Generated power prediction method
By combining environmental data and equipment parameters, a power generation prediction method constructed using LSTM and GA-BP neural network models solves the problems of accuracy and robustness in power generation prediction in wind, solar and energy storage systems, achieving higher accuracy prediction results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-27
- Publication Date
- 2026-03-17
AI Technical Summary
Existing technologies struggle to achieve high-precision and robust power generation prediction for wind, solar, and energy storage systems, especially when the intermittency and volatility of wind and solar power generation impact grid stability.
By acquiring environmental data and equipment parameters, and combining the Long Short-Term Memory (LSTM) network model with the Genetic Algorithm-Backpropagation (GA-BP) neural network model, a power generation prediction model is constructed. The model is then weighted and summed, taking into account both environmental data and equipment characteristics, and the weights and biases are optimized to improve prediction accuracy.
It improves the prediction accuracy of power generation in wind, solar and energy storage systems and the training efficiency of models, enhances the generalization ability of models, and improves the grid's ability to absorb new energy.
Smart Images
Figure CN121688807A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of power technology, and in particular to a method for predicting power generation. Background Technology
[0002] As the scale of wind and solar power generation in wind-solar-storage systems continues to increase, the intermittency and volatility of wind and solar power generation are having a growing impact on the stable operation of the power grid. To improve the grid's capacity to absorb new energy sources, the power output of wind and solar power can be predicted.
[0003] In related technologies, the future power generation of wind and solar power can be predicted by analyzing historical power generation data and environmental factors. However, this requires high-quality historical data to ensure prediction accuracy, making it difficult to achieve high-precision predictions and exhibiting low model robustness. Improving the prediction accuracy of power generation in wind, solar, and energy storage systems is a pressing technical problem that needs to be solved. Summary of the Invention
[0004] This application provides a power generation prediction method to improve the prediction accuracy of power generation in wind, solar and energy storage systems.
[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, a method for predicting power generation is provided. This method includes: acquiring prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters; the prediction environmental data includes: air density, wind speed at the hub height of the wind power generation equipment, irradiance, temperature coefficient, and photovoltaic module backsheet temperature; the wind power generation equipment parameters include: rotor swept area and wind energy utilization coefficient; the photovoltaic power generation equipment parameters include: number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, and photovoltaic panel area; based on the prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters, a first predicted power generation is determined; the first predicted power generation is the sum of the first predicted wind power generation and the first predicted photovoltaic power generation; the first predicted wind power generation is determined by half the product of the wind energy utilization coefficient, the cube of the wind speed at the hub height, and the rotor swept area; the first predicted photovoltaic power generation is determined by the product of the number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, photovoltaic panel area, irradiance, and temperature factor; the temperature factor W = 1 - s(Tc - 25). ; s represents the temperature coefficient; Tc represents the backsheet temperature of the photovoltaic module; the predicted environmental data is input into the trained power generation prediction model to obtain the second predicted power generation; the power generation prediction model is established based on the combination of the Long Short Time Memory Network (LSTM) model and the Genetic Algorithm-Backpropagation (GA-BP) neural network model; the first predicted power generation and the second predicted power generation are weighted and summed to obtain the target predicted power generation.
[0006] Optionally, the method further includes: constructing an LSTM neural network model; using the weights of the LSTM neural network model as initial weights to construct a BP neural network model; optimizing the initial weights and biases of the BP neural network model through selection, crossover, and mutation operations in a genetic algorithm to obtain optimized weights and biases; and using the optimized weights and biases as the weights and biases of the LSTM neural network model to obtain a power generation prediction model.
[0007] Optionally, the method further includes: acquiring multiple sets of sample data; each set of sample data includes historical power generation data and corresponding historical environmental data; performing preprocessing operations on the multiple sets of sample data; the preprocessing operations include normalization, smoothing, and removal of outliers; and using the multiple sets of sample data to train the power generation prediction model to obtain the trained power generation prediction model.
[0008] Optionally, the power generation prediction model can be trained using multiple sets of sample data to obtain the trained power generation prediction model. This includes: dividing the preprocessed sample data into a training set and a test set; training the power generation prediction model using the training set and adjusting the model parameters until a preset stopping condition is met; validating the power generation prediction model using the test set and optimizing the power generation prediction model based on the validation results to obtain the trained power generation prediction model.
