A new energy power prediction optimization system

Abstract

The application discloses a new energy power prediction optimization system, belongs to the technical field of new energy power, and comprises an information acquisition module, an environmental information data acquisition module, an LSTM network fusion analysis module, an association analysis module, a dynamic monitoring module and a monitoring terminal; wherein the power prediction model of new energy wind power generation is constructed through the LSTM network, the power prediction data of wind power generation is generated, the power correlation prediction data of wind power generation is generated after the power correlation prediction data of wind power generation is generated in combination with the association analysis module, dynamic decision is made, the result data of wind power generation power prediction optimization is generated and visualized, the power data under the current environment can be more accurately predicted, the wind power generation enterprise can realize more intelligent and automatic operation management, the power generation efficiency is improved, the resource allocation is optimized, and the safety and stability of the power generation process are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy power, in particular to a new energy power prediction optimization system. BACKGROUND

[0002] With the deep integration of new energy into the power system, it has become an important means to solve energy shortage and environmental pollution problems, among which wind energy is an important and effectively usable power generation energy. The use technology of wind power generation is also continuously improved. At present, the power prediction of wind power energy is mostly based on the collection of historical data to predict the smooth increase or decrease of wind power. At the present stage, linear or multiple linear regression is often used to analyze the correlation characteristics of wind power generation, which is not accurate enough and difficult to face the influence of current environmental and power generation data. In addition, the existing technology is difficult to analyze and predict the reference of the correlation change of the increase peak or decrease peak of power in some main months. If the increase peak or decrease peak of power exceeds the required safety range, it will cause noise hazards and mechanical hazards, and further trigger power generation risks, ultimately endangering the safe operation of the power grid. SUMMARY

[0003] In view of the above problems existing in the prior art new energy power prediction technology, the present application is proposed.

[0004] Therefore, one object of the present application is to provide a new energy power prediction optimization system which uses deep learning analysis technology to predict and optimize the power generation of wind turbine generators and environmental information data after hierarchical processing, and correlates the historical change characteristics to accurately control the power development trend of the preset wind turbine generator in the month, more accurately predict the power data under the current environment, and enable the system to self-alarm according to the current development trend to avoid unexpected accidents caused by power data changes exceeding the safety peak range.

[0005] To solve the above technical problems, the present application provides the following technical solutions:

[0006] On the one hand, the present application provides a new energy power prediction optimization system, comprising:

[0007] An information acquisition module for acquiring the regional position of the wind turbine generator and the power generation data of the wind turbine generator in the regional position;

[0008] An environmental information data acquisition module for acquiring the corresponding environmental information data of the wind turbine generator in the regional position, the environmental information data including wind speed, wind direction and environmental temperature and humidity;

[0009] An LSTM network fusion analysis module is configured to respond to the generated power data of the wind turbine generator set from the information acquisition module and the environmental information data from the environmental information data acquisition module, divide the data sets of the generated power data and the environmental information data into a training set, a validation set, and a test set, construct a power prediction model for new energy wind power generation through the LSTM network, determine a loss function for the power prediction model for new energy wind power generation, optimize the power prediction model for new energy wind power generation according to the loss function, and generate power prediction data for wind power generation; the fusion analysis module includes a division unit, an analysis unit, and an optimization unit;

[0010] a correlation analysis module for correlating the current wind power generation power prediction data with the historical power generation power prediction data, specifically by determining the months with the most occurrences of the maximum power generation data and the historical minimum power generation data in the historical power generation power prediction data, and performing integrated correlation analysis on the change trends of the environmental information data matching the months, so as to obtain power generation change characteristic data of the wind turbine generator set under different change trends of the environmental information data. The fusion analysis module generates wind power generation power correlation prediction data in response to the power generation change characteristic data;

[0011] a dynamic monitoring module, configured to respond to the wind power generation power correlation prediction data and the wind power generation power prediction data, and based on the historical power generation power prediction data and environmental information data, make dynamic decisions on the wind power generation power correlation prediction data and the wind power generation power prediction data while dynamically monitoring, generate result data of wind power generation power prediction optimization, and present them visually;

[0012] The monitoring terminal is used to record the communication base station matched with the wind turbine generator set, receive the field data uploaded from the preset area where the wind turbine generator set is located, and update the background data at the same time.

[0013] As a preferred solution of the present invention, the division unit extracts the change characteristics of the matched power generation data and environmental information data based on the power generation data and the environmental information data, and divides the change characteristics into primary change characteristics, secondary change characteristics, and tertiary change characteristics in a step-by-step manner through the division unit, wherein the change characteristic level is positively correlated with the power generation change characteristic;

[0014] The analysis unit analyzes the correlation change characteristics of wind speed and wind direction under different levels of change characteristics based on the hierarchical analysis in the LSTM network architecture, and simultaneously analyzes the change probability of the correlation change characteristics on the power generation data of the wind turbine generator set, and calculates the power generation change value of the wind turbine generator set under different levels of change characteristics based on the change probability of the power generation data of the wind turbine generator set;

[0015] The optimization unit responds to the change value of the generated power of the wind turbine generator set and generates corresponding wind power generation power prediction data after optimizing the power prediction model of the new energy wind power generation through a loss function.

