Wind power photovoltaic power prediction system and method in turning weather scene

Abstract

The invention discloses a wind power photovoltaic power prediction system and method in a turning weather scene, and relates to the technical field of photovoltaic power prediction, and the method comprises the following steps: obtaining various parameter information generated in the operation process of a data integration platform in real time during the operation of the data integration platform, and carrying out the feature extraction of the obtained parameter information; according to the invention, various parameter information of the data integration platform is acquired and analyzed in real time, the operation state of the system is identified and evaluated in time, the generated system response coefficient can effectively monitor the data processing capability, a power operator is helped to rapidly identify a high-risk operation process and take countermeasures, and the stability of power supply is ensured. A high-risk operation process is subdivided into a burst type, a continuous type and an unstable type, red, orange and yellow early warning signals are sent out, a more scientific risk management framework is provided, the capability of coping with emergencies is improved, economic losses are reduced, and the overall toughness and safety of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power prediction, and in particular to a wind power photovoltaic power prediction system and method under transitional weather scenarios. Background Art

[0002] Wind power and photovoltaic power forecasts under transitional weather scenarios are designed to cope with the impact of rapidly changing weather conditions (such as cloud movement, sudden rainfall, and changes in wind speed) on renewable energy generation capacity. Accurately predicting these changes is critical to the stability of the power system, because uncertain outputs may lead to an imbalance in the load on the power grid and affect the matching of overall power supply and demand. To meet these challenges, forecasting models usually combine meteorological data, historical power generation records, and machine learning algorithms for data analysis and modeling. These models need to quickly adapt to complex meteorological conditions and provide real-time power forecasts to improve the efficiency of renewable energy utilization, and ensure that the power system effectively dispatches backup power sources and maintains grid stability in sudden weather. Ultimately, accurate power forecasts not only support sustainable development goals, but also promote the integration of a higher proportion of renewable energy and reduce dependence on fossil fuels.

[0003] Data integration platforms play a vital role in forecasting wind and photovoltaic power under transitional weather scenarios. Such platforms can bring together data from multiple sources, including meteorological sensors, anemometers, photovoltaic power generation monitoring equipment, lidar, satellite remote sensing data, etc., to form a comprehensive database. By integrating and processing these heterogeneous data, the data integration platform can provide real-time and accurate weather and environmental conditions analysis, so that the forecasting model can better capture the impact of transitional weather on wind and photovoltaic power generation. In addition, this platform also supports advanced data analysis and machine learning algorithms to optimize the accuracy of power forecasts, improve the responsiveness of the power system, ensure that the grid dispatch can be adjusted in time in the event of sudden weather, and ensure the stability and reliability of power supply.

[0004] The prior art has the following deficiencies:

[0005] Under extreme weather conditions, such as severe storms, thunderstorms, and blizzards, sudden meteorological changes can trigger data collection from a large number of sensors and monitoring devices, resulting in a surge in data volume. This may cause the data processing capacity of the data integration platform to reach a bottleneck, limiting the processing of real-time data. When the data processing capacity of the data integration platform is limited, the prediction model cannot obtain the latest data input, resulting in inaccurate power forecasts. This inaccuracy will directly affect grid dispatch decisions, making it difficult for power operators to effectively manage grid loads, which may lead to insufficient or excessive power supply.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The purpose of the present invention is to provide a wind power photovoltaic power prediction system and method under transitional weather scenarios, which can timely identify and evaluate the system operation status by acquiring and analyzing various parameter information of the data integration platform in real time, and the generated system response coefficient can effectively monitor the data processing capacity, help power operators quickly identify high-risk operation processes and take countermeasures to ensure the stability of power supply. The high-risk operation process is subdivided into sudden, continuous and unstable types, and red, orange and yellow warning signals are issued to provide a more scientific risk management framework, enhance the ability to respond to emergencies, reduce economic losses, and improve the overall resilience and safety of the system to solve the problems in the above-mentioned background technology.

[0008] In order to achieve the above object, the present invention provides the following technical solution: a method for predicting wind power and photovoltaic power under transitional weather scenarios, comprising the following steps:

[0009] During the operation of the data integration platform, various parameter information generated during the operation of the data integration platform is obtained in real time, and feature extraction is performed on the obtained parameter information;

[0010] After feature extraction of the acquired parameters, the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests are analyzed, and the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests are input into the pre-learned machine learning model, and the system response coefficient is generated through the machine learning model, and the operation process of the data integration platform is intelligently evaluated through the system response coefficient;

[0011] Based on the evaluation results of the machine learning model, the data integration platform operation process is divided into low-risk operation process or high-risk operation process, and the abnormal operation process of the data integration platform is intelligently perceived;

[0012] For high-risk operation processes, continuously obtain parameter information during the operation of the data integration platform and divide the high-risk operation processes into sudden, continuous and unstable types;

[0013] Different warning signals are issued for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow warning is issued, for unstable high-risk operation processes, an orange warning is issued, and for continuous high-risk operation processes, a red warning is issued.

