Anchor cable prestress prediction method, device, equipment, storage medium and program product

Abstract

The invention relates to the technical field of wind power generation, and discloses an anchor cable prestress prediction method and device, equipment, a storage medium and a program product, and the method comprises the steps: obtaining fan operation data of a fan to which a target anchor cable belongs, environment data of an environment where the target anchor cable is located, and anchor cable parameters; establishing a time sequence characteristic matrix according to the environment data, the fan operation data and the anchor cable parameters; inputting the time sequence characteristic matrix into an LSTM encoder, and calculating a first hidden state of each time node through the LSTM encoder to form a first hidden state sequence; inputting the first hidden state sequence into a Transform encoder, calculating the first hidden state sequence by using a self-attention mechanism to obtain a second hidden state of each time node, and forming a second hidden state sequence; and inputting the second hidden state sequence into the full connection layer to obtain a pre-stress prediction value of the anchor cable. According to the method, the two encoders are combined, so that the prestress of the anchor cable can be predicted from the two dimensions of short-term characteristics and long-term characteristics, and the prediction is more accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power generation, in particular to an anchor cable prestress prediction method, device, equipment, storage medium and program product. BACKGROUND

[0002] In the technical field of wind power generation, as a key force component, the prestress state of the anchor cable is directly related to the safety and stability of the structure. The prestress of the anchor cable is comprehensively affected by many factors.

[0003] Accurate prediction of the prestress of the anchor cable is of great significance for timely discovering potential risks of the structure and formulating maintenance strategies. At present, the prestress of the anchor cable is mainly predicted by establishing a mechanical equation of force of the anchor cable through Hooke's law, the principle of thermal expansion and cold contraction, etc., so as to calculate the prestress according to the mechanical equation. However, this method is difficult to comprehensively consider the nonlinear influence of complex factors such as environment and operating conditions, resulting in limited prediction accuracy. SUMMARY

[0004] Therefore, the present application provides an anchor cable prestress prediction method, device, equipment, storage medium and program product to solve the problem that the nonlinear influence of complex factors such as environment and operating conditions cannot be comprehensively considered in the prior art, resulting in limited prediction accuracy.

[0005] In a first aspect, the present application provides an anchor cable prestress prediction method, comprising: obtaining wind turbine operating data of a target anchor cable belonging to a wind turbine, environment data of an environment in which the target anchor cable is located, and anchor cable parameters; establishing a time sequence feature matrix according to the environment data, the wind turbine operating data and the anchor cable parameters; inputting the time sequence feature matrix into an LSTM encoder, calculating a first hidden state of each time node through the LSTM encoder to form a first hidden state sequence; inputting the first hidden state sequence into a Transformer encoder, calculating the first hidden state sequence by using a self-attention mechanism of the Transformer encoder to obtain a second hidden state of each time node to form a second hidden state sequence; and inputting the second hidden state sequence into a full connection layer to obtain a prestress prediction value of the anchor cable.

[0006] The method provided in the embodiments of the present application combines the LSTM encoder and the Transformer encoder to predict the prestress of the anchor cable. The LSTM encoder can capture short-term dynamic fluctuations, and the Transformer encoder can mine global correlations. Therefore, combining the two encoders can predict the prestress of the anchor cable from two dimensions of short-term features and long-term features, solve the limitation of a single model, and make the prediction more accurate. In addition, in the embodiments of the present application, the features input into the encoder also include anchor cable parameters, which are physical parameters. Incorporating anchor cable physical parameters as constraints makes the prediction results conform to the mechanical principle and adapt to anchor cables and operating conditions of different specifications.

