Payload curve prediction method and device considering real-time payload monitoring data, electronic equipment and storage medium

Abstract

The invention discloses a net load curve prediction method and device considering real-time net load monitoring data, electronic equipment and a storage medium, and belongs to the field of net load prediction of a power grid, and the method comprises the steps: obtaining a net load original prediction curve of day-ahead scheduling of the power grid, and collecting the real-time net load monitoring data according to a preset time interval. And determining a target time point corresponding to the monitoring data after the monitoring data is acquired each time, and extracting a reference net load prediction value of the time point. And adjusting the predicted value of the future time period by calculating the error between the real-time monitoring data and the reference predicted value. Fusing the original prediction curve and the corrected prediction curve by adopting a fusion weight corresponding to a moment, so that a generated net load fusion prediction curve meets net load fluctuation constraint and reserve capacity constraint; therefore, by implementing the invention, the problems that the net load prediction method in the prior art is difficult to reflect the real-time load change and is difficult to dynamically adapt to the error distribution characteristics of different time scales can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid net load prediction, and particularly to a net load curve prediction method, device, electronic device and storage medium considering real-time net load monitoring data. Background Art

[0002] Net load curve prediction is a core link in the safe and economic operation of power systems. Especially in the context of the continuous increase in the penetration rate of renewable energy, accurately predicting the changing trend of net load is crucial for optimizing unit commitment, reducing the cost of reserve capacity, and avoiding power grid frequency fluctuations. Traditional prediction methods rely on historical data and meteorological models to generate daily dispatch plans, but the strong volatility of new energy output and load demand leads to significant prediction errors, directly affecting the reasonable allocation of peak shaving and frequency modulation resources, and even causing risks of wind and light abandonment or load shedding.

[0003] Currently, power grid dispatching generally uses the daily prediction curve as the decision-making basis. However, due to the dependence of daily prediction on historical data and statistical models, it is difficult to fully consider the impact of real-time load changes and emergencies, resulting in large uncertainties in the prediction results. In addition, although existing methods can reduce errors through short-term correction, the correction methods are usually relatively simple, such as adjusting based on a fixed ratio or simple linear regression, and it is difficult to dynamically adapt to the error distribution characteristics of different time scales, resulting in limited correction effects. Summary of the Invention

[0004] Embodiments of the present invention provide a net load curve prediction method, device, electronic device and storage medium considering real-time net load monitoring data, which can solve the problems in the existing net load prediction methods that it is difficult to reflect real-time load changes and it is difficult to dynamically adapt to the error distribution characteristics of different time scales.

[0005] An embodiment of the present invention provides a net load curve prediction method considering real-time net load monitoring data, including:

[0006] Obtain the original net load prediction curve of the power grid's daily dispatch;

[0007] Regularly collect the real-time net load monitoring data of the power grid at a preset time interval;

[0008] After obtaining each piece of the real-time net load monitoring data, determine the target time point corresponding to the real-time net load monitoring data;

[0009] Extract the original net load prediction value of the net load original prediction curve at the target time point, denoted as the reference net load prediction value;

[0010] Calculate the difference between the real-time net load monitoring data and the reference net load prediction value to generate a real-time prediction error value;

[0011] According to the real-time prediction error value, dynamically attenuate and adjust the original predicted net load values for future time periods in the original net load prediction curve to generate a corrected net load prediction curve; wherein, during the dynamic attenuation adjustment process, the adjustment amplitude of the real-time prediction error value for the recent time period is greater than that for the long-term time period.

[0012] Superpose and fuse the original net load prediction curve and the corrected net load prediction curve according to the fusion weights corresponding to each moment to generate a fused net load prediction curve, so that each fused net load prediction value in the fused net load prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

[0013] Further, the original predicted net load values for future time periods in the original net load prediction curve are dynamically attenuated and adjusted through the following formula:

[0014] P corr (t + k) = P pred (t + k) + γ k ·e(t)

[0015] Wherein, P corr (t + k) is the corrected predicted net load value for the kth future time period; P pred (t + k) is the original predicted net load value for the kth future time period; γ is the dynamic attenuation factor; k is the serial number of the future time period; e(t) is the real-time prediction error value.

[0016] Further, the step of superposing and fusing the original net load prediction curve and the corrected net load prediction curve according to the fusion weights corresponding to each moment to generate a fused net load prediction curve includes:

[0017] Obtain the first fusion weight of the original net load prediction curve and the second fusion weight of the corrected net load prediction curve;

[0018] According to the first fusion weight, the second fusion weight, the original net load prediction curve and the corrected net load prediction curve, calculate and generate the fused net load prediction value for the future time period through the following fused net load prediction value calculation formula:

[0019] P final (t + k) = α · P pred (t + k) + β · P corr (t + k)

[0020] Wherein, P final (t + k) is the fused net load prediction value for the kth future time period; α is the first fusion weight; β is the second fusion weight;

[0021] Generate a net load integration prediction curve based on the integrated prediction value of the net load for the future time period.