[0009] Optionally, the power generation prediction model is optimized based on the validation results, including: gradually increasing the number of LSTM layers in the power generation prediction model and determining the root mean square error (RMSE) corresponding to different LSTM layer numbers; the RMSE is determined based on the error between the predicted value and the actual observed value of the power generation prediction model; if the RMSE does not decrease continuously after gradually increasing the number of LSTM layers by N times, the number of LSTM layers corresponding to the N-times increase is determined as the target number of LSTM layers; based on the target number of LSTM layers in the power generation prediction model, the number of neurons in the power generation prediction model is gradually increased, and the RMSE corresponding to different number of neurons is determined; if the RMSE does not decrease continuously after gradually increasing the number of LSTM layers by M times, the number of neurons corresponding to the M-times increase is determined as the target number of neurons; the number of LSTM layers in the power generation prediction model is adjusted to the target number of LSTM layers, and the number of neurons in the power generation prediction model is adjusted to the target number of neurons, thereby optimizing the power generation prediction model.
[0010] Based on the technical solution provided in this application, a first predicted power generation is determined by predicting environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters. This allows for a comprehensive consideration of environmental data, the physical characteristics of wind power generation equipment, and the physical characteristics of photovoltaic power generation equipment, resulting in a highly accurate first predicted power generation. Furthermore, a second predicted power generation is determined based on the predicted environmental data and a power generation prediction model established using a combination of a Long Short-Term Memory (LSTM) network model and a Genetic Algorithm-Backpropagation (GA-BP) neural network model. Due to the global search capability, gradient-free nature, and flexible fitness function design of the GA-BP neural network model, it can provide better initial weights and thresholds for the LSTM, thereby improving the model's training efficiency and generalization ability. Finally, the first and second predicted power generation are weighted and summed to obtain the target predicted power generation. This combination of multiple highly accurate predicted power generation methods further improves the prediction accuracy of power generation in the wind-solar-storage system.
[0011] Secondly, a power generation prediction device is provided, comprising: an acquisition unit and a processing unit; the acquisition unit is used to acquire prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters; the prediction environmental data includes: air density, wind speed at the hub height of the wind power generation equipment, irradiance, temperature coefficient, and photovoltaic module backsheet temperature; the wind power generation equipment parameters include: rotor swept area and wind energy utilization coefficient; the photovoltaic power generation equipment parameters include: number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, and photovoltaic panel area; the processing unit is used to... The system predicts environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters to determine the first predicted power generation. The first predicted power generation is the sum of the first predicted wind power generation and the first predicted photovoltaic power generation. The first predicted wind power generation is determined by half the product of the wind energy utilization coefficient, the cube of the wind speed at the hub height, and the swept area of the impeller. The first predicted photovoltaic power generation is determined by the product of the number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, photovoltaic panel area, irradiance, and temperature factor. The temperature factor W = 1 - s(Tc - 25), where s represents the temperature coefficient and Tc represents the backsheet temperature of the photovoltaic module. The processing unit also inputs the predicted environmental data into the trained power generation prediction model to obtain the second predicted power generation. The power generation prediction model is established based on a combination of a Long Short Time Memory (LSTM) network model and a Genetic Algorithm-Backpropagation (GA-BP) neural network model. The processing unit also performs a weighted summation of the first and second predicted power generation to obtain the target predicted power generation.
[0012] Optionally, the processing unit is also used to construct an LSTM neural network model; the processing unit is also used to construct a BP neural network model using the weights of the LSTM neural network model as initial weights; the processing unit is also used to optimize the initial weights and biases of the BP neural network model through selection, crossover, and mutation operations in a genetic algorithm to obtain optimized weights and biases; the processing unit is also used to use the optimized weights and biases as the weights and biases of the LSTM neural network model to obtain a power generation prediction model.
[0013] Optionally, the acquisition unit is also used to acquire multiple sets of sample data; each set of sample data includes historical power generation data and corresponding historical environmental data; the processing unit is also used to perform preprocessing operations on the multiple sets of sample data; the preprocessing operations include normalization, smoothing, and removal of outliers; the processing unit is also used to train the power generation prediction model using the multiple sets of sample data to obtain the trained power generation prediction model.
