CN115759395A - Training of photovoltaic detection model, detection method of photovoltaic power generation and related device - Google Patents
Training of photovoltaic detection model, detection method of photovoltaic power generation and related device Download PDFInfo
- Publication number
- CN115759395A CN115759395A CN202211428658.3A CN202211428658A CN115759395A CN 115759395 A CN115759395 A CN 115759395A CN 202211428658 A CN202211428658 A CN 202211428658A CN 115759395 A CN115759395 A CN 115759395A
- Authority
- CN
- China
- Prior art keywords
- power
- time period
- photovoltaic
- weather
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a training method of a photovoltaic detection model, a detection method of photovoltaic power generation and a related device, wherein the method comprises the following steps: in the embodiment, the original power generation power recorded when the photovoltaic power station generates power in a plurality of time periods is obtained, and each time period is associated with weather data; completing the original power generation data in each time period to obtain target power generation power; dividing all weather data into a plurality of categories; and aiming at each category, training a photovoltaic detection model by taking weather data in the same time period as a sample and target power generation power as a label. On one hand, the weather types are more accurately divided by adopting the s-Kohonen neural network clustering, the weather type clustering device adapts to the change of different weathers, and has better generalization capability. On the other hand, a convolutional neural network is introduced on the basis of a long-term and short-term memory network model, so that the spatial characteristics of data can be effectively extracted under the condition of extreme weather conditions, and the prediction accuracy of photovoltaic power is remarkably improved.
Description
Technical Field
The invention relates to the field of microgrid photovoltaic power generation power prediction, in particular to a training method of a photovoltaic detection model, a photovoltaic power generation detection method and a related device.
Background
Green buildings and green parks supported by photovoltaic power generation as a power supply are gradually popularized.
At present, a physical model is established based on physical characteristics of a photovoltaic power generation system, a statistical prediction model is applied based on historical data of the photovoltaic power generation system, and photovoltaic power is predicted by utilizing learning of the model and a neural network algorithm.
At present, the weather state is a decisive factor for the photovoltaic power generation power, the weather type division is inaccurate, the generalization capability of the type is not strong, and the weather state is difficult to adapt to the change of the weather state, so that the photovoltaic power generation power has large prediction deviation and low prediction precision, and the safety and stability of the operation of the microgrid are influenced.
Disclosure of Invention
The invention provides a training method of a photovoltaic detection model, a detection method of photovoltaic power generation and a related device, aiming at solving the problem of how to finish the training of the model before predicting the power of the photovoltaic power generation.
In a first aspect, an embodiment of the present invention provides a training method for a photovoltaic detection model, where the method includes:
acquiring original power generation power recorded when a photovoltaic power station generates power in a plurality of time periods, wherein each time period is associated with weather data;
completing the original power generation data in each time period to obtain target power generation power;
classifying all of the weather data into a plurality of categories;
and aiming at each category, training a photovoltaic detection model by taking the weather data in the same time period as a sample and the target power generation power as a label.
In a second aspect, an embodiment of the present invention further provides a detection method for photovoltaic power generation, where the method includes:
acquiring weather data predicted by the photovoltaic power station in a future time period;
identifying a category to which the weather data belongs;
loading a photovoltaic detection model trained for the class by the method of any one of claims 1-7;
and inputting the weather data into the photovoltaic detection model for processing to obtain the generated power of the photovoltaic power station for generating within the time period.
In a third aspect, an embodiment of the present invention further provides a training apparatus for a photovoltaic detection model, where the apparatus includes:
the system comprises an original generating power acquisition module, a weather data acquisition module and a weather data acquisition module, wherein the original generating power acquisition module is used for acquiring original generating power recorded when a photovoltaic power station generates power in a plurality of time periods, and each time period is associated with weather data;
the original power generation data completion module is used for completing the original power generation data in each time period to obtain target power generation power;
the weather data dividing module is used for dividing all the weather data into a plurality of categories;
and the photovoltaic detection model training module is used for training a photovoltaic detection model by taking the weather data in the same time period as a sample and the target power generation power as a label for each category.
In a fourth aspect, an embodiment of the present invention further provides a detection apparatus for photovoltaic power generation, where the apparatus includes:
the prediction data acquisition module is used for acquiring weather data predicted by the photovoltaic power station in a future time period;
the category identification module is used for identifying the category to which the weather data belongs;
a model loading module for loading a photovoltaic detection model trained for the class by the method of any one of claims 1-7;
and the weather data processing module is used for inputting the weather data into the photovoltaic detection model for processing to obtain the power generation power of the photovoltaic power station for power generation in the time period.
In a fifth aspect, an embodiment of the present invention further provides a computer device, where the computer device includes:
one or more processors;
a memory for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of training a photovoltaic detection model according to the first aspect or a method of detecting photovoltaic power generation according to the second aspect.
In a sixth aspect, the embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for training a photovoltaic detection model according to the first aspect or the method for detecting photovoltaic power generation according to the second aspect.