[0022] Furthermore, determine the integration weight at the current moment through the following method:

[0023] Obtain the integrated prediction value of the net load at the previous moment and the integration weight at the previous moment;

[0024] Take the integration weight at the previous moment as the initial integration weight at the current moment; wherein, the integration weight includes the integration weight of the original prediction value of the net load and the integration weight of the corrected prediction value of the net load;

[0025] Calculate and generate the integrated prediction value of the net load for the current time period based on the initial integration weight, the original prediction value of the net load, and the corrected prediction value of the net load in the current time period;

[0026] Judge whether the integrated prediction value of the net load at the current moment meets the net load fluctuation constraint and the reserve capacity constraint according to the integrated prediction value of the net load at the previous moment and the integrated prediction value of the net load in the current time period;

[0027] If so, determine the integration weight of the original prediction value of the net load in the initial integration weight as the first integration weight for the current time period, and determine the integration weight of the corrected prediction value of the net load in the initial integration weight as the second integration weight for the current time period;

[0028] If not, adjust the initial integration weight until the integrated prediction value of the net load recalculated through the adjusted integration weight meets the net load fluctuation constraint and the reserve capacity constraint; determine the integration weight of the original prediction value of the net load in the adjusted integration weight as the first integration weight for the current time period, and determine the integration weight of the corrected prediction value of the net load in the adjusted integration weight as the second integration weight for the current time period.

[0029] Furthermore, the net load fluctuation constraint is specifically:

[0030] |P final (t + k)-P final (t + k - 1)|≤ΔP max

[0031] Wherein, P final (t + k - 1) is the integrated prediction value of the net load for the (k - 1)-th future time period; ΔP max is the maximum net load change rate;

[0032] The reserve capacity constraint is specifically:

[0033]

[0034] Wherein, P reserveis the spare capacity requirement; P load，max is the maximum net load value allowed by the power grid; P load，min is the minimum net load value allowed by the power grid.

[0035] Further, after superimposing and fusing the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment to generate a fused net load prediction curve, it further includes:

[0036] Generate an intraday unit output optimization plan and an intraday energy storage charge and discharge optimization plan according to the fused net load prediction curve;

[0037] Optimize the dispatching of the power grid according to the intraday unit output optimization plan and the intraday energy storage charge and discharge optimization plan.

[0038] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments.

[0039] An embodiment of the present invention provides a net load curve prediction device considering real-time net load monitoring data, including: a net load data acquisition module, a target time point determination module, a real-time prediction error value generation module, a net load dynamic attenuation adjustment module, and a net load superposition and fusion module;

[0040] The net load data acquisition module is used to acquire the original net load prediction curve of the power grid's day-ahead dispatching; regularly collect the real-time net load monitoring data of the power grid at a preset time interval;

[0041] The target time point determination module is used to determine the target time point corresponding to the real-time net load monitoring data after each acquisition of the real-time net load monitoring data;

[0042] The real-time prediction error value generation module is used to extract the original net load prediction value of the original net load prediction curve at the target time point, denoted as the reference net load prediction value; calculate the difference between the real-time net load monitoring data and the reference net load prediction value to generate a real-time prediction error value;

[0043] The net load dynamic attenuation adjustment module is used to dynamically attenuate and adjust the original net load prediction value in the future period of the original net load prediction curve according to the real-time prediction error value to generate a corrected net load prediction curve; wherein, during the dynamic attenuation adjustment process, the adjustment amplitude of the real-time prediction error value for the recent period is greater than that for the far future period;

[0044] The net load superposition and fusion module is used to superpose and fuse the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment, so as to generate a net load fusion prediction curve, so that each net load fusion prediction value in the net load fusion prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

[0045] Further, for the net load curve prediction considering real-time net load monitoring data, the net load dynamic attenuation adjustment module dynamically attenuates and adjusts the original net load prediction value in the future time period of the original net load prediction curve through the following formula:

[0046] P corr (t + k) = P pred (t + k) + γ k ·e(t)

[0047] Where, P corr (t + k) is the corrected net load prediction value in the kth future time period; P pred (t + k) is the original net load prediction value in the kth future time period; γ is the dynamic attenuation factor; k is the serial number of the future time period; e(t) is the real-time prediction error value.

[0048] Based on the above method item embodiments, the present invention correspondingly provides an electronic device item embodiment.

[0049] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it can implement the net load curve prediction method considering real-time net load monitoring data described in any one of the above method item embodiments.

[0050] Based on the above method item embodiments, the present invention correspondingly provides a storage medium item embodiment.