[0014] Optionally, the processing unit is specifically used for: dividing the preprocessed sample data into a training set and a test set; training the power generation prediction model using the training set, adjusting the model parameters until a preset stopping condition is met; validating the power generation prediction model using the test set, and optimizing the power generation prediction model based on the validation results to obtain the trained power generation prediction model.
[0015] Optionally, the processing unit is further configured to: progressively increase the number of LSTM layers in the power generation prediction model and determine the root mean square error (RMSE) corresponding to different LSTM layer numbers; the RMSE is determined based on the error between the predicted value and the actual observed value of the power generation prediction model; if the RMSE does not decrease continuously after progressively increasing the number of layers N times, the number of LSTM layers corresponding to the progressive increase N times is determined as the target number of LSTM layers; based on the target number of LSTM layers in the power generation prediction model, progressively increase the number of neurons in the power generation prediction model and determine the RMSE corresponding to different number of neurons; if the RMSE does not decrease continuously after progressively increasing the number of layers M times, the number of neurons corresponding to the progressive increase M times is determined as the target number of neurons; adjust the number of LSTM layers in the power generation prediction model to the target number of LSTM layers and adjust the number of neurons in the power generation prediction model to the target number of neurons to optimize the power generation prediction model.
[0016] Thirdly, a power generation prediction device is provided, which can realize the functions performed by the power generation prediction device in the above aspects or possible designs. The functions can be implemented by hardware. For example, in one possible design, the power generation prediction device may include a processor and a communication interface. The processor can be used to support the power generation prediction device in realizing the functions involved in the first aspect or any possible design of the first aspect.
[0017] In another possible design, the power generation prediction device may further include a memory for storing necessary computer execution instructions and data. When the power generation prediction device is running, the processor executes the computer execution instructions stored in the memory to cause the power generation prediction device to perform the first aspect or any of the possible power generation prediction methods described above.
[0018] Fourthly, a computer-readable storage medium is provided, which may be a readable non-volatile storage medium storing computer instructions or programs that, when executed on a computer, enable the computer to perform the power generation prediction method described in the first aspect or any of the possible methods described in the first aspect.
[0019] Fifthly, a computer program product containing instructions is provided, which, when run on a computer, enables the computer to execute the power generation prediction method of the first aspect or any possible design of the above aspects.
[0020] A sixth aspect provides an electronic device comprising one or more processors and one or more memories. The one or more memories are coupled to the one or more processors, and the one or more memories are used to store computer program code, including computer instructions, which, when executed by the one or more processors, cause the electronic device to perform a power generation prediction method as described in the first aspect or any possible design of the first aspect.
[0021] In a seventh aspect, a chip system is provided, including a processor and a communication interface, which can be used to implement the functions performed by the power generation prediction device in the first aspect or any possible design of the first aspect. In one possible design, the chip system further includes a memory for storing program instructions and / or data. The chip system can be composed of chips or may include chips and other discrete devices, without limitation. Attached Figure Description
[0022] Figure 1 A flowchart illustrating a power generation prediction method provided in an embodiment of this application; Figure 2 This is a schematic diagram of a power generation prediction device provided in an embodiment of this application. Detailed Implementation
[0023] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.
[0024] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.
[0025] It should also be understood that the term "comprising" indicates the presence of the described feature, whole, step, operation, element and / or component, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements and / or components.
[0026] As the scale of wind and solar power generation in wind-solar-storage systems continues to increase, the intermittency and volatility of wind and solar power generation are having a growing impact on the stable operation of the power grid. To improve the grid's capacity to absorb new energy sources, the power output of wind and solar power can be predicted.
[0027] In related technologies, the future power generation of wind and solar power can be predicted by analyzing historical power generation data and environmental factors. However, this requires high-quality historical data to ensure prediction accuracy, making it difficult to achieve high-precision predictions and exhibiting low model robustness. Improving the prediction accuracy of power generation in wind, solar, and energy storage systems is a pressing technical problem that needs to be solved.
