating an outer product space of the first vector and the second vector, and normalizing the outer product space; Perform nonlinear transformation on the normalized outer product space to obtain linear vector data; The linear vector data is multiplied by the third vector to obtain a first matrix.

4. The method according to claim 3, characterized in that The step of performing nonlinear transformation on the normalized outer product space to obtain linear vector data includes: The normalized outer product space is nonlinearly transformed by a nonlinear activation function, and normalized to obtain linear vector data, wherein the calculation formula of the linear vector data is as follows: Among them, Bi,j is the linear vector data, and ReLu(Ai,j) is the outer product space after nonlinear transformation.

5. The method according to claim 2, characterized in that The step of performing dimension transformation on the first matrix based on the hidden state data at the previous moment to obtain long-term memory data includes: Based on the data dimension of the hidden state data at the previous moment, a preset transformation matrix is ​​multiplied by the first matrix to obtain long-term memory data after dimension transformation, wherein the data dimension of the long-term memory data is the same as the data dimension of the hidden state data at the previous moment.

6. The method according to claim 1, characterized in that The step of fusing the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory comprises: The augmented vector is respectively processed by a forget gate and an input gate to obtain a first information flow in the forget gate and a second information flow in the input gate; Multiplying the long-term memory data and the first information flow with the hidden state data at the previous moment respectively to obtain a long-term memory data product and a forget gate data product; The long-term memory data product, the forget gate data product and the second information flow are weightedly summed with preset weight values ​​to obtain hidden state data including long-term memory.

7. A wind speed prediction device, characterized in that: The device comprises: An acquisition module, used for acquiring wind speed measurement data, and inputting the wind speed measurement data into a preset long short-term memory model, wherein the preset long short-term memory model includes a forget gate and an input gate; An extraction module, configured to extract long-term memory data from the wind speed measurement data based on the preset long-short term memory model, and fuse the long-term memory data, the information flow in the forget gate, and the information flow in the input gate to obtain hidden state data including long-term memory; The processing module is used to perform prediction processing on the hidden state data to obtain a wind speed prediction result.

8. A wind speed prediction device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the wind speed prediction method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the wind speed prediction method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the wind speed prediction method according to any one of claims 1 to 6 are implemented.

Images