[0007] In an optional embodiment, the environmental data, wind turbine operation data, and anchor cable parameters all contain multiple types of parameters; a time series feature matrix is ​​established based on the environmental data, wind turbine operation data, and anchor cable parameters, including: calculating the correlation coefficients between various parameters in the environmental data, wind turbine operation data, and anchor cable parameters and the historical prestress according to the historical environmental data, historical wind turbine operation data, historical anchor cable parameters, and historical prestress corresponding to the target anchor cable; and selecting parameters with correlation coefficients greater than preset values ​​from the environmental data, wind turbine operation data, and anchor cable parameters to construct a time series feature matrix.

[0008] In an optional embodiment, the correlation coefficient is calculated using the following formula:

[0009]

[0010] Among them, X i (t) is the value of the i-th parameter at time t, is the mean value of the i-th parameter in the current time window, F(t) is the prestress at time t, is the mean value of the prestress in the current time window.

[0011] In an optional embodiment, the time series feature matrix is ​​determined by the feature values ​​of multiple types of features at different time points, and the features include parameters with correlation coefficients greater than preset values, and composite features constructed based on various parameters in environmental data, wind turbine operation data, and anchor cable parameters.

[0012] In an optional embodiment, the composite feature includes one or more of wind load intensity, temperature stress factor, vibration energy of the wind turbine, and wet bulb temperature.

[0013] In an optional embodiment, the first hidden state sequence is input into the Transformer encoder, and the self-attention mechanism of the Transformer encoder is used to calculate the first hidden state sequence to obtain the second hidden state of each time node, forming a second hidden state sequence, including: mapping the first hidden state sequence into a query matrix, a key matrix, and a value matrix through a linear transformation; using a multi-head attention mechanism to divide the query matrix, the key matrix, and the value matrix into multiple heads, and calculating the attention in parallel; performing a linear transformation on the attention calculation results of each head to obtain a multi-head attention result; performing a nonlinear transformation on the multi-head attention to obtain enhanced features of the multi-head attention results; using residual connections and layer normalization to calculate the multi-head attention results and enhanced features to obtain a second hidden state sequence.

[0014] In a second aspect, the present application provides an anchor cable prestress prediction device, comprising: a data acquisition module, configured to acquire wind turbine operation data of a target anchor cable, environment data of an environment in which the target anchor cable is located, and anchor cable parameters; a time sequence feature matrix establishment module, configured to establish a time sequence feature matrix according to the environment data, the wind turbine operation data, and the anchor cable parameters; an LSTM encoder, configured to input the time sequence feature matrix into the LSTM encoder, calculate the time sequence feature matrix through the LSTM encoder, obtain first hidden states of each time node, and form a first hidden state sequence; a Transformer encoder, configured to calculate the first hidden state sequence through a self-attention mechanism of the Transformer encoder, obtain second hidden states of each time node, and form a second hidden state sequence; and a prediction module, configured to input the second hidden state sequence into a fully connected layer, and obtain a prestress prediction value of the anchor cable.

[0015] In a third aspect, the present application provides a computer device, comprising: a memory and a processor, which are communicatively connected with each other, and the memory stores computer instructions; the processor executes the computer instructions, thereby executing the anchor cable prestress prediction method of the first aspect or any of the corresponding embodiments thereof.

[0016] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer instructions, and the computer instructions are used to make a computer execute the anchor cable prestress prediction method of the first aspect or any of the corresponding embodiments thereof.

[0017] In a fifth aspect, the present application provides a computer program product, which comprises computer instructions, and the computer instructions are used to make a computer execute the anchor cable prestress prediction method of the first aspect or any of the corresponding embodiments thereof. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0019] Figure 1 is a flowchart of the anchor cable prestress prediction method according to an embodiment of the present application;

[0020] Figure 2 is a structural block diagram of the anchor cable prestress prediction device according to an embodiment of the present application;

[0021] Figure 3Fig. 1 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0023] In the embodiment, an anchor cable prestress prediction method is provided, which can be applied to the mobile terminal such as a mobile phone, a tablet computer and the like. Figure 1 Fig. 2 is a flowchart of the anchor cable prestress prediction method according to an embodiment of the present application, as shown in the figure, the flow includes the following steps: Figure 1

[0024] In step S101, the wind turbine operation data of the target anchor cable, the environmental data of the environment where the wind turbine is located and the anchor cable parameters are acquired.