[0051] An embodiment of the present invention provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, it can implement the net load curve prediction method considering real-time net load monitoring data described in any one of the above method item embodiments.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] An embodiment of the present invention provides a net load curve prediction method, device, electronic device, and storage medium considering real-time net load monitoring data. The method obtains the original predicted curve of the net load for the grid's day-ahead dispatch and regularly collects real-time net load monitoring data. After obtaining the monitoring data each time, the corresponding target time point is determined, and the baseline net load prediction value at this time point is extracted. Then, by calculating the error between the real-time monitoring data and the baseline prediction value, the prediction values for future time periods are dynamically adjusted, where the adjustment amplitude for the near-term time period is greater than that for the long-term time period, so as to more accurately correct the prediction curve. Finally, the original prediction curve and the corrected prediction curve are fused using a dynamic fusion weight, and the distribution of the weight is adjusted according to the real-time satisfaction of the net load fluctuation constraint and the reserve capacity constraint, so as to ensure that the final net load prediction curve can accurately reflect the real-time load change and meet the grid operation requirements.

[0054] By collecting real-time grid net load monitoring data and calculating the prediction error value, the present invention effectively overcomes the limitation that traditional day-ahead prediction is difficult to reflect real-time load changes; by using the method of dynamically decaying and adjusting the prediction values for future time periods, the correction process can adapt to the error distribution of different time scales, enhancing the timeliness and accuracy of the prediction; through the above improvements, the present invention effectively improves the accuracy of net load prediction and provides a more reliable load prediction result for the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a schematic flowchart of a net load curve prediction method considering real-time net load monitoring data provided by an embodiment of the present invention.

[0056] Figure 2 is a schematic structural diagram of a net load curve prediction device considering real-time net load monitoring data provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] As Figure 1 shown, to solve the problems in the existing net load prediction method that it is difficult to reflect real-time load changes and difficult to dynamically adapt to the error distribution characteristics of different time scales, an embodiment of the present invention provides a net load curve prediction method considering real-time net load monitoring data, which at least includes the following steps:

[0059] Step S1: Obtain the original predicted curve of the net load for the grid's day-ahead dispatch;

[0060] Specifically, in the process of power grid dispatching, it is first necessary to obtain the preliminary prediction curve of the net load for a future period (such as 24 hours). This step is achieved by integrating historical load data, new energy power generation predictions (such as wind power and photovoltaic), and meteorological forecast information. Specifically, the dispatching system retrieves the electricity load records for a past period from the database, combines the power generation prediction data reported by wind farms and photovoltaic power plants, and refers to the forecast information such as wind speed and sunlight provided by the meteorological department to comprehensively calculate the predicted values of the net load for each future period. For example, if the predicted photovoltaic output is high in a certain period, the net load (total electricity load minus new energy power generation) will decrease accordingly. In specific implementation, it can be generated using statistical modeling, machine learning, or deep learning methods, aiming to provide a forward-looking load change trend for power grid operation. These predicted values are arranged in chronological order to form a complete day-ahead prediction curve, which serves as the basis for subsequent corrections.

[0061] Step S2: Regularly collect the real-time net load monitoring data of the power grid at a preset time interval;

[0062] During the operation of the power grid, the system automatically collects the actual value of the net load at the current moment at a fixed time interval (such as every 15 minutes). This step is achieved by connecting to the measurement system (such as SCADA or PMU) of the power grid dispatching center to obtain real-time data including the total electricity load, actual new energy output (such as wind power and photovoltaic), and network losses, and calculate the current net load value (total load minus new energy output) in real time. This real-time data stream provides a basis for dynamic update for subsequent prediction corrections, ensuring that the prediction model can respond in a timely manner to changes in the actual operation state of the power grid.

[0063] Step S3: After obtaining each piece of the real-time net load monitoring data, determine the target time point corresponding to the real-time net load monitoring data;

[0064] Specifically, when the system obtains the real-time net load monitoring data, it will synchronously record the accurate time mark of the data (for example: 14:15:00 on July 20, 2024), and match the corresponding period in the day-ahead prediction curve based on this time mark. For example, if the prediction curve divides the periods at 15-minute intervals (such as 14:00 - 14:15 is period t1, 14:15 - 14:30 is period t2), and the acquisition time of the real-time data is 14:16:30, the system will assign the data to period t2 (14:15 - 14:30) according to the preset rule (such as "nearest neighbor matching") as the actual reference value for this period. This process is automatically completed through a timestamp alignment algorithm to ensure that the real-time data strictly corresponds to the periods of the prediction curve, avoiding time misalignment problems caused by communication delays or data disorder.