[0028] In view of this, this application provides a method for predicting power generation, including: acquiring prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters; the prediction environmental data includes: air density, wind speed at the hub height of the wind power generation equipment, irradiance, temperature coefficient, and photovoltaic module backsheet temperature; the wind power generation equipment parameters include: rotor swept area and wind energy utilization coefficient; the photovoltaic power generation equipment parameters include: number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, and photovoltaic panel area; determining a first predicted power generation based on the prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters; the first predicted power generation is the sum of the first predicted wind power generation and the first predicted photovoltaic power generation; the first predicted wind power generation is determined based on half the product of the wind energy utilization coefficient, the cube of the wind speed at the hub height, and the rotor swept area; the first predicted photovoltaic power generation is determined based on the product of the number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, photovoltaic panel area, irradiance, and temperature factor; the temperature factor W = 1 - s(Tc - 25). ; s represents the temperature coefficient; Tc represents the backsheet temperature of the photovoltaic module; the predicted environmental data is input into the trained power generation prediction model to obtain the second predicted power generation; the power generation prediction model is established based on the combination of the Long Short Time Memory Network (LSTM) model and the Genetic Algorithm-Backpropagation (GA-BP) neural network model; the first predicted power generation and the second predicted power generation are weighted and summed to obtain the predicted power generation.
[0029] The methods provided in the embodiments of this application will now be described in detail with reference to the accompanying drawings.
[0030] Figure 1This is a flowchart illustrating a power generation prediction method provided in an embodiment of this application, as shown below. Figure 1 As shown, the method includes the following steps S301-S304: S301. Obtain predicted environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters.
[0031] The predicted environmental data includes: air density, wind speed at the hub height of the wind turbine, irradiance, temperature coefficient, and photovoltaic module backsheet temperature. Wind turbine parameters include: rotor swept area and wind energy utilization coefficient; photovoltaic parameters include: number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, and photovoltaic panel area.
[0032] As one possible implementation, the power generation prediction device can periodically obtain future meteorological data from meteorological websites and determine the predicted environmental data based on the future meteorological data; and obtain the parameters of wind power generation equipment and photovoltaic power generation equipment through control commands.
[0033] It should be noted that control commands can refer to instructions generated in response to control operations performed by the operator. For example, control commands can be instructions input by the operator through the input device (such as a keyboard) of the power generation prediction device.
[0034] S302. Determine the first predicted power generation capacity based on the predicted environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters.
[0035] The first predicted power generation is the sum of the first predicted wind power generation and the first predicted photovoltaic power generation. The first predicted wind power generation is determined by half the product of the wind energy utilization coefficient, the cube of the wind speed at the hub height, and the swept area of the impeller. The first predicted photovoltaic power generation is determined by the product of the number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, photovoltaic panel area, irradiance, and temperature factor. The temperature factor W = 1 - s(Tc - 25), where s represents the temperature coefficient and Tc represents the backsheet temperature of the photovoltaic module.
[0036] As one possible implementation, the power generation prediction device can input predicted environmental data and wind power generation equipment parameters into Formula 1 to determine the first predicted wind power generation. It can also input predicted environmental data and photovoltaic power generation equipment parameters into Formula 2 to determine the first predicted photovoltaic power generation. The sum of the first and second predicted photovoltaic power generation is then determined as the first predicted power generation.
[0037] In one example, Formula 1 can be: P1= (Cp*ρ*v 3 *A1) / 2 Formula 1 Where P1 represents the first predicted wind power generation capacity. Cp represents the wind energy utilization coefficient. ρ represents the air density. v represents the wind speed at the hub height. A1 represents the rotor swept area.
[0038] In one example, Formula 2 can be: P2=N*n1*n2*n3*A2*Ra*[1-s(Tc-25)]Formula 2 Where P2 represents photovoltaic output power; N represents the number of photovoltaic modules; n1 represents photovoltaic conversion efficiency; n2 represents tracking efficiency (also known as peak power point); n3 represents inverter efficiency; A2 represents photovoltaic panel area; Ra represents photovoltaic system irradiance; s represents temperature coefficient; and Tc represents photovoltaic module backsheet temperature.
[0039] Furthermore, the first predicted power generation can also be expressed as P1+P2.
[0040] S303. Input the predicted environmental data into the trained power generation prediction model to obtain the second predicted power generation.