In the embodiment, the original power generation power recorded when the photovoltaic power station generates power in a plurality of time periods is obtained, and weather data is associated with each time period; completing the original power generation data in each time period to obtain target power generation power; dividing all weather data into a plurality of categories; and aiming at each category, training a photovoltaic detection model by taking weather data in the same time period as a sample and target power generation power as a label. On one hand, the weather types are more accurately divided by adopting the s-Kohonen neural network clustering, the weather type clustering device adapts to the change of different weathers, and has better generalization capability. On the other hand, a convolutional neural network is introduced on the basis of a long-term and short-term memory network model, so that the spatial characteristics of data can be effectively extracted under the condition of extreme weather conditions, and the prediction accuracy of photovoltaic power is remarkably improved.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present invention, nor do they necessarily limit the scope of the invention. Other features of the present invention will become apparent from the following description.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a training method of a photovoltaic detection model according to an embodiment of the present invention;
FIG. 2 is a flow chart of a detection method of photovoltaic power generation according to a second embodiment of the invention;
fig. 3 is a structural block diagram of a training apparatus for a photovoltaic detection model according to a third embodiment of the present invention;
FIG. 4 is a block diagram of a photovoltaic power generation detection apparatus according to a fourth embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to a fifth embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in other sequences than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example one
Fig. 1 is a flowchart of a training method for a photovoltaic detection model according to an embodiment of the present invention, where the embodiment is applicable to a situation of preparation for training the detection model before predicting photovoltaic power generation power in a microgrid, and the method may be executed by a training apparatus for the photovoltaic detection model, where the training apparatus for the photovoltaic detection model may be implemented by software and/or hardware, and may be configured in computer equipment, such as a server, a workstation, a personal computer, and the like, and specifically includes the following steps:
fig. 1 is a flowchart of a training method for a photovoltaic detection model according to an embodiment of the present invention, where this embodiment is applicable to find a situation of an optimal thickness of a metal shell of an electrical core, and the method may be executed by a training apparatus for the photovoltaic detection model, and the power battery safety apparatus may be implemented in a form of hardware and/or software, and may be configured in an electronic device. As shown in fig. 1, the method includes:
Photovoltaic power generation is a technology for directly converting light energy into electric energy by utilizing the photovoltaic effect of a semiconductor interface. The solar energy power generation system mainly comprises a solar panel (assembly), a controller and an inverter, and the main components of the system are electronic components.
The micro power grid belongs to a brand-new power grid structure, the photovoltaic power generation technology is used as the technical support of the micro power grid structure, the photovoltaic power generation technology can be well applied to the micro power grid, and compared with other energy sources, the photovoltaic power generation system has the following advantages: solar energy is sustainable energy, and can be recycled without consuming fuel. The power output can be more stable as long as the sun is irradiated. When the external grid is powered off, the photovoltaic power generation can continuously output power, and the micro power grid is switched to the isolated grid independent operation mode and supplies power to some important users continuously. Solar energy is clean energy, and the photovoltaic power generation technology can be built on a roof, so that the cost is low and the construction period is short. The micro power grid has a dispatching management function, and each unit can realize power balance of the power load, so that the system can be operated more highly, stably and safely.
The micro-grid is a concept relative to a traditional large grid, and refers to a network formed by a plurality of distributed power supplies and related loads according to a certain topological structure, and is associated to a conventional grid through a static switch. Before the micro-grid meets the requirements of stable operation and accurate control, the photovoltaic power generation power needs to be predicted.
At present, methods for predicting photovoltaic power generation power are divided into three types. The first type of photovoltaic power generation power prediction method is to establish a physical model based on physical characteristics of a photovoltaic power generation system. The second type of photovoltaic power generation power prediction method is based on historical data of a photovoltaic power generation system and applies a statistical prediction model. The third type of photovoltaic power generation power prediction method is based on historical data of a photovoltaic power generation system and utilizes machine learning and neural network algorithms.
However, the prediction accuracy of the three methods on the photovoltaic power generation power cannot meet the current requirement on the fine control of the micro power grid, and a finer weather type can be represented by improving a weather clustering method, so that the prediction accuracy of the photovoltaic power generation power is improved.
Before the prediction precision of the photovoltaic power generation power is improved, a model for photovoltaic power generation detection is trained. The method comprises the steps of firstly, obtaining original generating power recorded when a photovoltaic power station generates power in a plurality of time periods, wherein weather data are associated with each time period. The photovoltaic power station is a power generation system that directly converts solar radiation energy into electric energy by using photovoltaic effect of photovoltaic cells, and generally includes a transformer, an inverter, a photovoltaic guideline, and related auxiliary facilities. The method comprises the steps of obtaining original power generation power recorded in power generation of a photovoltaic power station in a plurality of time periods, wherein weather data are associated with each time period. The time period may be 24 hours. The weather data includes air temperature, light, precipitation, etc. The original power generation power is power data collected by the photovoltaic power generation system. And the original generated power is denoted by the letter Y.
And 102, completing the original power generation data in each time period to obtain target power generation power.
The historical data includes the raw generated power acquired, meteorological data (air temperature, light, precipitation, etc.). The acquired historical data may have partial deletion or abnormity in links of measurement, transmission, conversion and the like. The existence of missing or abnormal historical data can affect the prediction accuracy of the photovoltaic power generation prediction model on the photovoltaic power generation power. Therefore, by detecting the acquired history data, the history data having a missing or abnormal state is selected by the detection. The detection method comprises the following steps:
in one embodiment of the present invention, step 102 may comprise the steps of:
and step 1021, calculating the distance of the interval between the original generated powers in each time period.
Determining each time period, wherein each time period can be 24 hours, taking each time period as a sample for subsequent processing, firstly determining the number of the time periods in the sample to be represented by a letter k, and the value of k can be in a range of 10 to 50. Further, sample data Y in a certain time period is calculated respectively i And calculating the distance between the original generating power and all data in the sample data set Y by the following formula:
wherein, Y i Representing said raw generated power, Y, in the i-th time period j Representing said original generated power, y, in a j-th time period ir The r-th primary power generation function, y, representing the rate in the i-th time period jr Representing the r-th raw generated power in the j-th time period.
Step 1022, selecting a plurality of original generated powers within a plurality of other time periods with the smallest distance as candidate generated powers for the original generated power within the current time period.
After the distance between the original generating powers in each time period is calculated, a plurality of original generating powers in other time periods with the smallest distance are selected as candidate generating powers for the original generating power in the current time period. And using the candidate generated power for obtaining the target generated power.
And 1023, calculating an average value of the candidate generating power to obtain average generating power, and determining that the original generating power in the non-missing time period is complete as target generating power.