[0065] Step S4: Extract the original predicted value of the net load at the target time point from the original predicted curve of the net load, and denote it as the reference net load predicted value;

[0066] Specifically, when the system determines the target time point corresponding to the real-time monitoring data (for example, the time period from 14:15 to 14:30), it will accurately extract the original predicted value of this time period from the database storing the daily predicted curve. For example, if the original predicted value of this time period is 1000 MW, it will be marked as the reference net load predicted value and used as the reference benchmark for subsequent error calculation. This process is achieved through the fast query function of the time series database, ensuring that even in the face of a large amount of data (such as hundreds of thousands of nodes across the network), data retrieval can be completed within milliseconds. If the predicted value of a certain time period is missing due to communication anomalies, the system will automatically complete it through linear interpolation based on the predicted values of adjacent time periods (such as 995 MW in the previous time period and 1005 MW in the next time period) (for example, the reference value of the target time period is 1000 MW), to avoid interrupting the correction process due to data missing.

[0067] Step S5: Calculate the difference between the real-time net load monitoring data and the reference net load predicted value to generate a real-time prediction error value;

[0068] It should be noted here that when the system obtains the real-time monitoring data (for example, the actual net load in the time period from 14:15 to 14:30 is 850 MW) and extracts the corresponding reference predicted value (1000 MW), it will directly calculate the difference between the two to obtain the real-time prediction error value (such as 850 MW - 1000 MW = -150 MW). This error value not only includes the magnitude of the deviation but also indicates the deviation direction through the positive and negative signs (a negative value indicates that the actual value is lower than the predicted value, and the predicted value of the subsequent time period needs to be corrected downward). The system will perform a rationality check on the error value: if the deviation exceeds the preset threshold (such as ±30%), an abnormal alarm will be triggered (possibly due to measurement equipment failure or communication anomalies), and the error value will be automatically ignored and replaced with the moving window average (such as the average error of the previous 3 time periods) to avoid interference from extreme abnormal values on the correction logic.

[0069] Step S6: Dynamically decay and adjust the original predicted values of the net load in the future time periods in the original predicted curve of the net load according to the real-time prediction error value to generate a corrected predicted curve of the net load; among them, during the dynamic decay adjustment process, the adjustment amplitude of the real-time prediction error value for the recent time periods is greater than that for the far-time periods;

[0070] Specifically, after obtaining the real-time prediction error value (such as -150 MW), the system corrects the predicted values for each future time period in the original prediction curve according to a preset attenuation coefficient. During the correction process, the error has a higher adjustment weight for the recent time periods and decays exponentially over time. For example, for the time periods within the first hour in the future (such as the next 4 fifteen-minute time periods), the error correction weights are 80%, 60%, 40%, and 20% in sequence, while for the time periods after the fourth hour in the future, the weight decays to less than 5%. Through this mechanism, the system responds quickly to the actual deviation in the recent time periods and only makes fine-tuning in the long-term time periods, which not only suppresses short-term fluctuations but also avoids trend deviations caused by excessive correction.

[0071] In a preferred embodiment, the original predicted value of the net load for future time periods in the original net load prediction curve is dynamically attenuated and adjusted by the following formula:

[0072] P corr (t + k) = P pred (t + k) + γ k ·e(t)

[0073] Wherein, P corr (t + k) is the corrected predicted value of the net load for the kth future time period; P pred (t + k) is the original predicted value of the net load for the kth future time period; γ is the dynamic attenuation factor; k is the serial number of the future time period; e(t) is the real-time prediction error value.

[0074] Step S7: Superpose and fuse the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment to generate a fused net load prediction curve, so that each fused net load prediction value in the fused net load prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

[0075] In a preferred embodiment, the step of superposing and fusing the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment to generate a fused net load prediction curve includes:

[0076] Obtain the first fusion weight of the original net load prediction curve and the second fusion weight of the corrected net load prediction curve;

[0077] According to the first fusion weight, the second fusion weight, the original net load prediction curve, and the corrected net load prediction curve, calculate and generate the fused net load prediction value for the future time period through the following fused net load prediction value calculation formula:

[0078] P final (t + k) = α·P pred (t + k) + β·Pcorr (t + k)

[0079] where P final (t + k) is the predicted value of the integrated net load for the k-th future period; α is the first integration weight; β is the second integration weight;

[0080] Generate an integrated net load prediction curve based on the predicted value of the integrated net load for future periods.