[0041] Among them, the power generation prediction model is based on the combination of the Long Short Time Memory (LSTM) network model and the Genetic Algorithm-Backpropagation (GA-BP) neural network model.
[0042] As one possible implementation, the power generation prediction device can input the predicted environmental data into the trained power generation prediction model. After processing the predicted environmental data, the trained power generation prediction model can input the second predicted power generation.
[0043] S304. The first predicted power generation and the second predicted power generation are weighted and summed to obtain the target predicted power generation.
[0044] As one possible implementation, the power generation prediction device can pre-configure a first weight for the first predicted power generation and a second weight for the second predicted power generation. Furthermore, the product of the first weight and the first predicted power generation, and the product of the second weight and the second predicted power generation are summed to obtain the target predicted power generation.
[0045] In one example, the power generation prediction device can determine the predicted power generation according to the following formula three: P m =P a+ P b Formula 3 Among them, P m P represents the target predicted power generation. a P represents the first predicted power generation. b This indicates the second predicted power generation capacity.
[0046] Based on the technical solution provided in this application, a first predicted power generation is determined by predicting environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters. This allows for a comprehensive consideration of environmental data, the physical characteristics of wind power generation equipment, and the physical characteristics of photovoltaic power generation equipment, resulting in a highly accurate first predicted power generation. Furthermore, a second predicted power generation is determined based on the predicted environmental data and a power generation prediction model established using a combination of a Long Short-Term Memory (LSTM) network model and a Genetic Algorithm-Backpropagation (GA-BP) neural network model. Due to the global search capability, gradient-free nature, and flexible fitness function design of the GA-BP neural network model, it can provide better initial weights and thresholds for the LSTM, thereby improving the model's training efficiency and generalization ability. Finally, the first and second predicted power generation are weighted and summed to obtain the target predicted power generation. This combination of multiple highly accurate predicted power generation methods further improves the prediction accuracy of power generation in the wind-solar-storage system.
[0047] In one possible embodiment, in order to construct a power generation prediction model, this application may also include the following S401-S404.
[0048] S401. Construct an LSTM neural network model.
[0049] The LSTM neural network model can include an input layer, LSTM layers (there can be multiple layers), and an output layer.
[0050] LSTM layers, through memory cells and gating mechanisms, can effectively capture long-term dependencies in time-series data.
[0051] S402. Use the weights of the LSTM neural network model as initial weights to construct a BP neural network model.
[0052] The BP neural network model includes an input layer, a hidden layer, and an output layer.
[0053] S403. Optimize the initial weights and biases of the BP neural network model through selection, crossover, and mutation operations in the genetic algorithm to obtain the optimized weights and biases.
[0054] As one possible implementation, the power generation prediction device can design a genetic algorithm based on the process of individual encoding, fitness function, selection, crossover and mutation operations. The initial weights and biases of the BP neural network model can be optimized through the selection, crossover and mutation operations in the genetic algorithm to obtain the optimized weights and biases.
[0055] Individual encoding refers to flattening the weights and biases of the BP neural network into a one-dimensional array, which serves as the individual in the genetic algorithm.
[0056] The fitness function is used to evaluate the quality of each individual (i.e., weights and biases), and is usually measured by the error on the training set (such as mean squared error).
[0057] Selection can be done using roulette wheel selection or tournament selection. Crossover can be done using single-point crossover or uniform crossover, and mutation operations can randomly change the values of certain genes.
[0058] In one example, the genetic algorithm optimization process can be as follows: initializing the population, iteratively performing selection, crossover, and mutation operations to generate the next generation population, calculating the fitness of each individual, retaining the best individual, and assigning the optimal weights and biases obtained by the genetic algorithm to the BP neural network to obtain the optimized weights and biases.
[0059] S404. The optimized weights and biases are used as the weights and biases of the LSTM neural network model to obtain the power generation prediction model.
[0060] In one possible embodiment, in order to train the power generation prediction model, this application may also include the following S501-S503.
[0061] S501. Obtain multiple sets of sample data.
[0062] Each set of sample data includes historical power generation data and corresponding historical environmental data.
[0063] As one possible implementation, historical power generation data and historical environmental data can be stored in a database, and the power generation prediction device can draw from multiple sets of sample data in the database.
