A combined dispatching method and device for runoff prediction and a hydropower-hydrogen system

Abstract

The present invention relates to the technical field of hydropower generation, and discloses a combined scheduling method and device for runoff prediction and a hydropower-hydrogen system. The method includes: obtaining historical drought data of a target hydropower station, and performing enhancement processing on the historical drought data to obtain a runoff enhancement data set; training a preset prediction model using the runoff enhancement data set to obtain a trained prediction model; using the prediction model to obtain runoff prediction data of the target hydropower station according to a preset sampling time, where the runoff prediction data includes predicted runoff volumes at each preset sampling time, and performing energy scheduling on the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff volume. By enhancing the historical runoff data, the present invention increases the number of training samples of the prediction model, improves the performance of the prediction model, performs energy scheduling based on the annual revenue demand of the hydropower-hydrogen system, ensures the maximum annual revenue of the target hydropower station, and improves the resilience of the hydropower system under extreme drought conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of hydropower generation, and particularly relates to a combined scheduling method and device for runoff prediction and a hydropower-hydrogen system. Background Art

[0002] With the change of global climate conditions, drought has become one of the most destructive natural disasters affecting the reliability of hydropower supply. In recent years, the changes in hydro-climatic and socio-economic conditions have exacerbated the challenge of water resource shortage. Hydropower is the key to power supply, highly dependent on hydrological conditions. Severe drought disasters directly affect the power generation flow of hydropower stations, and insufficient long-term power generation flow often leads to a reduction in renewable energy power generation. The expansion of the supply-demand gap caused by insufficient power supply will trigger a series of chain reactions in the social and economic fields. Therefore, effectively enhancing the resilience of the power grid to extreme drought challenges has received wide attention. Active pre-disaster defense, emergency recovery during disasters, and post-disaster recovery are the key links in risk prevention and resilience improvement throughout the operation process.

[0003] First of all, runoff prediction under extreme drought conditions is a scientific and engineering challenge. The key to the problem lies in accurately establishing a prediction model for small-sample long-period runoff time series. Traditional physical models require precise assignment of a large number of hydrological parameters, rich expert knowledge, and high-quality data, which are difficult to popularize. Most data-driven time models are difficult to adapt to the characteristics of small samples and high randomness. Secondly, extreme drought disasters often last for a long seasonal cycle. Even if accurate prediction and early warning can be achieved, how to improve the power supply capacity of the system under the influence of drought disasters through resource allocation is also a major challenge. Summary of the Invention

[0004] In view of this, the present invention provides a combined scheduling method and device for runoff prediction and a hydropower-hydrogen system to solve the problem of improving the resilience of the hydropower system under extreme drought conditions through runoff prediction and resource allocation.

[0005] In a first aspect, the present invention provides a combined scheduling method for runoff prediction and a hydropower-hydrogen system, the method comprising:

[0006] Obtaining historical drought data of a target hydropower station, and performing data augmentation processing on the historical drought data to obtain a runoff augmentation dataset of the target hydropower station;

[0007] Training a preset prediction model using the runoff augmentation dataset to obtain a trained prediction model;

[0008] Obtaining runoff prediction data of the target hydropower station according to a preset sampling time using the prediction model, the runoff prediction data including predicted runoff volumes at each preset sampling time;

[0009] Based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff, the energy scheduling of the hydropower-hydrogen system is carried out.

[0010] The joint scheduling method of runoff prediction and hydropower-hydrogen system provided by the present invention enhances the historical runoff data, increases the number of training samples of the prediction model, improves the performance of the trained prediction model, and conducts energy scheduling based on the annual revenue demand of the hydropower-hydrogen system to ensure the maximum annual revenue of the target hydropower station and improve the resilience of the hydropower system under extreme drought conditions.

[0011] In an optional implementation manner, historical drought data of the target hydropower station is obtained, and the historical drought data is processed by data enhancement to obtain a runoff enhancement data set of the target hydropower station, including:

[0012] The hydrological frequency of the historical runoff data of the target hydropower station is obtained according to a preset period, and the drought years are determined according to the classification of the hydrological frequency of the historical runoff data;

[0013] The historical runoff data of the drought years is used as drought sample data, and the drought sample data is processed by data enhancement based on a preset generative adversarial network to obtain enhanced sample data. The enhanced sample data and the drought sample data form a runoff enhancement data set of the target hydropower station.

[0014] The joint scheduling method of runoff prediction and hydropower-hydrogen system provided by the present invention processes the historical runoff data of drought years by data enhancement, and forms a runoff enhancement data set with the enhanced sample data and the real sample data, analyzes the characteristics of the runoff data in drought years specifically, avoids the influence of few runoff samples and long cycle on the data prediction accuracy in extreme drought situations, and improves the prediction accuracy of the runoff data in drought years.

[0015] In an optional implementation manner, a preset prediction model is trained by using the runoff enhancement data set to obtain a trained prediction model, including:

[0016] The runoff enhancement data set is divided into a training set, a test set and a validation set according to a preset ratio;

[0017] Multiple preset prediction models are trained by using the training set and the test set to obtain multiple trained prediction models;

[0018] The performance of the multiple trained prediction models is evaluated by using the validation set, and the prediction model for runoff prediction is determined according to the evaluation results.

