Wind power plant unattended system based on deep learning

Abstract

The invention relates to the technical field of wind power plant unattended operation, in particular to a wind power plant unattended operation system based on deep learning, and the system comprises an environment collection unit which obtains the environment data of a wind power plant; the wind power plant state prediction unit is used for extracting historical environment data of a wind power plant, dividing the historical environment data into a training sample and a test sample, constructing a wind power plant state prediction model based on an MLP neural network, inputting the training sample into the wind power plant state prediction model for calculation, and calculating and predicting a loss value between a training result and a test set; iteratively updating parameters of the wind power plant state prediction model until training of the wind power plant state prediction model is completed; and the model optimization unit is used for optimizing the wind power plant state prediction model by taking balance among the safety, the risk, the environmental adaptability, the power generation efficiency and the maintenance efficiency of the wind power plant as an optimization target, outputting an optimal wind power plant state prediction model and executing unattended operation of the wind power plant based on the optimal wind power plant state prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned wind farms, and in particular to an unmanned wind farm system based on deep learning. Background Art

[0002] With the rapid development of renewable energy, the scale and complexity of wind farms are increasing. To ensure the safe and efficient operation of wind farms, it is necessary to monitor and predict the status of wind farms in real time. At present, traditional wind farm monitoring technology mainly relies on manual experience and simple threshold judgment, and has the following disadvantages: 1. Manual monitoring is inefficient and difficult to respond to complex environmental changes in real time; 2. It is difficult to strike a balance between safety, power generation efficiency and maintenance efficiency. Traditional monitoring technology is not only difficult to meet the needs of modern wind farms, but also seriously restricts the intelligent development of wind farms and the improvement of operational efficiency. Based on this, in response to the above problems, we designed an unmanned wind farm system based on deep learning. Summary of the invention

[0003] The purpose of the present invention is to provide an unmanned wind farm system based on deep learning, which uses an MLP neural network to build an intelligent prediction model, can accurately predict the state of the wind farm and realize unmanned operation. In addition, the present invention achieves a balance between safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency through multi-objective optimization, can significantly improve the operating efficiency and reliability of the wind farm, reduce labor costs, and provide strong support for the intelligent management of wind farms.

[0004] The embodiments of the present invention are implemented by the following technical solutions:

[0005] A wind farm unmanned system based on deep learning, comprising:

[0006] Environmental collection unit, which obtains environmental data of the wind farm;

[0007] The wind farm state prediction unit extracts the historical environmental data of the wind farm, divides it into training samples and test samples, builds a wind farm state prediction model based on the MLP neural network, inputs the training samples into the wind farm state prediction model for calculation, calculates the loss value between the prediction training results and the test set, and iteratively updates the parameters of the wind farm state prediction model until the training of the wind farm state prediction model is completed;

[0008] The model optimization unit takes the balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm as the optimization goal, optimizes the trained wind farm state prediction model, outputs the optimal wind farm state prediction model, and executes unmanned operation of the wind farm based on the optimal wind farm state prediction model.

[0009] Optionally, the environmental data of the wind farm includes: wind speed data, wind direction data, temperature data and humidity data.

[0010] Optionally, the wind farm state prediction model is composed of an input layer, a hidden layer and an output layer, wherein the calculation formula of the input layer is: X = [v, θ, T, H]; wherein v is wind speed, θ is wind direction, T is temperature, and H is humidity; the hidden layer is composed of a ReLu function, and the calculation formula of the output layer is: Y = [y 1 ,y 2 ,y 3 ]; where y 1 ,y 2 ,y 3 They represent the probability that the wind farm is in normal, warning and dangerous states respectively.

[0011] Optionally, the loss value between the calculated prediction training result and the test set is calculated using the following formula:

[0012]

[0013] Among them, L is the loss function, M is the number of samples, 3 is the number of categories, corresponding to the three states of normal, warning and danger, y mk is the true label of the mth sample belonging to the kth state, The model predicts the probability that the mth sample belongs to the kth state, that is, y 1 ,y 2 ,y 3 .

[0014] Optionally, the optimization goal is to strike a balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm, and the calculation formula is:

[0015]

[0016] Among them, J is the objective function, which is defined as the comprehensive performance index of the wind farm, N is the total number of wind turbines, S i is the safety factor of the ith fan, R i is the risk index of the i-th wind turbine, E i is the environmental adaptability index of the i-th fan, P i is the power generation of the i-th wind turbine, D i is the fault duration of the ith fan, w 1 ,w 2 ,w 3 ,w 4 ,w 5 is the weight.

