Electric power meteorology high influence weather trend prediction method based on meteorological numerical forecasting

Abstract

The invention discloses an electric power meteorological high-influence weather trend prediction method based on meteorological numerical forecasting, and particularly relates to the field of electric power meteorological data prediction under an end-cloud collaborative architecture, and the method comprises the steps: obtaining meteorological data through a meteorological observation device at an end side, and obtaining electric power equipment data through an electric power equipment monitoring device; performing timestamp calibration and interpolation processing on the meteorological data and the power equipment data according to a unified acquisition interval to form a time sequence data stream, transmitting the time sequence data stream to a cloud, correcting a time synchronization error through a cloud time reference signal, and outputting a synchronous data stream; a bidirectional coupling model of meteorological observation data and power equipment operation data is constructed under an end-cloud collaborative architecture, and dynamic fusion and adaptive prediction of meteorological change and power equipment response are realized in combination with causal relationship reasoning and game equilibrium calculation. The problem that trend prediction and real-time feedback adjustment are difficult to achieve under complex meteorological conditions is solved.

Description

Technical Field

[0001] This invention relates to the field of power meteorological data prediction technology under the edge-cloud collaborative architecture, and more specifically, to a method for predicting the trend of high-impact weather on power based on meteorological numerical forecasting. Background Technology

[0002] Current power weather forecasting systems generally adopt an edge-cloud collaborative architecture: the edge side is responsible for collecting data such as wind speed, radar echo, and line temperature in real time, while the cloud side centrally performs numerical weather forecasting and fusion calculations. The entire process forms a one-way link of edge data collection, cloud computing, and then issuing instructions. Although this improves information processing efficiency, it has inherent defects in terms of situational awareness. The situational awareness of existing systems mainly focuses on identifying weather anomalies, rather than understanding the interaction between weather and the power system. It can see that a storm is coming, but it cannot identify the reaction of equipment actions to the local meteorological environment. In other words, there is a lack of information feedback between cloud forecasting and edge response, which causes the cloud model to always make future judgments based on past data. In actual operation, this structure amplifies the problems layer by layer. First, under conditions of strong convection, extreme temperatures, or complex terrain, real-time actions at the end point directly alter the local electric field or energy distribution, but this change cannot be perceived by the cloud model. Second, the cloud model still follows a one-way causal chain of historical weather and equipment, ignoring the dynamic impact of power system feedback on the meteorological environment. Finally, although situational awareness results can trigger control commands, it is impossible to verify whether these commands have truly improved the situation. As a result, the entire system becomes disconnected between prediction and response. Therefore, the existing power weather situation perception under the existing edge-cloud collaborative architecture suffers from predictive game asymmetry, which means that the existing system can only passively predict the impact of weather on the power grid, but cannot simultaneously assess the reverse effect of power grid behavior on weather and risk situation. Summary of the Invention

[0003] To overcome the aforementioned deficiencies in the prior art, embodiments of the present invention provide a method for predicting the trend of high-impact weather on power systems based on numerical meteorological forecasting. By constructing a bidirectional coupling model of meteorological observation data and power equipment operation data under an edge-cloud collaborative architecture, and combining causal reasoning and game balance calculation, dynamic fusion and adaptive prediction of meteorological changes and power equipment responses are achieved, thereby solving the problem in the prior art that power systems are unable to achieve trend prediction and real-time feedback adjustment under complex meteorological conditions.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting the trend of high-impact weather in the power sector based on numerical weather prediction, comprising: S1. Meteorological data is acquired through the meteorological observation device on the end side, and power equipment data is acquired through the power equipment monitoring device; the meteorological data and power equipment data are timestamped and interpolated according to a unified acquisition interval to form a time-series data stream and transmit it to the cloud. After the time synchronization error is corrected by the cloud time reference signal, the synchronization data stream is output. S2. After receiving the synchronous data stream in the cloud, calculate the two-way influence relationship between meteorological data and power equipment data, statistically determine the coupling coefficient between meteorological changes and power equipment response, and construct a causal relationship model based on the coupling coefficient. S3. Apply the causal relationship model to the fusion calculation of real-time end-side meteorological data and cloud numerical forecast data, calculate the resonance factor of meteorological and power equipment response, dynamically adjust the fusion weight of meteorological and power equipment data, and generate a fusion vector. S4. Input the fusion vector into the cloud game balance calculation process, calculate the balance ratio of meteorological driving data and equipment reaction data, solve the control variables in the equilibrium state and generate a control command sequence, and combine the control command sequence with the balance ratio to form the balance result and output it. S5. After executing the control command sequence on the terminal side, meteorological data and power equipment data are collected again to form a feedback data stream and transmitted to the cloud. The difference between the balance result and the feedback data stream is calculated, the coupling coefficient and resonance factor in the causal relationship model are corrected, the fusion vector is updated and written back to the game balance calculation process.