[0081] It should be noted here that after the system obtains the original net load prediction curve and the corrected prediction curve, it first initializes two sets of integration weights for each future period (the integration weights at the previous moment can be used as the initial integration weights at the current moment for initialization): the sum of the first integration weight of the original prediction curve and the second integration weight of the corrected prediction curve is constant at 1 (for example, in the initial state, α = 0.3 and β = 0.7). The dynamic adjustment of the weights is based on the constraint verification result of the corrected predicted value: if the corrected value of a certain period passes the net load fluctuation constraint (the change rate between adjacent periods ≤ the maximum allowable value) and the spare capacity constraint (the predicted value ± the spare capacity is within the allowable range), then maintain or increase the second integration weight (such as increasing from 0.7 to 0.8) to preferentially adopt the corrected result; if the corrected value exceeds the limit, gradually reduce the second integration weight (such as decreasing from 0.7 to 0.4), and at the same time increase the first integration weight until the integrated predicted value meets all constraints. For example, when the corrected predicted value of a certain period is 1200 MW and the upper limit of the spare capacity constraint is 1150 MW, after the system detects the overlimit, it gradually adjusts the second integration weight from 0.7 to 0.4 (the corresponding first integration weight is 0.6), so that the integrated predicted value drops to 1200×0.4 + 1100×0.6 = 1140 MW (assuming the original predicted value is 1100 MW), and finally meets the upper limit requirement of 1150 MW. After the integration calculation is completed, the system summarizes the integrated predicted values of all periods in chronological order, generates a complete integrated net load prediction curve that meets the safety boundary, and pushes it to the dispatching system for execution.

[0082] In a preferred embodiment, the integration weights at the current moment are determined in the following manner;

[0083] Obtain the predicted value of the integrated net load and the integration weights at the previous moment;

[0084] Use the integration weights at the previous moment as the initial integration weights at the current moment; where the integration weights include the integration weights of the original net load prediction value and the corrected net load prediction value;

[0085] Calculate and generate the predicted value of the integrated net load for the current period according to the initial integration weights, the original net load prediction value, and the corrected net load prediction value for the current period;

[0086] According to the net load fusion prediction value at the previous moment and the net load fusion prediction value at the current time period, it is determined whether the net load fusion prediction value at the current moment meets the load net load fluctuation constraint and the reserve capacity constraint;

[0087] If yes, the fusion weight of the original prediction value of the net load in the initial fusion weight is determined as the first fusion weight of the current period, and the fusion weight of the modified prediction value of the net load in the initial fusion weight is determined as the second fusion weight of the current period;

[0088] If not, adjust the initial fusion weight until the net load fusion prediction value recalculated by the adjusted fusion weight meets the load net load fluctuation constraint and the reserve capacity constraint; determine the fusion weight of the net load original prediction value of the adjusted fusion weight as the first fusion weight for the current period, and determine the fusion weight of the net load revised prediction value of the adjusted fusion weight as the second fusion weight for the current period.

[0089] In actual operation, the system performs double safety checks on the revised forecast value of each future period: first, check whether the change rate of adjacent periods exceeds the maximum net load fluctuation value allowed by the power grid, and secondly verify whether the revised forecast value exceeds the upper and lower limits that the current reserve capacity can cover. If the revised value fully meets the constraints, a higher weight (for example, 70%) is given to the revised forecast curve to maximize the accuracy improvement brought by real-time correction; if the revised value causes the volatility to exceed the limit or the reserve capacity is insufficient, the weight of the revised curve is gradually reduced (for example, from 70% to 30%), and the weight ratio of the original forecast curve is increased. Through repeated iterative adjustments, the fused forecast value meets both the volatility and reserve capacity requirements. For example, when the revised forecast value of a certain period causes the load change rate of the adjacent period to reach 120MW (exceeding the maximum allowable value of 100MW), the system will automatically reduce the weight of the revised curve and increase the proportion of the original curve until the fused change rate falls back to within 100MW, and at the same time, the reserve capacity allocation strategy is adjusted in conjunction to ensure that the final output fused forecast value is both close to the actual operation trend and strictly complies with the safety boundary of the power grid.

[0090] In a preferred embodiment, the load net load fluctuation constraint is specifically:

[0091] |P final (t+k)-P final (t+k-1)|≤ΔP max

[0092] Among them, P final (t+k-1) is the net load fusion forecast value of the k-1th period in the future; ΔP max is the maximum net load change rate;

[0093] The spare capacity constraint is specifically as follows:

[0094]

[0095] Among them, P reserve is the spare capacity demand; P load，max is the maximum allowable net load value of the power grid; P load，min is the minimum allowable net load value of the power grid.

[0096] In a preferred embodiment, after superimposing and fusing the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment to generate a net load fusion prediction curve, it further includes:

[0097] Generating an intraday unit output optimization plan and an intraday energy storage charge and discharge optimization plan according to the net load fusion prediction curve;

[0098] Optimally dispatching the power grid according to the intraday unit output optimization plan and the intraday energy storage charge and discharge optimization plan.