Calculating the average value of the original generating power in a plurality of other time periods with the minimum distance, taking the calculated average value as the average generating power, determining the integrity of the original generating power in the non-missing time period according to the average generating power, and taking the original generating power in the non-missing time period as the target generating power.
Step 1024, determining that the original generated power in a plurality of time periods with the maximum average generated power is absent according to the original generated power in all time periods.
And determining the average value of all candidate generated power, namely the lack of the original generated power in a plurality of time periods with the maximum average generated power, in the original generated power in all time periods by taking the target generated power as a reference object. That is, among the raw generated power in all time periods, the raw generated power in a plurality of times at which the average generated power is maximum is determined to be absent.
And 1025, complementing the original generated power in the missing time period by taking the target generated power as a reference to obtain a new target generated power.
And completing the original power generation data in each time period to obtain target power generation, completing the original power in the missing time period by taking the target power generation as reference, and taking the completed original data as new target power generation.
In one embodiment of the present invention, step 1025 may comprise the steps of:
step 10251, calculating a maximum mutual information coefficient between the target generated power and the original generated power in the time period in which the loss exists.
The maximum mutual information coefficient can measure the correlation, the degree of correlation, and the linear or nonlinear strength between the target generated power and the original generated power which is absent in the time period.
In each time period, the historical data is sorted. And according to the calculation formula of the maximum mutual information coefficient, calculating the maximum mutual information coefficient between the target generating power and the original generating power which is lost in the time period. The maximum mutual information coefficient is calculated according to the following formula:
where x denotes a sequence in which the missing original generated power exists, and Y denotes a sequence of target generated power in the original generated power Y in the data set.
After the maximum mutual information coefficient between the target generating power and the original generating power with the deficiency in the time period is calculated, a plurality of time periods with the highest correlation degree between the target generating power and the original generating power with the deficiency in the time period are selected as reference according to the maximum mutual information coefficient. Based on the target generated power with high correlation and the original generated power with absence in the time period, a least square fitting method is used for curve fitting.
And step 10252, selecting a plurality of target generating powers without missing with the highest maximum mutual information coefficient as characteristic generating powers aiming at the original generating powers in the current missing time period.
After fitting is carried out on the basis of the target generated power with high correlation and the original generated power with deficiency in the time period by using a least square fitting curve, and a plurality of pieces of non-deficient target generated power with the highest maximum mutual information coefficient are selected as the characteristic generated power aiming at the original generated power in the time period with deficiency.
Step 10253, fitting the raw generated power in the time period in which the absence currently exists using a least squares method.
The least square fitting method is a kind of mathematical approximation and optimization, which uses the known data to obtain a curve, so that the square sum of the distance between the curve and the known data on the coordinate system is the minimum, and the data analysis of the fitted curve can be conveniently performed by using a function.
The raw generated power in the time period in which the absence currently exists is fitted using a least squares method. And judging whether the original generating power in the time period is completed or not through fitting.
Step 10254, if the square of the error between the original generated power and the target generated power in the fitted time period is minimum, then the original generated power in the fitted time period is confirmed to be completed and is used as the new target generated power.
And when the square of the error of the fitting curve is minimum, fitting the original generating power with the maximum mutual information coefficient being the highest by using a least square method by taking the target generating power as reference, namely, utilizing the fitting curve to calculate the value of the original generating power with missing in the time period to complement and obtain the target generating power.
When the square between the original generated power and the target generated power in the fitted time period is the minimum, it can be confirmed that the original generated power in the fitted time period is completed, and the completed original generated power is used as the new target generated power.
And step 1026, performing normalization processing on all the target generated power.
And performing normalization processing on all the target generating powers, wherein the normalization processing is to map the data of all the target generating powers into an interval from 0 to 1, so that all the target generating powers with different dimensions are converted into dimensionless data. The conversion formula is as follows:
wherein x' represents the history data after normalization, and x represents the history data before normalization.
And 103, dividing all weather data into a plurality of categories.
And carrying out normalization processing on the maximum mutual information coefficient obtained by calculation. The normalization process can unify all the characteristics in the data to be in a roughly same numerical value interval, and the normalization process ensures that the characteristics in different dimensions have certain comparability on numerical values, so that the classification accuracy and precision can be greatly improved. Dividing all weather data into multiple categories using normalization
In one embodiment of the present invention, step 103 may comprise the steps of:
step 1031, loading a self-organizing feature mapping neural network, wherein each node in the self-organizing feature mapping neural network has an initial weight.
The self-organizing feature mapping neural network mainly performs region classification on input vectors. The structure is similar to that of a competitive neural network. The self-organizing feature mapping neural network comprises an input layer network and an output layer network, but the existing side inhibition phenomenon can be better simulated by introducing a topological structure of the network into the output layer. The input neurons and the output neurons are connected through the weight, and meanwhile, the adjacent output neurons are connected through the weight vector. The output neurons are placed in one, two, or even multi-dimensional grid nodes, most commonly two-dimensional topologies. The self-organizing feature mapping neural network can automatically find out the similarity with the input weather data, and the similar weather input data is configured nearby on the network. It is therefore a possibility to construct a network that selectively gives weather input data to the respective network.
The self-organizing feature mapping learning step is divided into a learning stage and a clustering stage. In the learning stage, weather data is input and randomly selected as training data, winning neurons are selected according to Euclidean distance, and weights of the winning neurons and field neurons are updated. In the clustering stage, the tested weather data is mapped into neurons, and similar weather data will be mapped into neighboring neurons.
And when the weather data are divided into a plurality of categories, loading a self-organizing feature neural network and initializing the self-organizing feature neural network, wherein each node in the self-organizing feature mapping neural network has an initial weight.
Step 1032, calculating the distance between the weather data and the neurons of the competition layer in the self-organizing feature mapping neural network.