[0005] In a preferred embodiment, S1 further includes continuously collecting meteorological data through a meteorological observation device on the end side and continuously collecting power equipment data through a power equipment monitoring device. The meteorological data includes wind speed, temperature, radar echo intensity and local electric field intensity, and the power equipment data includes load, voltage, temperature rise and operation signal. Meteorological data and power equipment data are synchronized by triggering a unified collection interval. Meteorological collection points and equipment collection points are generated simultaneously in each collection cycle, and a collection index is established based on the collection time sequence to form the original collection sequence. The original acquisition sequence is input into the time calibration calculation process. During the time calibration calculation process, the acquisition time difference between adjacent acquisition points is calculated, and a time index sequence of meteorological acquisition points and equipment acquisition points is established. The time index sequence is compared with a unified time reference to obtain the time deviation of each acquisition point relative to the unified time reference. The correction coefficient is calculated based on the time deviation. The time index of all acquisition points is offset and corrected using the correction coefficient to form a time calibration sequence with time correction. The time-calibrated sequence is input into the interpolation calculation process. During the interpolation calculation process, a piecewise linear interpolation function is established with the time index as the independent variable and the collected value as the dependent variable. The intermediate interpolation points are generated within the time interval through this interpolation function, and the intermediate interpolation points are weighted and fused with the adjacent collected points according to the weight ratio to form a time-continuous and numerically smooth time series data stream. The time-series data stream is transmitted to the cloud. The difference between the arrival time of the data on the cloud computing end and the time reference signal on the cloud is used as a variable to establish a delay correction equation. The delay correction equation is solved within a continuous acquisition period to obtain a weighted average delay correction amount. The delay correction amount is then used to perform a remapping operation on all time indices of the time-series data stream. The corrected time-series data stream is then output as a synchronous data stream.

[0006] In a preferred embodiment, S2 further includes, after receiving the synchronous data stream in the cloud, extracting the time-corrected time calibration sequence from the synchronous data stream, inputting the time calibration sequence into the time window sliding calculation process, calculating the difference between adjacent meteorological data collection points and power equipment data collection points within each time window, and dividing by the time interval to obtain the rate of change of meteorological data and power equipment response. Based on the rate of change, the intensity of change between each meteorological data point and the power equipment response is statistically analyzed to generate a coupling coefficient between meteorological changes and power equipment responses; Based on the coupling coefficient, a regression analysis method is adopted, using the rate of change of meteorological data as the independent variable and the rate of change of power equipment response as the dependent variable. Regression analysis is performed to solve the corresponding regression coefficients, and a causal relationship model between meteorological factors and power equipment response is established based on the regression coefficients. The rate of change of meteorological data is input into the causal relationship model, the predicted rate of change of power equipment response is calculated using regression coefficients, and the predicted rate of change of power equipment response is compared with the actual feedback data of power equipment to calculate the prediction error. When the prediction error exceeds the preset threshold, the regression coefficients are corrected using the least squares method, and the parameters in the causal relationship model are updated. When the prediction error does not exceed the preset threshold, the current causal relationship model parameters remain unchanged, and the final causal relationship model is output.

[0007] In a preferred embodiment, S3 further includes receiving the causal relationship model in the cloud, performing registration calculations by combining meteorological data collected at the edge with meteorological numerical forecast data called from the cloud, and matching meteorological data collection points with power equipment data collection points within a preset time window to form a time-aligned meteorological dataset. Input the time-aligned meteorological dataset into the interactive computing process. In the interactive quantity calculation process, construct the energy mutual feedback equation with the meteorological data change rate as the independent variable and the power equipment response change rate as the dependent variable. In the energy feedback equation, the energy difference between meteorological data and power equipment data within the synchronous change interval is calculated, and the time derivative of the energy difference is performed to obtain the interactive energy change rate. Based on the interactive energy change rate, the resonance intensity between meteorological changes and power equipment response is determined, and the resonance factor is solved with the resonance intensity as a constraint. Using the relative difference between the rate of change of meteorological data and the rate of change of power equipment response as the input variable, and the resonance factor as the adjustment variable, a nonlinear weight correction calculation is performed to obtain the weighting coefficient of meteorological data and power equipment data. The weighting coefficients are adjusted based on the resonance factor. When the resonance factor is higher than the set threshold, the weighting coefficient of meteorological data is increased, and when the resonance factor is lower than the set threshold, the weighting coefficient of power equipment data is increased, forming a dynamically adjusted fusion weight. Meteorological data and power equipment data are multiplied by their respective weighting coefficients, and a weighted superposition calculation is performed. The result of the weighted superposition is then normalized to generate a fusion vector.

[0008] In a preferred embodiment, S4 further includes receiving the fusion vector in the cloud and then inputting the fusion vector into the game balance calculation process. The cloud extracts meteorological data and power equipment data from the fused vector, performs feature construction calculations separately, and reassembles the wind speed, temperature, radar echo intensity and local electric field intensity of the meteorological data into a meteorological feature vector according to a unified time index, and reassembles the load, voltage, temperature rise and action signal of the power equipment data into an equipment feature vector according to the same time index; The cloud performs amplitude normalization on meteorological feature vectors and equipment feature vectors, and then performs time index registration calculation after unifying the units, so that the meteorological feature vectors and equipment feature vectors form a synchronous correspondence in the time dimension. The cloud uses the time-synchronized meteorological feature vector as the driving input and the equipment feature vector as the reaction input to construct an energy balance equation between meteorological driving data and equipment reaction data. In each iteration cycle, the instantaneous energy difference between meteorological driving data and equipment reaction data is calculated, and differential calculation is performed on the energy difference between two consecutive iterations to obtain the change in energy difference. Divide the change in energy difference by the energy difference of the previous round to obtain the relative rate of change of energy difference, and use the relative rate of change as the criterion for convergence detection; when the relative rate of change is lower than the preset convergence threshold, determine the balance result of the current iteration cycle; when the relative rate of change is higher than the convergence threshold, continue to perform energy balance iteration until convergence. After the convergence condition is met, output the balance ratio between meteorological driving data and equipment reaction data.