[0099] Specifically, the system inputs the net load fusion prediction curve into the intraday dispatching optimization module to generate a unit output plan and an energy storage charge and discharge strategy in a rolling manner with a time granularity of 15 minutes. For traditional units such as thermal power and hydropower, an economic dispatch model (ED) is used to calculate the optimal output of each unit: based on the coal consumption characteristic curve, start-stop cost, and ramp rate limit, on the premise of meeting the net load demand, low coal consumption units are preferentially called and the operation duration of high-cost units is controlled to minimize the total fuel cost. For example, if the predicted net load value in a certain area in the next 2 hours is 800 - 950 MW, the system will combine the unit combination (such as 2 300 MW coal-fired units + 1 200 MW gas-fired unit) to generate the output instructions for each period (such as 750 MW for coal-fired units + 50 MW for gas-fired units), and ensure that the output change rate between adjacent periods ≤ the maximum ramp capacity of the unit (such as 5 MW / min for coal-fired units).

[0100] For the energy storage system, based on the charge and discharge efficiency (such as 90%) and the state of charge (SOC) limit (20% - 90%), the charge and discharge power is dynamically optimized with a cycle of 5 minutes: start the charging mode during the low net load period (such as when the predicted value is lower than the minimum output of the base load unit) to absorb the excess new energy power; switch to the discharging mode during the high net load period (such as when the predicted value is close to the grid carrying limit) to suppress the power gap. For example, when the prediction shows that the net load will suddenly increase by 200 MW from 10:00 to 11:00, the energy storage system starts discharging in advance at 09:55 and releases 50 MW of capacity at a rate of 10 MW / min to avoid calling high-cost gas peaking units.

[0101] After the optimization calculation is completed, the system issues the unit output plan and the energy storage strategy to each power station for execution through the scheduling instruction interface (such as the IEC 61850 protocol), and monitors the deviation between the actual output and the plan in real time. If the deviation exceeds the threshold (such as ±5%), the rolling correction mechanism is triggered: the prediction curve for the subsequent period is updated based on the latest net load monitoring data, the optimization plan is recalculated, and the execution instruction is dynamically adjusted to form a closed-loop control link of "prediction - optimization - execution - feedback".

[0102] In one embodiment, when generating the intraday unit output optimization plan and the intraday energy storage charge and discharge optimization plan according to the net load fusion prediction curve, the system power and energy balance constraint, the conventional unit operation constraint, the power transmission constraint, and the energy storage operation constraint should be satisfied to ensure that the intraday unit output optimization plan and the intraday energy storage charge and discharge optimization plan conform to the actual operation of the power grid.

[0103] The system power and energy balance constraint is specifically:

[0104] P final (t) = P G (t) + P SEC (t) + P ESS，DIS (t) - P ESS，CH (t)

[0105] Among them, P final (t) is the net load fusion prediction value; P G (t) is the total output of conventional units; P SEC (t) is the power exchanged between the current regional power grid and the external region, positive for incoming and negative for outgoing; P ESS,DIS (t) is the discharge power of the energy storage system; P ESS，CH (t) is the charging power of the energy storage system.

[0106] The conventional unit output constraint is specifically:

[0107] P G,min ≤P G (t) - P reserve ≤P G,max

[0108] Among them, P G，min is the lower limit value of the active power output of the conventional unit; P G，max is the upper limit value of the active power output of the conventional unit.

[0109] The power transmission constraint is specifically:

[0110] P SEC,min ≤P SEC (t) ≤ P SEC,max

[0111] Among them, PSEC,min is the lower limit value of the power transmitted out; P SEC,max is the upper limit value of the power transmitted out; P SEC (t) is the power exchanged between the current regional power grid and the external region.

[0112] The energy storage operation constraints include charge-discharge power constraints and state of charge constraints;

[0113] The charge-discharge power constraints are specifically as follows:

[0114]

[0115] wherein, P ch，max is the maximum charging power of the energy storage system; P dis，max is the maximum discharging power of the energy storage system.

[0116] The state of charge constraints are specifically as follows:

[0117] S soc，min ≤S soc (t)≤S soc，max

[0118] wherein, S soc (t) is the charge of the energy storage; S soc，min is the minimum charge limit value of the energy storage; S soc，max is the maximum charge limit value of the energy storage.

[0119] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments.

[0120] As Figure 2 shown, an embodiment of the present invention provides a net load curve prediction device considering real-time net load monitoring data, including: a net load data acquisition module, a target time point determination module, a real-time prediction error value generation module, a net load dynamic attenuation adjustment module, and a net load superposition and fusion module;

[0121] The net load data acquisition module is used to acquire the original predicted curve of the net load for the grid's day-ahead dispatch; regularly collect the real-time net load monitoring data of the grid at a preset time interval;

[0122] The target time point determination module is used to determine the target time point corresponding to the real-time net load monitoring data after each acquisition of the real-time net load monitoring data;

[0123] The real-time prediction error value generation module is used to extract the original predicted value of the net load on the target time point of the original predicted curve of the net load, denoted as the reference net load predicted value; calculate the difference between the real-time net load monitoring data and the reference net load predicted value to generate a real-time prediction error value;

[0124] The net load dynamic attenuation adjustment module is configured to dynamically attenuate and adjust the original net load prediction values in the future time periods of the original net load prediction curve according to the real-time prediction error value, and generate a corrected net load prediction curve; wherein, during the dynamic attenuation adjustment process, the adjustment amplitude of the real-time prediction error value for the recent time period is greater than that for the far future time period.