[0019] The joint scheduling method of runoff prediction and hydropower-hydrogen system provided by the present invention trains multiple preset prediction models by using the runoff enhancement data set, and selects the model with the best performance as the prediction model for runoff prediction by evaluating the performance of each model to improve the accuracy of runoff prediction.

[0020] In an alternative embodiment, runoff prediction data of a target hydropower station is obtained according to a preset sampling time by using a prediction model, including:

[0021] Obtain the input features of the prediction model, where the input features include: the preset historical time runoff of the target hydropower station, the runoff on the same day of the previous year, statistical features, and calendar features;

[0022] Input the input features into the prediction model to obtain the same-day runoff prediction data of the target hydropower station;

[0023] Adopt a rolling window method to obtain the runoff prediction data of the target hydropower station according to a preset sampling time interval.

[0024] The joint scheduling method of runoff prediction and hydropower-hydrogen system provided by the present invention extracts input features from three aspects: historical information, calendar features, and statistical features, obtains the same-day runoff prediction data according to the input features, and updates the runoff prediction data by using a rolling window method to ensure the effectiveness of the runoff prediction data.

[0025] In an alternative embodiment, energy scheduling of the hydropower-hydrogen system is performed based on the annual revenue demand and predicted runoff of the hydropower-hydrogen system, including:

[0026] Construct the objective function of the annual revenue of the hydropower-hydrogen system:

[0027]

[0028] Among them, Maximize represents maximizing the annual revenue, represents the load at time is the load shedding amount of the th hydropower station at time In addition, the parameter represents the electricity price at time represents the unit load shedding loss, is the discrete scheduling time period, is the unit scheduling time window, is the th hydropower station in the basin;

[0029] When the objective function is maximized, the output ratio of the hydrogen fuel cell to hydropower is obtained and used as the optimal scheduling result for energy scheduling. The load shedding amount of the th hydropower station at time is determined by the output ratio of the hydrogen fuel cell to hydropower at time

[0030] ​​The joint scheduling method for runoff prediction and hydropower-hydrogen system provided by the present invention determines the scheduling strategy when the annual revenue is maximized by setting the objective function of the hydropower-hydrogen system scheduling, calculating the annual revenue according to the scheduling result, aiming at maximizing the annual revenue, and ensuring the minimum load shedding of the hydropower-hydrogen system and the maximum annual revenue.

[0031] In an alternative embodiment, the method further includes:

[0032] According to the energy scheduling result, analyze the first sensitivity of different hydropower ratios to the load shedding loss, the second sensitivity of different hydrogen equipment capacities to the load shedding loss under different hydropower ratios, and the third sensitivity of the runoff under different hydropower ratios to the load shedding loss;

[0033] Obtain the runoff prediction data of the target hydropower station, and adjust the hydropower ratio and hydrogen equipment capacity of the target hydropower station according to the first sensitivity, the second sensitivity, and the third sensitivity.

[0034] The joint scheduling method for runoff prediction and hydropower-hydrogen system provided by the present invention determines the influence degree of different hydropower ratios, hydrogen equipment capacities, and runoff on the load shedding loss through sensitivity analysis from three aspects, which is beneficial to adjusting the energy scheduling according to the sensitivity analysis result, improving the accuracy of energy scheduling, and reducing the load shedding loss.

[0035] In the second aspect, the present invention provides a joint scheduling device for runoff prediction and hydropower-hydrogen system. The device includes:

[0036] A data enhancement module, configured to obtain the historical drought data of the target hydropower station and perform data enhancement processing on the historical drought data to obtain a runoff enhancement data set of the target hydropower station;

[0037] A model training module, configured to train a preset prediction model by using the runoff enhancement data set to obtain a trained prediction model;

[0038] A runoff prediction module, configured to obtain the runoff prediction data of the target hydropower station according to a preset sampling time by using the prediction model. The runoff prediction data includes the predicted runoff at each preset sampling time;

[0039] An energy scheduling module, configured to perform energy scheduling on the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff.

[0040] In the third aspect, the present invention provides a computer device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the method according to the first aspect or any corresponding embodiment thereof.

[0041] Fourthly, the present invention provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to cause a computer to execute the method according to the first aspect or any corresponding embodiment thereof.

[0042] Fifthly, the present invention provides a computer program product, including computer instructions, and the computer instructions are used to cause a computer to execute the method according to the first aspect or any corresponding embodiment thereof. Description of the Drawings

[0043] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 is a schematic flowchart of a method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0045] Figure 2 is a bar chart of the hydrogen equipment capacity required to eliminate load shedding under different hydropower ratios in the method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0046] Figure 3 is a schematic diagram of the reduction of load shedding losses under different hydrogen energy capacities and hydropower ratios in the method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0047] Figure 4 is a schematic diagram of the monthly load losses caused by drought under different hydropower ratios in the method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0048] Figure 5 is a schematic diagram of the monthly load shedding losses during drought under different runoff levels and hydropower ratios in the method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0049] Figure 6 is a schematic flowchart of another method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0050] Figure 7 is a schematic diagram of the process of joint scheduling in the method for runoff prediction and joint scheduling of a hydropower-hydrogen system according to an embodiment of the present invention;