[0009] In a preferred embodiment, S4 further includes, after obtaining the balance ratio, the cloud uses the balance ratio as a constraint parameter and inputs the meteorological driving data and the equipment reaction data into the balance constraint calculation process. In the process of balance constraint calculation in the cloud, a set of balance constraint equations is established with meteorological driving data as independent variables and equipment reaction data as dependent variables, and the balance ratio is embedded in the coefficient matrix of the equation set to form a balance constraint matrix. The cloud performs matrix decomposition and inverse product solution on the balance constraint matrix. In each iteration cycle, it calculates the energy deviation between meteorological driving data and equipment reaction data under the balance constraint. The energy deviation is used as feedback to adjust the value of the control variable so that the control variable gradually tends to the solution that satisfies the balance constraint condition. When the update amount of the control variable solved in three consecutive iterations is lower than the convergence threshold, the cloud determines that the control variable has reached a balanced state and defines the control variable as a balanced control variable. During the control command generation process, the cloud takes the balanced control variable as input, performs time discretization calculation, divides the balanced control variable into multiple time slices according to a unified acquisition period, calculates the instantaneous rate of change of the control variable in each time slice, and performs weighted superposition of the instantaneous rate of change with the balance ratio of the corresponding time slice to generate a control command sequence. During the stability verification process, the cloud uses the time index as a benchmark to compare and analyze the time change curves of the control command sequence and the balance ratio, and calculates the stability index. When the stability index is lower than the error threshold, the control command sequence is output to the end side; when the stability index is higher than the error threshold, the balance constraint calculation process is returned.

[0010] In a preferred embodiment, S5 further includes, after the execution of the control command sequence is completed on the terminal side, the terminal side re-collects meteorological data and power equipment data, and establishes a collection index with a uniform collection interval to form a feedback data stream; The edge performs acquisition continuity correction processing on the feedback data stream. In the correction processing, interpolation calculation is performed with the acquisition time as the index variable and the acquisition value as the dependent variable to restore the missing acquisition points, generate a feedback data stream with complete time and continuous acquisition, and transmit it to the cloud. After receiving the feedback data stream, the cloud inputs it into the balance result comparison calculation process, and extracts the balance result generated by the previous game balance calculation during the balance result comparison calculation process; The cloud takes meteorological data and power equipment data from the feedback data stream as input, performs differential calculations with the corresponding data in the balance results, and generates an error data stream with the time index as a reference. Perform sliding difference calculation on the error data stream within a continuous time window to obtain the rate of change between adjacent acquisition points, which is defined as the feedback error rate of change. The amplitude difference of the error data stream is calculated within the same time window and defined as the change in the amplitude of the feedback error.

[0011] In a preferred embodiment, S5 further includes a process of correcting the input parameters of the feedback error change rate and the feedback error amplitude change after obtaining them in the cloud. In the parameter correction calculation, the feedback error change rate is used as input to perform a weighted correction operation on the coupling coefficient in the causal relationship model; the feedback error amplitude change is used as a constraint to perform a smoothing adjustment operation on the resonance factor to obtain the corrected coupling coefficient and resonance factor; the corrected coupling coefficient and resonance factor are weighted and superimposed to generate a comprehensive correction parameter. After obtaining the comprehensive correction parameters, the cloud inputs the correction parameters into the fusion update calculation process. During the fusion update calculation process, the fusion ratio of meteorological data and power equipment data is recalculated with the comprehensive correction parameters as weights to generate an updated fusion vector. The cloud writes the updated fusion vector into the game balance calculation process and re-executes the balance solution calculation.

[0012] The technical effects and advantages of this invention are as follows: 1. This method introduces a feedback loop in the situational awareness process, enabling the cloud to not only identify the impact of weather changes on the power system, but also to perceive the reverse effect of power equipment actions on the local meteorological field in real time. This achieves dynamic co-evolution reasoning between meteorology and the power system, solving the core problem of the disconnect between prediction and response and the inability to form a game-theoretic symmetric situational awareness in existing technologies. 2. This invention forms a causal relationship model between meteorological changes and power equipment response by performing dynamic calculations of coupling coefficients and resonance factors in the cloud, enabling the system to identify the strong and weak coupling relationship between meteorological-driven changes and equipment response in real time, thereby achieving quantitative prediction of equipment operating status under high-impact weather conditions. 3. In the process of fusion calculation, the present invention introduces a resonance factor as a nonlinear adjustment variable, and obtains the dynamic balance between meteorological data and equipment response through the energy feedback equation, so as to realize the adaptive weight allocation of meteorological and equipment data, enabling the system to automatically adjust the fusion weight under different meteorological disturbance intensities, and avoid the prediction bias dominated by a single data source. 4. In the game equilibrium calculation process, this invention establishes an energy balance equation between meteorological driving data and equipment reaction data, obtains the balance ratio through an iterative convergence mechanism, and uses this ratio as a constraint parameter to generate a control command sequence, thereby achieving the coordinated unity of control strategy and prediction process, and thus maintaining the controllability of system operation under extreme weather conditions. Attached Figure Description