[0125] The net load superposition and fusion module is configured to superpose and fuse the original net load prediction curve and the corrected net load prediction curve according to the fusion weight corresponding to each moment, and generate a fused net load prediction curve, so that each fused net load prediction value in the fused net load prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

[0126] In a preferred embodiment, the net load dynamic attenuation adjustment module in the net load curve prediction device considering real-time net load monitoring data dynamically attenuates and adjusts the original net load prediction values in the future time periods of the original net load prediction curve through the following formula:

[0127] P corr (t + k) = P pred (t + k) + γ k ·e(t)

[0128] Wherein, P corr (t + k) is the corrected net load prediction value for the k-th future time period; P pred (t + k) is the original net load prediction value for the k-th future time period; γ is the dynamic attenuation factor; k is the serial number of the future time period; e(t) is the real-time prediction error value.

[0129] It should be noted that the embodiments of the device described above correspond to the above embodiments of the present invention, and can implement the net load curve prediction method considering real-time net load monitoring data described in any one of the above of the present invention. In addition, the embodiments of the above device are merely illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement without creative labor.

[0130] Based on the above method embodiment of the present invention, an embodiment of an electronic device is correspondingly provided.

[0131] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method for predicting a net load curve considering real-time net load monitoring data according to any one of the present invention is implemented, or when the processor executes the computer program, the functions of each module in the above device embodiments are implemented.

[0132] Exemplarily, the computer program may be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0133] The terminal device may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0134] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device and connects various parts of the entire terminal device through various interfaces and lines.

[0135] The memory can be used to store the computer program and / or modules. By running or executing the computer program and / or modules stored in the memory, and invoking the data stored in the memory, the processor realizes various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0136] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments;

[0137] Another embodiment of the present invention provides a storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the storage medium is located to execute any of the above net load curve prediction methods considering real-time net load monitoring data of the present invention.

[0138] Among them, the above storage medium is a computer-readable storage medium. The computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0139] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of this application. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0140] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can still be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A net load curve prediction method considering real-time net load monitoring data, characterized in that, Including: Obtaining the original prediction curve of the net load for the grid's day-ahead dispatch; Regularly collecting the real-time net load monitoring data of the grid at a preset time interval; After obtaining each piece of the real-time net load monitoring data, determining the target time point corresponding to the real-time net load monitoring data; Extracting the original net load prediction value of the net load original prediction curve at the target time point, denoted as the reference net load prediction value; Calculating the difference between the real-time net load monitoring data and the reference net load prediction value to generate a real-time prediction error value; According to the real-time prediction error value, dynamically attenuating and adjusting the original net load prediction values in the future time period of the net load original prediction curve to generate a net load corrected prediction curve; wherein, during the dynamic attenuation adjustment process, the adjustment amplitude of the real-time prediction error value for the recent time period is greater than that for the far future time period; Superimposing and fusing the net load original prediction curve and the net load corrected prediction curve according to the fusion weights corresponding to each moment to generate a net load fused prediction curve, so that each net load fused prediction value in the net load fused prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

2. The net load curve prediction method considering real-time net load monitoring data according to claim 1, characterized in that Dynamically attenuating and adjusting the original net load prediction values in the future time period of the net load original prediction curve through the following formula: P corr (t + k)=P pred (t + k)+γ k ·e(t) Among them, P corr (t + k) is the corrected predicted value of the net load for the k-th future period; P pred (t + k) is the original predicted value of the net load for the k-th future period; γ is the dynamic decay factor; k is the serial number of the future period; e(t) is the real-time prediction error value.

3. The net load curve prediction method considering real-time net load monitoring data according to claim 2, wherein The step of superimposing and fusing the net load original prediction curve and the net load corrected prediction curve according to the fusion weights corresponding to each moment to generate a net load fused prediction curve includes: Obtaining the first fusion weight of the net load original prediction curve and the second fusion weight of the net load corrected prediction curve; According to the first fusion weight, the second fusion weight, the net load original prediction curve and the net load corrected prediction curve, calculating and generating the net load fused prediction value for the future time period through the following net load fused prediction value calculation formula: P final (t + k) = α·P pred (t + k) + β·P corr (t + k) Among them, P final (t + k) is the predicted value of the integrated net load for the k-th future period; α is the first integration weight; β is the second integration weight; Generating a net load fused prediction curve according to the net load fused prediction value for the future time period.