[0014] Preferably, after analyzing the cumulative amount of unprocessed data and the average response time of the platform to requests, a data backlog index and a system response time index are generated respectively, and the data backlog index and the system response time index are input into a pre-learned machine learning model, and a system response coefficient is generated through the machine learning model, and the operation process of the data integration platform is intelligently evaluated through the system response coefficient.

[0015] Preferably, the specific steps of analyzing the accumulated amount of unprocessed data in the detection window and generating the data backlog index are as follows:

[0016] In the detection window, the input data streams from various sensors and monitoring devices are collected in real time to ensure the integrity and accuracy of the data source, and the acquired input data streams are calibrated as D(t), which represents the amount of new data received by the system at time t;

[0017] Calculate the amount of real-time data that the system can process at each time point. The calculation expression is as follows:

[0018]

[0019] , where P(t) is the instantaneous processing capacity, C is the maximum processing capacity of the system, L(t) is the current system load, and L max is the maximum load threshold of the system, e -α·t is the time decay term, α is the data processing efficiency decay coefficient, and e is the natural base;

[0020] Evaluate the ratio of the system's processing capacity to the current load. The calculation expression is as follows:

[0021]

[0022] , where R(t) is the ratio of processing capacity to load, ∈ is a small constant;

[0023] In each time interval, the cumulative amount of unprocessed data is calculated, and the calculation expression is as follows:

[0024] I(t)=U(t-Δt)+D(t)-P(t)

[0025] , where U(t) is the cumulative amount of unprocessed data, indicating the total amount of unprocessed data at time point t, Δt is the time interval, and U(t-Δt) is the amount of unprocessed data at the previous time point;

[0026] According to the accumulated amount of unprocessed data, the data backlog index is generated, and the calculation expression is as follows:

[0027]

[0028] , where DBI is the data backlog index, Uthreshold is the preset unprocessed data volume threshold, D max is the maximum data flow, β is the nonlinear coefficient, and γ is the load attenuation coefficient, which describes the influence of system load on the backlog index.

[0029] Preferably, under the detection window, a response load function is defined. The response load function is used to describe the request load level of the system at time point t, reflecting the pressure of the current request volume on the system processing capacity. The calculation expression is as follows:

[0030]

[0031] , where Q(t) is the response load function, A is the reference constant of the response load, which is used to calibrate the amplitude of the load function, H(t) is the system demand at time t, and H max is the maximum number of requests that the system can bear, e is the natural base, and μ is the time decay coefficient, which is used to simulate the gradual decay of requests over time;

[0032] Evaluate the response rate of the system when processing requests and define the response rate function. The calculation expression is as follows:

[0033]

[0034] , where V(t) is the response rate function, B is the rate adjustment constant used to control the initial value of the response rate, φ(t) is the load adjustment factor, K(t) is the actual response time of the system at time t, and K limit is the system response time limit constant, ensuring within a reasonable range;

[0035] In order to reflect the real-time adaptability of the system response under different loads and requests, a dynamic adjustment factor is introduced. The calculation expression is as follows:

[0036]

[0037] , where θ(t) is the dynamic adjustment factor, η is the adjustment amplitude constant, which controls the overall amplitude of the dynamic adjustment factor, κ is the adjustment rate constant, which controls the impact of the load on the adjustment factor, and V ref is the reference rate constant, ∈ is a small constant;

[0038] Combining the response load function Q(t), the response rate function V(t) and the dynamic adjustment factor θ(t), the system response time index is generated. The calculation expression is as follows:

[0039] SRTI=ln(Q(t)·θ(t)+V(t)+δ)

[0040] , where SRTI is the system response time index and δ is a small constant.

[0041] Preferably, the data backlog index DBI and the system response time index SRTI are input into a pre-learned machine learning model, and the system response coefficient SRC is generated by the machine learning model according to the following formula:

[0042]

[0043] , where f1 and f2 are the preset proportional coefficients of the data backlog index DBI and the system response time index SRTI, respectively, and both f1 and f2 are greater than 0.

[0044] Preferably, the generated system response coefficient is compared and analyzed with a preset system response coefficient reference threshold, and the operation process of the data integration platform is divided. The division process is as follows:

[0045] If the system response coefficient is greater than or equal to the preset system response coefficient reference threshold, the data integration platform operation process is classified as a high-risk operation process;

[0046] If the system response coefficient is less than a preset system response coefficient reference threshold, the data integration platform operation process is classified as a low-risk operation process.