[0025] In an optional embodiment, the wind turbine operation data of the target anchor cable includes one or more of power output, blade pitch angle, rotating speed, vibration data, yaw angle and the like.

[0026] The power output can reflect the running state and load condition of the wind turbine. The change of the blade pitch angle can affect the aerodynamic performance of the wind turbine and the stress of the tower. The rotating speed can reflect the running efficiency and mechanical load of the wind turbine. The vibration data can reflect the dynamic stress condition of the tower and the anchor cable. The change of the yaw angle can affect the stress distribution of the tower.

[0027] In an optional embodiment, the environmental data includes one or more of environmental temperature, humidity, wind speed, wind direction, air pressure and the like.

[0028] The environmental temperature can be collected in real time by a high-precision temperature sensor installed around the tower. The temperature data can reflect the influence of the environment on the thermal expansion and contraction of the anchor cable material, and the thermal expansion and contraction is one of the important factors causing the change of the prestress.

[0029] The humidity can be collected by a humidity sensor installed around the tower. The change of the humidity can affect the corrosion resistance and material properties of the anchor cable, the material properties include the elastic modulus. The moisture in the environment can corrode the anchor cable, and the elastic modulus of the corroded part will change. Therefore, the environmental humidity needs to be collected when the prestress of the anchor cable is predicted.

[0030] The wind speed can be measured by a wind speed meter installed at the top of the wind turbine. The wind speed is a key factor affecting the running state of the wind turbine and the stress of the anchor cable.​

[0031] The wind direction can be measured by a wind vane, and the change of the wind direction can affect the stress distribution of the tower, and further affect the prestress of the anchor cable.

[0032] The air pressure can be collected by an air pressure sensor, and the change of the air pressure can indirectly affect the stress of the anchor cable.

[0033] In an optional embodiment, the anchor cable parameters include the diameter, length, anchoring mode, elastic modulus, thermal expansion coefficient, yield strength, initial installation tension, and the like of the anchor cable.

[0034] In an optional embodiment, the environmental data and the wind turbine operation data are time series data that need to be collected at different time points.

[0035] Step S102, establishing a time series feature matrix according to the environmental data, the wind turbine operation data, and the anchor cable parameters.

[0036] In an optional embodiment, when the time series feature matrix is established according to the environmental data, the wind turbine operation data, and the anchor cable parameters, the data is first cleaned, and the cleaned data is normalized or standardized, and the time series feature matrix is established based on the processed data.

[0037] Step S103, inputting the time series feature matrix into an LSTM encoder, calculating a first hidden state of each time node by the LSTM encoder, and forming a first hidden state sequence.

[0038] In an optional embodiment, the LSTM encoder is calculated by the following formula:

[0039] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0040] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0041]

[0042] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0043]

[0044] ht =o t ⊙tanh(C t )

[0045] Among them, f t 、i t 、o t are the outputs of the forget gate, input gate, and output gate respectively, σ is the sigmoid activation function, tanh is the hyperbolic tangent activation function, and W f 、W i 、W C 、W o are the weight matrices of the forget gate, input gate, cell state, and output gate, respectively, b f 、b i 、b C 、b o are the bias vectors of the forget gate, input gate, cell state and output gate respectively, h t-1 is the hidden state of the previous moment, x t is the input at the current moment, ⊙ represents element-wise multiplication, C t is the cell state at the current moment, h t is the hidden state at the current moment, which is also the output of the LSTM encoder.

[0046] The above is the process of processing the features of a moment in the time series feature matrix. After calculating the features of multiple moments in the time series feature matrix, the first hidden states of all time nodes can be obtained, thereby forming a first hidden state sequence.