[0017] Optionally, the safety factor is calculated as follows:

[0018]

[0019] Among them, T normal is the normal operation time, T total is the total running time.

[0020] Optionally, the risk index is calculated as follows:

[0021] R i =αy 2 +βy 3

[0022] Among them, α and β are the weights of warning and danger states.

[0023] Optionally, the environmental adaptability index is calculated as follows:

[0024]

[0025] Among them, k is the wind speed sensitivity parameter, v rated is the rated wind speed, T optimal is the optimal operating temperature, H max is the maximum allowable humidity.

[0026] Optionally, the power generation is calculated as follows:

[0027]

[0028] Among them, η is the power generation efficiency, v is the air density, A is the wind sweeping area of ​​the wind wheel, C p is the power factor, T operation The actual running time.

[0029] Optionally, the fault duration is calculated as follows:

[0030] D i =T total -T operation

[0031] Among them, T total is the total running time, T operation The actual running time.

[0032] The technical solution of the embodiment of the present invention has at least the following advantages and beneficial effects:

[0033] The embodiment of the present invention utilizes the MLP neural network to construct an intelligent prediction model, which can accurately predict the status of a wind farm and realize unmanned operation. In addition, the present invention achieves a balance between safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency through multi-objective optimization, which can significantly improve the operating efficiency and reliability of the wind farm, reduce labor costs, and provide strong support for the intelligent management of wind farms. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A schematic diagram of the principle of an unmanned wind farm system based on deep learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0036] like Figure 1 As shown, the present invention provides one embodiment: an unmanned wind farm system based on deep learning, comprising:

[0037] Environmental collection unit, which obtains environmental data of the wind farm;

[0038] The wind farm state prediction unit extracts the historical environmental data of the wind farm, divides it into training samples and test samples, builds a wind farm state prediction model based on the MLP neural network, inputs the training samples into the wind farm state prediction model for calculation, calculates the loss value between the prediction training results and the test set, and iteratively updates the parameters of the wind farm state prediction model until the training of the wind farm state prediction model is completed;

[0039] The model optimization unit takes the balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm as the optimization goal, optimizes the trained wind farm state prediction model, outputs the optimal wind farm state prediction model, and executes unmanned operation of the wind farm based on the optimal wind farm state prediction model.

[0040] In this embodiment, the environmental data of the wind farm includes: wind speed data, wind direction data, temperature data and humidity data.

[0041] In the specific application of this embodiment, the wind farm state prediction model is composed of an input layer, a hidden layer and an output layer, wherein the calculation formula of the input layer is: X = [v, θ, T, H]; wherein v is wind speed, θ is wind direction, T is temperature, and H is humidity; the hidden layer is composed of ReLu function, and the calculation formula of the output layer is: Y = [y 1 ,y 2 ,y 3 ]; where y 1 ,y 2 ,y 3 They represent the probability that the wind farm is in normal, warning and dangerous states respectively.

[0042] The calculation formula for the loss value between the training result and the test set is:

[0043]

[0044] Among them, L is the loss function, M is the number of samples, 3 is the number of categories, corresponding to the three states of normal, warning and danger, y mk is the true label of the mth sample belonging to the kth state, The model predicts the probability that the mth sample belongs to the kth state, that is, y 1 ,y 2 ,y 3 .

[0045] In the implementation, this embodiment is provided with corresponding y 1 ,y 2 ,y 3 The warning state execution logic, where y 1 The state continues to execute the data collection of the wind farm without additional processing; 2 The wind farm issues an alarm and dispatches staff to handle the situation; 3 The wind farm in this state implements safety protection mechanisms, including but not limited to shutdown maintenance and other measures.

[0046] Specifically, the optimization goal is to strike a balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm, and the calculation formula is:

[0047]

[0048] Among them, J is the objective function, which is defined as the comprehensive performance index of the wind farm, N is the total number of wind turbines, S i is the safety factor of the ith fan, R i is the risk index of the i-th wind turbine, E i is the environmental adaptability index of the i-th fan, P i is the power generation of the i-th wind turbine, Di is the fault duration of the ith fan, w 1 ,w 2 ,w 3 ,w 4 ,w 5 is the weight.

[0049] Wherein, the safety factor is calculated as follows:

[0050]

[0051] Among them, T normal is the normal operation time, T total is the total running time.

[0052] The risk index is calculated as follows:

[0053] R i =αy 2 +βy 3

[0054] Among them, α and β are the weights of warning and danger states.