[0047] Preferably, for high-risk operation processes, a plurality of system response coefficients generated in a plurality of detection windows during operation of the data integration platform are continuously acquired to establish a data set, and a comprehensive analysis is performed;

[0048] The average values ​​and standard deviations of several system response coefficients in the data set are obtained, and the obtained average value and standard deviation of the system response coefficient are compared and analyzed with the preset reference threshold value of the system response coefficient average value and the preset reference threshold value of the system response coefficient standard deviation, respectively, and the high-risk operation status is further divided. The division results are as follows:

[0049] If the average value of the system response coefficient is less than a preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than a preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as a sudden type;

[0050] If the average value of the system response coefficient is greater than or equal to the preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than the preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as unstable;

[0051] If the system response coefficient average value is greater than or equal to the preset system response coefficient average value reference threshold, the high-risk operation state is further divided into a continuous type.

[0052] The wind power and photovoltaic power prediction system under transitional weather scenarios includes data acquisition and feature extraction modules, feature analysis and model evaluation modules, risk classification and intelligent perception modules, high-risk type classification modules, and early warning signal issuance modules:

[0053] The data collection and feature extraction module acquires various parameter information generated during the operation of the data integration platform in real time, and performs feature extraction on the acquired parameter information;

[0054] The feature analysis and model evaluation module extracts features from the acquired parameters, analyzes the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests, and inputs the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests into the pre-learned machine learning model, generates a system response coefficient through the machine learning model, and intelligently evaluates the operation process of the data integration platform through the system response coefficient;

[0055] The risk classification and intelligent perception module divides the operation process of the data integration platform into low-risk operation process or high-risk operation process based on the evaluation results of the machine learning model, and intelligently perceives the abnormal operation process of the data integration platform;

[0056] The high-risk type classification module continuously obtains parameter information during the operation of the data integration platform for high-risk operation processes, and divides high-risk operation processes into sudden, continuous and unstable types;

[0057] The warning signal issuing module issues different warning signals for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow warning is issued, for unstable high-risk operation processes, an orange warning is issued, and for continuous high-risk operation processes, a red warning is issued.

[0058] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0059] The present invention can timely identify and evaluate the system operation status by acquiring and analyzing various parameter information of the data integration platform in real time. Through the application of feature extraction and machine learning models, the generated system response coefficient can quickly reflect the changes in data processing capabilities, thereby realizing effective monitoring of data backlogs and response time. This real-time perception capability enables power operators to quickly identify high-risk operation processes and take timely countermeasures to avoid inaccurate power forecasts and failed grid load management due to insufficient data processing capabilities, thereby ensuring the stability and reliability of power supply.

[0060] The present invention provides a more detailed and scientific risk management framework for power operators by subdividing high-risk operation processes into sudden, continuous and unstable types, and issuing corresponding warning signals according to different types. The graded prompts of red, orange and yellow warning signals help the operation team take targeted response measures in different situations and improve the system's ability to respond to emergencies. This optimized early warning mechanism not only enhances the ability to prevent potential risks, but also reduces economic losses caused by untimely response, and improves the overall resilience and safety of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0062] Figure 1 The present invention is a flow chart of a method for predicting wind power and photovoltaic power under transitional weather scenarios.

[0063] Figure 2 It is a module schematic diagram of the wind power and photovoltaic power prediction system under the transitional weather scenario of the present invention. DETAILED DESCRIPTION

[0064] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0065] The present invention provides Figure 1 The wind power and photovoltaic power prediction method under the transitional weather scenario shown includes the following steps:

[0066] During the operation of the data integration platform, various parameter information generated during the operation of the data integration platform is obtained in real time, and feature extraction is performed on the obtained parameter information;

[0067] After feature extraction of the acquired parameters, the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests are analyzed, and the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests are input into the pre-learned machine learning model, and the system response coefficient is generated through the machine learning model, and the operation process of the data integration platform is intelligently evaluated through the system response coefficient;

[0068] After analyzing the accumulated amount of unprocessed data and the average response time of the platform to requests, the data backlog index and system response time index are generated respectively. The data backlog index and system response time index are input into the pre-learned machine learning model, and the system response coefficient is generated through the machine learning model. The operation process of the data integration platform is intelligently evaluated through the system response coefficient.

[0069] The continued surge in the accumulation of unprocessed data does indicate that the data processing capacity of the data integration platform has reached a bottleneck, thus limiting the processing of real-time data. This is because the data integration platform needs to process a large amount of real-time data in a rapidly changing environment, especially under extreme weather conditions, when the acquisition frequency of sensors and monitoring equipment may increase significantly. When the backlog of unprocessed data continues to increase, it means that the system cannot respond to and process the newly arrived data in a timely manner, resulting in the data inflow speed exceeding the processing speed. This situation usually stems from a variety of factors, such as insufficient computing resources, inefficient processing algorithms, or improper system architecture design. When the processing capacity is insufficient, the system latency increases and the response time becomes longer, which further affects the real-time and reliability of the data, resulting in a decrease in the accuracy of power forecasts and difficult grid dispatch decisions. Therefore, the continued accumulation of unprocessed data not only reflects the shortcomings of the current system performance, but may also cause more serious problems, including unstable power supply, increased risk of equipment failure, and potential safety hazards. The existence of this situation urgently requires system optimization and resource adjustment to restore the efficiency and real-time performance of data processing.