[0047] In step S104, the first hidden state sequence is input into the Transformer encoder, and the self-attention mechanism of the Transformer encoder is used to calculate the first hidden state sequence to obtain the second hidden state of each time node, thereby forming a second hidden state sequence.

[0048] In an optional embodiment, the Transformer encoder is computed using a self-attention mechanism and a feed-forward neural network.

[0049] Step S105: input the second hidden state sequence into the fully connected layer to obtain the prestress prediction value of the anchor cable.

[0050] In an optional embodiment, the calculation formula of the fully connected layer is as follows:

[0051]

[0052] in, is the predicted prestress value, W out is the weight matrix of the output layer, b out is the bias vector of the output layer, ht The hidden state output by the Transformer encoder.

[0053] The method provided by the embodiment of the application combines the LSTM encoder and the Transformer encoder to predict the prestress of the anchor cable. The LSTM encoder can capture short-term dynamic fluctuations, and the Transformer encoder can mine global correlations. Therefore, the combination of the two encoders can predict the prestress of the anchor cable from two dimensions of short-term features and long-term features, solves the limitation of a single model, and is more accurate in prediction. In addition, in the embodiment of the application, the features in the input encoder further include anchor cable parameters, which are physical parameters. The incorporation of anchor cable physical parameters as constraints makes the prediction results conform to the mechanical principle and adapt to anchor cables and working conditions of different specifications.

[0054] In an optional embodiment, as described in step S101, the environmental data, the fan operation data, and the anchor cable parameters all include multiple types of parameters. In step S102, the step of establishing the time sequence feature matrix according to the environmental data, the fan operation data, and the anchor cable parameters specifically includes:

[0055] In step a1, the correlation coefficients between each type of parameter in the environmental data, the fan operation data, and the anchor cable parameters and the historical prestress are calculated according to the historical environmental data, the historical fan operation data, the historical anchor cable parameters, and the historical prestress corresponding to the target anchor cable.

[0056]

[0057] wherein, X i (t) is the value of the i-th parameter at time t, is the mean value of the i-th parameter in the current time window, F(t) is the prestress at time t, is the mean value of the prestress in the current time window.

[0058] In step a2, the parameters with correlation coefficients greater than a preset value are selected from the environmental data, the fan operation data, and the anchor cable parameters to construct the time sequence feature matrix.

[0059] In an optional embodiment, the time sequence feature matrix can be constructed according to the parameters with absolute values of the correlation coefficients greater than a preset value.

[0060] In an optional embodiment, the time-series feature matrix is determined by feature values of multiple types of features at different time points, the features including parameters with a correlation coefficient greater than a preset value, and composite features constructed from each type of parameters in the environmental data, the fan operation data, and the anchor cable parameters. That is, in the embodiment of the application, the parameters with a correlation coefficient greater than a preset value can be added to the time-series feature matrix as separate features, and in addition, one or more parameters in the environmental data, the fan operation data, and the anchor cable parameters can be further fused and calculated to obtain composite features that will affect the prestress of the anchor cable.

[0061] In an optional embodiment, the composite features include one or more of wind load intensity, temperature stress factor, vibration energy of the fan, and wet bulb temperature.

[0062] In an optional embodiment, the wind load intensity is used to represent the impact force of the wind speed on the anchor cable, and the greater the impact force, the greater the influence on the prestress. In the specific implementation, the wind load intensity can be calculated by the following formula:

[0063] W load =0.5·ρ·v(t) 2 ·C d ·S,

[0064] wherein, ρ is the air density, C d is the drag coefficient, S is the windward area, and v(t) is the wind speed.

[0065] In an optional embodiment, the temperature stress factor is used to represent the thermal stress caused by temperature change. In the specific implementation, the temperature stress factor can be calculated by the following formula:

[0066] T stress =E·α·(T(t)-T0)

[0067] wherein, α is the linear expansion coefficient, T0 is the initial temperature, T(t) is the environmental temperature collected at time t, and E is the elastic modulus of the anchor cable material.