Weather data are input into the self-organizing feature neural network, and the distance between the weather data and the neurons of a competition layer in the self-organizing feature neural network is calculated. Namely calculating the distance between the neuron of the distance mapping layer and the weather input data vector of the weight vector of the mapping layer and the weather input data vector.
Step 1033, selecting the neuron of the output layer in the self-organizing feature mapping neural network as the target neuron when the distance is the minimum value.
The neuron that minimizes the distance between the weather input data vector and the weight vector is calculated and selected, and is referred to as the winning neuron. Further, when the distance is the minimum value, the neuron of the output layer in the self-organizing feature mapping neural network is selected, namely the winning neuron. The winning neuron is set as a target neuron.
Step 1034, correcting the target neuron and the weight of the nodes contained in the neighborhood of the target neuron, wherein the neighborhood of the target neuron is a range of which the Euclidean distance from the target neuron is smaller than the radius of the neighborhood.
And correcting the weights of the target neurons and nodes contained in the field of the target neurons by taking the obtained target neurons as a reference, wherein the field of the target neurons is a range in which Euclidean distance between the target neurons is smaller than the radius of the field.
1035, judging whether the weight is smaller than a preset threshold value, if so, determining the type of the self-organizing feature mapping neural network output; if not, returning to execute steps 1031-1034.
The weights of the target neuron and the nodes included in the field of the target neuron are corrected. And judging whether the weight of the modified node is smaller than a preset threshold value. And when the weight value of the node is smaller than a preset threshold value, determining the type of the self-organizing feature mapping neural network output. And if the weight of the node after correction is not less than the preset threshold, returning to execute steps 1031 to 1034.
And 104, aiming at each category, training a photovoltaic detection model by taking weather data in the same time period as a sample and target power generation power as a label.
And training the photovoltaic detection model by using the classified weather data as a sample and the target power generation power as a label in the same time period.
The photovoltaic detection model comprises a convolutional neural network and a long-term and short-term memory network.
The convolutional neural network is a feedforward neural network which comprises convolutional calculation and has a deep structure, and is one of representative algorithms of deep learning. The convolutional neural network has the characteristic learning ability and can carry out translation invariant classification on input information according to the hierarchical structure of the convolutional neural network.
The long-term and short-term memory network is a time-cycle neural network and is specially designed for solving the long-term dependence problem of the common cyclic neural network, and all the cyclic neural networks have a chain form of repeated neural network modules.
Historical data corresponding to the classified weather data are used as training data of a convolutional neural network for extracting data features, the fact that irradiance, temperature and humidity in the weather data are large to photovoltaic power generation power is considered, feature information implicit in the weather data is extracted through the convolutional neural network, and the change process of original photovoltaic power can be reflected more remarkably.
And aiming at each category, training a photovoltaic detection model by taking weather data in the same time period as a sample and target power generation power as a label.
In one embodiment of the present invention, step 104 may include the steps of:
step 1041, inputting the weather data into the convolutional neural network for processing according to each category to obtain weather features.
The convolutional neural network mainly comprises an input layer, a convolutional layer, a pooling layer, a full-link layer and the like. By adding the layers together, a complete convolutional neural network is constructed, and the learning capability of the neural network on long-distance dependence is improved by adopting expansion convolution and hopping connection. And inputting the weather data into the convolutional neural network for processing aiming at each weather category to obtain weather characteristics. The weather data mainly comprises irradiance, temperature and humidity, and the influence of the irradiance, the temperature and the humidity on the photovoltaic power generation power is large.
In one embodiment of the present invention, step 1041 may comprise the steps of:
step 10411, calculating Pearson correlation coefficient between each meteorological factor of weather data and original generated power.
The pearson correlation coefficient is used for measuring whether two data sets are on a line or not and for measuring the linear relation between distance variables, the closer the pearson correlation coefficient is to 1, the more the two data sets are related, and otherwise, the closer the pearson correlation coefficient is to 0, the more the two data sets are not related.
Calculating a Pearson correlation coefficient between each weather data and the target generated power in each time period in the historical data, wherein the Pearson correlation coefficient is calculated according to the following formula:
wherein A represents the meteorological factors of the weather data in the historical data, and B represents the target generated power.
Factor data with a Pearson correlation coefficient close to 1 is selected from the natural gas data in the historical data to serve as an influence factor of the photovoltaic power generation power. And inputting the influence factors into a preset S-Kohonen network to cluster the weather data to obtain a plurality of categories.
The S-Kohonen network is a self-organizing feature mapping neural network, is a meta-supervised learning network, and can identify weather data features and automatically cluster the weather data features.
Illustratively, the weather data is clustered by using an S-Kohonen network, namely, a self-organizing feature mapping neural network clustering method, so as to obtain a plurality of categories. The method comprises the following steps:
1) Initializing the network weight w ij (i =1,2, \8230;, n; j =1,2, \8230;, m), w is the interval [0,1]An inner n x m order random number matrix. w is a ij The constraint condition is satisfied as
2) Calculating the distance between weather data X and competition layer neuron
Wherein epsilon thre (b) Is the interval [0,1]Random number within.
3) And selecting the neurons, and taking the output layer neurons b' meeting the condition that the distance between the weather data and the competition layer neurons is the minimum value as the optimal matching output neurons.
4) Modified neuron b' and its neighborhood N c (t ') the range of the neighborhood is all nodes t ' whose Euclidean distance from the neuron b ' is smaller than the radius of the neighborhood.
5) And returning to the step 2) until the weight adjustment quantity is smaller than the set training threshold.
6) Categorized weather data is obtained.
Step 10412, selecting meteorological factors having a correlation with the original generated power as influence factors by using the pearson correlation coefficient, wherein the influence factors include irradiance, temperature and humidity.