[0013] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Refer to the instruction manual appendix Figure 1 An embodiment of the present invention provides a method for predicting the trend of high-impact weather in the power sector based on numerical weather forecasting, comprising: S1. Meteorological data is acquired through the meteorological observation device on the end side, and power equipment data is acquired through the power equipment monitoring device; the meteorological data and power equipment data are timestamped and interpolated according to a unified acquisition interval to form a time-series data stream and transmit it to the cloud. After the time synchronization error is corrected by the cloud time reference signal, the synchronization data stream is output. S2. After receiving the synchronous data stream in the cloud, calculate the two-way influence relationship between meteorological data and power equipment data, statistically determine the coupling coefficient between meteorological changes and power equipment response, and construct a causal relationship model based on the coupling coefficient. S3. Apply the causal relationship model to the fusion calculation of real-time end-side meteorological data and cloud numerical forecast data, calculate the resonance factor of meteorological and power equipment response, dynamically adjust the fusion weight of meteorological and power equipment data, and generate a fusion vector. S4. Input the fusion vector into the cloud game balance calculation process, calculate the balance ratio of meteorological driving data and equipment reaction data, solve the control variables in the equilibrium state and generate a control command sequence, and combine the control command sequence with the balance ratio to form the balance result and output it. S5. After executing the control command sequence on the terminal side, meteorological data and power equipment data are collected again to form a feedback data stream and transmitted to the cloud. The difference between the balance result and the feedback data stream is calculated, the coupling coefficient and resonance factor in the causal relationship model are corrected, the fusion vector is updated and written back to the game balance calculation process.

[0016] S1 also includes the continuous collection of meteorological data through the meteorological observation device on the end side and the continuous collection of power equipment data through the power equipment monitoring device. The meteorological data includes wind speed, temperature, radar echo intensity and local electric field intensity, and the power equipment data includes load, voltage, temperature rise and operation signal. Meteorological data and power equipment data are synchronized by triggering a unified collection interval. Meteorological collection points and equipment collection points are generated simultaneously in each collection cycle, and a collection index is established based on the collection time sequence to form the original collection sequence. The original acquisition sequence is input into the time calibration calculation process. During the time calibration calculation process, the acquisition time difference between adjacent acquisition points is calculated, and a time index sequence of meteorological acquisition points and equipment acquisition points is established. The time index sequence is compared with a unified time reference to obtain the time deviation of each acquisition point relative to the unified time reference. The correction coefficient is calculated based on the time deviation. The time index of all acquisition points is offset and corrected using the correction coefficient to form a time calibration sequence with time correction. The time-calibrated sequence is input into the interpolation calculation process. During the interpolation calculation process, a piecewise linear interpolation function is established with the time index as the independent variable and the collected value as the dependent variable. The intermediate interpolation points are generated within the time interval through this interpolation function, and the intermediate interpolation points are weighted and fused with the adjacent collected points according to the weight ratio to form a time-continuous and numerically smooth time series data stream. The time-series data stream is transmitted to the cloud. The difference between the arrival time of the data on the cloud computing end and the time reference signal on the cloud is used as a variable to establish a delay correction equation. The delay correction equation is solved within a continuous acquisition period to obtain a weighted average delay correction amount. The delay correction amount is then used to perform a remapping operation on all time indices of the time-series data stream. The corrected time-series data stream is then output as a synchronous data stream.

[0017] In S2, after receiving the synchronous data stream in the cloud, the time-corrected time calibration sequence is extracted from the synchronous data stream, and the time calibration sequence is input into the time window sliding calculation process. In each time window, the difference between adjacent meteorological data collection points and power equipment data collection points is calculated and divided by the time interval to obtain the rate of change of meteorological data and power equipment response. Based on the rate of change, the intensity of change between each meteorological data point and the power equipment response is statistically analyzed to generate a coupling coefficient between meteorological changes and power equipment responses. The coupling coefficient is a single value that represents the degree of influence of meteorological data on power equipment responses and reflects the correlation strength between meteorological factors and power equipment responses. Based on the coupling coefficient, a regression analysis method is adopted, using the rate of change of meteorological data as the independent variable and the rate of change of power equipment response as the dependent variable. Regression analysis is performed to calculate the corresponding regression coefficients, and a causal relationship model between meteorological factors and power equipment response is established based on the regression coefficients. The regression coefficients represent the quantitative impact of the rate of change of meteorological data on the rate of change of power equipment response, and describe the causal relationship between meteorological data and power equipment response. The rate of change of meteorological data is input into the causal relationship model, the predicted rate of change of power equipment response is calculated using regression coefficients, and the predicted rate of change of power equipment response is compared with the actual feedback data of power equipment to calculate the prediction error. When the prediction error exceeds the preset threshold, the regression coefficients are corrected using the least squares method, and the parameters in the causal relationship model are updated. When the prediction error does not exceed the preset threshold, the current causal relationship model parameters remain unchanged, and the final causal relationship model is output.