[0070] Under the detection window, the accumulated amount of unprocessed data is analyzed and the specific steps to generate the data backlog index are as follows:

[0071] In the detection window, the input data streams from various sensors and monitoring devices are collected in real time to ensure the integrity and accuracy of the data source, and the acquired input data streams are calibrated as D(t), which represents the amount of new data received by the system at time t;

[0072] Calculate the amount of real-time data that the system can process at each time point. The calculation expression is as follows:

[0073]

[0074] , where P(t) is the instantaneous processing capacity, which means the instantaneous data processing capacity of the system at time point t, C is the maximum processing capacity of the system, L(t) is the current system load, which means the amount of data being processed by the system at time point t, and L max is the maximum load threshold of the system, e -α·t is the time decay term, indicating that the processing capacity decays over time, α is the data processing efficiency decay coefficient, and e is the natural base;

[0075] In the actual operation of data processing systems, although the formula is used to calculate the instantaneous processing capacity at time point t, the system usually performs calculations at each time interval in order to continuously track and update the dynamic changes in processing capacity. This is because in data-intensive environments, system load and processing capacity will fluctuate rapidly over time, so the current processing capacity needs to be frequently evaluated to ensure that the system can adapt to new load conditions.

[0076] Calculating the amount of real-time data at each time interval can provide continuous status monitoring and help capture fluctuations and changing trends in system performance in a timely manner. If evaluation is not performed at each time interval, the system may miss critical load peaks or processing bottlenecks, resulting in data backlogs, response delays and other problems. Therefore, although the formula itself calculates instantaneous processing capacity, frequent time interval calculations are still necessary to ensure the real-time responsiveness and stability of the system.

[0077] Evaluate the ratio of the system's processing capacity to the current load. The calculation expression is as follows:

[0078]

[0079] , where R(t) is the ratio of processing capacity to load, ∈ is a small constant used to avoid the denominator being zero;

[0080] The purpose of evaluating the ratio of the system's processing capacity to the current load is to measure the system's processing capacity under the current load conditions. This ratio can directly reflect whether the system's ability to process data is sufficient, especially when the load increases, helping to identify whether the system is in a bottleneck state.

[0081] In each time interval, the cumulative amount of unprocessed data is calculated, and the calculation expression is as follows:

[0082] U(t)=U(t-Δt)+D(t)-P(t)

[0083] , where U(t) is the cumulative amount of unprocessed data, indicating the total amount of unprocessed data at time point t, Δt is the time interval, and U(t-Δt) is the amount of unprocessed data at the previous time point;

[0084] According to the accumulated amount of unprocessed data, the data backlog index is generated, and the calculation expression is as follows:

[0085]

[0086] , where DBI is the data backlog index, U threshold is the preset unprocessed data volume threshold, D maxis the maximum data flow, which indicates the maximum amount of data that may be received in a certain period of time, β is the nonlinear coefficient, which indicates the nonlinear coefficient of the influence of the amount of new data on the backlog index, and γ is the load attenuation coefficient, which describes the influence of the system load on the backlog index;

[0087] The nonlinear coefficient β indicates the nonlinear influence of the amount of new data on the data backlog index. In the formula, β is used to adjust the impact of the amount of new data relative to the system pressure. Especially when the amount of data inflow is large, this nonlinear coefficient can amplify the impact of data flow. By introducing β, the backlog index can respond more sensitively to the impact of large data flow and reflect the potential pressure of the system when data surges. This enables the model to dynamically adjust the response of the backlog index when facing data flows of different sizes, thereby more accurately warning of backlog risks under high load conditions.

[0088] The load attenuation coefficient γ reflects the degree of suppression of the system load on the data backlog index, and describes the gradual upward trend of the system backlog risk when the load increases. In the formula, the load attenuation coefficient is expressed by the exponential term e -γ·L(t) The backlog index is attenuated. When the system load is high, this item will significantly reduce the growth rate of the backlog index. The introduction of this coefficient can balance the load impact and make the calculation of the backlog index more accurate. Especially under high load conditions, it can prevent the backlog index from expanding rapidly, providing system administrators with smoother and more predictable system status monitoring.

[0089] The larger the performance value of the data backlog index generated after analyzing the cumulative amount of unprocessed data in the detection window, the more likely it is that the data processing capacity of the data integration platform has reached a bottleneck, limiting the processing of real-time data. This is because the performance value of the data backlog index directly reflects the system's ability to process new data. If the index continues to rise, it means that the platform cannot process the incoming data stream in a timely manner, causing data retention and accumulation. On the contrary, when the performance value of the data backlog index is low, it means that the system can efficiently process real-time data, and the amount of unprocessed data remains within a controllable range, indicating that the processing capacity of the data integration platform has not reached a bottleneck and can stably respond to data requests.