[0016] As a preferred solution of the present invention, the probability of change of the associated change feature to the power generation data of the wind turbine generator set is analyzed and calculated according to the following formula:

[0017] Where P(X) represents the probability of change of wind turbine power generation data;

[0018] Where X represents the wind turbine power data collected when the wind speed value appears the most times, K represents the change value of the wind direction angle data at the current wind speed value, and ω i represents the weight of the wind turbine power data collected for the i-th time at the current wind speed value, m is the total number of items collected for the i-th time, and N represents the mixed weight, which represents the weight of the wind turbine power data relative to the data at different wind direction angles under the same wind speed value;

[0019] Based on the change probability of the power generation data of the wind turbine generator set, the power generation change value of the wind turbine generator set under different levels of change characteristics is calculated. Specifically, the change probability of the power generation data of the wind turbine generator set and the different levels of change characteristic data are converted into serialized form, and the architecture of the LSTM network is defined, including an input layer, three LSTM layers and an output layer using a softmax function. The LSTM model is trained using a training data set, and the validation set is used to adjust the power prediction model parameters of the new energy wind power generation. At the same time, the trained LSTM model is used to analyze the correlation between the level change characteristics of wind speed and wind direction and the power generation. For different wind speed and wind direction level change characteristics, the statistics and confidence intervals of the power generation are calculated respectively. The confidence interval is the confidence interval for calculating the power generation, which is determined by assuming a normal distribution.

[0020] As a preferred solution of the present invention, the loss function adopts the MSE loss function or the MAE loss function as the standard loss function for the regression analysis of the change value of the generated power of the wind turbine generator set.

[0021] As a preferred solution of the present invention, the dynamic monitoring module makes dynamic decisions on the power-related prediction data of the wind power generation and the power prediction data of the wind power generation. Specifically, the similarity value of the power-related prediction data of the wind power generation and the power prediction data of the wind power generation is calculated through the Euclidean distance algorithm, and a dynamic decision is made based on the similarity value to determine the result data of the wind power generation power prediction optimization and visualize it in a graphical format.

[0022] As a preferred solution of the present invention, the correlation analysis module obtains the minimum wind speed value data of the month with the most occurrences of the minimum power generation power data in the historical power forecast data, and analyzes the unit time period with the most occurrences of the minimum wind speed value data. At the same time, a safety threshold is preset based on the historical minimum safe power generation power data within the unit time period, and wind speed value data corresponding to the safety threshold is collected. When the wind speed value data appearing in the unit time period of the month in the future period is less than the wind speed value data corresponding to the safety threshold, the system determines that the power generation power of the preset wind turbine generator set is lower than the historical minimum safe power generation power data, and issues a safety warning. Otherwise, no determination is made.

[0023] As a preferred solution of the present invention, the correlation analysis module obtains the first three digits of power value data higher than the minimum safe power generation power data in the power forecast data of the historical power generation based on the power value data lower than the minimum safe power generation power data, and obtains the wind speed value data change characteristics corresponding to the three digits of power value data to generate a data model, and presets a risk threshold based on the median data of the three digits of power value data. When the power generation power of the preset wind turbine generator set in the future time period is lower than the risk threshold, the system determines that the preset wind turbine generator set is in a risky state, otherwise, it does not make a determination.

[0024] As a preferred solution of the present invention, the correlation analysis module collects the minimum wind speed value and the maximum wind speed value in the month with the most occurrences of the maximum power generation data in the historical power forecast data, analyzes the change trend of the minimum wind speed value toward the maximum wind speed value, and simultaneously intercepts the median wind speed value in the change trend, and presets a critical threshold based on the median wind speed value. When the wind speed change value in the month in the future period exceeds the critical threshold, the system determines that the power generation power of the preset wind turbine generator set will exceed the historical maximum power generation data and issues a warning. Otherwise, no determination is made.

[0025] Based on the critical threshold, the safe power generation power value in the historical maximum power generation power data is obtained, and the Monte Carlo method is used to sample the safe power generation power value after reliability assessment. At the same time, the sampled data is divided into high-safety power generation power value, medium-safety power generation power value and low-safety power generation power value in a step-by-step manner, and the historical maximum wind speed value data corresponding to the three safety power generation power values ​​are collected. When the wind speed value data collected in the future period in the month with the most occurrences of the historical maximum power generation power data exceeds the historical maximum wind speed value data corresponding to the medium-safety power generation power value, the system determines that the power generation power of the preset wind turbine generator set is converted to the low-safety power generation power value, and issues a risk warning.

