Time-sharing power supply and demand intelligent regulation and control system and method based on large model
By building a large-scale intelligent control system for time-sharing power supply and demand, utilizing multi-dimensional feature extraction and correlation analysis, dynamically classifying event types, and building a supply and demand change model, we have achieved precise control of power supply and demand, solved the problem of large prediction deviation in traditional control methods, and improved the economy and stability of the power grid.
Patent Information
- Application Number
- CN202510821310.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-19
- Publication Date
- 2025-09-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional power control methods are unable to accurately capture the time-divided and multi-source supply and demand fluctuation characteristics, especially during periods of frequent fluctuation events, resulting in large forecast deviations, low control efficiency and insufficient economy.
By building a time-sharing power supply and demand intelligent control system based on a large model, using multi-dimensional feature extraction and correlation analysis, dynamically dividing event types, building a supply and demand change model, and capturing fluctuation events in real time and performing precise control.
It significantly improves the recognition accuracy and classification efficiency of complex fluctuation events, solves the prediction deviation problem of traditional time series extrapolation method in multi-source heterogeneous data scenarios, and improves the economy and stability of the power grid.
Smart Images
Figure CN120675284A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power regulation and control, and in particular to a system and method for intelligently regulating time-sharing power supply and demand based on a large model. Background Art
[0002] With the large-scale integration of new energy sources and the diversification of electricity loads, the supply and demand balance of the power system faces severe challenges. Traditional power regulation relies mainly on historical data statistics and manual experience, which makes it difficult to accurately capture the characteristics of supply and demand fluctuations in different time periods and from multiple sources. This is particularly limited in scenarios where fluctuations occur frequently. Traditional control methods mainly rely on the time-series extrapolation prediction of historical load curves, which makes it difficult to accurately capture the implicit correlation between multi-source heterogeneous data and power fluctuations, resulting in large prediction deviations. Especially during periods of high incidence of fluctuation events, the existing system has a delayed response to sudden power shortages or surpluses and often needs to rely on manual experience intervention, resulting in low control efficiency and insufficient economic efficiency. Summary of the Invention
[0003] The purpose of the present invention is to provide a time-sharing power supply and demand intelligent control system and method based on a large model to solve the problems raised in the prior art.
[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently controlling time-sharing power supply and demand based on a large model, the control method comprising the following steps: Step S100: collecting fluctuations in power supply and demand every unit cycle to generate supply and demand fluctuation records for each unit cycle; presetting several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extracting each fluctuation interval to match the corresponding fluctuation event; Step S200: Conduct multi-dimensional analysis on any fluctuation event to extract several event features of the fluctuation event; perform correlation analysis on the event features of the fluctuation event and the power supply and demand situation to classify the event types of each fluctuation event; Step S300: performing differential analysis on the fluctuations of different fluctuation events within any same event type, obtaining a number of differential features of the same event type, and analyzing the impact of each differential feature; constructing a supply and demand change model to analyze the expected power supply and demand of any fluctuation event; Step S400: Identify the real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; based on the impact of each difference feature, use the supply and demand change model to predict the expected supply and demand situation, and adjust the current power supply and demand situation accordingly based on the prediction results.
[0005] Furthermore, step S100 includes the following steps: Step S101: Pre-build a power supply and demand management platform to collect power supply and power demand at each time point, and present the changes in power supply and power demand in chronological order. Pre-set a unit period, capture the changes in power supply and power demand in each unit period, and generate corresponding supply and demand fluctuation records. Step S102: arbitrarily selecting a supply and demand fluctuation record, presetting a corresponding event identification rule for each channel, and obtaining a plurality of fluctuation events of the selected supply and demand fluctuation record; arbitrarily selecting a fluctuation event from the plurality of fluctuation events, obtaining a channel source of the selected fluctuation event, and obtaining an occurrence time interval of the selected fluctuation event based on the start time point and end time point of the selected fluctuation event in the channel source; obtaining the occurrence time interval of each fluctuation event, and dividing the unit period corresponding to the selected supply and demand fluctuation record into a plurality of time intervals; Several channels are preset, including meteorological monitoring channels, load monitoring channels, social event channels, etc.; for example, the meteorological monitoring channel can determine whether it is in an extreme environment by setting intelligent rules such as temperature range or light intensity, which will affect the changes in electricity supply and demand; for example, the social event channel can identify characteristic events by extracting keywords from various power notification messages, such as electricity price changes and regional power outages.