[0051] Figure 8 、 Figure 9 、Figure 10 Schematic diagrams of the daily runoff data of three hydropower stations in the same river basin in the past ten years (from 2010 to 2022) in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0052] Figure 11 、 Figure 12 、 Figure 13 Schematic diagrams of the Q-Q plots and probability distribution fitting results of the generated runoff data and the real runoff data of three hydropower stations in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0053] Figure 14 、 Figure 15 、 Figure 16 Schematic diagrams of the prediction results of the XGBoost model for three power generation stations in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0054] Figure 17 Schematic diagram of the typical load curves of three cascaded hydropower stations considering seasonal variations in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0055] Figure 18 Schematic diagram of the optimized dispatching result in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0056] Figure 19 Schematic diagram of the optimal dispatching scheme for hydrogen fuel cell discharge when three cascaded hydropower stations are under load reduction in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0057] Figure 20 Schematic diagram of the recovery ability of the hydropower system in an extremely dry year in the runoff prediction and combined dispatching method of the hydropower-hydrogen system according to the embodiments of the present invention;

[0058] Figure 21 Block diagram of the structure of the runoff prediction and combined dispatching device of the hydropower-hydrogen system according to the embodiments of the present invention;

[0059] Figure 22 Schematic diagram of the hardware structure of the computer device according to the embodiments of the present invention. Detailed implementation manners

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. 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.

[0061] Given the urgency of climate change and the growing energy demand, it is crucial to develop clean, efficient, and sustainable energy supply systems. Dynamic water resource redistribution is the basis for balancing the multi-purpose use of hydropower stations, and coordination with other renewable energy sources can enhance the resilience of power systems that mainly rely on hydropower. However, such coordination requires a comprehensive consideration of the resource endowments of renewable energy sources. Although most studies focus on addressing short-term flood disasters and long-term sustainability, scholars have also explored the use of various forms of energy and seasonal energy storage to promote cross-scale conversion and regulation of water, hydropower resources, and energy to cope with the frequent extreme drought disasters occurring globally in recent years.

[0062] Hydrogen energy is widely recognized as a clean and low-carbon energy source that can play a key role in decarbonizing challenging industries and achieving carbon emission reduction targets. Compared with pumped storage, it provides a more adaptable form of energy storage, enriches the methods of seasonal renewable energy storage, and contributes to the stability of energy supply.

[0063] Runoff prediction has been widely studied in various fields such as meteorology, hydrology, and environmental science, thus developing various runoff prediction models, which can be roughly divided into two categories: physical models and data-driven models. Physical models are constructed based on the operating mechanism of runoff, clearly representing hydrological state variables and fluxes. Although physical models are widely used and have good interpretability, physical models require specific catchment parameters, expert knowledge, and high-quality data. Data-driven models can learn complex behaviors from runoff data. Existing models are mainly designed for conventional scenarios and are not applicable to unconventional situations such as extreme drought. The runoff patterns during extreme drought are intricate and difficult to extract from historical data. To solve this problem, researchers have attempted to incorporate meteorological drought into hydrological drought prediction and have partially demonstrated that combining antecedent precipitation can enhance runoff prediction during drought. However, given that runoff involves multiple regions, it is very difficult to establish the relationship between the runoff in these regions and meteorological factors.

[0064] To improve the economy of the system, a large number of studies have explored the optimal scheduling of hydropower-hydrogen combined systems. These studies mainly use hydrogen energy for peak shaving and arbitrage, participate in the optimal scheduling of hydropower through water electrolysis hydrogen production technology and hydrogen energy storage conversion technology, and consider the volatility of electricity prices at the same time. To cope with the impact of runoff randomness in the environment, common optimal scheduling methods include stochastic optimization methods considering short-term or long-term runoff distributions, online and multi-stage adaptive robust optimization methods, and model predictive control. Most of these studies are based on large samples of historical runoff data or strong assumptions about the runoff data distribution, and mainly focus on the unit commitment strategy in hydropower scheduling and how to solve the nonlinear non-convex programming problem of hydropower generation more efficiently and accurately.

[0065] Runoff prediction involves analyzing complex time series data with high randomness. However, few studies have combined runoff time series prediction methods with hydropower scheduling decisions. Related research mainly focuses on applying grey system theory to hydropower generation, but this method cannot accurately predict short-term events, especially extreme events.

[0066] To solve the above problems, an embodiment of the present invention provides a combined scheduling method for runoff prediction and hydropower-hydrogen systems, which achieves the effect of enhancing system resilience through pre-event prediction and early warning of extreme drought disasters and optimal scheduling during the event.

[0067] According to an embodiment of the present invention, an embodiment of a combined scheduling method for runoff prediction and hydropower-hydrogen systems is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0068] In this embodiment, a combined scheduling method for runoff prediction and hydropower-hydrogen systems is provided, which can be used in the above computer system. Figure 1 is a flowchart of a combined scheduling method for runoff prediction and hydropower-hydrogen systems according to an embodiment of the present invention, as Figure 1 shown, the process includes the following steps:

[0069] Step S101, obtain historical drought data of the target hydropower station, and perform data augmentation processing on the historical drought data to obtain a runoff augmentation dataset of the target hydropower station.