[0064] S502. Perform preprocessing operations on multiple sets of sample data.
[0065] The preprocessing operations include normalization, smoothing, and removal of outliers.
[0066] This ensures the stability and consistency of the data.
[0067] S503. Use multiple sets of sample data to train the power generation prediction model to obtain the trained power generation prediction model.
[0068] In one example, the power generation prediction device can train the power generation prediction model according to the following S1-S3.
[0069] S1. Divide the preprocessed sample data into training set and test set.
[0070] As one possible implementation, the power generation prediction device can divide the pre-processed sample data into a training set and a test set according to a preset ratio.
[0071] The preset ratio can be set as needed. For example, it can be 1:2, etc.
[0072] S2. Train the power generation prediction model using the training set, and adjust the model parameters until the preset stopping conditions are met.
[0073] The preset stopping condition can be set as needed. For example, it can be set to reach a preset number of training iterations. Or, it can be set to the accuracy of the power generation prediction model exceeding an accuracy threshold.
[0074] S3. Validate the power generation prediction model using the test set, and optimize the power generation prediction model based on the validation results to obtain the trained power generation prediction model.
[0075] This step can be referred to in the following sections, and will not be repeated here.
[0076] In one possible embodiment, in order to optimize the power generation prediction model based on the verification results and obtain the trained power generation prediction model, this application may further include the following S601-S605.
[0077] S601. Gradually increase the number of LSTM layers in the power generation prediction model and determine the root mean square error corresponding to different numbers of LSTM layers.
[0078] The root mean square error is determined based on the error between the predicted value of the power generation prediction model and the actual observed value.
[0079] As one possible implementation, the power generation prediction device can increase the number of LSTM layers in the power generation prediction model according to the prediction step size, and determine the root mean square error corresponding to different numbers of LSTM layers.
[0080] It should be noted that the prediction step size can be set as needed. For example, it can be set to 1 layer, etc.
[0081] S602. If the root mean square error does not decrease continuously after increasing the number of layers by N, determine the number of LSTM layers corresponding to the number of layers after increasing the number of layers by N as the target number of LSTM layers.
[0082] Where N is a preset positive integer. For example, it can be 4.
[0083] For example, if the root mean square error of the LSTM layer does not decrease continuously after increasing the number of layers by 4, the number of LSTM layers corresponding to the number of layers after increasing the number of layers by 4 is determined as the target number of LSTM layers.
[0084] S603. Based on the target number of LSTM layers in the power generation prediction model, gradually increase the number of neurons in the power generation prediction model and determine the root mean square error corresponding to different numbers of neurons.
[0085] As one possible implementation, the power generation prediction device can increase the number of neurons in the power generation prediction model according to the prediction step size, and determine the root mean square error corresponding to different numbers of neurons.
[0086] For example, the prediction step size can be 64 neurons, etc.
[0087] It should be noted that the prediction step size can be set as needed. For example, it can be set to 1 layer, etc.
[0088] S604. If the root mean square error does not decrease continuously after increasing the number of neurons by M times, the number of neurons corresponding to the increase after M times is determined as the target number of neurons.
[0089] Where M is a preset positive integer. For example, it can be 5.
[0090] S605. Adjust the number of LSTM layers in the power generation prediction model to the target number of LSTM layers, and adjust the number of neurons in the power generation prediction model to the target number of neurons, in order to optimize the power generation prediction model.
[0091] Understandably, the above methods can be used to systematically adjust the number of LSTM layers and neurons, find the optimal balance between prediction accuracy and stability, and improve the accuracy and stability of time series prediction models.
[0092] This application embodiment can divide the power generation prediction device into functional modules or functional units according to the above method example. For example, each function can be divided into a separate functional module or functional unit, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or in software functional modules or functional units. The module or unit division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.