[0006] Furthermore, step S200 includes the following steps: Step S201: Pre-train a feature association database, which stores feature sets of several dimensions. A fluctuation event is randomly selected, and features of the selected fluctuation event in various dimensions are extracted using the feature association database to obtain several features of the selected fluctuation event. The dimensions include time dimension features such as duration, occurrence period, and periodic changes; spatial dimension features such as regional range; and intensity dimension features such as fluctuation amplitude and change rate. Step S202: randomly selecting a feature from the selected fluctuation event, obtaining a value variation range of the selected feature within the occurrence time interval of the selected fluctuation event; matching the arbitrary feature with the corresponding value variation range to obtain an event feature of the selected fluctuation event; Step S203: arbitrarily select an event feature and extract the occurrence time interval corresponding to the event feature, obtain the power demand and power supply at each time point in the occurrence time interval, calculate the power surplus at each time point, and set the power surplus at the i-th time point as S i The power surplus at time point i+1 is S i+1 , we get the surplus difference ΔS between the i-th time point and the i+1-th time point (i,i+1); Get the profit difference between any two adjacent time points, and calculate the average value to get the average profit difference ΔS in the time interval ave , if ΔS ave ≥0, the selected fluctuation event is set as the first characteristic event. If ΔS ave <0, the selected fluctuation event is set as the second characteristic event; the first characteristic event represents an event that can promote the power supply to exceed the demand, and the second characteristic event represents an event that makes the power demand exceed the supply; Step S204: Obtain the value variation range of the feature in the occurrence time interval, and set the feature value at the i-th time point as D i And the characteristic value of the i+1th time point is D i+1 , then the numerical difference ΔD between the i-th time point and the i+1-th time point is obtained (i,i+1) , if ΔS (i,i+1) ×ΔD (i,i+1) >0, then the numerical difference ΔD (i,i+1) Set as feature difference; count the number of feature differences in the occurrence time interval as B, set the number of time points included in the occurrence time interval as A, and calculate the number of feature differences as η=B / A; preset a number ratio threshold η th , if η≥η th , then the selected event feature is set as the first associated feature group of the selected fluctuation event, if η<η th , the selected event features are set as the second associated feature group of the selected fluctuation event; the first associated feature group represents the features that can promote the event itself. For example, if the event itself needs to help supply exceed demand, then the first associated feature group needs to promote the occurrence of this situation, while the second associated feature group is to inhibit the occurrence of this situation; Step S205: Count and select the number of first correlation feature groups in the fluctuation event as C1, and select another fluctuation event, and count and obtain the number of first correlation feature groups of the other fluctuation event as C1. ’ , perform similarity comparison on the first associated feature groups in the two fluctuation events respectively. If the similarity obtained by comparison exceeds the preset similarity threshold, the two compared first associated feature groups are set as the same feature group, and the number of the same feature groups is counted as C sim , pre-set a quantity threshold C th and the same proportion threshold X th , if C sim >C th And C sim / max(C1,C1 ’ )>X th , where max() is the maximum value function, the two fluctuation events are classified as the same event type.
[0007] Furthermore, step S300 includes the following steps: Step S301: arbitrarily select an event of the same type, and arbitrarily extract two fluctuation events from the selected event type, and obtain all event features of the two fluctuation events respectively; compare the event features of the two fluctuation events to obtain a set of difference features between the two fluctuation events; Step S302: Perform a union operation on the event features in all the difference feature sets of the same event type to obtain a complete difference feature set; arbitrarily select an event feature from the complete difference feature set, extract the target fluctuation event where the selected event feature is located, and set the power surplus of the target fluctuation event at the starting time point as S start and the power surplus at the end time is S end , the comprehensive surplus of the target volatility event in the time interval is ΔS=S end -S start ; Step S303: Randomly select two adjacent time points from the occurrence time interval, obtain the power surplus at the two adjacent time points and the characteristic value of the selected time feature group, set the i-th time point and the i+1-th time point to obtain the surplus difference ΔS between the two adjacent time points. (i,i+1) and the numerical difference ΔD (i,i+1) ; The number of event features included in the complete difference feature set in the target fluctuation event is n, and the influence of the selected event features on the power surplus between two adjacent time points is calculated to be Y (i,i+1) =ΔS (i,i+1) / [n×ΔD (i,i+1) ]; take the average of the influence amplitudes of the selected event features at any adjacent time points to obtain the average influence amplitude Y of the selected event features in the same event type ave ; Step S304: arbitrarily select a fluctuation event in a supply and demand fluctuation record, extract each event feature in the selected fluctuation event, and compare it with each event feature in the complete difference feature set to obtain several difference event features of the selected fluctuation event; set the average impact amplitude of the e-th difference event feature as Y e , construct a supply and demand change model to obtain the expected power surplus ΔS of the fluctuation event ex : ; Among them, r is the number of difference event features contained in the selected fluctuation event, and g is a constant; Step S305: Obtain the comprehensive surplus ΔS of the selected fluctuation event in the occurrence time interval. If ΔS < 0, then , similarly if ΔS≥0, then , and obtain the value range of the constant g; input each fluctuation event into the supply and demand change model to calculate the expected power surplus, and extract the comprehensive surplus of each fluctuation event to obtain the value range of the constant coefficient g, perform an intersection operation on the value ranges of each constant coefficient g, and obtain the revised value range of the constant g, and arbitrarily select a constant from the revised value range to adjust the supply and demand change model.