[0026] As a preferred solution of the present invention, the following steps are performed: obtaining the month with the most historical maximum wind speed data corresponding to the medium safety power generation value, analyzing the wind speed change trend of the month, and collecting wind direction angle data under the wind speed change trend, and dividing the wind direction angle data into θ1, θ2, ..., θ n , where n represents the nth wind direction angle data collected. The power value data difference of the wind turbine generator set generated power due to the change of different wind direction angle data under the same wind speed value data is analyzed, and the proportion of different wind direction angle data to the power value data difference is calculated as follows:

[0027] δ t vj is the proportion of different wind direction angle data in the concentrated period;

[0028] Among them, t represents the concentrated period of time in a month when the wind direction angle data has the largest change; R vj represents the jth historical maximum wind direction angle data collected in the vth unit period, and R represents the wind speed value data collected in different time periods;

[0029] At the same time, based on the wind speed data collected in different unit time periods, the wind direction angle with the most occurrences corresponding to the data is obtained, and the wind force combinations are preset. The wind force combinations are divided into ▽1 wind force combination, ▽2 wind force combination, and ▽3 wind force combination according to the number of occurrences. The weights of different wind force combinations in different unit time periods are calculated as follows:

[0030]

[0031] Among them, ▽ represents the wind force combination, u v It represents the wind force combination with the longest duration collected in the vth unit period, and E represents the total number of items in the vth unit period.

[0032] As a preferred solution of the present invention, the change characteristics of the environmental information data are divided according to different unit time periods, the change characteristics of the environmental information data in different unit time periods are obtained, and the environmental information data are divided into several evaluation indicators using a cross-validation method, and a corresponding evaluation report is generated based on the evaluation indicators.

[0033] The beneficial effects of the present invention are as follows: the present invention utilizes deep learning analysis technology to perform predictive analysis and optimization on the power generation power and environmental information data of the wind turbine generator set after layered processing, and at the same time associates historically related change characteristics, which can accurately control the power development trend of the preset wind turbine generator set in the said month, and can more accurately predict the power data under the current environment, and enable the system to automatically issue early warnings according to the current development trend to avoid unexpected events caused by changes in power data exceeding the safe peak range. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort. Among them:

[0035] Figure 1 This is a schematic diagram of the modular structure of a new energy power prediction and optimization system according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the process structure of the new energy power prediction and optimization system according to an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of the wind power combination system process structure according to an embodiment of the present invention;

[0038] Numbers in the figure: 10 - information acquisition module; 20 - environmental information data acquisition module; 30 - LSTM network fusion analysis module; 40 - correlation analysis module; 50 - dynamic monitoring module; 60 - monitoring terminal. DETAILED DESCRIPTION

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0040] The present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0041] Reference Figures 1 to 3 , is an embodiment of the present invention, which provides a new energy power prediction and optimization system, including:

[0042] The information acquisition module 10 is used to obtain the regional location of the wind turbine generator set and the power generation data of the wind turbine generator set in the regional location;

[0043] Environmental information data acquisition module 20, used to collect environmental information data corresponding to the wind turbine generator set in the regional location, the environmental information data including wind speed, wind direction, and ambient temperature and humidity;

[0044] The LSTM network fusion analysis module 30 is used to respond to the power generation data of the wind turbine generator set from the information acquisition module and the environmental information data from the environmental information data acquisition module, divide the data sets of the power generation data and the environmental information data into a training set, a validation set, and a test set, construct a power prediction model for new energy wind power generation through the LSTM network, determine the loss function of the power prediction model for new energy wind power generation, optimize the power prediction model for new energy wind power generation according to the loss function, and generate wind power generation power prediction data; the fusion analysis module includes a division unit, an analysis unit, and an optimization unit;

[0045] The correlation analysis module 40 is configured to correlate the current wind power generation power forecast data with the historical power generation power forecast data. Specifically, the module determines the months with the most occurrences of the maximum power generation data and the historical minimum power generation data in the historical power generation power forecast data, and integrates the change trends of the environmental information data corresponding to the months, and performs correlation analysis to obtain the power generation change characteristic data of the wind turbine generator set under different change trends of the environmental information data. The fusion analysis module generates wind power generation power correlation prediction data in response to the power generation change characteristic data.

[0046] The dynamic monitoring module 50 is configured to respond to the wind power generation power correlation prediction data and the wind power generation power prediction data, and based on the historical power generation power prediction data and environmental information data, make dynamic decisions on the wind power generation power correlation prediction data and the wind power generation power prediction data while dynamically monitoring, generate wind power generation power prediction optimization result data, and present them visually;

[0047] The monitoring terminal 60 is used to record the communication base station matched with the wind turbine generator set, receive the field data uploaded from the preset area where the wind turbine generator set is located, and update the background data at the same time.

[0048] Based on the above, it can be seen that this embodiment constructs a power prediction model for new energy wind power generation through an LSTM network, generates power prediction data for wind power generation, and combines the correlation analysis module 40 to generate power correlation prediction data for wind power generation, and then makes dynamic decisions, generates result data for wind power generation power prediction optimization and presents it visually. The present invention can more accurately predict power data under the current environment, and can enable wind power generation companies to achieve more intelligent and automated operation and management, improve power generation efficiency, optimize resource allocation, and ensure the safety and stability of the power generation process.

[0049] Specifically, in this embodiment, the division unit extracts the change characteristics of the matched power generation data and environmental information data based on the power generation data and the environmental information data, and divides the change characteristics into primary change characteristics, secondary change characteristics, and tertiary change characteristics in a step-by-step manner through the division unit, wherein the change characteristic level is positively correlated with the power generation change characteristic;

[0050] The analysis unit analyzes the correlation change characteristics of wind speed and wind direction under different levels of change characteristics based on the hierarchical analysis in the LSTM network architecture. It also analyzes the probability of change of the wind turbine generator set's power generation data due to the correlation change characteristics, and calculates the wind turbine generator set power generation change value under different levels of change characteristics based on the change probability of the wind turbine generator set's power generation data.