[0018] In S3, after receiving the causal relationship model in the cloud, the meteorological data collected at the edge and the meteorological numerical forecast data called in the cloud are matched according to a unified time index to form a time-aligned meteorological dataset. Input the time-aligned meteorological dataset into the interactive computing process. In the interactive quantity calculation process, construct the energy mutual feedback equation with the meteorological data change rate as the independent variable and the power equipment response change rate as the dependent variable. In the energy feedback equation, the energy difference between meteorological data and power equipment data within the synchronous change interval is calculated, and the time derivative of the energy difference is performed to obtain the interactive energy change rate. Based on the interactive energy change rate, the resonance intensity between meteorological changes and power equipment response is determined, and the resonance factor is solved with the resonance intensity as a constraint. Using the relative difference between the rate of change of meteorological data and the rate of change of power equipment response as the input variable, and the resonance factor as the adjustment variable, a nonlinear weight correction calculation is performed to obtain the weighting coefficient of meteorological data and power equipment data. The weighting coefficients are adjusted based on the resonance factor. When the resonance factor is higher than the set threshold, the weighting coefficient of meteorological data is increased, and when the resonance factor is lower than the set threshold, the weighting coefficient of power equipment data is increased, forming a dynamically adjusted fusion weight. Meteorological data and power equipment data are multiplied by their respective weighting coefficients, and a weighted superposition calculation is performed. The result of the weighted superposition is then normalized to generate a fusion vector.

[0019] S4 also includes receiving the fusion vector in the cloud and inputting the fusion vector into the game balance calculation process; The cloud extracts meteorological data and power equipment data from the fused vector, performs feature construction calculations separately, and reassembles the wind speed, temperature, radar echo intensity and local electric field intensity of the meteorological data into a meteorological feature vector according to a unified time index, and reassembles the load, voltage, temperature rise and action signal of the power equipment data into an equipment feature vector according to the same time index; The cloud performs amplitude normalization on meteorological feature vectors and equipment feature vectors, and then performs time index registration calculation after unifying the units, so that the meteorological feature vectors and equipment feature vectors form a synchronous correspondence in the time dimension. The cloud uses the time-synchronized meteorological feature vector as the driving input and the equipment feature vector as the reaction input to construct an energy balance equation between meteorological driving data and equipment reaction data. In each iteration cycle, the instantaneous energy difference between meteorological driving data and equipment reaction data is calculated, and differential calculation is performed on the energy difference between two consecutive iterations to obtain the change in energy difference. Divide the change in energy difference by the energy difference of the previous round to obtain the relative rate of change of energy difference, and use the relative rate of change as the criterion for convergence detection; when the relative rate of change is lower than the preset convergence threshold, determine the balance result of the current iteration cycle; when the relative rate of change is higher than the convergence threshold, continue to perform energy balance iteration until convergence. After the convergence condition is met, the balance ratio between meteorological driving data and equipment reaction data is output, and the balance ratio is passed into the balance constraint calculation process for subsequent control variable solving.

[0020] In S4, after obtaining the balance ratio, the cloud uses the balance ratio as a constraint parameter and inputs the meteorological driving data and equipment reaction data into the balance constraint calculation process. In the process of balance constraint calculation in the cloud, a set of balance constraint equations is established with meteorological driving data as independent variables and equipment reaction data as dependent variables, and the balance ratio is embedded in the coefficient matrix of the equation set to form a balance constraint matrix. The cloud performs matrix decomposition and inverse product solution on the balance constraint matrix. In each iteration cycle, it calculates the energy deviation between meteorological driving data and equipment reaction data under the balance constraint. The energy deviation is used as feedback to adjust the value of the control variable so that the control variable gradually tends to the solution that satisfies the balance constraint condition. When the update amount of the control variable solved in three consecutive iterations is lower than the convergence threshold, the cloud determines that the control variable has reached a balanced state and defines the control variable as a balanced control variable. During the control command generation process, the cloud takes the balanced control variable as input, performs time discretization calculation, divides the balanced control variable into multiple time slices according to a unified acquisition period, calculates the instantaneous rate of change of the control variable in each time slice, and performs weighted superposition of the instantaneous rate of change with the balance ratio of the corresponding time slice to generate a control command sequence. During the stability verification process, the cloud uses the time index as a benchmark to compare and analyze the time change curves of the control command sequence and the balance ratio, and calculates the stability index. When the stability index is lower than the error threshold, the control command sequence is output to the end side; when the stability index is higher than the error threshold, the balance constraint calculation process is returned.