[0068] In an optional embodiment, the vibration energy of the fan is the product of the vibration acceleration and the rotating speed, and can reflect the cumulative influence of the fan operation vibration on the anchor cable. In the specific implementation, the vibration energy can be calculated by the following formula:

[0069] A energy =A(t) 2 ·R(t)

[0070] wherein, A(t) is the vibration acceleration at time t, and R(t) is the rotating speed at time t.

[0071] In an optional embodiment, the wet-bulb temperature can be used to integrate the environmental comfort index of temperature and humidity and correlate with the material aging rate. The aging of the material will affect the prestressing of the anchor cable. In a specific embodiment, the wet-bulb temperature can be calculated using the following formula:

[0072] T wet =T(t)-(1-H(t) / 100)·(14.55+0.114T(t))

[0073] Wherein, T(t) is the ambient temperature at time t, and H(t) is the ambient humidity at time t.

[0074] In an optional embodiment, after executing step S101 to collect data at multiple time points, a sliding window may be used to extract historical statistical features and trend features. The historical statistical features may include the mean, variance, maximum, minimum, range, and fluctuation frequency of various parameters. Trend features may include the linear trend slope, first-order difference mean, and cumulative change.

[0075] In the embodiment of the present invention, time series features are extracted through correlation analysis and composite feature construction, static parameters and dynamic data are integrated, the influencing mechanism on prestress is fully reflected, and feature fragmentation is avoided.

[0076] In an optional embodiment, both the wind turbine operation data and the environmental data are time series data, and the changes in the anchor cable parameters at different times can be ignored. However, when constructing the time series feature matrix, the anchor cable parameters need to be expanded into a matrix that is consistent with other parameters, that is, the same anchor cable parameters are repeatedly input at each time step to ensure consistency with the time dimension of other time series features.

[0077] Because of the introduction of anchor parameters, LSTM will pass the cell state C when processing time series data. t and hidden state h t It also captures the temporal dependence of dynamic features, such as the hysteresis effect of wind speed trends on prestress, and the fundamental role of static features, such as the overall prestress level determined by the elastic modulus. For example, the cell state update formula In the example, static parameters such as anchor parameters are input through gate i t Influence the weight of new information so that the model can adjust its sensitivity to dynamic characteristics under different anchor specifications.

[0078] Because the anchor cable parameter is introduced, the self-attention mechanism calculates the correlation weight of the features at different time points, and the consistency of the same anchor cable prestress change at different times is kept, which avoids unreasonable prediction caused by excessive fluctuation of dynamic features. For example, when calculating the attention score Attention(Q, K, V), the static parameter is used as the reference feature, so that the model pays more attention to the deviation of dynamic features relative to the reference.

[0079] The static parameter such as the anchor cable parameter is transmitted to the full connection layer through the hidden state of the pre-sequence network, and finally affects the reference value of the prediction result. For example, the anchor cable with high elastic modulus has a higher predicted prestress average, and the dynamic feature affects the fluctuation amplitude.

[0080] Therefore, the anchor cable parameter is the core input feature of the prestress prediction model, which provides physical basic constraints for the model, distinguishes the inherent characteristics of different anchor cables, and cooperates with dynamic time sequence features to explain the prestress change rule. By extending the static feature to the time sequence dimension and splicing it with the dynamic feature, the input requirements of the LSTM-Transformer hybrid model can be seamlessly adapted, ensuring that the prediction result has both data-driven accuracy and physical mechanism rationality.