[0051] The optimization unit responds to the change in the power generation value of the wind turbine generator set, and generates corresponding wind power generation power prediction data after optimizing the power prediction model of the new energy wind power generation through the loss function.

[0052] Specifically, in this embodiment, the probability of change of the wind turbine generator power generation data due to the associated change characteristics is analyzed and calculated according to the following formula:

[0053] Where P(X) represents the probability of change of wind turbine power generation data;

[0054] Where X represents the wind turbine power data collected when the wind speed value appears the most times, K represents the change value of the wind direction angle data at the current wind speed value, and ω i represents the weight of the wind turbine power data collected for the i-th time at the current wind speed value, m is the total number of items collected for the i-th time, and N represents the mixed weight, which represents the weight of the wind turbine power data relative to the data at different wind direction angles under the same wind speed value;

[0055] Based on the change probability of the wind turbine power generation data, the change value of the wind turbine power generation under different levels of change characteristics is calculated. Specifically, the change probability of the wind turbine power generation data and the different levels of change characteristic data are converted into serialized form, and the architecture of the LSTM network is defined, including an input layer, three LSTM layers, and an output layer using a softmax function. The LSTM model is trained using a training data set, and the validation set is used to adjust the parameters of the power prediction model for renewable energy wind power generation. At the same time, the trained LSTM model is used to analyze the correlation between the level change characteristics of wind speed and wind direction and the power generation. For different wind speed and wind direction level change characteristics, the statistics and confidence intervals of the power generation are calculated respectively. The confidence interval is the confidence interval for calculating the power generation, which is determined by assuming a normal distribution.

[0056] Furthermore, in this embodiment, the loss function adopts the MSE loss function or the MAE loss function as the standard loss function for the regression analysis of the change value of the generated power of the wind turbine generator set.

[0057] Preferably, in this embodiment, the dynamic monitoring module 50 makes dynamic decisions on the power-related prediction data of wind power generation and the power prediction data of wind power generation. Specifically, the similarity value of the power-related prediction data of wind power generation and the power prediction data of wind power generation is calculated through the Euclidean distance algorithm, and a dynamic decision is made based on the similarity value to determine the result data of the wind power generation power prediction optimization and visualize it in a graphical format.

[0058] Specifically, in this embodiment, the correlation analysis module 40 obtains the minimum wind speed value data of the month with the most occurrences of the minimum power generation power data in the historical power prediction data, and analyzes the unit time period with the most occurrences of the minimum wind speed value data. At the same time, a safety threshold is preset based on the historical minimum safe power generation power data within the unit time period, and wind speed value data corresponding to the safety threshold is collected. When the wind speed value data appearing in the unit time period of the month in the future period is less than the wind speed value data corresponding to the safety threshold, the system determines that the power generation power of the preset wind turbine generator set is lower than the historical minimum safe power generation power data, and issues a safety warning. Otherwise, no determination is made.

[0059] Specifically, in this embodiment, the correlation analysis module 40 obtains the first three power value data higher than the minimum safe power generation power data in the power prediction data of historical power generation based on the power value data, and obtains the wind speed value data change characteristics corresponding to the three power value data to generate a data model, and presets a risk threshold based on the median data of the three power value data. When the power generation power of the preset wind turbine generator set in the future time period is lower than the risk threshold, the system determines that the preset wind turbine generator set is in a risk state, otherwise, it does not make a determination.

[0060] Specifically, in this embodiment, the correlation analysis module 40 collects the minimum wind speed value and the maximum wind speed value of the month with the most maximum power generation data in the historical power forecast data, analyzes the change trend of the minimum wind speed value toward the maximum wind speed value, and simultaneously intercepts the median wind speed value in the change trend. A critical threshold is preset based on the median wind speed value. When the wind speed change value in a month in the future period exceeds the critical threshold, the system determines that the power generation of the preset wind turbine generator set will exceed the historical maximum power generation data and issues a warning. Otherwise, no determination is made.

[0061] Based on the critical threshold, the safe power generation power value in the historical maximum power generation power data is obtained, and the Monte Carlo method is used to sample the safe power generation power value after reliability assessment. At the same time, the sampled data is divided into high-safety power generation power value, medium-safety power generation power value and low-safety power generation power value in a step-by-step manner, and the historical maximum wind speed value data corresponding to the three safety power generation values ​​are collected. When the wind speed value data collected in the future period in the month with the most occurrences of the historical maximum power generation data exceeds the historical maximum wind speed value data corresponding to the medium-safety power generation value, the system determines that the power generation power of the preset wind turbine generator set is shifting to the low-safety power generation value and issues a risk warning.