[0070] Specifically, due to the changes in meteorological conditions, there are various hydrological scenarios, such as high water and low water. To distinguish extremely dry years, the Pearson type III distribution, a hydrological classification method, is used to calculate the hydrological frequency of historical runoff data (the probability that a hydrological characteristic value appears greater than or less than a specified value), and the high-water years, dry years, and extremely dry years are determined based on the hydrological frequency in the historical runoff data. In this embodiment, the runoff data of dry years and extremely dry years are studied, so the historical drought data includes the historical runoff data of dry years and extremely dry years. The runoff data of extremely dry years is usually very scarce and difficult to support accurate runoff prediction for future extremely dry years. Therefore, it is necessary to enhance the small-sample runoff data. The runoff data of extremely dry years is used as the real runoff data, and data enhancement is performed using the real runoff data to obtain enhanced data. The real runoff data and the enhanced data together form the runoff enhancement dataset of the target hydropower station.

[0071] Step S102: Train a preset prediction model using the runoff enhancement dataset to obtain a trained prediction model. Specifically, perform an outlier test on the runoff enhancement dataset and process the abnormal data to obtain an effective runoff enhancement dataset. The effective runoff enhancement dataset is divided into two parts: a training set and a test set. Generally, the training set data accounts for 80% of the total data, and the test set data accounts for 20% of the total data. This is only an example and is not limited thereto. To avoid the influence of data dimensions, use the formula to perform normalization processing on the training set, where is the maximum value in the training set, and is the minimum value in the training set.

[0072] Step S103: Use the prediction model to obtain the runoff prediction data of the target hydropower station according to the preset sampling time. The runoff prediction data includes the predicted runoff volumes at each preset sampling time.

[0073] Specifically, accurate runoff prediction depends on feature engineering. For a single-variable runoff sequence, input features of the prediction model are extracted from three aspects: historical information, calendar features, and statistical features. The input features include, but are not limited to, the historical runoff data of the previous seven days, the historical runoff data of the same day in the previous year, and the average value, variance, median, maximum value, and minimum value of the past three days.

[0074] Input the input features into the trained prediction model to obtain the runoff prediction data for the current day. To obtain the runoff prediction values for upcoming dry years, the rolling window method can be used to update the runoff prediction data every seven days.

[0075] Step S104: Perform energy scheduling for the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff. Specifically, in extremely dry years, the hydropower generation cannot meet the electricity demand of users, resulting in a decrease in the revenue of the hydropower station. To increase the revenue of the hydropower station, both hydropower and hydrogen energy equipment are used for power generation to increase the annual revenue of the hydropower station. In addition to using hydrogen energy equipment, other methods such as thermal power generation and nuclear power generation can also be combined with hydropower generation to meet the electricity demand and increase the revenue. This is only an example and not limited thereto.

[0076] The joint scheduling method of runoff prediction and hydropower-hydrogen system provided in this embodiment enhances the historical runoff data, increases the number of training samples of the prediction model, improves the performance of the trained prediction model, performs energy scheduling based on the annual revenue demand of the hydropower-hydrogen system, ensures the maximum annual revenue of the target hydropower station, and improves the resilience of the hydropower system under extremely dry conditions.

[0077] In some alternative embodiments, the method further includes:

[0078] Step S105: According to the energy scheduling results, analyze the first sensitivity of the loss of load due to different hydropower ratios, the second sensitivity of the loss of load due to different hydrogen equipment capacities under different hydropower ratios, and the third sensitivity of the loss of load due to runoff under different hydropower ratios.

[0079] Specifically, the hydropower ratio refers to the proportion of the total electricity demand that the hydropower generation needs to bear. The loss of load is the loss caused by the system's inability to meet the load supply due to a shortage of generation capacity or other reasons. The greater the loss of load, the lower the revenue. Therefore, by analyzing the sensitivities of different hydropower ratios, different hydrogen equipment capacities under different hydropower ratios, and runoff under different hydropower ratios to the loss of load, adjustments can be made according to the sensitivity analysis results to reduce the loss of load and increase the revenue.

[0080] (1) First sensitivity analysis: As shown in Table 1, which presents the quantitative relationship between the hydropower ratio of the hydropower station and the loss of load, the loss of load factor is selected as an indicator to measure the overall economic impact of drought on the system. It can be seen from Table 1 that when the hydropower ratio exceeds 30%, load shedding occurs at the third hydropower station (HP3), that is, load shedding occurs in the hydropower-hydrogen system; the first hydropower station (HP1), the second hydropower station (HP2), and the third hydropower station (HP3) are arranged from upstream to downstream in sequence, and the load shedding phenomenon of the downstream hydropower station is more serious than that of the upstream hydropower station, indicating that compared with large hydropower stations, small hydropower stations are more sensitive to runoff changes and have higher requirements for runoff conditions. In the face of extreme drought disasters, the power generation capacity of large hydropower stations becomes an important pillar for ensuring power supply.