[0090] The significant increase in the average response time of the platform to requests does indicate that the data processing capacity of the data integration platform has reached a bottleneck, limiting the processing of real-time data. This is because when the platform faces excessive data input, the system's computing resources (such as CPU, memory, and I / O bandwidth) will be occupied by high load, resulting in insufficient processing capacity. At this time, the system needs longer to complete data processing and feedback when receiving requests, which is reflected in the significant increase in response time. At the same time, the extension of response time may also cause task queuing, that is, new data requests need to wait for unfinished processing tasks, further increasing the burden on the system. This situation not only reduces the frequency of real-time data updates, but may also cause the prediction model to fail to obtain timely information, affecting the accuracy and timeliness of decision-making. In addition, the longer response time may also lead to a decline in user experience, which in turn affects the reliability and trust of the system. In short, the significantly increased average response time is a direct reflection of the bottleneck of data processing capacity, reflecting that the system cannot effectively manage data flow under high load, and it is urgent to optimize the processing capacity to ensure the smooth processing of real-time data.

[0091] In the detection window, the average response time of the platform to requests is analyzed. The specific steps to generate the system response time index are as follows:

[0092] Under the detection window, the response load function is defined. The response load function is used to describe the request load level of the system at time point t, reflecting the pressure of the current request volume on the system processing capacity. The calculation expression is as follows:

[0093]

[0094] , where Q(t) is the response load function, A is the reference constant of the response load, which is used to calibrate the amplitude of the load function, H(t) is the system demand at time t, and H max is the maximum number of requests that the system can sustain, ensuring e is the natural base, and μ is the time decay coefficient, which is used to simulate the gradual decay of requests over time;

[0095] The response load function Q(t) is a function used to describe the degree of request load that the system is subjected to at a specific point in time. Its function is to quantify the load pressure of the current request volume relative to the maximum carrying capacity of the system, thereby reflecting the load status of the system in processing requests at that moment. Through the response load function, the system can evaluate the impact of the current request volume on the processing capacity in real time, thereby providing information support at the load level for the calculation of the system response time index. If the request load is large, the value of Q(t) will also increase, indicating that the system load is high and the processing capacity is close to the limit; otherwise, it means that the system is in a low load state and the processing pressure is small. Therefore, the response load function plays a key role in measuring the load pressure here, so that the system response time index can more accurately reflect the impact of the load on the response capacity.

[0096] Evaluate the response rate of the system when processing requests and define the response rate function. The calculation expression is as follows:

[0097]

[0098] , where V(t) is the response rate function, B is the rate adjustment constant used to control the initial value of the response rate, φ(t) is the load adjustment factor, which is related to the current load of the system, K(t) is the actual response time of the system at time t, and K limit is the system response time limit constant, ensuring within a reasonable range;

[0099] The response rate function V(t) is a function used to evaluate the dynamic response speed of the system in processing requests at a specific point in time. It combines the current response time of the system and the load adjustment factor to reflect the response efficiency of the system under different load conditions. The role of the response rate function is to measure the processing speed of the system for requests when the load fluctuates, and to reflect the actual response capability of the system under the current load pressure by dynamically adjusting the response rate. When the system response time increases or the load pressure is high, the value of V(t) will decrease, indicating that the system response rate slows down and the efficiency of processing requests decreases; on the contrary, if the load is small and the response time is normal, V(t) will increase, indicating that the system responds quickly. The response rate function plays a role in dynamically measuring the response efficiency of the system, so that the system response time index can more comprehensively reflect the response status of the system in actual operation.

[0100] In order to reflect the real-time adaptability of the system response under different loads and requests, a dynamic adjustment factor is introduced. The calculation expression is as follows:

[0101]

[0102] , where θ(t) is the dynamic adjustment factor, η is the adjustment amplitude constant, which controls the overall amplitude of the dynamic adjustment factor, κ is the adjustment rate constant, which controls the impact of the load on the adjustment factor, and V ref is the reference rate constant used to prevent the rate value from approaching zero, ∈ is a small constant used to avoid the denominator being zero;

[0103] The role of the dynamic adjustment factor θ(t) is to reflect the adaptive adjustment ability of the system under different request loads and response rates, so that the system can flexibly respond to changes in load and response time. It dynamically adjusts the response ability of the system by combining load pressure and response rate, thereby adapting to instantaneous load fluctuations and response requirements. Specifically, when the load is high or the response rate is low, the value of the dynamic adjustment factor will decrease, indicating that the adaptability of the system is reduced and the response ability is limited; when the load is low and the response rate is fast, the dynamic adjustment factor increases, indicating that the system has high adaptability and adjustment space. Therefore, the dynamic adjustment factor plays a key role in adaptive regulation here, so that the system response time index can more accurately reflect the system's flexible response ability under instantaneous changes.