[0062] The embodiment should emphasize that the month with the most historical maximum wind speed data exceeding the medium safety power generation value is obtained, and the wind speed change trend of the month is analyzed. At the same time, the wind direction angle data under the wind speed change trend is collected, and the wind direction angle data is divided into θ1, θ2, ..., θ n , where n represents the nth wind direction angle data collected. Analyze the power value data difference of the wind turbine generator set generated power due to the change of different wind direction angle data under the same wind speed value data, and calculate the proportion of different wind direction angle data to the power value data difference, as follows:

[0063] δ t vj is the proportion of different wind direction angle data in the concentrated period;

[0064] in, t Indicates the concentrated period of time in a month when the wind direction angle data has the largest change; R vj represents the jth historical maximum wind direction angle data collected in the vth unit period, and R represents the wind speed value data collected in different time periods;

[0065] At the same time, based on the wind speed data collected in different unit time periods, the wind direction angle with the most occurrences corresponding to the data is obtained, and the wind force combinations are preset. The wind force combinations are divided into ▽1 wind force combination, ▽2 wind force combination, and ▽3 wind force combination according to the number of occurrences. The weights of different wind force combinations in different unit time periods are calculated as follows:

[0066]

[0067] Among them, ▽ represents the wind force combination, u v It represents the wind force combination with the longest duration collected in the vth unit period, and E represents the total number of items in the vth unit period.

[0068] This embodiment divides the change characteristics of environmental information data into different unit time periods, obtains the change characteristics of environmental information data in different unit time periods, and uses a cross-validation method to divide the environmental information data into several evaluation indicators, and generates corresponding evaluation reports based on the evaluation indicators.

[0069] In summary, this invention uses advanced deep learning analysis technology to perform refined, layered processing of wind turbine power generation and related environmental information. This approach not only enables in-depth predictive analysis and optimization, but also correlates changing patterns in historical data, enabling precise control of wind turbine power trends in specific months.

[0070] This technology provides more accurate power generation forecasts based on current environmental conditions. More importantly, the system automatically issues warnings based on real-time data and the evolving trends of the forecast model. This self-warning mechanism effectively prevents abnormal fluctuations in power generation, ensuring it remains within a safe peak range and minimizing unforeseen risks.

[0071] Through the application of the present invention, wind power generation companies will be able to achieve more intelligent and automated operation and management, improve power generation efficiency, optimize resource allocation, and ensure the safety and stability of the power generation process.

[0072] Some code examples:

[0073] Information acquisition module 110

[0074] import numpy as np

[0075] import matplotlib.pyplot as plt

[0076] fromsklearn.datasetsimportmake_classification

[0077] from sklearn.svm import SVC

[0078] # Generate GPS satellite positioning navigation data

[0079] X, y = make_classification(n_samples=100, n_features=2, n_redundant=0, n_clusters_per_class=1, random_state=42)

[0080] # Get power data

[0081] svm_linear = SVC(kernel='linear')

[0082] svm_linear.fit(X, y)

[0083] # Generate decision boundary

[0084] plt.figure(figsize=(8, 6))

[0085] plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Paired, marker='o', edgecolors='k')

[0086] ax = plt.gca()

[0087] xlim = ax.get_xlim()

[0088] ylim = ax.get_ylim()

[0089] xx, yy = np.meshgrid(np.linspace(xlim[0], xlim[1], 100), np.linspace(ylim[0], ylim[1], 100))

[0090] Z = svm_linear.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)

[0091] # Generate decision boundary and support vector data

[0092] plt.contourf(xx, yy, Z, alpha=0.2, cmap=plt.cm.Paired)

[0093] plt.scatter(svm_linear.support_vectors_[:,0],svm_linear.support_vectors_[:,1],s=100,facecolors='none',edgecolors='k')

[0094] plt.title('LinearSVMDecisionBoundary')

[0095] plt.xlabel('Feature1')

[0096] plt.ylabel('Feature2')

[0097] plt.show()

[0098] Data acquisition module 130

[0099] import numpy asnp

[0100] importmatplotlib.pyplotasplt

[0101] fromsklearn.datasetsimportmake_classification

[0102] fromsklearn.svmimportSVC

[0103] #Get wind speed, wind direction, ambient temperature and humidity data

[0104] X,y=make_classification(n_samples=100,n_features=2,n_redundant=0,n_clusters_per_class=1,random_state=42)

[0105] #Get the historical average power generation data of wind turbines

[0106] svm_poly=SVC(kernel='poly',degree=3) #3rd degree polynomial kernel function

[0107] svm_poly.fit(X,y)

[0108] #Preset algorithm

[0109] plt.figure(figsize=(8,6))

[0110] plt.scatter(X[:,0],X[:,1],c=y,cmap=plt.cm.Paired,marker='o',edgecolors='k')

[0111] #Generate grid data for drawing decision boundaries

[0112] ax = plt.gca()

[0113] xlim = ax.get_xlim()

[0114] ylim = ax.get_ylim()

[0115] xx,yy=np.meshgrid(np.linspace(xlim[0],xlim[1],100),np.linspace(ylim[0],ylim[1],100))

[0116] Z=svm_poly.predict(np.c_[xx.ravel(),yy.ravel()]).reshape(xx.shape)

[0117] # draw the decision boundary and support vector

[0118] plt.contourf(xx,yy,Z,alpha=0.2,cmap=plt.cm.Paired)