[0021] In S5, after the execution of the control command sequence is completed on the end side, the end side re-collects meteorological data and power equipment data, and establishes a collection index with a uniform collection interval to form a feedback data stream. The edge performs acquisition continuity correction processing on the feedback data stream. In the correction processing, interpolation calculation is performed with the acquisition time as the index variable and the acquisition value as the dependent variable to restore the missing acquisition points, generate a feedback data stream with complete time and continuous acquisition, and transmit it to the cloud. After receiving the feedback data stream, the cloud inputs it into the balance result comparison calculation process, and extracts the balance result generated by the previous game balance calculation during the balance result comparison calculation process; The cloud takes meteorological data and power equipment data from the feedback data stream as input, performs differential calculations with the corresponding data in the balance results, and generates an error data stream with the time index as a reference. Perform sliding difference calculation on the error data stream within a continuous time window to obtain the rate of change between adjacent acquisition points, which is defined as the feedback error rate of change. The amplitude difference of the error data stream is calculated within the same time window and defined as the change in feedback error amplitude, which is used to characterize the error intensity.

[0022] S5 also includes a process of correcting the input parameters of the feedback error change rate and feedback error amplitude change after obtaining them in the cloud. In the parameter correction calculation, the feedback error change rate is used as input to perform a weighted correction operation on the coupling coefficients in the causal relationship model; the feedback error amplitude change is used as a constraint to perform a smoothing adjustment operation on the resonance factor to obtain the corrected coupling coefficients and resonance factor. The corrected coupling coefficients and resonance factor are weighted and superimposed to generate a comprehensive correction parameter. The weights are determined by the previous round of balancing ratio and are used for subsequent fusion update calculations. The comprehensive correction parameter is used to reflect the overall adjustment effect of the coupling relationship change on the fusion weights. After obtaining the comprehensive correction parameters, the cloud inputs the correction parameters into the fusion update calculation process. During the fusion update calculation process, the fusion ratio of meteorological data and power equipment data is recalculated with the comprehensive correction parameters as weights to generate an updated fusion vector. The cloud writes the updated fusion vector into the game balance calculation process and re-executes the balance solution calculation to achieve dynamic adaptive cyclic updates based on feedback data streams.

[0023] In practical implementation, this scheme first continuously collects meteorological data and power equipment data through end-side meteorological observation devices and power equipment monitoring devices. Under a unified time reference, time calibration, interpolation correction, and delay compensation are performed to form a time-continuous and synchronous data stream. This ensures the comparability and consistency of data from different sources in the time dimension, providing an accurate data foundation for subsequent coupling analysis and causal modeling. Next, in a cloud computing environment, statistical analysis is performed on the synchronous data stream to calculate the bidirectional impact relationship between meteorological changes and power equipment response, generate coupling coefficients, and establish a causal relationship model. This enables the system to quantitatively describe the mechanism by which meteorological factors affect power response from a data perspective. Building upon this foundation, the solution further integrates the causal relationship model with real-time weather forecast data. By constructing an energy feedback equation, it calculates the resonance factor between weather changes and power equipment responses, and dynamically adjusts the data fusion weights of the two to generate a fusion vector. This fusion vector is then input into the game equilibrium calculation process. The cloud-based system uses feature vector recombination, amplitude normalization, time index registration, and energy difference iterative calculation to gradually obtain the equilibrium ratio between weather-driven data and equipment reaction data, thereby solving for the control variables in equilibrium. The control variables are then discretized and weighted to generate a control command sequence, which is then used for the adjustment and execution of the end-side equipment. After the end-side execution is completed, the newly acquired feedback data stream is compared with the previous round's equilibrium results to extract the feedback error change rate and amplitude change, thereby correcting the coupling coefficient and resonance factor in the causal relationship model. Based on this, the fusion vector is updated and written back to the equilibrium calculation process to achieve adaptive closed-loop adjustment. This solution integrates data-driven causal modeling and control feedback into a unified architecture, enabling weather forecast information to not only predict external changes but also influence the operation and scheduling of the power system through feedback mechanisms, thereby achieving proactive defense and intelligent response under high-impact weather conditions.

[0024] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the trend of high-impact weather in the power sector based on numerical weather prediction, characterized in that, include: S1. Obtain meteorological data through the meteorological observation device on the end side, and obtain power equipment data through the power equipment monitoring device; Meteorological data and power equipment data are timestamped and interpolated according to a unified collection interval to form a time-series data stream and transmitted to the cloud. After the time synchronization error is corrected by the cloud time reference signal, a synchronized data stream is output. S2. After receiving the synchronous data stream in the cloud, calculate the two-way influence relationship between meteorological data and power equipment data, statistically determine the coupling coefficient between meteorological changes and power equipment response, and construct a causal relationship model based on the coupling coefficient. S3. Apply the causal relationship model to the fusion calculation of real-time end-side meteorological data and cloud numerical forecast data, calculate the resonance factor of meteorological and power equipment response, dynamically adjust the fusion weight of meteorological and power equipment data, and generate a fusion vector. S4. Input the fusion vector into the cloud game balance calculation process, calculate the balance ratio of meteorological driving data and equipment reaction data, solve the control variables in the equilibrium state and generate a control command sequence, and combine the control command sequence with the balance ratio to form the balance result and output it. S5. After executing the control command sequence on the terminal side, meteorological data and power equipment data are collected again to form a feedback data stream and transmitted to the cloud. The difference between the balance result and the feedback data stream is calculated, the coupling coefficient and resonance factor in the causal relationship model are corrected, the fusion vector is updated and written back to the game balance calculation process.