[0081] Table 1

[0082]

[0083] (2) Second sensitivity analysis: Assume that the hydrogen fuel cells equipped in each power station are exactly the same. As Figure 10 shown, it is the hydrogen equipment capacity required to eliminate load shedding under different hydropower ratios. The vertical axis represents the capacity of the hydrogen fuel cell, and the horizontal axis represents the hydropower ratio, with an increment of 10%. The bar chart shows the hydrogen energy equipment capacity required for each power station to achieve zero load shedding. Table 2 shows the exact capacity values of the hydrogen fuel cells corresponding to the bar chart in Figure 2 . The results show that starting from a hydropower ratio of 40%, the system requires a certain amount of hydrogen energy equipment to ensure zero load shedding for the corresponding power station. Figure 2 It can also be seen that when the hydrogen energy equipment capacity of the three power stations is three times the required maximum value, zero load shedding of all power stations can be guaranteed. When the hydropower ratio reaches 100%, the total hydrogen energy equipment capacity required is 78 MW.

[0084] Table 2

[0085]

[0086] Figure 3 shows the reduction of load shedding loss (VOLL) under different hydrogen energy capacities and hydropower ratios. The z-axis represents the reduction of load shedding loss compared to the scenario without hydrogen energy equipment under the corresponding hydropower ratio, the y-axis represents the change in hydropower ratio, and the x-axis represents the hydrogen energy equipment capacity under different hydropower ratios. It can be seen from the figure that at the same hydropower ratio, the larger the hydrogen energy equipment capacity, the greater the reduction in VOLL, but the marginal utility will decrease, and the threshold of marginal utility is proportional to the hydropower ratio. In addition, when the hydropower ratio remains at a low level, only the third hydropower station (blue bar chart) shows a reduction in VOLL, which is consistent with the situation in Figure 2 .

[0087] (3) Third sensitivity analysis: According to the runoff prediction, December is usually the month with the lowest runoff and relatively high load demand. Therefore, December is selected as the representative dry month for monthly sensitivity analysis. Figure 4 shows the monthly load losses caused by drought under different hydropower ratios. It can be clearly seen that the trend shown by the bar chart is extremely similar to the situation described in Figure 2 . Figure 5 The 3D bar chart in shows the monthly VOLL during drought under different runoff levels and hydropower ratios. It can be seen that when the runoff is in the range of 5 m³ / s to 42 m³ / s and the hydropower ratio is between 40% and 100%, the VOLL of the system experiences an approximately exponential level of growth. In addition, as the hydropower ratio increases and the runoff decreases, the VOLL of the system increases.

[0088] Step S106: Obtain the runoff prediction data of the target hydropower station, and adjust the hydropower proportion and hydrogen equipment capacity of the target hydropower station according to the first sensitivity, the second sensitivity, and the third sensitivity.

[0089] Specifically, according to the impact of different runoff volumes on the system's VOLL, it can better assist in prediction and early warning. For example, when the hydropower accounts for 100% of the supply, if the goal is to keep the VOLL below 100 million kWh, the runoff volume needs to be monitored monthly. If the runoff volume is less than 23 cubic meters per second, hydrogen equipment needs to be introduced to maintain the normal operation of the system. This is only an example and not limited thereto.

[0090] The joint dispatching method of runoff prediction and hydropower-hydrogen system provided in this embodiment determines the influence degree of different hydropower proportions, hydrogen equipment capacities, and runoff volumes on the loss of load by conducting sensitivity analysis from three aspects. It is beneficial to adjust the energy dispatching according to the sensitivity analysis results, improve the accuracy of energy dispatching, and reduce the loss of load.

[0091] In this embodiment, a joint dispatching method of runoff prediction and hydropower-hydrogen system is provided, which can be used in the above computer system. Figure 6 It is a flowchart of the joint dispatching method of runoff prediction and hydropower-hydrogen system according to the embodiment of the present invention. As Figure 6 shown, this process includes the following steps:

[0092] Step S201: Obtain the historical drought data of the target hydropower station, and perform data enhancement processing on the historical drought data to obtain the runoff enhancement data set of the target hydropower station.

[0093] Specifically, the above step S201 includes:

[0094] Step S2011: Obtain the hydrological frequency of the historical runoff data of the target hydropower station according to a preset period, and determine the drought years according to the hydrological frequency of the historical runoff data.

[0095] Specifically, as Figure 7 shown, it is a schematic diagram of the joint dispatching process of runoff prediction and hydropower-hydrogen system provided in this embodiment. The Pearson type III distribution is used to analyze the hydrological frequency of the historical runoff data, and the historical years are divided into five types according to the size of the hydrological frequency: normal years, drought years, extremely drought years, wet years, and extremely wet years.

[0096] The expression of the Pearson type III distribution is:

[0097] (1)

[0098] where F represents the hydrological frequency value, and x represents the value of the runoff volume. , is the mean value of x, are the coefficient of variation and the coefficient of skewness respectively.

[0099] The classification standard formula is:

[0100] (2)

[0101] Calculate the hydrological frequency of the annual runoff data for each year according to formula (1), and divide each year into corresponding types according to formula (2).

[0102] Step S2012: Use the historical runoff data of drought years as drought sample data, and perform data augmentation on the drought sample data based on a preset generative adversarial network to obtain augmented sample data. The augmented sample data and the drought sample data form the runoff augmentation dataset of the target hydropower station.