[0119] plt.scatter(svm_poly.support_vectors_[:,0],svm_poly.support_vectors_[:,1],s=100,facecolors='none',edgecolors='k')

[0120] plt.title('PolynomialSVMDecisionBoundary')

[0121] plt.xlabel('Feature1')

[0122] plt.ylabel('Feature2')

[0123] plt.show()

[0124] Fusion analysis unit 140

[0125] import numpy asnp

[0126] importmatplotlib.pyplotasplt

[0127] fromsklearn.datasetsimportmake_classification

[0128] fromsklearn.svmimportSVC

[0129] #Generate calculation data

[0130] X,y=make_classification(n_samples=100,n_features=2,n_redundant=0,n_clusters_per_class=1,random_state=42)

[0131] #Analyze the data

[0132] svm_rbf=SVC(kernel='rbf',gamma=0.5)#gamma controls the width of the Gaussian kernel

[0133] svm_rbf.fit(X,y)

[0134] #Generate decision boundary

[0135] plt.figure(figsize=(8,6))

[0136] plt.scatter(X[:,0],X[:,1],c=y,cmap=plt.cm.Paired,marker='o',edgecolors='k')

[0137] #Predict data changes

[0138] ax = plt.gca()

[0139] xlim = ax.get_xlim()

[0140] ylim = ax.get_ylim()

[0141] xx,yy=np.meshgrid(np.linspace(xlim[0],xlim[1],100),np.linspace(ylim[0],ylim[1],100))

[0142] Z=svm_rbf.predict(np.c_[xx.ravel(),yy.ravel()]).reshape(xx.shape)

[0143] plt.contourf(xx,yy,Z,alpha=0.2,cmap=plt.cm.Paired)

[0144] plt.scatter(svm_rbf.support_vectors_[:,0],svm_rbf.support_vectors_[:,1],s=100,facecolors='none',edgecolors='k')

[0145] plt.title('RBFSVMDecisionBoundary')

[0146] plt.xlabel('Feature1')

[0147] plt.ylabel('Feature2')

[0148] plt.show()

[0149] Data integration module 150

[0150] import numpy asnp

[0151] importmatplotlib.pyplotasplt

[0152] fromsklearn.datasetsimportmake_classification

[0153] fromsklearn.svmimportSVC

[0154] #Integrate data change trends

[0155] X,y=make_classification(n_samples=100,n_features=2,n_redundant=0,n_clusters_per_class=1,random_state=42)

[0156] #Use MonteCarlo method to sample data

[0157] svm_sigmoid = SVC(kernel = 'sigmoid', coef0 = 1.0) #coef0 controls the bias of the Sigmoid function

[0158] svm_sigmoid.fit(X,y)

[0159] plt.figure(figsize=(8,6))

[0160] plt.scatter(X[:,0],X[:,1],c=y,cmap=plt.cm.Paired,marker='o',edgecolors='k')

[0161] #Determine data transformation

[0162] ax = plt.gca()

[0163] xlim = ax.get_xlim()

[0164] ylim = ax.get_ylim()

[0165] xx,yy=np.meshgrid(np.linspace(xlim[0],xlim[1],100),np.linspace(ylim[0],ylim[1],100))

[0166] Z=svm_sigmoid.predict(np.c_[xx.ravel(),yy.ravel()]).reshape(xx.shape)

[0167] #Get different trend change characteristics

[0168] plt.contourf(xx,yy,Z,alpha=0.2,cmap=plt.cm.Paired)

[0169] plt.scatter(svm_sigmoid.support_vectors_[:,0],svm_sigmoid.support_vectors_[:,1],s=100,facecolors='none',edgecolors='k')

[0170] plt.title('SigmoidSVMDecisionBoundary')

[0171] plt.xlabel('Feature1')

[0172] plt.ylabel('Feature2')

[0173] plt.show()

[0174] Dynamic monitoring module 160

[0175] import numpy asnp

[0176] importmatplotlib.pyplotasplt

[0177] fromsklearn.datasetsimportmake_classification

[0178] fromsklearn.svmimportSVC

[0179] #Collect data for a specific month

[0180] X,y=make_classification(n_samples=100,n_features=2,n_redundant=0,n_clusters_per_class=1,random_state=42)

[0181] #Use Laplacian data classifier

[0182] svm_laplacian = SVC(kernel = 'rbf', gamma = 0.5) # gamma controls Laplacian

[0183] svm_laplacian.fit(X,y)

[0184] #Data Fusion

[0185] plt.figure(figsize=(8,6))

[0186] plt.scatter(X[:,0],X[:,1],c=y,cmap=plt.cm.jet,marker='o',edgecolors='k')

[0187] #Generate wind force combination

[0188] ax = plt.gca()

[0189] xlim = ax.get_xlim()

[0190] ylim = ax.get_ylim()

[0191] xx, yy = np.meshgrid(np.linspace(xlim[0], xlim[1], 100), np.linspace(ylim[0], ylim[1], 100))

[0192] Z = svm_laplacian.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)

[0193] # Generate data unit period

[0194] plt.contourf(xx, yy, Z, alpha=0.2, cmap=plt.cm.jet)

[0195] plt.scatter(svm_laplacian.support_vectors_[:, 0], svm_laplacian.support_vectors_[:, 1], s=100, facecolors='none', edgecolors='limegreen', linewidths=2)

[0196] plt.title('LaplacianSVMDecisionBoundary', fontsize=14)

[0197] plt.xlabel('Feature1', fontsize=12)

[0198] plt.ylabel('Feature2', fontsize=12)

[0199] plt.show()

[0200] In the above embodiments, all or part can be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium.