[0104] Combining the response load function Q(t), the response rate function V(t) and the dynamic adjustment factor θ(t), the system response time index is generated. The calculation expression is as follows:

[0105] SRTI=ln(q(t)·θ(t)+V(t)+δ)

[0106] , where SRTI is the system response time index and δ is a small constant used to prevent calculation errors of zero or negative values ​​in the logarithmic function;

[0107] In the detection window, the larger the system response time index performance value generated after analyzing the average response time of the platform to requests, the greater the value, indicating that the data processing capacity of the data integration platform is close to the bottleneck, limiting the processing of real-time data. This situation usually means that the system cannot maintain high efficiency when processing high-concurrency requests, resulting in longer response time and affecting the timely update and accuracy of data. On the contrary, if the index performance value is small, it means that the data processing capacity of the platform has not yet reached the bottleneck, and it can efficiently process real-time data requests and maintain good system performance and response speed.

[0108] The data backlog index DBI and the system response time index SRTI are input into the pre-learned machine learning model, and the system response coefficient SRC is generated by the machine learning model according to the following formula:

[0109]

[0110] , where f1 and f2 are the preset proportional coefficients of the data backlog index DBI and the system response time index SRTI, respectively, and both f1 and f2 are greater than 0.

[0111] It can be seen from the calculation expression of the system response coefficient that, under the detection window, the larger the performance value of the data backlog index generated after analyzing the cumulative amount of unprocessed data, the larger the performance value of the system response time index generated after analyzing the average response time of the platform to the request, indicating that the larger the performance value of the system response coefficient generated under the detection window, indicating that the data processing capacity of the data integration platform has approached the bottleneck, which limits the processing of real-time data. Conversely, it means that the data processing capacity of the data integration platform has not yet reached the bottleneck, and can efficiently process real-time data requests and maintain good system performance and response speed.

[0112] Based on the evaluation results of the machine learning model, the data integration platform operation process is divided into low-risk operation process or high-risk operation process, and the abnormal operation process of the data integration platform is intelligently perceived;

[0113] The generated system response coefficient is compared and analyzed with the preset system response coefficient reference threshold, and the operation process of the data integration platform is divided. The division process is as follows:

[0114] If the system response coefficient is greater than or equal to the preset system response coefficient reference threshold, the data integration platform operation process is classified as a high-risk operation process;

[0115] If the system response coefficient is less than the preset system response coefficient reference threshold, the data integration platform operation process is classified as a low-risk operation process;

[0116] For high-risk operation processes, continuously obtain parameter information during the operation of the data integration platform and divide the high-risk operation processes into sudden, continuous and unstable types;

[0117] For high-risk operation processes, continuously obtain several system response coefficients generated under several detection windows during the operation of the data integration platform to establish a data set and conduct comprehensive analysis;

[0118] The average values ​​and standard deviations of several system response coefficients in the data set are obtained, and the obtained average value and standard deviation of the system response coefficient are compared and analyzed with the preset reference threshold value of the system response coefficient average value and the preset reference threshold value of the system response coefficient standard deviation, respectively, and the high-risk operation status is further divided. The division results are as follows:

[0119] If the average value of the system response coefficient is less than a preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than a preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as a sudden type;

[0120] If the average value of the system response coefficient is greater than or equal to the preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than the preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as unstable;

[0121] If the average value of the system response coefficient is greater than or equal to the preset reference threshold value of the average value of the system response coefficient, the high-risk operation state is further classified as continuous;

[0122] Different early warning signals are issued for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow early warning is issued, and regular monitoring and in-depth analysis are carried out to ensure continuous attention to the system operation status so as to promptly discover and deal with potential problems. For unstable high-risk operation processes, an orange early warning is issued, prompting the operation team to strengthen monitoring and adjustment and be prepared for possible fluctuations to maintain system stability; and for continuous high-risk operation processes, a red early warning is issued, indicating that immediate action should be taken and relevant personnel are required to respond quickly and conduct system inspections to prevent the occurrence of emergencies.

[0123] The present invention can timely identify and evaluate the system operation status by acquiring and analyzing various parameter information of the data integration platform in real time. Through the application of feature extraction and machine learning models, the generated system response coefficient can quickly reflect the changes in data processing capabilities, thereby realizing effective monitoring of data backlogs and response time. This real-time perception capability enables power operators to quickly identify high-risk operation processes and take timely countermeasures to avoid inaccurate power forecasts and failed grid load management due to insufficient data processing capabilities, thereby ensuring the stability and reliability of power supply.