[0093] When dividing each function into modules according to its corresponding function. Figure 2 A schematic diagram of another power generation prediction device 700 is shown. This power generation prediction device 700 can also be a chip, processor, etc., used in power generation prediction devices. This power generation prediction device 700 can be used to perform the functions of the power generation prediction devices involved in the above embodiments. Figure 2The power generation prediction device 700 shown may include: an acquisition unit 701 and a processing unit 702; the acquisition unit 701 is used to acquire prediction environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters; the prediction environmental data includes: air density, wind speed at the hub height of the wind power generation equipment, irradiance, temperature coefficient, and photovoltaic module backsheet temperature; the wind power generation equipment parameters include: rotor swept area and wind energy utilization coefficient; the photovoltaic power generation equipment parameters include: number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, and photovoltaic panel area; the processing unit 702, used for... Based on predicted environmental data, wind power generation equipment parameters, and photovoltaic power generation equipment parameters, a first predicted power generation is determined. The first predicted power generation is the sum of the first predicted wind power generation and the first predicted photovoltaic power generation. The first predicted wind power generation is determined by half the product of the wind energy utilization coefficient, the cube of the wind speed at the hub height, and the swept area of the impeller. The first predicted photovoltaic power generation is determined by the product of the number of photovoltaic modules, photovoltaic conversion efficiency, tracking efficiency, inverter efficiency, photovoltaic panel area, irradiance, and temperature factor. The temperature factor W = 1 - s(Tc - 25), where s represents the temperature coefficient and Tc represents the backsheet temperature of the photovoltaic module. The processing unit 702 is also used to input the predicted environmental data into the trained power generation prediction model to obtain a second predicted power generation. The power generation prediction model is established based on a combination of a Long Short Time Memory (LSTM) network model and a Genetic Algorithm-Backpropagation (GA-BP) neural network model. The processing unit 702 is also used to perform a weighted summation of the first predicted power generation and the second predicted power generation to obtain the target predicted power generation.
[0094] Optionally, the processing unit 702 is further configured to construct an LSTM neural network model; the processing unit 702 is further configured to use the weights of the LSTM neural network model as initial weights to construct a BP neural network model; the processing unit 702 is further configured to optimize the initial weights and biases of the BP neural network model through selection, crossover, and mutation operations in a genetic algorithm to obtain optimized weights and biases; the processing unit 702 is further configured to use the optimized weights and biases as the weights and biases of the LSTM neural network model to obtain a power generation prediction model.
[0095] Optionally, the acquisition unit 701 is further configured to acquire multiple sets of sample data; each set of sample data includes historical power generation data and corresponding historical environmental data; the processing unit 702 is further configured to perform preprocessing operations on the multiple sets of sample data; the preprocessing operations include normalization, smoothing, and removal of outliers; the processing unit 702 is further configured to train the power generation prediction model using the multiple sets of sample data to obtain the trained power generation prediction model.
[0096] Optionally, the processing unit 702 is specifically used for: dividing the preprocessed sample data into a training set and a test set; training the power generation prediction model using the training set, adjusting the model parameters until a preset stopping condition is met; validating the power generation prediction model using the test set, and optimizing the power generation prediction model based on the validation results to obtain the trained power generation prediction model.
[0097] Optionally, the processing unit 702 is further configured to: gradually increase the number of LSTM layers in the power generation prediction model and determine the root mean square error (RMSE) corresponding to different LSTM layer numbers; the RMSE is determined based on the error between the predicted value and the actual observed value of the power generation prediction model; if the RMSE does not decrease continuously after increasing the number of layers N times, the number of LSTM layers corresponding to the N-times increase is determined as the target number of LSTM layers; based on the target number of LSTM layers in the power generation prediction model, gradually increase the number of neurons in the power generation prediction model and determine the RMSE corresponding to different number of neurons; if the RMSE does not decrease continuously after increasing the number of layers M times, the number of neurons corresponding to the M-times increase is determined as the target number of neurons; adjust the number of LSTM layers in the power generation prediction model to the target number of LSTM layers and adjust the number of neurons in the power generation prediction model to the target number of neurons to optimize the power generation prediction model.
[0098] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by a computer program instructing related hardware. This program can be stored in the computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be an internal storage unit of the power generation prediction device (including a data transmitter and / or a data receiver) of any of the foregoing embodiments, such as the hard disk or memory of the power generation prediction device. The computer-readable storage medium can also be an external storage device of the terminal device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device. Further, the computer-readable storage medium can include both the internal storage unit of the power generation prediction device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the power generation prediction device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.
[0099] It should be noted that the terms "first" and "second," etc., in the specification, claims, and drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.