2. The method for predicting the trend of high-impact weather in power generation based on numerical weather prediction as described in claim 1, characterized in that: S1 also includes the continuous collection of meteorological data through the meteorological observation device on the end side and the continuous collection of power equipment data through the power equipment monitoring device. The meteorological data includes wind speed, temperature, radar echo intensity and local electric field intensity, and the power equipment data includes load, voltage, temperature rise and operation signal. Meteorological data and power equipment data are synchronized by triggering a unified collection interval. Meteorological collection points and equipment collection points are generated simultaneously in each collection cycle, and a collection index is established based on the collection time sequence to form the original collection sequence. The original acquisition sequence is input into the time calibration calculation process. During the time calibration calculation process, the acquisition time difference between adjacent acquisition points is calculated, and a time index sequence of meteorological acquisition points and equipment acquisition points is established. The time index sequence is compared with a unified time reference to obtain the time deviation of each acquisition point relative to the unified time reference. The correction coefficient is calculated based on the time deviation. The time index of all acquisition points is offset and corrected using the correction coefficient to form a time calibration sequence with time correction. The time-calibrated sequence is input into the interpolation calculation process. During the interpolation calculation process, a piecewise linear interpolation function is established with the time index as the independent variable and the collected value as the dependent variable. The intermediate interpolation points are generated within the time interval through this interpolation function, and the intermediate interpolation points are weighted and fused with the adjacent collected points according to the weight ratio to form a time-continuous and numerically smooth time series data stream. The time-series data stream is transmitted to the cloud. The difference between the arrival time of the data on the cloud computing end and the time reference signal on the cloud is used as a variable to establish a delay correction equation. The delay correction equation is solved within a continuous acquisition period to obtain a weighted average delay correction amount. The delay correction amount is then used to perform a remapping operation on all time indices of the time-series data stream. The corrected time-series data stream is then output as a synchronous data stream.

3. The method for predicting the trend of high-impact weather in power generation based on numerical weather prediction according to claim 2, characterized in that: In S2, after receiving the synchronous data stream in the cloud, the time-corrected time calibration sequence is extracted from the synchronous data stream, and the time calibration sequence is input into the time window sliding calculation process. In each time window, the difference between adjacent meteorological data collection points and power equipment data collection points is calculated and divided by the time interval to obtain the rate of change of meteorological data and power equipment response. Based on the rate of change, the intensity of change between each meteorological data point and the power equipment response is statistically analyzed to generate a coupling coefficient between meteorological changes and power equipment responses; Based on the coupling coefficient, a regression analysis method is adopted, using the rate of change of meteorological data as the independent variable and the rate of change of power equipment response as the dependent variable. Regression analysis is performed to solve the corresponding regression coefficients, and a causal relationship model between meteorological factors and power equipment response is established based on the regression coefficients. The rate of change of meteorological data is input into the causal relationship model, the predicted rate of change of power equipment response is calculated using regression coefficients, and the predicted rate of change of power equipment response is compared with the actual feedback data of power equipment to calculate the prediction error. When the prediction error exceeds the preset threshold, the regression coefficients are corrected based on the least squares method to update the parameters in the causal relationship model. When the prediction error does not exceed the preset threshold, the current causal relationship model parameters remain unchanged, and the final causal relationship model is output.

4. The method for predicting the trend of high-impact weather in power generation based on numerical weather prediction according to claim 3, characterized in that: In S3, after receiving the causal relationship model in the cloud, the meteorological data collected at the edge and the meteorological numerical forecast data called in the cloud are matched according to a unified time index to form a time-aligned meteorological dataset. Input the time-aligned meteorological dataset into the interactive computing process. In the interactive quantity calculation process, construct the energy mutual feedback equation with the meteorological data change rate as the independent variable and the power equipment response change rate as the dependent variable. In the energy feedback equation, the energy difference between meteorological data and power equipment data within the synchronous change interval is calculated, and the time derivative of the energy difference is performed to obtain the rate of change of interactive energy. Based on the rate of change of interactive energy, the resonance intensity between meteorological changes and power equipment response is determined, and the resonance factor is solved using the resonance intensity as a constraint condition. Using the relative difference between the rate of change of meteorological data and the rate of change of power equipment response as the input variable, and the resonance factor as the adjustment variable, a nonlinear weight correction calculation is performed to obtain the weighting coefficient of meteorological data and power equipment data. The weighting coefficients are adjusted based on the resonance factor. When the resonance factor is higher than the set threshold, the weighting coefficient of meteorological data is increased, and when the resonance factor is lower than the set threshold, the weighting coefficient of power equipment data is increased, forming a dynamically adjusted fusion weight. Meteorological data and power equipment data are multiplied by their respective weighting coefficients, and a weighted superposition calculation is performed. The result of the weighted superposition is then normalized to generate a fusion vector.