And selecting a meteorological factor having a correlation with the original generated power by using the pearson correlation coefficient obtained by calculation, and using the meteorological factor as an influence factor. Wherein the influence factors include irradiance, temperature, humidity. The fact that the irradiance, the temperature and the humidity in the weather data are large to photovoltaic power generation power is considered, characteristic information implicit in the weather data is extracted through a convolutional neural network, and the change process of the original photovoltaic power can be reflected more remarkably.
Step 10413, for each category, concatenating irradiance, temperature, and humidity into a weather parameter sequence.
And splicing the irradiance, the temperature and the perusal into a weather sequence aiming at each weather category, and obtaining the weather characteristics of the corresponding weather sequence.
And 10414, inputting the weather parameter sequence into the convolutional neural network for processing to obtain weather characteristics.
And splicing the irradiance, the temperature and the humidity into a weather parameter sequence aiming at each category in the classified weather data. And inputting the spliced weather parameter sequence into a convolutional neural network for processing, and further obtaining weather characteristics.
And 1042, inputting the weather characteristics into a long-term and short-term memory network for processing to obtain the predicted generated power generated by the photovoltaic power station under the weather data.
And inputting the obtained weather characteristics into the long-term and short-term memory network for processing in the long-term and short-term memory network to obtain the predicted generated power of the photovoltaic power station under the weather data.
Step 1043, calculating a difference between the predicted generated power and the target generated power as a loss value.
And comparing by using three difference evaluation indexes, namely calculating a root mean square error, an average absolute error and a decision coefficient, wherein when the equivalence classes are different, the calculated root mean square error, the average absolute error and the decision coefficient are difficult to measure the performance of the convolutional neural network and the long-short term memory network, and the decision coefficient is used to reflect the proportion that all the variation of the dependent variable can be explained by an independent variable through a regression relation. The predicted generated power is equal to the value of the target generated power when the decision coefficient is equal to 1.
And step 1044, updating the long-term and short-term memory network and the convolutional neural network according to the loss value.
And updating the long-term memory network and the convolutional neural network according to the difference between the predicted generated power obtained by calculation and the target generated power.
Step 1045, judging whether a preset training condition is met; if yes, determining the long-short term memory network and the convolutional neural network corresponding to the category to finish training; if not, the step 1041 to the step 1044 are executed in a returning mode.
And updating the long-term and short-term memory network and the convolutional neural network according to the loss value, and judging whether the photovoltaic detection model meets a preset training condition. And when the photovoltaic detection model meets the preset training condition, determining the long-short term memory network and the convolutional neural network corresponding to the categories to finish training. And if the photovoltaic detection model does not meet the preset training condition, returning to execute the steps 1041 to 1044.
In the embodiment, the original power generation power recorded when the photovoltaic power station generates power in a plurality of time periods is obtained, and each time period is associated with weather data; completing the original power generation data in each time period to obtain target power generation power; dividing all weather data into a plurality of categories; and aiming at each category, training a photovoltaic detection model by taking weather data in the same time period as a sample and target power generation power as a label. On one hand, the weather types are divided more accurately by adopting the s-Kohonen neural network clustering, the weather type classification method is suitable for the change of different weathers, and the generalization capability is better. On the other hand, a convolutional neural network is introduced on the basis of a long-term and short-term memory network model, so that the spatial characteristics of data can be effectively extracted under the condition of extreme weather conditions, and the prediction accuracy of photovoltaic power is remarkably improved.
Example two
Fig. 2 is a flowchart of a photovoltaic power generation detection method according to a second embodiment of the present invention, where this embodiment is applicable to a situation of detecting photovoltaic power generation power in a microgrid, and the method may be executed by a photovoltaic power generation detection apparatus, where the photovoltaic power generation detection apparatus may be implemented by software and/or hardware, and may be configured in a computer device, such as a server, a workstation, a personal computer, and the like, and specifically includes the following steps:
After the long-short term memory network and the convolutional neural network corresponding to the category are determined to finish training, weather data predicted by the photovoltaic power station in a future time period is obtained,
And identifying the category to which the weather data belongs according to the maximum mutual information coefficient. The self-organizing feature mapping neural network is loaded first. And inputting weather data into the self-organizing feature mapping neural network, and clustering the weather data through the self-organizing feature mapping neural network to obtain the category to which the corresponding weather data belongs.
And step 203, loading the photovoltaic detection model trained aiming at the maximum mutual information coefficient type.
In specific implementation, the method of the first embodiment may be applied to train the photovoltaic detection models in advance for each category, and when the category to which the current weather data belongs is identified, the photovoltaic detection model corresponding to the category is loaded to the memory to operate, so that the detection of the photovoltaic power generation power is completed through the photovoltaic detection model.
In one embodiment of the present invention, step 203 may comprise the steps of:
step 2031, inputting the weather data with the maximum mutual information coefficient into the photovoltaic detection model with the maximum mutual information coefficient for processing, and obtaining the power generation power of the photovoltaic power station with the maximum mutual information coefficient for power generation within the time period of the maximum mutual information coefficient.
The maximum mutual information coefficient photovoltaic detection model comprises a convolutional neural network and a long-term and short-term memory network.
And inputting the weather data with the maximum mutual information coefficient into the photovoltaic detection model with the maximum mutual information coefficient for processing to obtain the power generation power of the photovoltaic power station with the maximum mutual information coefficient for generating power in the time period with the maximum mutual information coefficient, and further inputting the weather data into the convolutional neural network for processing to obtain the weather characteristics.
And step 204, inputting the weather characteristics into the long-term and short-term memory network for processing to obtain the predicted generated power generated by the photovoltaic power station in the time period.
And inputting the obtained weather characteristics into a long-term and short-term memory network for processing to obtain the predicted generated power generated by the photovoltaic power station in the practice period.