[0008] Furthermore, step S400 includes the following steps: Step S401: identifying and collecting fluctuation events in real time through several preset channels. When a real-time fluctuation event is collected, feature extraction is performed on the real-time fluctuation event to obtain several difference features. Step S402: Compare the features extracted from the real-time fluctuation event with various event types to determine the event type of the real-time fluctuation event, and obtain the average impact amplitude of each difference feature under the corresponding event type, and call the supply and demand change model to obtain the expected power surplus of the real-time fluctuation event; if the expected power surplus is less than zero, send a risk alert to the power supply side.
[0009] In order to better implement the above method, a time-sharing power supply and demand intelligent control system is also proposed. The control system includes a historical supply and demand analysis module, an event feature association module, a change model analysis module and a real-time supply and demand analysis module. The historical supply and demand analysis module is used to collect fluctuations in power supply and demand every unit period and generate supply and demand fluctuation records for each unit period. Several channels are preset to identify fluctuation events that occur in any supply and demand fluctuation record, and each fluctuation interval is extracted and matched with the corresponding fluctuation event. The event feature association module is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; associate the event features of any fluctuation event with the power supply and demand situation and classify the event types of each fluctuation event; The fluctuation model analysis module is used to perform differential analysis on the fluctuation conditions of different fluctuation events within any same event type, obtain several differential features of the same event type, and analyze the impact of each differential feature; construct a supply and demand fluctuation model to analyze the expected power supply and demand of any fluctuation event; The real-time supply and demand analysis module is used to identify real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; based on the impact of each difference feature, the supply and demand change model is used to predict the expected supply and demand situation, and the current power supply and demand situation is adjusted accordingly based on the prediction results.
[0010] Furthermore, the historical supply and demand analysis module includes a time-sharing supply and demand collection unit and a fluctuation event identification unit; The time-sharing supply and demand collection unit is used to collect the fluctuation of power supply and demand every unit cycle and generate a supply and demand fluctuation record for each unit cycle; the fluctuation event identification unit is used to preset several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extract each fluctuation interval and match it with the corresponding fluctuation event.
[0011] Furthermore, the event feature association module includes an event feature extraction unit and an event association analysis unit; The event feature extraction unit is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; the event correlation analysis unit is used to correlate the event features of any fluctuation event with the power supply and demand situation and classify the event types of each fluctuation event.
[0012] Furthermore, the change model analysis module includes a difference feature impact unit and a change model construction unit; The difference feature impact unit is used to perform difference analysis on the fluctuation conditions of different fluctuation events in any same event type, obtain several difference features of the same event type, and analyze the impact of each difference feature; the change model construction unit is used to construct a supply and demand change model to analyze the expected power supply and demand of any fluctuation event.
[0013] Furthermore, the real-time supply and demand analysis module includes a real-time event analysis unit and an expected supply and demand prediction unit; The real-time event analysis unit is used to identify real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; the expected supply and demand prediction unit is used to predict the expected supply and demand situation based on the influence of each difference feature and use the supply and demand change model to adjust the current power supply and demand situation accordingly based on the prediction results.