[0124] The present invention provides a more detailed and scientific risk management framework for power operators by subdividing high-risk operation processes into sudden, continuous and unstable types, and issuing corresponding warning signals according to different types. The graded prompts of red, orange and yellow warning signals help the operation team take targeted response measures in different situations and improve the system's ability to respond to emergencies. This optimized early warning mechanism not only enhances the ability to prevent potential risks, but also reduces economic losses caused by failure to respond in a timely manner, and improves the overall resilience and safety of the power system.

[0125] The present invention provides Figure 2The wind power and photovoltaic power prediction system under the transitional weather scenario shown in the figure includes a data acquisition and feature extraction module, a feature analysis and model evaluation module, a risk classification and intelligent perception module, a high-risk type classification module, and an early warning signal issuance module:

[0126] The data collection and feature extraction module acquires various parameter information generated during the operation of the data integration platform in real time, and performs feature extraction on the acquired parameter information;

[0127] The feature analysis and model evaluation module extracts features from the acquired parameters, analyzes the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests, and inputs the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests into the pre-learned machine learning model, generates a system response coefficient through the machine learning model, and intelligently evaluates the operation process of the data integration platform through the system response coefficient;

[0128] The risk classification and intelligent perception module divides the operation process of the data integration platform into low-risk operation process or high-risk operation process based on the evaluation results of the machine learning model, and intelligently perceives the abnormal operation process of the data integration platform;

[0129] The high-risk type classification module continuously obtains parameter information during the operation of the data integration platform for high-risk operation processes, and divides high-risk operation processes into sudden, continuous and unstable types;

[0130] The warning signal issuing module issues different warning signals for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow warning is issued, for unstable high-risk operation processes, an orange warning is issued, and for continuous high-risk operation processes, a red warning is issued.

[0131] The wind power and photovoltaic power prediction method under the transitional weather scenario provided by the embodiment of the present invention is realized by the wind power and photovoltaic power prediction system under the transitional weather scenario mentioned above. The specific method and process of the wind power and photovoltaic power prediction system under the transitional weather scenario are detailed in the embodiment of the wind power and photovoltaic power prediction method under the transitional weather scenario mentioned above, which will not be repeated here.

[0132] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for predicting wind and photovoltaic power under transitional weather scenarios, characterized in that: The following steps are involved: During the operation of the data integration platform, various parameter information generated during the operation of the data integration platform is obtained in real time, and feature extraction is performed on the obtained parameter information; After feature extraction of the acquired parameters, the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests are analyzed, and the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests are input into the pre-learned machine learning model, and the system response coefficient is generated through the machine learning model, and the operation process of the data integration platform is intelligently evaluated through the system response coefficient; Based on the evaluation results of the machine learning model, the data integration platform operation process is divided into low-risk operation process or high-risk operation process, and the abnormal operation process of the data integration platform is intelligently perceived; For high-risk operation processes, continuously obtain parameter information during the operation of the data integration platform and divide the high-risk operation processes into sudden, continuous and unstable types; Different warning signals are issued for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow warning is issued, for unstable high-risk operation processes, an orange warning is issued, and for continuous high-risk operation processes, a red warning is issued.

2. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 1 is characterized by: After analyzing the cumulative amount of unprocessed data and the average response time of the platform to requests, the data backlog index and system response time index are generated respectively. The data backlog index and system response time index are input into the pre-learned machine learning model, and the system response coefficient is generated by the machine learning model. The system response coefficient is used to perform an intelligent evaluation of the operation process of the data integration platform.

3. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 2 is characterized in that: Under the detection window, the accumulated amount of unprocessed data is analyzed and the specific steps to generate the data backlog index are as follows: In the detection window, the input data streams from various sensors and monitoring devices are collected in real time to ensure the integrity and accuracy of the data source, and the acquired input data streams are calibrated as D(t), which represents the amount of new data received by the system at time t; Calculate the amount of real-time data that the system can process at each time point. The calculation expression is as follows: , where P(t) is the instantaneous processing capacity, C is the maximum processing capacity of the system, L(t) is the current system load, and L max is the maximum load threshold of the system, e -α·t is the time decay term, α is the data processing efficiency decay coefficient, and e is the natural base; Evaluate the ratio of the system's processing capacity to the current load. The calculation expression is as follows: , where R(t) is the ratio of processing capacity to load, ∈ is a small constant; In each time interval, the cumulative amount of unprocessed data is calculated, and the calculation expression is as follows: U(t)=U(t-Δt)+D(t)-P(t), where U(t) is the cumulative amount of unprocessed data, indicating the total amount of unprocessed data at time point t, Δt is the time interval, and U(t-Δt) is the amount of unprocessed data at the previous time point; According to the accumulated amount of unprocessed data, the data backlog index is generated, and the calculation expression is as follows: , where DBI is the data backlog index, U threshold is the preset unprocessed data volume threshold, D max is the maximum data flow, β is the nonlinear coefficient, and γ is the load attenuation coefficient, which describes the influence of system load on the backlog index.

4. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 2 is characterized in that: Under the detection window, the response load function is defined. The response load function is used to describe the request load level of the system at time point t, reflecting the pressure of the current request volume on the system processing capacity. The calculation expression is as follows: , where Q(t) is the response load function, A is the reference constant of the response load, which is used to calibrate the amplitude of the load function, H(t) is the system demand at time t, and H max is the maximum number of requests that the system can bear, e is the natural base, and μ is the time decay coefficient, which is used to simulate the gradual decay of requests over time; Evaluate the response rate of the system when processing requests and define the response rate function. The calculation expression is as follows: , where V(t) is the response rate function, B is the rate adjustment constant used to control the initial value of the response rate, φ(t) is the load adjustment factor, K(t) is the actual response time of the system at time t, and K limit is the system response time limit constant, ensuring within a reasonable range; In order to reflect the real-time adaptability of the system response under different loads and requests, a dynamic adjustment factor is introduced. The calculation expression is as follows: , where θ(t) is the dynamic adjustment factor, η is the adjustment amplitude constant, which controls the overall amplitude of the dynamic adjustment factor, κ is the adjustment rate constant, which controls the impact of the load on the adjustment factor, and V ref is the reference rate constant, ∈ is a small constant; Combining the response load function Q(t), the response rate function V(t) and the dynamic adjustment factor θ(t), the system response time index is generated. The calculation expression is as follows: SRTI = ln(Q(t)·θ(t)+V(t)+δ), where SRTI is the system response time index and δ is a small constant.

5. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 2 is characterized by: The data backlog index DBI and the system response time index SRTI are input into the pre-learned machine learning model, and the system response coefficient SRC is generated by the machine learning model according to the following formula: , where f1 and f2 are the preset proportional coefficients of the data backlog index DBI and the system response time index SRTI, respectively, and both f1 and f2 are greater than 0.

6. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 5 is characterized by: The generated system response coefficient is compared and analyzed with the preset system response coefficient reference threshold, and the operation process of the data integration platform is divided. The division process is as follows: If the system response coefficient is greater than or equal to the preset system response coefficient reference threshold, the data integration platform operation process is classified as a high-risk operation process; If the system response coefficient is less than a preset system response coefficient reference threshold, the data integration platform operation process is classified as a low-risk operation process.

7. The method for predicting wind and photovoltaic power under transitional weather scenarios according to claim 6 is characterized by: For high-risk operation processes, continuously obtain several system response coefficients generated under several detection windows during the operation of the data integration platform to establish a data set and conduct comprehensive analysis; The average values ​​and standard deviations of several system response coefficients in the data set are obtained, and the obtained average value and standard deviation of the system response coefficient are compared and analyzed with the preset reference threshold value of the system response coefficient average value and the preset reference threshold value of the system response coefficient standard deviation, respectively, and the high-risk operation status is further divided. The division results are as follows: If the average value of the system response coefficient is less than a preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than a preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as a sudden type; If the average value of the system response coefficient is greater than or equal to the preset reference threshold value of the average value of the system response coefficient and the standard deviation of the system response coefficient is less than the preset reference threshold value of the standard deviation of the system response coefficient, the high-risk operation state is further classified as unstable; If the system response coefficient average value is greater than or equal to the preset system response coefficient average value reference threshold, the high-risk operation state is further divided into a continuous type.

8. A wind power and photovoltaic power prediction system under transitional weather scenarios, used to implement the wind power and photovoltaic power prediction method under transitional weather scenarios described in any one of claims 1 to 7, characterized in that: It includes data collection and feature extraction module, feature analysis and model evaluation module, risk classification and intelligent perception module, high-risk type classification module and early warning signal issuance module: The data collection and feature extraction module acquires various parameter information generated during the operation of the data integration platform in real time, and performs feature extraction on the acquired parameter information; The feature analysis and model evaluation module extracts features from the acquired parameters, analyzes the extracted features including the cumulative amount of unprocessed data and the average response time of the platform to requests, and inputs the analyzed cumulative amount of unprocessed data and the average response time of the platform to requests into the pre-learned machine learning model, generates a system response coefficient through the machine learning model, and intelligently evaluates the operation process of the data integration platform through the system response coefficient; The risk classification and intelligent perception module divides the operation process of the data integration platform into low-risk operation process or high-risk operation process based on the evaluation results of the machine learning model, and intelligently perceives the abnormal operation process of the data integration platform; The high-risk type classification module continuously obtains parameter information during the operation of the data integration platform for high-risk operation processes, and divides high-risk operation processes into sudden, continuous and unstable types; The warning signal issuing module issues different warning signals for different types of high-risk operation processes. For sudden high-risk operation processes, a yellow warning is issued, for unstable high-risk operation processes, an orange warning is issued, and for continuous high-risk operation processes, a red warning is issued.

Images