Illustratively, weather data predicted in a future time period is obtained, the category to which the weather data belongs is identified according to the maximum mutual information coefficient, a trained maximum mutual information coefficient photovoltaic detection model is loaded, the maximum mutual information coefficient weather data is input into the maximum mutual information coefficient photovoltaic detection model for processing, and then the power generation power of the maximum mutual information coefficient photovoltaic power station in the time period corresponding to the maximum mutual information coefficient is obtained. Further, the photovoltaic detection model comprises a convolutional neural network and a long-term and short-term memory network. The weather characteristics obtained after the weather data are input into the convolutional neural network are respectively constructed for the weather characteristics corresponding to different weather data, and the long-term and short-term memory networks process the weather characteristics to obtain the predicted power generation power of the photovoltaic power station for power generation in a time period.
In the embodiment, weather data predicted for a photovoltaic power plant in a future time period is acquired; identifying a category to which the weather data belongs; loading a photovoltaic detection model trained for the category; and inputting the weather data into the photovoltaic detection model for processing to obtain the power generation power of the photovoltaic power station for power generation in a time period. Aiming at the problem of unstable photovoltaic power generation power prediction accuracy, a convolutional neural network is introduced on the basis of a long-term and short-term memory network, so that the characteristics of weather data can be more effectively extracted under the condition of extreme weather conditions, and the photovoltaic power generation power prediction accuracy is remarkably improved.
EXAMPLE III
Fig. 3 is a block diagram of a training apparatus for a photovoltaic detection model according to a third embodiment of the present invention, where the apparatus may include the following modules:
the original generating power acquiring module 301 is configured to acquire original generating power recorded when the photovoltaic power station generates power in multiple time periods, where each time period is associated with weather data;
an original power generation data completion module 302, configured to complete the original power generation data in each time period to obtain a target power generation power;
a weather data dividing module 303, configured to divide all the weather data into multiple categories;
and a photovoltaic detection model training module 304, configured to train a photovoltaic detection model by using the weather data in the same time period as a sample and the target generated power as a tag for each category.
In one embodiment of the present invention, the raw power generation data completion module 302 comprises:
the distance calculation module is used for calculating the distance of the interval between the original generating powers in each time period;
a candidate generated power obtaining module, configured to select, as candidate generated power, the original generated power in a plurality of other time periods with a smallest distance, for the original generated power in the current time period;
the candidate generating power average value calculating module is used for calculating an average value of the candidate generating power to obtain average generating power, and determining that the original generating power in the time period without loss is complete as target generating power;
an original generating power deficiency determining module, configured to determine, for the original generating power in all the time periods, that the original generating power in a plurality of the time periods in which the average generating power is maximum is deficient;
and the original generating power complementing module is used for complementing the original generating power in the missing time period by taking the target generating power as a reference to obtain new target generating power. And the target generating power normalization processing module is used for performing normalization processing on all the target generating power.
In one embodiment of the present invention, the distance calculation module includes:
calculating a distance of separation between the raw generated power for each of the time periods by the following formula:
wherein Y is i Representing said raw generated power, Y, in the i-th time period j Represents the raw generated power, y, in the jth time period ir The r-th primary power generation function, y, representing the rate in the i-th time period jr Representing the r-th raw generated power in the j-th time period. In one embodiment of the invention, the raw power generation completion module comprises:
maximum mutual information coefficient calculation module for
Calculating a maximum mutual information coefficient between a target generated power and the original generated power in the time period with the deficiency;
a characteristic generated power obtaining module, configured to select, as characteristic generated power, a plurality of non-missing target generated powers with a highest maximum mutual information coefficient for the original generated power in the current missing time period;
the original generating power fitting module is used for fitting the original generating power in the time period with the current missing by using a least square method; and a power complement determination module, configured to determine that the fitted original generated power in the time period is complemented as a new target generated power if a square of an error between the fitted original generated power and the target generated power in the time period is minimum.
In one embodiment of the present invention, the weather data partitioning module 303 includes:
the neural network loading module is used for loading a self-organizing feature mapping neural network, and each node in the self-organizing feature mapping neural network has an initial weight;
the neuron distance calculation module is used for calculating the distance between the weather data and neurons of a competition layer in the self-organizing feature mapping neural network;
a distance maximum value selection module for selecting the neuron of the output layer in the self-organizing feature mapping neural network as a target neuron when the distance is the minimum value;
the weight correction module is used for correcting the weight of the target neuron and nodes contained in the neighborhood of the target neuron, wherein the neighborhood of the target neuron is a range of which the Euclidean distance from the target neuron is smaller than the radius of the neighborhood;
the weight judgment module is used for judging whether the weight is smaller than a preset threshold value or not, and if so, determining the type of the self-organizing feature mapping neural network output; if not, returning to execute the neural network loading module, namely the weight correction module.
In one embodiment of the invention, the photovoltaic detection model comprises a convolutional neural network, a long-short term memory network;
the photovoltaic detection model training module 304 includes:
a weather feature obtaining module, configured to, for each category, input the weather data into the convolutional neural network for processing to obtain a weather feature;
the predicted generating power obtaining module is used for inputting the weather characteristics into the long-term and short-term memory network for processing to obtain the predicted generating power generated by the photovoltaic power station under the weather data;
a loss value calculation module for calculating a difference between the predicted generated power and the target generated power as a loss value;
the loss value updating module is used for updating the long-short term memory network and the convolutional neural network according to the loss value;
the training condition judging module is used for judging whether preset training conditions are met or not; if so, determining that the long-short term memory network and the convolutional neural network corresponding to the category complete training; if not, returning to the execution of the weather characteristic obtaining module, namely the loss value updating module.
In one embodiment of the present invention, the weather feature obtaining module includes:
a coefficient calculation module for calculating Pearson correlation coefficients between each meteorological factor of the weather data and the original generated power;
a meteorological factor selection module for selecting the meteorological factor having a correlation with the original power generation power as influence factors using the Pearson correlation coefficient, wherein the influence factors include irradiance, temperature and humidity;
a weather sequence splicing module for splicing the irradiance, the temperature and the humidity into a weather parameter sequence for each category;
and the weather parameter processing module is used for inputting the weather parameter sequence into the convolutional neural network for processing to obtain weather characteristics.