[0201] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0202] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0203] Any process or method description in a flow chart or otherwise described herein can be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a specific logical function or process. The scope of the preferred embodiments of the present application includes additional implementations in which the functions may be performed in a different order than shown or discussed, including in a substantially simultaneous manner or in a reverse order depending on the functions involved.

[0204] The logic and / or steps represented in the flowchart or otherwise described herein may be considered, for example, as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device).

[0205] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above embodiment method can be completed by instructing the relevant hardware through a program, which can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0206] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the aforementioned integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. The storage medium may be a read-only memory, a magnetic disk, or an optical disk, etc.

[0207] The above are only specific embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A new energy power prediction and optimization system, characterized in that: include: An information acquisition module (10) is used to acquire the regional location of the wind turbine generator set and the power generation data of the wind turbine generator set in the regional location; An environmental information data acquisition module (20) is used to acquire environmental information data corresponding to the wind turbine generator set in the regional location, wherein the environmental information data includes wind speed, wind direction, and ambient temperature and humidity; An LSTM network fusion analysis module (30) is used to respond to the power generation data of the wind turbine generator set from the information acquisition module and the environmental information data from the environmental information data acquisition module, divide the data sets of the power generation data and the environmental information data into a training set, a validation set and a test set, construct a power prediction model of new energy wind power generation through the LSTM network, determine the loss function of the power prediction model of new energy wind power generation, optimize the power prediction model of new energy wind power generation according to the loss function, and generate power prediction data of wind power generation; the fusion analysis module includes a division unit, an analysis unit and an optimization unit; The division unit extracts the change characteristics of the matched power generation data and environmental information data based on the power generation data and the environmental information data, and divides the change characteristics into primary change characteristics, secondary change characteristics, and tertiary change characteristics in a step-by-step manner through the division unit, wherein the change characteristic level is positively correlated with the power generation change characteristic; The analysis unit analyzes the correlation change characteristics of wind speed and wind direction under different levels of change characteristics based on the hierarchical analysis in the LSTM network architecture, and simultaneously analyzes the change probability of the correlation change characteristics on the power generation data of the wind turbine generator set, and calculates the power generation change value of the wind turbine generator set under different levels of change characteristics based on the change probability of the power generation data of the wind turbine generator set; The optimization unit responds to the power generation change value of the wind turbine generator set and generates corresponding wind power generation power prediction data after optimizing the power prediction model of the new energy wind power generation through the loss function; A correlation analysis module (40) is used to correlate the current wind power generation power prediction data with the historical power generation power prediction data, specifically by determining the month with the most occurrences of the maximum power generation data and the historical minimum power generation data in the historical power generation power prediction data, and performing correlation analysis after integration, so as to obtain the power generation change characteristic data of the wind turbine generator set under different change trends of the environmental information data. The fusion analysis module generates the wind power generation power correlation prediction data in response to the power generation change characteristic data; A dynamic monitoring module (50) is used to respond to the wind power generation power correlation prediction data and the wind power generation power prediction data, and based on the historical power generation power prediction data and environmental information data, perform dynamic decision-making on the wind power generation power correlation prediction data and the wind power generation power prediction data while dynamically monitoring, generate wind power generation power prediction optimization result data and visualize the result data; The monitoring terminal (60) is used to record the communication base station matched with the wind turbine generator set, receive the field data uploaded from the area where the preset wind turbine generator set is located, and update the background data at the same time.

2. A new energy power prediction and optimization system according to claim 1, characterized in that: The probability of change of the wind turbine generator power generation data due to the associated change characteristics is analyzed and calculated according to the following formula: Where P(X) represents the probability of change of wind turbine power generation data; Where X represents the wind turbine power data collected when the wind speed value appears the most times, K represents the change value of the wind direction angle data at the current wind speed value, and ω i represents the weight of the wind turbine power data collected for the i-th time at the current wind speed value, m is the total number of items collected for the i-th time, and N represents the mixed weight, which represents the weight of the wind turbine power data relative to the data at different wind direction angles under the same wind speed value; Based on the change probability of the power generation data of the wind turbine generator set, the power generation change value of the wind turbine generator set under different levels of change characteristics is calculated. Specifically, the change probability of the power generation data of the wind turbine generator set and the different levels of change characteristic data are converted into serialized form, and the architecture of the LSTM network is defined, including an input layer, three LSTM layers and an output layer using a softmax function. The LSTM model is trained using a training data set, and the validation set is used to adjust the power prediction model parameters of the new energy wind power generation. At the same time, the trained LSTM model is used to analyze the correlation between the level change characteristics of wind speed and wind direction and the power generation. For different wind speed and wind direction level change characteristics, the statistics and confidence intervals of the power generation are calculated respectively. The confidence interval is the confidence interval for calculating the power generation, which is determined by assuming a normal distribution.