[0081] In an optional embodiment, in the step S104, the first hidden state sequence is input into the Transformer encoder, and the self-attention mechanism of the Transformer encoder is used to calculate the first hidden state sequence to obtain the second hidden state of each time node, forming a second hidden state sequence, comprising:

[0082] Step b1: mapping the first hidden state sequence to a query matrix, a key matrix, and a value matrix through linear transformation:

[0083] Q = H LSTM · W Q

[0084] K = H LSTM · W K

[0085] V = H LSTM · W V

[0086] Wherein, Q is the query matrix, K is the key matrix, V is the value matrix, H LSTM is the first hidden state sequence output by the LSTM encoder, W Q is the weight matrix of Q, W K is the weight matrix of K, W V is the weight matrix of V, and W Q , WK ,W V ∈R dlstm×dk ,d k for Q, K, V, d k is a divisor of d lstm , such as d k =d lstm / h is the dimension of input embedding.

[0087] Step b2, the query matrix, key matrix, value matrix are divided into multiple heads by using multi-head attention mechanism, and attention is calculated in parallel.

[0088] Step b3, the attention calculation results of each head are linearly transformed to obtain multi-head attention results.

[0089] Step b4, the multi-head attention is nonlinearly transformed to obtain enhanced features of the multi-head attention results.

[0090] Step b5, using residual connection and layer normalization, the multi-head attention results and enhanced features are calculated to obtain a second hidden state sequence.

[0091] In an optional embodiment, the LSTM encoder, the Transformer encoder and the fully connected layer constitute a complete model, and when the model is trained, the mean square error can be selected as the loss function:

[0092]

[0093] wherein, is the predicted prestress value, y t is the true prestress value, and N is the number of samples.

[0094] If the loss value calculated according to the predicted prestress does not meet the condition, the model parameters are updated by using the Adam optimization algorithm:

[0095] m t =β1m t-1 +(1-β1)g t

[0096]

[0097] wherein, m t and v t are the first moment estimate and the second moment estimate respectively, β1 and β2 are the exponential decay rates of the first moment and the second moment respectively, usually β1=0.9 and β2=0.999, g t is the gradient, η is the learning rate, ∈ is the numerical stability constant, usually ∈=10 -8 .

[0098] In an optional embodiment, after performing the above steps S1001-S105 multiple times using the trained model, the model can be evaluated by calculating the mean absolute error, the root mean square error, and the determination coefficient.

[0099] In an optional embodiment, the mean absolute error is:

[0100]

[0101] wherein, is the predicted prestress value, y t is the true prestress value, and N is the number of samples.

[0102] In an optional embodiment, the root mean square error is:

[0103]

[0104] wherein, is the predicted prestress value, y t is the true prestress value, and N is the number of samples.

[0105] In an optional embodiment, the determination coefficient (R 2 ) is:

[0106]

[0107] wherein, is the predicted prestress value, y t is the true prestress value, is the average value of the true prestress value, and N is the number of samples.

[0108] In an optional embodiment, the trend of the predicted anchor cable prestress can be analyzed. By observing the curve of the predicted value and the true value over time, it can be found that the model can better capture the overall trend of the anchor cable prestress, including the rising and falling stages.

[0109] In an optional embodiment, the anchor cable prestress predicted multiple times can be analyzed. By displaying the linear relationship between the predicted value and the true value through a scatter plot, R 2 indicates the true value variation that can be explained by the model. R 2 The larger the R

[0110] In this embodiment, an anchor cable prestress prediction device is also provided, which is used to implement the above embodiments and preferred embodiments, and will not be described again. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware can also be implemented and conceived.

[0111] The embodiment provides an anchor cable prestress prediction device, which comprises the following components as shown in the figure: Figure 2 The data acquisition module 201 is configured to acquire wind turbine operation data of a target anchor cable, environment data of an environment where the target anchor cable is located, and anchor cable parameters.

[0112] The time sequence feature matrix establishment module 202 is configured to establish a time sequence feature matrix according to the environment data, the wind turbine operation data, and the anchor cable parameters.

[0113] The LSTM encoder 203 is configured to input the time sequence feature matrix into an LSTM encoder, perform calculation on the time sequence feature matrix by using the LSTM encoder, obtain a first hidden state of each time node, and form a first hidden state sequence.