[0103] Specifically, use the DoppelGANger model of the data augmentation technology based on the Generative Adversarial Networks (GAN) to implement data augmentation. GAN is a popular scenario generation method that involves the interaction between a generator and a discriminator, aiming to refine the artificially generated data distribution until the artificially generated data distribution is close to the real data distribution, and finally making the two indistinguishable. By carefully adjusting the parameters that control the generator and the discriminator, this model seeks to optimize the performance of the discriminator on the real runoff sequence during extreme droughts, while ensuring that the synthetic samples obtain relatively high scores. This iterative process promotes the model's ability to internalize the unique distribution characteristics of the actual sample data representing extreme drought conditions. The DoppelGANger model is a variant of GAN, which is more suitable for time series such as runoff data. It combines domain-specific insights with the latest progress of GAN to address the fidelity challenge. First, to construct a correlation model between the measured values and their metadata, the DoppelGANger model separates the generation of metadata from the time series and inputs the corresponding metadata into the time series generator at each time step. At the same time, an auxiliary discriminator for generating metadata is also introduced. Second, to solve the mode collapse problem, the GAN architecture of the DoppelGANger model generates randomized maximum and minimum boundaries and normalized time series respectively, which can then be rescaled to the actual range. Third, to capture temporal correlations, the DoppelGANger model outputs batch samples instead of single samples. In addition, the DoppelGANger model can efficiently capture the temporal correlations in long time series data.

[0104] The runoff prediction and combined dispatching method of the hydropower-hydrogen system provided by this embodiment performs data enhancement processing on the historical runoff data of dry years, and forms a runoff enhancement data set with the enhanced sample data and the real sample data, specifically analyzes the characteristics of the runoff data in dry years, avoids the influence of few runoff samples and long cycle on the data prediction accuracy in extremely dry situations, and improves the prediction accuracy of the runoff data in dry years.

[0105] Step S202: Use the runoff enhancement data set to train a preset prediction model to obtain a trained prediction model.

[0106] Specifically, the above step S202 includes:

[0107] Step S2021: Divide the runoff enhancement data set into a training set, a validation set and a test set according to a preset ratio.

[0108] Specifically, the runoff enhancement data set is divided, for example: 70% as the training set, 10% as the validation set, and 20% as the test set, only as an example, but not limited to this order.

[0109] Step S2022: Use the training set and the validation set to train multiple preset prediction models to obtain multiple trained prediction models. Specifically, the preset prediction models may include: XGBoost, ARIMA and Prophet. The ARIMA model is a statistical method widely used in time series prediction, which is good at capturing various time structure features in the data; the Prophet model is built based on the additive principle, including modules such as trend, seasonality and holiday effects, and is especially suitable for processing time series data with significant seasonal fluctuations, and shows strong adaptability to missing data and trend changes.

[0110] Use the same runoff enhancement data set to train three preset prediction models with different architectures to obtain the corresponding trained prediction models. The specific training process is mature existing technology and will not be elaborated here.

[0111] Step S2023: Use the test set to evaluate the performance of the multiple trained prediction models, and determine the prediction model for runoff prediction according to the evaluation results.

[0112] Specifically, in order to evaluate the performance of ARIMA, Prophet and XGBoost, two commonly used evaluation indicators, the normalized root mean square error (NRMSE) and the normalized mean absolute error (NMAE), are used. The evaluation results show that the NRMSE and NMAE values of XGBoost are the smallest under mixed data, indicating that XGBoost has the best prediction effect. Therefore, XGBoost is used as the prediction model for runoff prediction.

[0113] The XGBoost model is characterized by low computational complexity, fast running speed, and high accuracy, and is widely used in time series prediction. The XGBoost model consists of multiple decision trees. By continuously adding decision trees and splitting features until the growth conditions are met, a new function is learned to fit the residuals of the final prediction, thus growing a tree completely. After completing the model training process and obtaining multiple trees, according to the characteristics of historical runoff data, each sample is classified into a leaf node, and each tree corresponds to a score. The sum of the scores of all trees is the predicted value of the sample.

[0114] The goal of the XGBoost model is to make the predicted runoff values of all trees as close as possible to the corresponding true runoff values while maximizing the generalization ability. The predicted runoff value can be expressed by the following formula:

[0115] (3)

[0116] where, represents the predicted runoff value, represents the weight function of the k-th tree.

[0117] To fit the residuals between the predicted values of the previous several trees and the corresponding true values, a new decision tree is introduced. By combining the loss function and the regularization function, the objective function (equivalent to the decision tree) is introduced. The expression of the objective function is:

[0118] (4)

[0119] where, represents the true runoff value corresponding to the predicted runoff value, is used to quantify the difference between the predicted outflow value and the true value. The regularization term is used to regulate the complexity of the model, and its calculation formula is as follows:

[0120] (5)

[0121] where, represents the complexity of the leaf node of the decision tree, represents the number of leaf nodes of the decision tree, represents the regularization coefficient, represents the weight.