4. The net load curve prediction method considering real-time net load monitoring data according to claim 3, characterized in that, Determining the fusion weight at the current moment in the following manner; Obtaining the net load fused prediction value at the previous moment and the fusion weight at the previous moment; Taking the fusion weight at the previous moment as the initial fusion weight at the current moment; wherein, the fusion weight includes the fusion weight of the original net load prediction value and the fusion weight of the corrected net load prediction value; Calculating and generating the net load fused prediction value for the current time period according to the initial fusion weight, the original net load prediction value and the corrected net load prediction value in the current time period; Judging whether the net load fused prediction value at the current moment meets the net load fluctuation constraint and the reserve capacity constraint according to the net load fused prediction value at the previous moment and the net load fused prediction value for the current time period; If so, determining the fusion weight of the original net load prediction value in the initial fusion weight as the first fusion weight for the current time period, and determining the fusion weight of the corrected net load prediction value in the initial fusion weight as the second fusion weight for the current time period; Otherwise, adjust the initial fusion weights until the net load fusion prediction value recalculated with the adjusted fusion weights satisfies the load net load fluctuation constraint and the reserve capacity constraint; determine the fusion weight of the net load original prediction value with the adjusted fusion weights as the first fusion weight for the current time period, and determine the fusion weight of the net load corrected prediction value with the adjusted fusion weights as the second fusion weight for the current time period.

5. The net load curve prediction method considering real-time net load monitoring data according to claim 4, wherein the load net load fluctuation constraint is specifically: |P final (t + k)-P final (t + k - 1)| ≤ ΔP max Among them, P final (t + k - 1) is the predicted value of the integrated net load for the (k - 1)-th future period; ΔP max is the maximum net load change rate; the reserve capacity constraint is specifically: Among them, P reserve is the spare capacity requirement; P load，max is the maximum net load value allowed by the power grid; P load，min is the minimum net load value allowed by the power grid.

6. The net load curve prediction method considering real-time net load monitoring data according to claim 5, characterized in that, After superimposing and fusing the net load original prediction curve and the net load corrected prediction curve according to the fusion weights corresponding to each moment to generate a net load fusion prediction curve, it further includes: Generating an intraday unit output optimization plan and an intraday energy storage charge and discharge optimization plan according to the net load fusion prediction curve; Optimally dispatching the power grid according to the intraday unit output optimization plan and the intraday energy storage charge and discharge optimization plan.

7. A net load curve prediction device considering real-time net load monitoring data, characterized in that, It includes: A net load data acquisition module, a target time point determination module, a real-time prediction error value generation module, a net load dynamic attenuation adjustment module, and a net load superimposing and fusing module; The net load data acquisition module is used to acquire the net load original prediction curve of the power grid's day-ahead dispatch; regularly collect the real-time net load monitoring data of the power grid at a preset time interval; The target time point determination module is used to determine the target time point corresponding to the real-time net load monitoring data after each acquisition of the real-time net load monitoring data; The real-time prediction error value generation module is used to extract the net load original prediction value of the net load original prediction curve at the target time point, denoted as the reference net load prediction value; calculate the difference between the real-time net load monitoring data and the reference net load prediction value to generate a real-time prediction error value; The net load dynamic attenuation adjustment module is used to dynamically attenuate and adjust the net load original prediction values in the future time period of the net load original prediction curve according to the real-time prediction error value to generate a net load corrected prediction curve; wherein, during the dynamic attenuation adjustment process, the adjustment amplitude of the real-time prediction error value for the recent time period is greater than that for the far future time period; The net load superimposing and fusing module is used to superimpose and fuse the net load original prediction curve and the net load corrected prediction curve according to the fusion weights corresponding to each moment to generate a net load fusion prediction curve, so that each net load fusion prediction value in the net load fusion prediction curve satisfies the net load fluctuation constraint and the reserve capacity constraint.

8. The net load curve prediction device considering real-time net load monitoring data according to claim 7, characterized in that The net load dynamic attenuation adjustment module dynamically attenuates and adjusts the net load original prediction values in the future time period of the net load original prediction curve through the following formula: P corr (t + k) = P pred (t + k) + γ k ·e(t) Among them, P corr (t + k) is the corrected predicted value of the net load for the k-th future time period; P pred (t + k) is the original predicted value of the net load for the k-th future time period; γ is the dynamic decay factor; k is the serial number of the future time period; e(t) is the real-time prediction error value.

9. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the net load curve prediction method considering real-time net load monitoring data according to any one of claims 1 to 6.

10. A storage medium, characterized in that, The storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the storage medium is located to execute the net load curve prediction method considering real-time net load monitoring data according to any one of claims 1 to 6.

Images