[0055] The environmental adaptability index is calculated as follows:

[0056]

[0057] Among them, k is the wind speed sensitivity parameter, v rated is the rated wind speed, T optimal is the optimal operating temperature, H max is the maximum allowable humidity.

[0058] The power generation is calculated as follows:

[0059]

[0060] Among them, η is the power generation efficiency, v is the air density, A is the wind sweeping area of ​​the wind wheel, C p is the power factor, T operation The actual running time.

[0061] The fault duration is calculated as follows:

[0062] D i =T total -T operation

[0063] Among them, T total is the total running time, T operation The actual running time.

[0064] In the above implementation, the objective function designed in this embodiment performs a weighted summation of the safety, risk and environmental adaptability of all wind turbines, and characterizes the gain of power generation with logarithmic and exponential functions, reflects the diminishing marginal effect, penalizes the fault duration, emphasizes the importance of rapid repair, and can comprehensively evaluate the performance of the wind farm while taking into account safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency, thereby achieving efficient and safe operation of the wind farm.

[0065] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A wind farm unattended system based on deep learning, characterized in that: include: Environmental collection unit, which obtains environmental data of the wind farm; The wind farm state prediction unit extracts the historical environmental data of the wind farm, divides it into training samples and test samples, builds a wind farm state prediction model based on the MLP neural network, inputs the training samples into the wind farm state prediction model for calculation, calculates the loss value between the prediction training results and the test set, and iteratively updates the parameters of the wind farm state prediction model until the training of the wind farm state prediction model is completed; The model optimization unit takes the balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm as the optimization goal, optimizes the trained wind farm state prediction model, outputs the optimal wind farm state prediction model, and executes unmanned operation of the wind farm based on the optimal wind farm state prediction model.

2. The deep learning-based unattended wind farm system according to claim 1, characterized in that: The environmental data of the wind farm includes: wind speed data, wind direction data, temperature data and humidity data.

3. The deep learning-based unmanned wind farm system according to claim 2, characterized in that: The wind farm state prediction model consists of an input layer, a hidden layer and an output layer, wherein the calculation formula of the input layer is: X=[v,θ,T,H]; wherein v is wind speed, θ is wind direction, T is temperature, and H is humidity; the hidden layer is composed of ReLu functions, and the calculation formula of the output layer is: Y=[y1,y2,y3]; wherein y1, y2, y3 represent the probabilities of the wind farm being in normal, warning and dangerous states, respectively.

4. The deep learning-based unmanned wind farm system according to claim 3, characterized in that: The calculation formula for the loss value between the training result and the test set is: Among them, L is the loss function, M is the number of samples, 3 is the number of categories, corresponding to the three states of normal, warning and danger, y mk is the true label of the mth sample belonging to the kth state, The model predicts the probability that the mth sample belongs to the kth state, that is, y1, y2, y3.

5. The deep learning-based unattended wind farm system according to claim 4, characterized in that: The optimization goal is to strike a balance between the safety, risk, environmental adaptability, power generation efficiency and maintenance efficiency of the wind farm, and the calculation formula is: Among them, J is the objective function, which is defined as the comprehensive performance index of the wind farm, N is the total number of wind turbines, S i is the safety factor of the ith fan, R i is the risk index of the i-th wind turbine, E i is the environmental adaptability index of the i-th fan, P i is the power generation of the i-th wind turbine, D i is the fault duration of the i-th wind turbine, and w1,w2,w3,w4,w5 are weights.

6. The deep learning-based unmanned wind farm system according to claim 5, characterized in that: The safety factor is calculated as follows: Among them, T normal is the normal operation time, T total is the total running time.

7. The deep learning-based unattended wind farm system according to claim 5, characterized in that: The risk index is calculated as follows: R i =αy2+βy3 Among them, α and β are the weights of warning and danger states.

8. The deep learning-based unattended wind farm system according to claim 5, characterized in that: The environmental adaptability index is calculated as follows: Among them, k is the wind speed sensitivity parameter, v rated is the rated wind speed, T optimal is the optimal operating temperature, H max is the maximum allowable humidity.

9. The deep learning-based unmanned wind farm system according to claim 5, characterized in that: The power generation is calculated as follows: Among them, η is the power generation efficiency, ρ is the air density, A is the wind sweeping area of ​​the wind wheel, C p is the power factor, T operation The actual running time.

10. The deep learning-based unattended wind farm system according to claim 5, characterized in that: The fault duration is calculated as follows: D i =T total -T operation Among them, T total is the total running time, T operation The actual running time.

Images