Example four
Fig. 4 is a block diagram of a detection apparatus for photovoltaic power generation according to a fourth embodiment of the present invention, where the apparatus may include the following modules:
a forecast data acquisition module 401, configured to acquire forecast weather data of the photovoltaic power plant in a future time period;
a category identification module 402, configured to identify a category to which the weather data belongs;
a model loading module 403, configured to load the photovoltaic detection model trained by the training apparatus of the photovoltaic detection model for the category;
and a weather data processing module 404, configured to input the weather data into the photovoltaic detection model for processing, so as to obtain power generation power of the photovoltaic power plant for power generation in the time period.
In one embodiment of the present invention, the category identification module 402 comprises:
the self-organizing feature mapping neural network loading module is used for loading a self-organizing feature mapping neural network;
and the weather data clustering module is used for inputting the weather data into the self-organizing feature mapping neural network to cluster the weather data to obtain the category to which the weather data belongs.
In one embodiment of the invention, the photovoltaic detection model comprises a convolutional neural network, a long-short term memory network;
the weather data processing module 404 includes:
the weather data input module is used for inputting the weather data into the convolutional neural network for processing to obtain weather characteristics;
and the data processing module is used for inputting the weather characteristics into the long-term and short-term memory network for processing to obtain the predicted generating power generated by the photovoltaic power station in the time period.
In one embodiment of the invention, the weather data includes irradiance, temperature, humidity;
the weather data input module includes:
a parameter sequence obtaining module for splicing the irradiance, the temperature and the humidity into a weather parameter sequence;
and the parameter sequence processing module is used for inputting the weather parameter sequence into the convolutional neural network for processing to obtain weather characteristics.
EXAMPLE five
FIG. 5 illustrates a block diagram of an electronic device 10 that may be used to implement an embodiment of the invention. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
As shown in fig. 5, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, and the like, wherein the memory stores a computer program executable by the at least one processor, and the processor 11 can perform various suitable actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from a storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data necessary for the operation of the electronic apparatus 10 may also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.
A number of components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, or the like; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
In some embodiments, the power cell safety method may be implemented as a computer program tangibly embodied on a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the power cell safety method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the power cell safety method by any other suitable means (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
Computer programs for implementing the methods of the present invention can be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be performed. A computer program can execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.
The computing system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical host and VPS service are overcome.
EXAMPLE six
Embodiments of the present invention further provide a computer program product, which includes a computer program, and when executed by a processor, the computer program implements the power battery safety method provided in any embodiment of the present invention.
Computer program product in implementing the computer program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C + +, and including conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It should be understood that various forms of the flows shown above, reordering, adding or deleting steps, may be used. For example, the steps described in the present invention may be executed in parallel, sequentially, or in different orders, and are not limited herein as long as the desired results of the technical solution of the present invention can be achieved.
The above-described embodiments should not be construed as limiting the scope of the invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A training method of a photovoltaic detection model is characterized by comprising the following steps:
acquiring original generating power recorded when a photovoltaic power station generates power in a plurality of time periods, wherein each time period is associated with weather data;
completing the original power generation data in each time period to obtain target power generation power;
classifying all of the weather data into a plurality of categories;
and aiming at each category, training a photovoltaic detection model by taking the weather data in the same time period as a sample and the target power generation power as a label.
2. The method according to claim 1, wherein the complementing the raw power generation data for each of the time periods to obtain a target power generation comprises:
calculating the distance of the interval between the original generating powers in each time period;
selecting a plurality of original generating powers with the minimum distance in a plurality of other time periods as candidate generating powers aiming at the original generating power in the current time period;
calculating an average value of the candidate generating power to obtain an average generating power, and determining that the original generating power in the time period which is not lost is complete to be used as a target generating power;
determining, for the raw generated power over all of the time periods, that the raw generated power is missing for a plurality of the time periods in which the average generated power is maximum;
complementing the original generating power in the missing time period by taking the target generating power as a reference to obtain new target generating power;
and carrying out normalization processing on all the target generating powers.
3. The method of claim 2, wherein said calculating a distance of separation between said raw generated power for each of said time periods comprises:
calculating a distance of separation between the raw generated power for each of the time periods by the following formula:
wherein Y is i Representing said raw generated power, Y, in the i-th time period j Represents the raw generated power, y, in the jth time period ir The r-th primary power generation function, y, representing the rate in the i-th time period jr Representing the r-th said raw generated power in the j-th time period.
4. The method according to claim 2, wherein the complementing the original generated power in the time period in which the deficiency exists with reference to the target generated power to obtain a new target generated power comprises:
calculating a maximum mutual information coefficient between a target generated power and the original generated power in the time period in which the loss exists;
selecting a plurality of un-missing target generated powers with the highest maximum mutual information coefficient as characteristic generated powers aiming at the original generated power in the time period with missing at present;
fitting the original generated power in the time period in which the absence currently exists using a least squares method;
and if the square of the error between the original generating power and the target generating power in the fitted time period is minimum, determining that the original generating power in the fitted time period is completed and the completed original generating power is used as new target generating power.
5. The method of claim 1, wherein the classifying all of the weather data into a plurality of categories comprises:
loading a self-organizing feature mapping neural network, wherein each node in the self-organizing feature mapping neural network has an initial weight;
calculating the distance between the weather data and neurons of a competition layer in the self-organizing feature mapping neural network;
selecting a neuron of an output layer in the self-organizing feature mapping neural network when the distance is the minimum value as a target neuron; a (c)
Correcting the target neuron and the weight of a node contained in the neighborhood of the target neuron, wherein the neighborhood of the target neuron is a range of which the Euclidean distance from the target neuron is smaller than the radius of the neighborhood;
judging whether the weight is smaller than a preset threshold value, if so, determining the type of the self-organizing feature mapping neural network output; and if not, returning to execute the calculation of the distance between the weather data and the neurons of the competition layer in the self-organizing feature mapping neural network.