[0114] The Transformer encoder 204 is configured to perform calculation on the first hidden state sequence by using a self-attention mechanism of a Transformer encoder, obtain a second hidden state of each time node, and form a second hidden state sequence.

[0115] The prediction module 205 is configured to input the second hidden state sequence into a full connection layer, and obtain a prestress prediction value of the anchor cable.

[0116] Further function descriptions of the above modules and units are the same as those of the above corresponding embodiments, and thus are not described herein.

[0117] The anchor cable prestress prediction device in the embodiment is presented in the form of a functional unit, and the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0118] The embodiment of the present application further provides a computer device with the above anchor cable prestress prediction device.

[0119] Figure 2 The computer device is shown in the figure.

[0120] Please refer to Figure 3 , Figure 3 which is a structural schematic diagram of a computer device provided in an optional embodiment of the present application. Figure 3 ​As shown, the computer device includes one or more processors 10, memory 20, and interfaces 50 for external devices such as modems and network interfaces. The one or more processors 10 can be implemented as one or more central processing units (CPUs), one or more microprocessors, microcontrollers, digital signal processors, specialized processors or controller, or one or more processors of any equivalent known in the art. In some embodiments, the one or more processors 10 can be implemented as a combination of one or more of the above physical processors and / or one or more software or firmware modules. The software or firmware can be stored in a non-transitory computer readable medium such as memory 20, a storage device or any equivalent medium known in the art. Figure 3 The processor 10 is taken as an example in the embodiments.

[0121] The processor 10 can be a central processing unit, a network processing unit or a combination thereof. The processor 10 can further include a hardware chip. The hardware chip can be an application specific integrated circuit, a programmable logic device or a combination thereof. The programmable logic device can be a complex programmable logic device, a field programmable logic device, a generic array logic or any combination thereof.

[0122] The memory 20 stores instructions that are executable by the at least one processor 10, so as to enable the at least one processor 10 to perform the method shown in the above embodiments.

[0123] The memory 20 can include a program storage area and a data storage area. The program storage area can store an operating system, application programs required by at least one function; and the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can further include a non-transitory memory such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some alternative embodiments, the memory 20 can optionally include a memory that is remotely arranged with respect to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0124] The memory 20 can include a volatile memory such as a random access memory, and can also include a non-volatile memory such as a flash memory, a hard disk or a solid state disk, and can further include a combination of the above kinds of memories.

[0125] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 can be connected through a bus or other means,Figure 3 The bus connection is taken as an example.

[0126] The input device 30 can receive inputted digital or character information, and generate key signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touchpad, a pointing stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), a tactile feedback device (e.g., a vibration motor), etc. The display device includes, but is not limited to, a liquid crystal display, a light-emitting diode, a display, and a plasma display. In some alternative embodiments, the display device can be a touch screen.

[0127] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded from a network and stored in a local storage medium, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer code, which, when accessed and executed by the computer, the processor, or the hardware, implements the method shown in the above embodiments.

[0128] Part of the present application can be applied as a computer program product, for example, computer program instructions, when executed by a computer, the operation of the computer can invoke or provide the method and / or technical solutions according to the present application. Those skilled in the art should understand that the form of computer program instructions in computer readable medium includes but is not limited to source file, executable file, installation package file, etc. Correspondingly, the way of computer program instructions executed by computer includes but is not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Here, the computer readable medium can be any available computer readable storage medium or communication medium accessible to the computer.

[0129] While embodiments of the application have been described in connection with the preferred embodiments of the various figures, those of ordinary skill in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the application, and that such modifications and changes fall within the scope of the appended claims.