[0122] The XGBoost model uses the gradient boosting method to optimize the objective function. The second-order Taylor expansion is applied in the loss function to accelerate the optimization process. The objective function of the first iteration after being expressed by the second-order Taylor expansion is:

[0123] (6)

[0124] where, is the first derivative, is the second derivative, is the constant term.

[0125] For formula (6), it can be simplified to:

[0126] (7)

[0127] Among them, formula (7) is obtained by deleting all constant terms in formula (6).

[0128] The joint dispatching method of runoff prediction and hydropower - hydrogen system provided by the embodiments of the present invention trains multiple preset prediction models by using a runoff enhanced dataset, and selects the model with the best performance as the prediction model for runoff prediction by evaluating the performance of each model, so as to improve the accuracy of runoff prediction.

[0129] Step S203, obtain the runoff prediction data of the target hydropower station according to the preset sampling time by using the prediction model, and calculate the predicted hydropower generation according to the runoff prediction data.

[0130] Specifically, in the above step S203, obtaining the runoff prediction data of the target hydropower station according to the preset sampling time by using the prediction model includes:

[0131] Step S2031, obtain the input features of the prediction model, and the input features include: the preset historical time runoff of the target hydropower station, the runoff on the same day of the previous year, statistical features and calendar features.

[0132] Specifically, the preset historical time can be the previous seven days, and the statistical features include the average value, variance and median of the runoff data of the previous three days. Then the input features are the runoff values of the previous seven days, the runoff value on the same day of the previous year, the average value, variance and median of the runoff data of the previous three days, and calendar features.

[0133] Step S2032, input the input features into the prediction model to obtain the same - day runoff prediction data of the target hydropower station. Specifically, input the input features into the prediction model, and use formula (7) to predict the current runoff prediction data of the target hydropower station.

[0134] Step S2033, adopt the rolling window method to obtain the runoff prediction data of the target hydropower station according to the preset sampling time interval. Specifically, adopt the rolling window method to update the runoff prediction data every seven days to determine the runoff prediction values in future dry years.

[0135] The joint dispatching method of runoff prediction and hydropower - hydrogen system provided by this embodiment extracts input features from three aspects of historical information, calendar features and statistical features, obtains the same - day runoff prediction data according to the input features, and adopts the rolling window method to update the runoff prediction data to ensure the effectiveness of the runoff prediction data.

[0136] Step S204, perform energy scheduling on the hydropower-hydrogen system based on the annual revenue requirement of the hydropower-hydrogen system and the predicted runoff.

[0137] Specifically, the above-mentioned step S204 includes:

[0138] Step S2041, construct the objective function of the annual revenue of the hydropower-hydrogen system:

[0139] (8)

[0140] Among them, Maximize represents maximizing the annual revenue, represents the load at time is the rd hydropower station's load shedding volume at time , in addition, the parameter represents the electricity price at time represents the unit load shedding loss, is the discrete scheduling time period, is the unit scheduling time window, is the th

[0141] hydropower station in the basin.

[0142] Perform energy scheduling according to the power of the hydrogen fuel cell and the predicted hydropower generation, and the following constraints need to be satisfied:

[0143] (1) Power balance constraint condition: , where represents the th hydropower station's output at time represents the output of the hydrogen fuel cell equipped with the th hydropower station at time

[0144] (2) The reservoir capacity of the target hydropower station is equal to the difference between the inflow and the outflow, and the reservoir capacity constraint condition: , where and respectively represent at time Upper and lower limits of the water storage capacity of a reservoir;

[0145] 3 Discharge constraint conditions: , represent the minimum and maximum discharge of the th reservoir respectively;

[0146] 4 Power generation flow constraint: wherein, represents the maximum overcurrent capacity that the hydroelectric generating units of the th power station can withstand, is the maximum overflow capacity of the aqueduct of the th power station;

[0147] 5 Hydropower generation constraint: wherein, and are respectively the maximum and minimum allowable power generation of the th level hydropower station at time

[0148] wherein, is the output coefficient of the th hydropower station, is the water head of the th hydropower station, is the power generation flow of the th power station at time ;

[0149] 6 The power and hydrogen quantity of the hydrogen fuel cell follow the following constraints:

[0150]

[0151] wherein, and respectively represent the upper limit of the hydrogen fuel cell power and the maximum hydrogen production per unit time.

[0152] Step S2042, when the objective function is maximized, obtain the output ratio of the hydrogen fuel cell to hydropower, and use the output ratio as the optimal scheduling result for energy scheduling. The load shedding amount of the th hydropower station at time is determined by the output ratio of the hydrogen fuel cell to hydropower at time .

[0153] Specifically, by adjusting the power of the hydrogen fuel cell, changing the hydrogen power generation, and combining with the hydropower generation amount obtained based on the runoff prediction value, change the th hydropower station in the objective function formula (8) at time The load shedding amount at a moment. By adjusting the power of the hydrogen fuel cell and changing the hydrogen power generation, the output ratio of the hydrogen fuel cell and hydropower is adjusted to reduce the load shedding amount and maximize the objective function of the annual income.

[0154] The joint dispatching method of runoff prediction and hydropower-hydrogen system provided in this embodiment calculates the annual income according to the dispatching result by setting the objective function of the hydropower-hydrogen system dispatching. With the goal of maximizing the annual income, the dispatching strategy when the annual income is the largest is determined to ensure the minimum load shedding and the maximum annual income of the hydropower-hydrogen system.