[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention automatically extracts multi-dimensional feature events from power data and dynamically classifies event types based on the correlation between power surplus and features. Compared with traditional manual experience-based control, this method significantly improves the recognition accuracy and classification efficiency of complex fluctuation events, laying the foundation for precise control. 2. By constructing a supply and demand forecasting model based on differential characteristics, the present invention can effectively solve the prediction bias problem of traditional time series extrapolation methods in multi-source heterogeneous data scenarios, and is particularly suitable for sudden power shortage or oversupply warning situations; 3. The present invention quickly generates the expected power surplus prediction results by capturing fluctuation events in real time, matching the historical event type library, and calling the supply and demand change model. It can respond to the power supply and demand situation very quickly and significantly improve the economy and stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 A schematic diagram of the steps of a method for intelligent control of time-sharing power supply and demand based on a large model; Figure 2 This is a structural diagram of a time-sharing power supply and demand intelligent control system based on a large model. DETAILED DESCRIPTION
[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0017] Example: Figures 1 to 2 As shown, the present invention provides a method for intelligently controlling time-sharing power supply and demand based on a large model, the control method comprising the following steps: Step S100: collecting fluctuations in power supply and demand every unit cycle to generate supply and demand fluctuation records for each unit cycle; presetting several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extracting each fluctuation interval to match the corresponding fluctuation event; Wherein, step S100 includes the following steps: Step S101: Pre-build a power supply and demand management platform to collect power supply and power demand at each time point, and present the changes in power supply and power demand in chronological order. Pre-set a unit period, capture the changes in power supply and power demand in each unit period, and generate corresponding supply and demand fluctuation records. Step S102: arbitrarily select a supply and demand fluctuation record, preset corresponding event identification rules for each channel, and obtain several fluctuation events of the selected supply and demand fluctuation record; arbitrarily select a fluctuation event from the several fluctuation events, obtain the channel source of the selected fluctuation event, and obtain the occurrence time interval of the selected fluctuation event for the starting time point and the ending time point of the selected fluctuation event in the channel source; obtain the occurrence time interval of each fluctuation event, and divide the unit period corresponding to the selected supply and demand fluctuation record into several time intervals.
[0018] Step S200: Conduct multi-dimensional analysis on any fluctuation event to extract several event features of the fluctuation event; perform correlation analysis on the event features of the fluctuation event and the power supply and demand situation to classify the event types of each fluctuation event; Wherein, step S200 includes the following steps: Step S201: Pre-train a feature association database, which stores feature sets of several dimensions. A fluctuation event is randomly selected, and features of the selected fluctuation event in various dimensions are extracted using the feature association database to obtain several features of the selected fluctuation event. Step S202: randomly selecting a feature from the selected fluctuation event, obtaining a value variation range of the selected feature within the occurrence time interval of the selected fluctuation event; matching the arbitrary feature with the corresponding value variation range to obtain an event feature of the selected fluctuation event; Step S203: arbitrarily select an event feature and extract the occurrence time interval corresponding to the event feature, obtain the power demand and power supply at each time point in the occurrence time interval, calculate the power surplus at each time point, and set the power surplus at the i-th time point as S i The power surplus at time point i+1 is S i+1 , we get the surplus difference ΔS between the i-th time point and the i+1-th time point (i,i+1) ; Get the profit difference between any two adjacent time points, and calculate the average value to get the average profit difference ΔS in the time interval ave , if ΔS ave ≥0, the selected fluctuation event is set as the first characteristic event. If ΔS ave <0, the selected fluctuation event is set as the second characteristic event; Step S204: Obtain the value variation range of the feature in the occurrence time interval, and set the feature value at the i-th time point as D i And the characteristic value of the i+1th time point is D i+1 , then the numerical difference ΔD between the i-th time point and the i+1-th time point is obtained (i,i+1) , if ΔS(i,i+1) ×ΔD (i,i+1) >0, then the numerical difference ΔD (i,i+1) Set as feature difference; count the number of feature differences in the occurrence time interval as B, set the number of time points included in the occurrence time interval as A, and calculate the number of feature differences as η=B / A; preset a number ratio threshold η th , if η≥η th , then the selected event feature is set as the first associated feature group of the selected fluctuation event, if η<η th , then the selected event feature is set as the second associated feature group of the selected fluctuation event; Example 1: Set a fluctuation event as the first characteristic event, namely ΔS ave ≥0, then extract one of the features and the corresponding value range, and get the value difference between two adjacent time points as ΔD (i,i+1) =3, because ΔS (i,i+1) ×ΔD (i,i+1) >0, this feature is set as the feature difference; statistics show that there are 10 feature differences in the time interval of the fluctuation event, and there are 20 time points in the time interval, so the number of feature differences accounts for 50%, and the preset number ratio threshold is 40%. Therefore, this feature is the first associated feature group of the fluctuation event; Step S205: Count and select the number of first correlation feature groups in the fluctuation event as C1, and select another fluctuation event, and count and obtain the number of first correlation feature groups of the other fluctuation event as C1. ’ , perform similarity comparison on the first associated feature groups in the two fluctuation events respectively. If the similarity obtained by comparison exceeds the preset similarity threshold, the two compared first associated feature groups are set as the same feature group, and the number of the same feature groups is counted as C sim , pre-set a quantity threshold C th and the same proportion threshold X th , if C sim >C th And C sim / max(C1,C1 ’ )>X th , where max() is the maximum value function, the two fluctuation events are classified as the same event type.