Claims

1. A method for predicting the pre-stress of an anchor cable, characterized in that, The method comprises the steps of: obtaining wind turbine operation data of a target anchor cable, environmental data of an environment in which the target anchor cable is located, and anchor cable parameters; establishing a time sequence feature matrix according to the environmental data, the wind turbine operation data, and the anchor cable parameters; inputting the time sequence feature matrix into an LSTM encoder, calculating a first hidden state of each time node through the LSTM encoder, and forming a first hidden state sequence; inputting the first hidden state sequence into a Transformer encoder, calculating the first hidden state sequence through a self-attention mechanism of the Transformer encoder, obtaining a second hidden state of each time node, and forming a second hidden state sequence; inputting the second hidden state sequence into a fully connected layer to obtain a prestress prediction value of the anchor cable.

2. The method of claim 1, wherein: the environmental data, the wind turbine operation data, and the anchor cable parameters each comprise multiple types of parameters; establishing the time sequence feature matrix according to the environmental data, the wind turbine operation data, and the anchor cable parameters comprises: calculating a correlation coefficient between each type of parameter in the environmental data, the wind turbine operation data, and the anchor cable parameters and a historical prestress corresponding to the target anchor cable according to the historical environmental data, the historical wind turbine operation data, the historical anchor cable parameters, and the historical prestress; selecting parameters with a correlation coefficient greater than a preset value from the environmental data, the wind turbine operation data, and the anchor cable parameters to construct the time sequence feature matrix.

3. The method of claim 2, wherein: the correlation coefficient is calculated by the following formula: wherein X i (t) is the value of the i-th parameter at time t, is the mean value of the i-th parameter within the current time window, F(t) is the prestress at time t, is the mean value of the prestress within the current time window.

4. The method of claim 2, wherein: the time sequence feature matrix is determined by feature values of multiple types of features at different time points, the features include the parameters with a correlation coefficient greater than a preset value, and a composite feature constructed according to each type of parameter in the environmental data, the wind turbine operation data, and the anchor cable parameters.

5. The method of claim 4, wherein: the composite feature includes one or more of wind load intensity, temperature stress factor, vibration energy of the wind turbine, and wet-bulb temperature.

6. The method of claim 1, wherein: the inputting the first hidden state sequence into the Transformer encoder, calculating the first hidden state sequence through the self-attention mechanism of the Transformer encoder to obtain a second hidden state of each time node, and forming a second hidden state sequence comprises: mapping the first hidden state sequence into a query matrix, a key matrix, and a value matrix through linear transformation; dividing the query matrix, the key matrix, and the value matrix into multiple heads through a multi-head attention mechanism, and calculating attention in parallel; performing linear transformation on the attention calculation results of each head to obtain a multi-head attention result; performing nonlinear transformation on the multi-head attention to obtain enhanced features of the multi-head attention result; calculating the multi-head attention result and the enhanced features through residual connection and layer normalization to obtain the second hidden state sequence.

7. An anchor cable prestress prediction device, characterized by, The method comprises the steps of: The data acquisition module is configured to acquire wind turbine operation data of a target anchor cable, environment data of an environment in which the target anchor cable is located, and anchor cable parameters; The time sequence feature matrix establishment module is configured to establish a time sequence feature matrix according to the environment data, the wind turbine operation data, and the anchor cable parameters; The LSTM encoder is configured to input the time sequence feature matrix into an LSTM encoder, and to calculate the time sequence feature matrix by using the LSTM encoder to obtain a first hidden state of each time node and form a first hidden state sequence; The Transformer encoder is configured to calculate the first hidden state sequence by using a self-attention mechanism of a Transformer encoder to obtain a second hidden state of each time node and form a second hidden state sequence; The prediction module is configured to input the second hidden state sequence into a fully connected layer to obtain a predicted value of the anchor cable prestress.

8. A computer device, comprising: The method comprises: A memory and a processor are communicatively connected, and the memory stores computer instructions. The processor executes the computer instructions to perform the anchor cable prestress prediction method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing a computer to perform the anchor cable prestress prediction method according to any one of claims 1 to 6.

10. A computer program product, characterised in that, The computer instructions are configured to cause a computer to perform the anchor cable prestress prediction method according to any one of claims 1 to 6.

Images