3. A new energy power prediction and optimization system according to claim 1, characterized in that: The loss function adopts the MSE loss function or the MAE loss function as the standard loss function for regression analysis of the change value of the generated power of the wind turbine generator set.

4. A new energy power prediction and optimization system according to claim 1, characterized in that: The dynamic monitoring module (50) makes a dynamic decision on the power correlation prediction data of the wind power generation and the power prediction data of the wind power generation, specifically, calculates the similarity value of the power correlation prediction data of the wind power generation and the power prediction data of the wind power generation through the Euclidean distance algorithm, makes a dynamic decision based on the similarity value, determines the result data of the wind power generation power prediction optimization and presents it visually in a chart format.

5. The new energy power prediction and optimization system according to claim 1, characterized in that: The correlation analysis module (40) obtains the minimum wind speed value data of the month with the most occurrences of the minimum power generation data in the power prediction data of the historical power generation, and analyzes the unit time period with the most occurrences of the minimum wind speed value data. At the same time, a safety threshold is preset based on the historical minimum safe power generation data within the unit time period, and wind speed value data corresponding to the safety threshold is collected. When the wind speed value data appearing in the unit time period of the month in the future period is less than the wind speed value data corresponding to the safety threshold, the system determines that the power generation power of the preset wind turbine generator set is lower than the historical minimum safe power generation data and issues a safety warning. Otherwise, no determination is made.

6. A new energy power prediction and optimization system according to claim 5, characterized in that: The correlation analysis module (40) obtains the first three digits of power value data higher than the minimum safe power generation data in the power prediction data of the historical power generation based on the power value data lower than the minimum safe power generation data, and obtains the wind speed value data change characteristics corresponding to the three digits of power value data to generate a data model, and presets a risk threshold based on the median data of the three digits of power value data. When the power generation power of the preset wind turbine generator set in the future period is lower than the risk threshold, the system determines that the preset wind turbine generator set is in a risk state, otherwise, no determination is made.

7. A new energy power prediction and optimization system according to claim 6, characterized in that: The correlation analysis module (40) collects the minimum wind speed value and the maximum wind speed value of the month with the most occurrences of the maximum power generation data in the power prediction data of the historical power generation, analyzes the change trend of the minimum wind speed value toward the maximum wind speed value, and simultaneously intercepts the median wind speed value in the change trend, and presets a critical threshold value based on the median wind speed value. When the wind speed change value in the month in the future period exceeds the critical threshold value, the system determines that the power generation of the preset wind turbine generator set will exceed the historical maximum power generation data and issues a warning. Otherwise, no determination is made. Based on the critical threshold, the safe power generation power value in the historical maximum power generation power data is obtained, and the Monte Carlo method is used to sample the safe power generation power value after reliability assessment. At the same time, the sampled data is divided into high-safety power generation power value, medium-safety power generation power value and low-safety power generation power value in a step-by-step manner, and the historical maximum wind speed value data corresponding to the three safety power generation power values ​​are collected. When the wind speed value data collected in the future period in the month with the most occurrences of the historical maximum power generation power data exceeds the historical maximum wind speed value data corresponding to the medium-safety power generation power value, the system determines that the power generation power of the preset wind turbine generator set is converted to the low-safety power generation power value, and issues a risk warning.

8. A new energy power prediction and optimization system according to claim 7, characterized in that: Obtain the month with the most historical maximum wind speed data corresponding to the medium safety power generation value, analyze the wind speed change trend of the month, and collect wind direction angle data under the wind speed change trend, and divide the wind direction angle data into θ1, θ2, ..., θ n , where n represents the nth wind direction angle data collected. The power value data difference of the wind turbine generator set generated power due to the change of different wind direction angle data under the same wind speed value data is analyzed, and the proportion of different wind direction angle data to the power value data difference is calculated as follows: δ t vj is the proportion of different wind direction angle data in the concentrated period; Among them, t represents the concentrated period of time in a month when the wind direction angle data has the largest change; R vj represents the jth historical maximum wind direction angle data collected in the vth unit period, and R represents the wind speed value data collected in different time periods; At the same time, based on the wind speed data collected in different unit time periods, the wind direction angle with the most occurrences corresponding to the data is obtained, and the wind force combinations are preset. The wind force combinations are divided into ▽1 wind force combination, ▽2 wind force combination, and ▽3 wind force combination according to the number of occurrences. The weights of different wind force combinations in different unit time periods are calculated as follows: Among them, ▽ represents the wind force combination, u v It represents the wind force combination with the longest duration collected in the vth unit period, and E represents the total number of items in the vth unit period.

9. A new energy power prediction and optimization system according to claim 8, characterized in that: The change characteristics of the environmental information data are divided into different unit time periods, the change characteristics of the environmental information data in different unit time periods are obtained, and the environmental information data are divided into several evaluation indicators using a cross-validation method, and a corresponding evaluation report is generated according to the evaluation indicators.

Images