[0019] Step S300: performing differential analysis on the fluctuations of different fluctuation events within any same event type, obtaining a number of differential features of the same event type, and analyzing the impact of each differential feature; constructing a supply and demand change model to analyze the expected power supply and demand of any fluctuation event; Wherein, step S300 includes the following steps: Step S301: arbitrarily select an event of the same type, and arbitrarily extract two fluctuation events from the selected event type, and obtain all event features of the two fluctuation events respectively; compare the event features of the two fluctuation events to obtain a set of difference features between the two fluctuation events; Step S302: Perform a union operation on the event features in all the difference feature sets of the same event type to obtain a complete difference feature set; arbitrarily select an event feature from the complete difference feature set, extract the target fluctuation event where the selected event feature is located, and set the power surplus of the target fluctuation event at the starting time point as S start and the power surplus at the end time is S end , the comprehensive surplus of the target volatility event in the time interval is ΔS=S end -S start ; Step S303: Randomly select two adjacent time points from the occurrence time interval, obtain the power surplus at the two adjacent time points and the characteristic value of the selected time feature group, set the i-th time point and the i+1-th time point to obtain the surplus difference ΔS between the two adjacent time points. (i,i+1) and the numerical difference ΔD (i,i+1) ; The number of event features included in the complete difference feature set in the target fluctuation event is n, and the influence of the selected event features on the power surplus between two adjacent time points is calculated to be Y (i,i+1) =ΔS (i,i+1) / [n×ΔD (i,i+1) ]; take the average of the influence amplitudes of the selected event features at any adjacent time points to obtain the average influence amplitude Y of the selected event features in the same event type ave ; Step S304: arbitrarily select a fluctuation event in a supply and demand fluctuation record, extract each event feature in the selected fluctuation event, and compare it with each event feature in the complete difference feature set to obtain several difference event features of the selected fluctuation event; set the average impact amplitude of the e-th difference event feature as Y e , construct a supply and demand change model to obtain the expected power surplus ΔS of the fluctuation event ex : ; Among them, r is the number of difference event features contained in the selected fluctuation event, and g is a constant; Example 2: Assume that there are four different event characteristics in a fluctuation event, with average impact amplitudes of 20%, 30%, -15% and 10% respectively. Set the constant g to 0 and calculate the expected power surplus ΔS ex=45%>0, assuming that the comprehensive surplus of the volatility event in the time interval is greater than 0, then it holds true; if there is another volatility event with a comprehensive surplus of ΔS<0, and the average impact of the characteristics of each difference event is 10%, 5%, -8% and -2% respectively, then it does not hold if g=0. Therefore, it is necessary to set the constant g=-6%, so that both volatility events can hold true; Step S305: Obtain the comprehensive surplus ΔS of the selected fluctuation event in the occurrence time interval. If ΔS < 0, then , similarly if ΔS≥0, then , and obtain the value range of the constant g; input each fluctuation event into the supply and demand change model to calculate the expected power surplus, and extract the comprehensive surplus of each fluctuation event to obtain the value range of the constant coefficient g, perform an intersection operation on the value ranges of each constant coefficient g, and obtain the revised value range of the constant g, and arbitrarily select a constant from the revised value range to adjust the supply and demand change model.
[0020] Step S400: Identifying real-time fluctuation events at the current time point and extracting differential features from the real-time fluctuation events; predicting the expected supply and demand situation using a supply and demand change model based on the impact of each differential feature, and adjusting the current power supply and demand situation accordingly based on the prediction results; Step S400 includes the following steps: Step S401: identifying and collecting fluctuation events in real time through several preset channels. When a real-time fluctuation event is collected, feature extraction is performed on the real-time fluctuation event to obtain several difference features. Step S402: Compare the features extracted from the real-time fluctuation event with various event types to determine the event type of the real-time fluctuation event, and obtain the average impact amplitude of each difference feature under the corresponding event type, and call the supply and demand change model to obtain the expected power surplus of the real-time fluctuation event; if the expected power surplus is less than zero, send a risk alert to the power supply side.
[0021] An intelligent control system for time-sharing power supply and demand, the control system includes a historical supply and demand analysis module, an event feature correlation module, a change model analysis module and a real-time supply and demand analysis module; The historical supply and demand analysis module is used to collect fluctuations in power supply and demand every unit period and generate supply and demand fluctuation records for each unit period. Several channels are preset to identify fluctuation events that occur in any supply and demand fluctuation record, and each fluctuation interval is extracted and matched with the corresponding fluctuation event. The event feature association module is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; associate the event features of any fluctuation event with the power supply and demand situation and classify the event types of each fluctuation event; The fluctuation model analysis module is used to perform differential analysis on the fluctuation conditions of different fluctuation events within any same event type, obtain several differential features of the same event type, and analyze the impact of each differential feature; construct a supply and demand fluctuation model to analyze the expected power supply and demand of any fluctuation event; The real-time supply and demand analysis module is used to identify real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; based on the impact of each difference feature, the supply and demand change model is used to predict the expected supply and demand situation, and the current power supply and demand situation is adjusted accordingly based on the prediction results.
[0022] Among them, the historical supply and demand analysis module includes a time-sharing supply and demand collection unit and a fluctuation event identification unit; The time-sharing supply and demand collection unit is used to collect the fluctuation of power supply and demand every unit cycle and generate a supply and demand fluctuation record for each unit cycle; the fluctuation event identification unit is used to preset several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extract each fluctuation interval and match it with the corresponding fluctuation event.
[0023] Among them, the event feature association module includes an event feature extraction unit and an event association analysis unit; The event feature extraction unit is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; the event correlation analysis unit is used to correlate the event features of any fluctuation event with the power supply and demand situation and classify the event types of each fluctuation event.
[0024] Among them, the change model analysis module includes a difference feature impact unit and a change model construction unit; The difference feature impact unit is used to perform difference analysis on the fluctuation conditions of different fluctuation events in any same event type, obtain several difference features of the same event type, and analyze the impact of each difference feature; the change model construction unit is used to construct a supply and demand change model to analyze the expected power supply and demand of any fluctuation event.
[0025] Among them, the real-time supply and demand analysis module includes a real-time event analysis unit and an expected supply and demand forecasting unit; The real-time event analysis unit is used to identify real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; the expected supply and demand prediction unit is used to predict the expected supply and demand situation based on the influence of each difference feature and use the supply and demand change model to adjust the current power supply and demand situation accordingly based on the prediction results.
[0026] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.
Claims
1. A method for intelligent control of time-sharing power supply and demand based on a large model, characterized by: The control method comprises the following steps: Step S100: collecting fluctuations in power supply and demand every unit cycle to generate supply and demand fluctuation records for each unit cycle; presetting several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extracting each fluctuation interval to match the corresponding fluctuation event; Step S200: Conduct multi-dimensional analysis on any fluctuation event to extract several event features of the fluctuation event; perform correlation analysis on the event features of the fluctuation event and the power supply and demand situation to classify the event types of each fluctuation event; Step S300: performing differential analysis on the fluctuations of different fluctuation events within any same event type, obtaining a number of differential features of the same event type, and analyzing the impact of each differential feature; constructing a supply and demand change model to analyze the expected power supply and demand of any fluctuation event; Step S400: Identify the real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; based on the impact of each difference feature, use the supply and demand change model to predict the expected supply and demand situation, and adjust the current power supply and demand situation accordingly based on the prediction results.
2. The method for intelligent control of time-sharing power supply and demand based on a large model according to claim 1, characterized in that: The step S100 includes the following steps: Step S101: Pre-build a power supply and demand management platform to collect power supply and power demand at each time point, and present the changes in power supply and power demand in chronological order. Pre-set a unit period, capture the changes in power supply and power demand in each unit period, and generate corresponding supply and demand fluctuation records. Step S102: arbitrarily select a supply and demand fluctuation record, preset corresponding event identification rules for each channel, and obtain several fluctuation events of the selected supply and demand fluctuation record; arbitrarily select a fluctuation event from the several fluctuation events, obtain the channel source of the selected fluctuation event, and obtain the occurrence time interval of the selected fluctuation event for the starting time point and the ending time point of the selected fluctuation event in the channel source; obtain the occurrence time interval of each fluctuation event, and divide the unit period corresponding to the selected supply and demand fluctuation record into several time intervals.
3. The method for intelligent control of time-sharing power supply and demand based on a large model according to claim 2, characterized in that: The step S200 includes the following steps: Step S201: Pre-train a feature association database, which stores feature sets of several dimensions. A fluctuation event is randomly selected, and features of the selected fluctuation event in various dimensions are extracted using the feature association database to obtain several features of the selected fluctuation event. Step S202: randomly selecting a feature from the selected fluctuation event, obtaining a value variation range of the selected feature within the occurrence time interval of the selected fluctuation event; matching the arbitrary feature with the corresponding value variation range to obtain an event feature of the selected fluctuation event; Step S203: arbitrarily select an event feature and extract the occurrence time interval corresponding to the event feature, obtain the power demand and power supply at each time point in the occurrence time interval, calculate the power surplus at each time point, and set the power surplus at the i-th time point as S i The power surplus at time point i+1 is S i+1 , we get the surplus difference ΔS between the i-th time point and the i+1-th time point (i,i+1) ; Get the profit difference between any two adjacent time points, and calculate the average value to get the average profit difference ΔS in the time interval ave , if ΔS ave ≥0, the selected fluctuation event is set as the first characteristic event. If ΔS ave <0, the selected fluctuation event is set as the second characteristic event; Step S204: Obtain the value variation range of the feature in the occurrence time interval, and set the feature value at the i-th time point as D i And the characteristic value of the i+1th time point is D i+1 , then the numerical difference ΔD between the i-th time point and the i+1-th time point is obtained (i,i+1) , if ΔS (i,i+1) ×ΔD (i,i+1) >0, then the numerical difference ΔD (i,i+1) Set as feature difference; count the number of feature differences in the occurrence time interval as B, set the number of time points included in the occurrence time interval as A, and calculate the number of feature differences as η=B / A; preset a number ratio threshold η th , if η≥η th , then the selected event feature is set as the first associated feature group of the selected fluctuation event, if η<η th , then the selected event feature is set as the second associated feature group of the selected fluctuation event; Step S205: Count and select the number of first correlation feature groups in the fluctuation event as C1, and select another fluctuation event, and count and obtain the number of first correlation feature groups of the other fluctuation event as C1. ’ , perform similarity comparison on the first associated feature groups in the two fluctuation events respectively. If the similarity obtained by comparison exceeds the preset similarity threshold, the two compared first associated feature groups are set as the same feature group, and the number of the same feature groups is counted as C sim , pre-set a quantity threshold C th and the same proportion threshold X th , if C sim >C th And C sim / max(C1,C1 ’ )>X th , where max() is the maximum value function, the two fluctuation events are classified as the same event type.
4. The method for intelligent control of time-sharing power supply and demand based on a large model according to claim 3 is characterized by: The step S300 includes the following steps: Step S301: arbitrarily select an event of the same type, and arbitrarily extract two fluctuation events from the selected event type, and obtain all event features of the two fluctuation events respectively; compare the event features of the two fluctuation events to obtain a set of difference features between the two fluctuation events; Step S302: Perform a union operation on the event features in all the difference feature sets of the same event type to obtain a complete difference feature set; arbitrarily select an event feature from the complete difference feature set, extract the target fluctuation event where the selected event feature is located, and set the power surplus of the target fluctuation event at the starting time point as S start and the power surplus at the end time is S end , the comprehensive surplus of the target volatility event in the time interval is ΔS=S end -S start ; Step S303: Randomly select two adjacent time points from the occurrence time interval, obtain the power surplus at the two adjacent time points and the characteristic value of the selected time feature group, set the i-th time point and the i+1-th time point to obtain the surplus difference ΔS between the two adjacent time points. (i,i+1) and the numerical difference ΔD (i,i+1) ; The number of event features included in the complete difference feature set in the target fluctuation event is n, and the influence of the selected event features on the power surplus between two adjacent time points is calculated to be Y (i,i+1) =ΔS (i,i+1) / [n×ΔD (i,i+1) ]; take the average of the influence amplitudes of the selected event features at any adjacent time points to obtain the average influence amplitude Y of the selected event features in the same event type ave ; Step S304: arbitrarily select a fluctuation event in a supply and demand fluctuation record, extract each event feature in the selected fluctuation event, and compare it with each event feature in the complete difference feature set to obtain several difference event features of the selected fluctuation event; set the average impact amplitude of the e-th difference event feature as Y e , construct a supply and demand change model to obtain the expected power surplus ΔS of the fluctuation event ex : ； Among them, r is the number of difference event features contained in the selected fluctuation event, and g is a constant; Step S305: Obtain the comprehensive surplus ΔS of the selected fluctuation event in the occurrence time interval. If ΔS < 0, then , similarly if ΔS≥0, then , and obtain the value range of the constant g; input each fluctuation event into the supply and demand change model to calculate the expected power surplus, and extract the comprehensive surplus of each fluctuation event to obtain the value range of the constant coefficient g, perform an intersection operation on the value ranges of each constant coefficient g, and obtain the revised value range of the constant g, and arbitrarily select a constant from the revised value range to adjust the supply and demand change model.
5. The method for intelligent control of time-sharing power supply and demand based on a large model according to claim 4 is characterized by: The step S400 includes the following steps: Step S401: identifying and collecting fluctuation events in real time through several preset channels. When a real-time fluctuation event is collected, feature extraction is performed on the real-time fluctuation event to obtain several difference features. Step S402: Compare the features extracted from the real-time fluctuation event with various event types to determine the event type of the real-time fluctuation event, and obtain the average impact amplitude of each difference feature under the corresponding event type, and call the supply and demand change model to obtain the expected power surplus of the real-time fluctuation event; if the expected power surplus is less than zero, send a risk alert to the power supply side.
6. A time-sharing power supply and demand intelligent control system, configured to execute a large model-based time-sharing power supply and demand intelligent control method according to any one of claims 1 to 5, characterized in that: The control system includes a historical supply and demand analysis module, an event feature association module, a change model analysis module and a real-time supply and demand analysis module; The historical supply and demand analysis module is used to collect fluctuations in power supply and demand every unit period and generate supply and demand fluctuation records for each unit period; several channels are preset to identify fluctuation events occurring in any supply and demand fluctuation record, and each fluctuation interval is extracted and matched with the corresponding fluctuation event; The event feature association module is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; The event characteristics of any fluctuation event are correlated with the power supply and demand situation, and the event types of each fluctuation event are classified; The change model analysis module is used to perform differential analysis on the fluctuation conditions of different fluctuation events within any same event type, obtain several differential features of the same event type, and analyze the impact of each differential feature; construct a supply and demand change model to analyze the expected power supply and demand of any fluctuation event; The real-time supply and demand analysis module is used to identify real-time fluctuation events at the current time point and extract differential features from the real-time fluctuation events; Based on the impact of various difference characteristics, the supply and demand change model is used to predict the expected supply and demand situation, and the current electricity supply and demand situation is adjusted accordingly based on the prediction results.
7. The time-sharing power supply and demand intelligent control system according to claim 6, characterized in that: The historical supply and demand analysis module includes a time-sharing supply and demand collection unit and a fluctuation event identification unit; The time-sharing supply and demand collection unit is used to collect the fluctuation of power supply and demand every unit cycle and generate a supply and demand fluctuation record for each unit cycle; The fluctuation event identification unit is used to preset several channels to identify fluctuation events occurring in any supply and demand fluctuation record, and extract each fluctuation interval and match it with the corresponding fluctuation event.
8. The time-sharing power supply and demand intelligent control system according to claim 6, characterized in that: The event feature association module includes an event feature extraction unit and an event association analysis unit; The event feature extraction unit is used to conduct multi-dimensional analysis on any fluctuation event and extract several event features of any fluctuation event; The event correlation analysis unit is used to perform correlation analysis on the event characteristics of any fluctuation event and the power supply and demand situation, and to classify the event types of each fluctuation event.
9. The intelligent control system for time-sharing power supply and demand according to claim 6, characterized in that: The change model analysis module includes a difference feature impact unit and a change model construction unit; The difference feature impact unit is used to perform difference analysis on the fluctuation conditions of different fluctuation events in any same event type, obtain several difference features of the same event type, and analyze the impact of each difference feature; the change model construction unit is used to construct a supply and demand change model to analyze the expected power supply and demand of any fluctuation event.
10. The time-sharing power supply and demand intelligent control system according to claim 6, characterized in that: The real-time supply and demand analysis module includes a real-time event analysis unit and an expected supply and demand prediction unit; The real-time event analysis unit is used to identify real-time fluctuation events at the current time point and extract the difference features in the real-time fluctuation events; the expected supply and demand prediction unit is used to predict the expected supply and demand situation based on the influence of each difference feature and using the supply and demand change model, and to adjust the current power supply and demand situation accordingly based on the prediction results.