6. The method of any one of claims 1-5, wherein the photovoltaic detection model comprises a convolutional neural network, a long-short term memory network;
the training of the photovoltaic detection model by taking the weather data in the same time period as a sample and the target generated power as a label for each category comprises: inputting the weather data into the convolutional neural network for processing aiming at each category to obtain weather characteristics;
inputting the weather characteristics into the long-term and short-term memory network for processing to obtain the predicted generating power generated by the photovoltaic power station under the weather data;
calculating a difference between the predicted generated power and the target generated power as a loss value;
updating the long-short term memory network and the convolutional neural network according to the loss value;
judging whether preset training conditions are met or not; if so, determining that the long-short term memory network and the convolutional neural network corresponding to the category complete training; if not, returning to execute the operation aiming at each category, and inputting the weather data into the convolutional neural network for processing to obtain weather characteristics.
7. The method according to claim 6, wherein the inputting the weather data into the convolutional neural network for processing for each of the categories to obtain weather features comprises:
calculating Pearson correlation coefficients between each meteorological factor of the weather data and the original generated power;
selecting the meteorological factors having a correlation with the original generated power as influence factors by using the Pearson correlation coefficient, wherein the influence factors comprise irradiance, temperature and humidity;
for each of the categories, stitching the irradiance, the temperature, and the humidity into a sequence of weather parameters;
and inputting the weather parameter sequence into the convolutional neural network for processing to obtain weather characteristics.
8. A detection method for photovoltaic power generation is characterized by comprising the following steps:
acquiring weather data predicted for a photovoltaic power station in a future time period;
identifying a category to which the weather data belongs;
loading a photovoltaic detection model trained for the class by the method of any one of claims 1-7;
and inputting the weather data into the photovoltaic detection model for processing to obtain the generated power of the photovoltaic power station for generating within the time period.
9. A computer device, characterized in that the computer device comprises:
one or more processors;
a memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method of training a photovoltaic detection model as claimed in any one of claims 1 to 7 or a method of detecting photovoltaic power generation as claimed in claim 8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, implements a method of training a photovoltaic detection model as claimed in any one of claims 1 to 7 or a method of detecting photovoltaic power generation as claimed in claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211428658.3A CN115759395A (en) | 2022-11-15 | 2022-11-15 | Training of photovoltaic detection model, detection method of photovoltaic power generation and related device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211428658.3A CN115759395A (en) | 2022-11-15 | 2022-11-15 | Training of photovoltaic detection model, detection method of photovoltaic power generation and related device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115759395A true CN115759395A (en) | 2023-03-07 |
Family
ID=85371777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211428658.3A Pending CN115759395A (en) | 2022-11-15 | 2022-11-15 | Training of photovoltaic detection model, detection method of photovoltaic power generation and related device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115759395A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504005A (en) * | 2023-05-09 | 2023-07-28 | 齐鲁工业大学(山东省科学院) | Perimeter security intrusion signal identification method based on improved CDIL-Bi-LSTM |
-
2022
- 2022-11-15 CN CN202211428658.3A patent/CN115759395A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504005A (en) * | 2023-05-09 | 2023-07-28 | 齐鲁工业大学(山东省科学院) | Perimeter security intrusion signal identification method based on improved CDIL-Bi-LSTM |
| CN116504005B (en) * | 2023-05-09 | 2024-02-20 | 齐鲁工业大学(山东省科学院) | Perimeter security intrusion signal identification method based on improved CDIL-Bi-LSTM |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962364B (en) | Multi-factor power load prediction method based on deep learning | |
| CN110909912B (en) | Park electric power system net load combination prediction method based on self-adaptive error feedback | |
| Feng et al. | A taxonomical review on recent artificial intelligence applications to PV integration into power grids | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN107944594B (en) | Short-term load prediction method based on spearman grade and RKELM microgrid | |
| CN114792156B (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| CN110135635A (en) | A kind of region electric power saturation load forecasting method and system | |
| CN116307652A (en) | Artificial intelligent resource allocation method for intelligent power grid | |
| Liu et al. | Photovoltaic generation power prediction research based on high quality context ontology and gated recurrent neural network | |
| CN113627735A (en) | Early warning method and system for safety risk of engineering construction project | |
| CN115358437A (en) | Power supply load prediction method based on convolutional neural network | |
| CN116578551A (en) | GRU-GAN-based power grid data restoration method | |
| CN115759395A (en) | Training of photovoltaic detection model, detection method of photovoltaic power generation and related device | |
| CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
| CN115660232A (en) | Ultra-short-term prediction method, device and system for wind power | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| CN109858667A (en) | It is a kind of based on thunder and lightning weather to the short term clustering method of loading effects | |
| CN115860190A (en) | Training of load detection model, method for detecting power load and related device | |
| CN106816871B (en) | State similarity analysis method for power system | |
| KR20230072219A (en) | A system for determining the optimal bid amount of a stochastic scenario technique using weather forecast data | |
| Silva et al. | Application of a hybrid neural fuzzy inference system to forecast solar intensity | |
| Shen et al. | An interval analysis scheme based on empirical error and MCMC to quantify uncertainty of wind speed | |
| El Marghichi | A Solar PV Model Parameter Estimation Based on the Enhanced Self-Organization Maps | |
| CN112801356A (en) | Power load prediction method based on MA-LSSVM | |
| Bazine et al. | Prediction of photovoltaic production for smart grid energy management using hidden Markov model: a study case |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |