CN-122026328-A - New energy generated power prediction method and system based on meteorological data

Abstract

The invention provides a new energy power generation power prediction method and a new energy power generation power prediction system based on meteorological data, which relate to the technical field of new energy power generation, firstly, a meteorological data set (including continuous data acquisition of multiple types of meteorological elements) and a working condition data set (including operation parameter data of the equipment in different operation states) corresponding to new energy power generation equipment are obtained. And then carrying out association analysis on the meteorological data set, and establishing a meteorological element time sequence association diagram to represent the time dimension mutual influence relationship of the meteorological elements. And then, the meteorological element time sequence association diagram is mapped with the working condition data set in an association mode, and working condition response association characteristics are generated. And constructing a power generation power linkage prediction model based on the working condition response association characteristics, and predicting the real-time meteorological element data to obtain an initial prediction result. And finally, the initial prediction result and the historical power generation data are subjected to correlation correction to generate a final prediction result, so that the accurate prediction of the power generation power of the new energy is realized.

Inventors

• HUAN JIAFEI
• JIANG JIXUAN
• Jiang Shangguang
• LIU YU
• LI HAOXING

Assignees

• 国家电网有限公司华北分部

Dates

Publication Date

20260512

Application Date

20260203

Claims (10)

1. 1. The new energy generated power prediction method based on meteorological data is characterized by comprising the following steps: Acquiring a meteorological data set corresponding to new energy power generation equipment and a working condition data set of the new energy power generation equipment, wherein the meteorological data set comprises continuous acquisition data of multiple types of meteorological elements, and the working condition data set comprises operation parameter data of the new energy power generation equipment in different operation states; performing association analysis processing on continuous acquired data of multiple types of meteorological elements in the meteorological data set, and establishing a meteorological element time sequence association diagram, wherein the meteorological element time sequence association diagram is used for representing the mutual influence relationship of different meteorological elements in a time dimension; Performing association mapping processing on the meteorological element time sequence association diagram and the working condition data set to generate working condition response association characteristics of the new energy power generation equipment, wherein the working condition response association characteristics are used for representing the corresponding relation between meteorological element changes and equipment operation parameter changes; constructing a power generation linkage prediction model based on the working condition response association characteristics, and performing prediction analysis processing on real-time meteorological element data in the meteorological data set through the power generation linkage prediction model to obtain a new energy power generation initial prediction result; and carrying out association correction processing on the initial prediction result of the new energy generated power and the historical generated power data in the working condition data set to generate a final prediction result of the new energy generated power.
2. 2. The new energy generated power prediction method based on meteorological data according to claim 1, wherein the performing association analysis processing on continuous collected data of multiple types of meteorological elements in the meteorological data set, and establishing a meteorological element time sequence association diagram, includes: extracting time sequence information of continuous data acquisition of multiple types of meteorological elements in the meteorological data set, determining time acquisition intervals and time sequence lengths of continuous data acquisition of each type of meteorological elements, and marking abnormal time nodes in the time sequence information, wherein the abnormal time nodes are time nodes corresponding to data acquisition interruption or data missing; the marked time sequence information is subjected to complement processing, and weather element data corresponding to the abnormal time node is supplemented according to the change trend of weather element data before and after the abnormal time node to form complete time sequence information; calculating a degree of correlation parameter of the same time node between different meteorological elements continuously acquired data based on the completed time sequence information, wherein the degree of correlation parameter is used for quantifying the degree of interaction between different meteorological elements; According to the relevancy parameters, an initial meteorological element relevancy graph is constructed, nodes in the initial meteorological element relevancy graph are meteorological elements of different types, connecting lines among the nodes are corresponding relevancy parameters, invalid connecting lines in the initial meteorological element relevancy graph are deleted, and the invalid connecting lines are connecting lines with relevancy parameters lower than a preset relevancy threshold value; performing time sequence dimension expansion processing on the initial weather element association graph after invalid connection is deleted, and connecting the initial weather element association graphs of different time nodes in series according to time sequence to form a weather element time sequence association frame containing time dimension information; In the meteorological element time sequence association frame, supplementing time dimension association edges according to the change trend of meteorological element association degree parameters among different time nodes, and marking corresponding time span information for each time dimension association edge, wherein the time dimension association edges are used for representing influence relationship information of the same meteorological element among different time nodes; Integrating the initial weather element association diagram, the weather element time sequence association frame and the time dimension association side to generate the weather element time sequence association diagram, wherein the weather element time sequence association diagram comprises node type information, inter-node association degree parameter information, time sequence association side information and time span information.
3. 3. The method for predicting new energy generated power based on meteorological data according to claim 2, wherein calculating a correlation parameter between continuously collected data of different meteorological elements at the same time node based on the completed time series information comprises: Extracting all the same time nodes from the completed time sequence information, determining the numerical value of different meteorological elements continuously collected data corresponding to each same time node, and establishing a time node-meteorological element numerical value correspondence table; Selecting two kinds of different continuous acquisition data of meteorological elements as a group of analysis objects, selecting the continuous acquisition data of the meteorological elements of the wind type and the continuous acquisition data of the meteorological elements of the light type as a first group of analysis objects, selecting the continuous acquisition data of the meteorological elements of the wind type and the continuous acquisition data of the meteorological elements of the air pressure type as a second group of analysis objects, and selecting the continuous acquisition data of the meteorological elements of the light type and the continuous acquisition data of the meteorological elements of the air pressure type as a third group of analysis objects; For each group of analysis objects, extracting the numerical value of the group of analysis objects at each same time node from the time node-meteorological element numerical value corresponding table to form a numerical value sequence pair of the group of analysis objects; calculating covariance of each group of analysis object value sequences under each same time node, wherein the covariance is used for reflecting the change trend consistency of two types of meteorological element data values, carrying out validity judgment on the calculated covariance, and eliminating covariance values with absolute values exceeding a preset covariance range; According to the covariance after the validity judgment, further calculating a correlation coefficient, wherein the correlation coefficient is the ratio of the covariance to the standard deviation product of two types of meteorological element data, and recording the correlation coefficient of each group of analysis objects at each same time node; Carrying out normalization processing on the correlation coefficient, and converting the value range of the correlation coefficient into between 0 and 1 by adopting a linear normalization method, wherein the normalized correlation coefficient is the correlation parameter; According to the steps, the association degree parameters of all the groups of analysis objects at each same time node are calculated to form an association degree parameter sequence of each group of analysis objects.
4. 4. The method for predicting new energy generated power based on meteorological data according to claim 1, wherein the performing the association mapping processing on the meteorological element time sequence association diagram and the working condition data set to generate the working condition response association characteristic of the new energy generating device includes: extracting operation parameter change information in the working condition data set, filtering abnormal change data in the operation parameter change information, wherein the abnormal change data is data with numerical value change amplitude exceeding a preset amplitude range, the operation parameter change information comprises numerical value change amplitude and numerical value change speed of the operation parameter, the numerical value change amplitude is a difference value of the operation parameter values of different time nodes, and the numerical value change speed is a ratio of the numerical value change amplitude to a time interval; Calculating the change rate of the meteorological element value and the change rate of the operation parameter value, wherein the change rate is defined as the relative change quantity of the value relative to a reference value; establishing a mapping relation table, correspondingly matching the weather element numerical value change rate with the operating parameter numerical value change rate, and labeling corresponding time span information for each matching item; Optimizing the mapping relation table, calculating the matching degree of each matching item, deleting the matching item with the matching degree lower than a preset matching threshold value, and obtaining an optimized mapping relation table, wherein the matching degree is the corresponding fitting degree of the weather element numerical value change rate and the operating parameter numerical value change rate; calculating probability distribution of the numerical change rate of the operation parameter corresponding to the numerical change rate of each meteorological element based on the optimized mapping relation table, wherein the probability distribution is used for representing the possibility of different numerical change rates of the operation parameter under the numerical change rate of the specific meteorological element; weighting the running parameter change information according to the probability distribution and the relevancy parameters in the meteorological element time sequence relevancy graph, and obtaining weighted running parameter change information after weighting, wherein the weight value is the corresponding relevancy parameter; And combining the weighted operation parameter change information with the time sequence association direction and the time span information in the meteorological element time sequence association diagram to generate the working condition response association characteristic containing the time dimension change trend, wherein the working condition response association characteristic contains a meteorological element change sequence, a corresponding operation parameter change sequence, an association weight sequence and a time span sequence of the meteorological element change sequence and the corresponding operation parameter change sequence.
5. 5. The method for predicting new energy generated power based on meteorological data according to claim 4, wherein the calculating the probability distribution of the variation amplitude of the numerical value of the operation parameter corresponding to the variation range of the numerical value of each meteorological element based on the optimized mapping relation table comprises: Extracting a meteorological element numerical value change rate from the optimized mapping relation table as a current analysis change rate, determining all operation parameter numerical value change rates corresponding to the current analysis change rate, and forming an operation parameter numerical value change rate set of the current analysis change rate; Grouping the operation parameter numerical value change rate set, and dividing the operation parameter numerical value change rate into a plurality of change rate sections according to equal intervals of the change rate; counting the number of the operating parameter numerical change rates contained in each change rate interval, and calculating the proportion of the number in each change rate interval to the total number of the operating parameter numerical change rate sets, wherein the proportion is the probability value that the operating parameter numerical change rate is in the change rate interval; Recording each change rate interval and the corresponding probability value thereof correspondingly to form a probability distribution table corresponding to the change rate of the meteorological element numerical value; integrating all probability distribution tables, marking the corresponding weather element type and time span information for each probability distribution table, and forming a probability distribution set, wherein the probability distribution set comprises probability distribution information, weather element type information and time span information corresponding to the numerical change rate of each weather element.
6. 6. The method for predicting the new energy generated power based on the meteorological data according to claim 1, wherein the constructing a generated power linkage prediction model based on the condition response association features, performing prediction analysis processing on real-time meteorological element data in the meteorological data set through the generated power linkage prediction model, and obtaining an initial prediction result of the new energy generated power comprises: Extracting a meteorological element change sequence, an operation parameter change sequence, an association weight sequence and a time sequence span sequence in the working condition response association characteristic, taking the meteorological element change sequence as a model input characteristic, the operation parameter change sequence as a model intermediate characteristic, the association weight sequence as a model weight parameter, and the time sequence span sequence as a model time dimension adjustment parameter; The method comprises the steps of constructing a network structure of a power generation linkage prediction model, wherein the network structure comprises an input layer, an intermediate feature processing layer, a time dimension adjusting layer and an output layer, the input layer is used for receiving input features of the model, the intermediate feature processing layer is used for carrying out association operation processing on the intermediate features of the model, the time dimension adjusting layer is used for adjusting the sensitivity of the model to time dimension information according to time dimension adjusting parameters, and the output layer is used for outputting a power generation prediction value; The associated weight sequence is imported into the intermediate feature processing layer and used as a weighting coefficient in the intermediate feature operation process, so that the intermediate feature processing layer performs feature strengthening or weakening processing according to the weight value of the associated weight sequence when processing the operation parameter change sequence; importing the time sequence span sequence into the time dimension adjusting layer, and adjusting the response weight of the model to the weather element change under different time spans through the time dimension adjusting layer; Training the power generation power linkage prediction model, wherein training data are historical meteorological element data and corresponding historical power generation power data in the meteorological data set, and the training target is to enable deviation between a power generation power prediction value output by the model and the historical power generation power data to be in a preset range; After training is completed, inputting real-time meteorological element data in the meteorological data set into an input layer of the power generation linkage prediction model, performing operation processing on operation parameter changes corresponding to the real-time meteorological element data through the intermediate feature processing layer, adjusting time dimension information in an operation process through the time dimension adjusting layer, outputting corresponding power generation predicted values through the output layer, and arranging the power generation predicted values corresponding to the real-time meteorological element data according to time sequence to form a new energy power generation initial predicted result.
7. 7. The method for predicting the power generated by new energy based on meteorological data according to claim 6, wherein the training the power generation linkage prediction model, wherein the training data is historical meteorological element data and corresponding historical power generation data in the meteorological data set, and the training target is to make the deviation between the power generation predicted value output by the model and the historical power generation data be within a preset range, and the method comprises the following steps: Extracting historical meteorological element data from the meteorological data set, wherein the historical meteorological element data are continuously acquired data of meteorological elements acquired in a preset historical time period, the historical meteorological element data are divided into a training data set, a verification data set and a test data set according to a time sequence, and the dividing proportion is determined according to a preset data set distribution proportion; Extracting historical power generation data corresponding to the historical meteorological element data from the working condition data set, enabling the historical power generation data to correspond to the historical meteorological element data one by one on a time node, and dividing the historical power generation data into a training tag set, a verification tag set and a test tag set according to the data set distribution proportion; Performing data enhancement processing on the training data set and the training label set, generating newly-added training data and corresponding newly-added training labels by disturbing meteorological element data in the training data set, and expanding the scale of the training data set and the training label set; inputting the expanded training data set into the power generation linkage prediction model, and calculating through an input layer, an intermediate feature processing layer, a time dimension adjusting layer and an output layer of the power generation linkage prediction model to obtain a power generation predicted value in the training process; Calculating a deviation value of the generated power predicted value in the training process and the historical generated power data corresponding to the extended training label set, wherein the deviation value is an absolute difference value of the predicted value and the label value, and counting an average deviation value in the training process; according to the average deviation value and the deviation value of each predicted value, adjusting network parameters of a power generation power linkage prediction model, wherein an adjustment object comprises a weighting coefficient of an intermediate feature processing layer, a sensitivity parameter of a time dimension adjusting layer and an operation parameter of an output layer, and the adjustment direction is that the average deviation value is reduced; After the primary parameter adjustment is completed, inputting the verification data set into the adjusted power generation power linkage prediction model to obtain a power generation power predicted value in the verification process, and calculating a deviation value between the predicted value in the verification process and the verification tag set to obtain a verification average deviation value; Judging whether the verification average deviation value is smaller than a preset deviation threshold value, if so, continuing to use the test data set to carry out model test, and if not, returning to continue to adjust model network parameters until the verification average deviation value is smaller than the preset deviation threshold value; inputting the test data set into a verified power generation power linkage prediction model to obtain a power generation power predicted value in the test process, and calculating a predicted value in the test process and a deviation value of the test tag set to obtain a test average deviation value; and if not, readjusting the dividing proportion of the data set and the data enhancement mode, and repeating the training steps until the test average deviation value is smaller than the preset deviation threshold value.
8. 8. The method for predicting new energy generated power based on meteorological data according to claim 1, wherein the performing the association correction processing on the initial prediction result of the new energy generated power and the historical generated power data in the working condition data set to generate the final prediction result of the new energy generated power comprises: Extracting a predicted power sequence in the initial predicted result of the new energy generated power, and marking an abnormal predicted value in the predicted power sequence, wherein the predicted power sequence comprises generated power predicted values corresponding to a plurality of time nodes, and the abnormal predicted value is a value exceeding a preset power predicted range; Extracting historical power generation power data in the working condition data set to form a historical power sequence, and eliminating abnormal historical values in the historical power sequence, wherein the historical power sequence comprises a plurality of historical power generation values with the same time interval as the predicted power sequence, and the abnormal historical values are power generation values recorded in a device fault state; Performing time dimension alignment processing on the predicted power sequence and the historical power sequence, and deleting an abnormal predicted value marked in the predicted power sequence and a corresponding historical power generation power value after alignment; Calculating deviation parameters between each predicted power value and the corresponding historical power generation value in the aligned predicted power sequences, and counting the distribution condition of all deviation parameters; analyzing the change rule of the deviation parameter in the time dimension, determining the trend characteristic of the change of the deviation parameter along with the time, and labeling a corresponding time interval for each trend characteristic, wherein the trend characteristic comprises an increasing trend, a decreasing trend or a fluctuation trend of the deviation parameter; constructing a deviation correction function according to the trend characteristics and the corresponding time interval, wherein the input of the deviation correction function is a predicted power value, corresponding time node information and trend characteristic type, and the output of the deviation correction function is a corrected power value; inputting each predicted power value, corresponding time node information and matched trend feature type in the initial predicted result of the new energy generated power into the deviation correction function to obtain a corrected power value; and arranging all the corrected power values according to a time sequence to form a final prediction result of the new energy generated power.
9. 9. The method for predicting the generated power of new energy based on meteorological data according to claim 8, wherein the constructing a deviation correction function according to the trend feature and the corresponding time interval, wherein the input of the deviation correction function is a predicted power value, the corresponding time node information and the trend feature type, and the output is a corrected power value, includes: classifying and coding the trend features, respectively distributing unique trend codes for increasing trend, decreasing trend and fluctuation trend, and establishing a corresponding relation table of the trend features and the trend codes; Dividing a plurality of time subintervals according to the time intervals corresponding to the trend characteristics, keeping the variation rule of the deviation parameter in each time subinterval consistent, and marking the corresponding time code for each time subinterval; Collecting deviation parameters, corresponding predicted power values and historical power generation data in each time subinterval, and establishing a corresponding database; extracting a plurality of groups of deviation parameters, predicted power values and historical power generation data under the same trend characteristic in the same time subinterval from the corresponding database, and constructing a deviation correction sample set, wherein each sample in the sample set comprises the predicted power values, time node information, trend codes and corresponding correction target values, and the correction target values are the historical power generation data; modeling the deviation correction sample set by adopting a regression analysis method, taking a predicted power value, time node information and trend codes as input variables, and taking a correction target value as an output variable to construct an initial deviation correction function; parameter optimization is carried out on the initial deviation correction function, and coefficient parameters of the initial deviation correction function are adjusted by minimizing the sum of squares of errors of a function predicted value and a correction target value, so that an optimized deviation correction function is obtained; Integrating the optimized deviation correction functions corresponding to different time subintervals and different trend features, establishing a function index table, wherein index items comprise time codes and trend codes, and positioning the corresponding deviation correction functions through the index items; And performing performance verification on the integrated deviation correction function, selecting a verification sample set to input the deviation correction function, calculating the deviation between the corrected power value output by the deviation correction function and the historical power generation data in the verification sample set, if the deviation is smaller than a preset verification deviation, completing the construction of the deviation correction function, and if the deviation is not smaller than the preset verification deviation, re-optimizing the parameters of the deviation correction function until the deviation is smaller than the preset verification deviation.
10. 10. The new energy generation power prediction system based on meteorological data is characterized by comprising a processor and a memory, wherein the memory is connected with the processor, the memory is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the memory so as to realize the new energy generation power prediction method based on meteorological data according to any one of claims 1-9.

Description

New energy generated power prediction method and system based on meteorological data Technical Field The invention relates to the technical field of new energy power generation, in particular to a new energy power generation power prediction method and system based on meteorological data. Background In the field of new energy power generation, such as wind power generation, solar power generation and the like, accurate prediction of power generation power is important for stable operation of a power grid, reasonable allocation of power resources and efficient utilization of new energy power generation equipment. The traditional new energy generated power prediction method is mainly used for modeling prediction based on single meteorological data or simple equipment operation parameters. On the one hand, when only single meteorological data is considered, because complex mutual influence and time sequence change relation exist between meteorological elements, omitting the internal relation can lead to inaccurate overall grasp of meteorological conditions, and further influence the accuracy of the prediction of the generated power. For example, in wind power generation, it is difficult to accurately predict the actual power generated by a wind turbine generator system by omitting the synergistic effect of other meteorological elements such as wind direction and air pressure in consideration of only wind speed. On the other hand, the prediction is only performed according to the operation parameters of the equipment, and the current operation state of the equipment can be reflected, but the potential influence of the change of the meteorological conditions on the operation of the equipment cannot be fully considered. Moreover, the traditional method lacks of deep mining and effective utilization of historical data in the prediction process, and fails to correlate and correct the real-time prediction result and the historical data, so that a large deviation exists between the prediction result and the actual power generation, and the high-precision requirement of the modern power grid on the prediction of the new energy power generation cannot be met. Disclosure of Invention In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a new energy generated power prediction method based on meteorological data. In still another aspect, the embodiment of the invention further provides a new energy generated power prediction system based on meteorological data. Based on the above aspects, according to the embodiment of the invention, through acquiring the meteorological data set and the working condition data set corresponding to the new energy power generation equipment, carrying out association analysis processing on the meteorological data set and establishing a meteorological element time sequence association diagram, describing the mutual influence relationship of different meteorological elements in the time dimension, carrying out association mapping processing on the meteorological element time sequence association diagram and the working condition data set, generating working condition response association characteristics, effectively revealing the inherent correspondence between the meteorological element change and the equipment operation parameter change, and realizing the deep fusion of the meteorological conditions and the equipment operation. The power generation power linkage prediction model is constructed based on the condition response association characteristics, so that the power generation power can be predicted by fully utilizing the synergistic effect of meteorological elements and equipment conditions, and the accuracy and reliability of the prediction are improved. And finally, carrying out association correction processing on the initial prediction result of the new energy generated power and the historical generated power data, further eliminating prediction errors, generating a final prediction result which is closer to the actual new energy generated power, and obviously improving the accuracy of new energy generated power prediction. Drawings Fig. 1 is a schematic diagram of an execution flow of a new energy generated power prediction method based on meteorological data according to an embodiment of the present invention. FIG. 2 is a schematic diagram of exemplary hardware and software components of a new energy generated power prediction system based on meteorological data provided by an embodiment of the present invention. Detailed Description The present invention is specifically described below with reference to the accompanying drawings, and fig. 1 is a schematic flow chart of a new energy generated power prediction method based on meteorological data according to an embodiment of the present invention, and the new energy generated power prediction method based on meteorological data is de