5. The method for predicting the trend of high-impact weather in power generation based on numerical weather prediction according to claim 4, characterized in that: S4 also includes receiving the fusion vector in the cloud and inputting the fusion vector into the game balance calculation process; The cloud extracts meteorological data and power equipment data from the fused vector, performs feature construction calculations separately, and reassembles the wind speed, temperature, radar echo intensity and local electric field intensity of the meteorological data into a meteorological feature vector according to a unified time index, and reassembles the load, voltage, temperature rise and action signal of the power equipment data into an equipment feature vector according to the same time index; The cloud performs amplitude normalization on meteorological feature vectors and equipment feature vectors, and then performs time index registration calculation after unifying the units, so that the meteorological feature vectors and equipment feature vectors form a synchronous correspondence in the time dimension; The cloud uses the time-synchronized meteorological feature vector as the driving input and the equipment feature vector as the reaction input to construct an energy balance equation between meteorological driving data and equipment reaction data. In each iteration cycle, the instantaneous energy difference between meteorological driving data and equipment reaction data is calculated, and differential calculation is performed on the energy difference between two consecutive iterations to obtain the change in energy difference. Divide the change in energy difference by the energy difference of the previous round to obtain the relative rate of change of energy difference, and use the relative rate of change as the criterion for convergence detection; when the relative rate of change is lower than the preset convergence threshold, determine the balance result of the current iteration cycle. When the relative rate of change is higher than the convergence threshold, continue to perform energy balance iterations until convergence. After the convergence condition is met, output the balance ratio between meteorological driving data and equipment reaction data.

6. The method for predicting the trend of high-impact weather in power generation based on numerical weather prediction as described in claim 5, characterized in that: In S4, after obtaining the balance ratio, the cloud uses the balance ratio as a constraint parameter and inputs the meteorological driving data and equipment reaction data into the balance constraint calculation process. In the process of balance constraint calculation in the cloud, a set of balance constraint equations is established with meteorological driving data as independent variables and equipment reaction data as dependent variables, and the balance ratio is embedded in the coefficient matrix of the equation set to form a balance constraint matrix. The cloud performs matrix decomposition and inverse product solution on the balance constraint matrix. In each iteration cycle, it calculates the energy deviation between meteorological driving data and equipment reaction data under the balance constraint. The energy deviation is used as feedback to adjust the value of the control variable so that the control variable gradually tends to the solution that satisfies the balance constraint condition. When the update amount of the control variable solved in three consecutive iterations is lower than the convergence threshold, the cloud determines that the control variable has reached a balanced state and defines the control variable as a balanced control variable. During the control command generation process, the cloud takes the balanced control variable as input, performs time discretization calculation, divides the balanced control variable into multiple time slices according to a unified acquisition period, calculates the instantaneous rate of change of the control variable in each time slice, and performs weighted superposition of the instantaneous rate of change with the balance ratio of the corresponding time slice to generate a control command sequence. During the stability verification process, the cloud uses the time index as a benchmark to compare and analyze the time change curves of the control command sequence and the balance ratio, and calculates the stability index. When the stability index is lower than the error threshold, the control command sequence is output to the end side; when the stability index is higher than the error threshold, the balance constraint calculation process is returned.

7. A method for predicting the trend of high-impact weather in power generation based on numerical weather prediction, as described in claim 6, characterized in that: In S5, after the execution of the control command sequence is completed on the end side, the end side re-collects meteorological data and power equipment data, and establishes a collection index with a uniform collection interval to form a feedback data stream. The edge performs acquisition continuity correction processing on the feedback data stream. In the correction processing, interpolation calculation is performed with the acquisition time as the index variable and the acquisition value as the dependent variable to restore the missing acquisition points, generate a feedback data stream with complete time and continuous acquisition, and transmit it to the cloud. After receiving the feedback data stream, the cloud inputs it into the balance result comparison calculation process, and extracts the balance result generated by the previous game balance calculation during the balance result comparison calculation process; The cloud takes meteorological data and power equipment data from the feedback data stream as input, performs differential calculations with the corresponding data in the balance results, and generates an error data stream with the time index as a reference. Perform sliding difference calculation on the error data stream within a continuous time window to obtain the rate of change between adjacent acquisition points, which is defined as the feedback error rate of change. The amplitude difference of the error data stream is calculated within the same time window and defined as the change in the amplitude of the feedback error.

8. A method for predicting the trend of high-impact weather in power generation based on numerical weather prediction, as described in claim 7, characterized in that: S5 also includes a process of correcting the input parameters of the feedback error change rate and feedback error amplitude change after obtaining them in the cloud. In the parameter correction calculation, the feedback error change rate is used as input to perform a weighted correction operation on the coupling coefficient in the causal relationship model; the feedback error amplitude change is used as a constraint to perform a smoothing adjustment operation on the resonance factor to obtain the corrected coupling coefficient and resonance factor; the corrected coupling coefficient and resonance factor are weighted and superimposed to generate a comprehensive correction parameter. After obtaining the comprehensive correction parameters, the cloud inputs the correction parameters into the fusion update calculation process. During the fusion update calculation process, the fusion ratio of meteorological data and power equipment data is recalculated with the comprehensive correction parameters as weights to generate an updated fusion vector. The cloud writes the updated fusion vector into the game balance calculation process and re-executes the balance solution calculation.

Images