[0100] It should be understood that in this application, "at least one (item)" means one or more, "more than one" means two or more, "at least two (items)" means two or three or more, and "and / or" is used to describe the relationship between related objects, indicating that there can be three relationships. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the related objects before and after are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.
[0101] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.
[0102] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0103] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0104] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0105] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.
[0106] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method of power generation forecast, characterized by, The method comprises: obtaining predicted environment data, wind power generation equipment parameters and photovoltaic power generation equipment parameters; the predicted environment data comprises air density, wind speed at a hub height of the wind power generation equipment, irradiance, a temperature coefficient and a backboard temperature of a photovoltaic component; the wind power generation equipment parameters comprise a swept wind area of a blade and a wind energy utilization coefficient; and the photovoltaic power generation equipment parameters comprise a number of photovoltaic groups, a photovoltaic conversion efficiency, a tracking work efficiency, an inverter efficiency and a photovoltaic panel area; determining a first predicted power generation according to the predicted environment data, the wind power generation equipment parameters and the photovoltaic power generation equipment parameters; the first predicted power generation is a sum of a first predicted wind power generation and a first predicted photovoltaic power generation; the first predicted wind power generation is determined according to half of a product of the wind energy utilization coefficient, the cube of the wind speed at the hub height and the swept wind area of the blade; and the first predicted photovoltaic power generation is determined according to a product of the number of photovoltaic groups, the photovoltaic conversion efficiency, the tracking work efficiency, the inverter efficiency, the photovoltaic panel area, the irradiance and a temperature factor; the temperature factor W = 1 - s (Tc - 25) ; s represents the temperature coefficient; and Tc represents the backboard temperature of the photovoltaic component; inputting the predicted environment data into a trained power generation prediction model to obtain a second predicted power generation; the power generation prediction model is established based on a combination of a long short-term memory (LSTM) model and a genetic algorithm-back propagation (GA-BP) neural network model; performing weighted summation on the first predicted power generation and the second predicted power generation to obtain a target predicted power generation.
2. The method of claim 1, wherein, The method further comprises: constructing an LSTM neural network model; using the weights of the LSTM neural network model as initial weights to construct a BP neural network model; optimizing the initial weights and bias of the BP neural network model through selection, crossover and mutation operations in a genetic algorithm to obtain optimized weights and bias; using the optimized weights and bias as the weights and bias of the LSTM neural network model to obtain the power generation prediction model.
3. The method of claim 1, wherein, The method further comprises: obtaining a plurality of groups of sample data; each group of sample data comprises historical power generation data and corresponding historical environment data; performing a preprocessing operation on the plurality of groups of sample data; the preprocessing operation comprises normalization, smoothing and removal of abnormal values; training the power generation prediction model using the plurality of groups of sample data to obtain the trained power generation prediction model.
4. The method of claim 3, wherein, The training of the power generation prediction model using the plurality of groups of sample data to obtain the trained power generation prediction model comprises: dividing the plurality of groups of sample data after the preprocessing operation into a training set and a test set; training the power generation prediction model using the training set, adjusting model parameters until a preset stopping condition is met; verifying the power generation prediction model using the test set and optimizing the power generation prediction model according to a verification result to obtain the trained power generation prediction model.
5. The method of claim 4, wherein, The power generation power prediction model is optimized according to the verification result, and the optimization comprises the following steps: Step 1: gradually increase the LSTM layer number of the power generation power prediction model, and determine the root mean square error corresponding to different LSTM layer numbers; the root mean square error is determined according to the error between the predicted value of the power generation power prediction model and the actual observation value; Step 2: in the case that the root mean square error does not continuously decrease after gradually increasing N times, determine the LSTM layer number corresponding to the Nth time as the target LSTM layer number; Step 3: on the basis of the LSTM layer number of the power generation power prediction model being the target LSTM layer number, gradually increase the neuron number of the power generation power prediction model, and determine the root mean square error corresponding to different neuron numbers; Step 4: in the case that the root mean square error does not continuously decrease after gradually increasing M times, determine the neuron number corresponding to the Mth time as the target neuron number; Step 5: adjust the LSTM layer number of the power generation power prediction model to the target LSTM layer number, and adjust the neuron number of the power generation power prediction model to the target neuron number, so as to optimize the power generation power prediction model.