[0155] In a specific embodiment, the first hydropower station, the second hydropower station, and the third hydropower station are all diversion-type hydropower stations. Among them, the first hydropower station is connected to a reservoir with a certain regulation capacity. The basic parameters of the three hydropower stations are shown in Table 3.

[0156] Table 3

[0157]

[0158] Figure 8 , Figure 9 , Figure 10 are the daily runoff data of three hydropower stations in the same river basin in the past ten years (from 2010 to 2022), respectively. Generally speaking, May to August is the rainy season, and December to April is the dry season, but there are differences in the daily and monthly runoff in different years. From the perspective of runoff, the runoff of the first hydropower station (HP1) and the second hydropower station (HP2) is greater than that of the third hydropower station (HP3).

[0159] For each hydropower station, the Pearson type III curve is used to fit the ten-year runoff data, the optimal curve fitting method is used for calibration, and the runoff frequency value in the past ten years is calculated. The coefficient of variation and skewness coefficient are set in advance as: , and the hydrological frequency is calculated using the expression of the Pearson type III distribution. The years are divided into five categories, and the classification criteria are shown in formula (2).

[0160] (1) Hydrological classification: Taking the first hydropower station as an example, the classification results of its historical annual runoff are shown in Table 4 in detail. Since each station is located in the same river basin, the classification results are basically the same. The years with an annual average runoff frequency greater than 90% are defined as extremely dry years (Extremely wet year), and such years are called those that have experienced extremely dry events. The results show that since 2010, there have been six years in the river basin where the three hydropower stations are located with an average runoff in the dry and extremely dry scenarios. Among them, 2016 is the only year classified as extremely dry.

[0161] During the dry season from December to February, hydropower is generally not the main source of power supply. Only during the wet season from March to November does hydropower become the main power source for regional consumption. Under these conditions, extreme drought disasters occurring during the wet season have a greater impact on regional power supply. Therefore, in subsequent case studies, the focus is mainly on runoff prediction and coordinated scheduling of the hydropower-hydrogen system under extreme drought disasters during the wet season.

[0162] Table 4

[0163]

[0164] (2)Sample data augmentation: Use quantiles to evaluate the distribution of data, and at the same time use the double quantile-quantile plot (double Q-Q plot) to compare the similarity of the probability distributions between two variables, so as to ensure the effectiveness and reliability of small sample data. In the Q-Q plot, the vertical axis and the horizontal axis represent the quantiles of the generated sample and the original sample respectively. If the points on the graph are distributed near the diagonal line passing through the origin, it is considered that the probability distributions of the two variables are similar. In addition, by comparing the probability distributions of the generated sample and the original sample, the effectiveness and accuracy of the generated sample are further confirmed.

[0165] Use the DoppelGANger algorithm to augment the runoff data of three hydropower stations. Figure 11 、 Figure 12 、 Figure 13 Show the Q-Q plots and probability distribution fitting results of the generated runoff data and the real runoff data of these three hydropower stations. The results show that the real samples and the synthetic samples generated by the DoppelGANger algorithm are generally consistent in distribution, indicating that the DoppelGANger algorithm has successfully learned the runoff characteristics under extreme drought conditions and generated new samples that can reflect the characteristics of the real samples accordingly.

[0166] (3)Runoff prediction: To predict runoff more accurately, combine the above-generated data with the real runoff data to create the initial dataset for the prediction model (a total of 24 years of data are available in this embodiment). Select the data of the first 22 years as the training set, and use the remaining 2 years of data as the validation set and the test set respectively. For input features, select historical runoff (including runoff for the previous seven days), runoff on the same day of the previous year, statistical features (including the mean, variance, and median of runoff data for the past three days), and calendar features.

[0167] To verify the effectiveness of XGBoost, ARIMA and Prophet can also be selected as comparison benchmarks. Use two commonly used evaluation metrics: the normalized root mean square error (NRMSE) and the normalized mean absolute error (NMAE).

[0168] Figure 14, Figure 15 , Figure 16 shows the prediction results of the XGBoost model for three power stations. In this figure, the black line represents the predicted values of the model, while the red line represents the actual runoff. It can be found that the predicted values are similar to and relatively close to the actual runoff change trends of the three hydropower stations, indicating that the XGBoost model is a good choice for runoff prediction in extremely dry years.

[0169] Table 5 lists the prediction results of different models. It can be seen from it that the NRMSE and NMAE values of XGBoost are the smallest under mixed data, indicating that the XGBoost model has the best prediction effect. Compared with the prediction models using original data, the prediction models using mixed data have better prediction effects for the three hydropower stations, indicating that more information can be discovered from the mixed data with a larger sample size. In addition, for the original data, ARIMA is usually better than Prophet and XGBoost because, as a statistical model, the parameter training of the ARIMA model does not depend on a large amount of data.

[0170] Table 5

[0171]

[0172] (4) Energy dispatch: Use the runoff prediction results as input data. Figure 17 shows the typical load curves of three cascade hydropower stations considering seasonal variations. To verify the effectiveness of the prediction method and its impact on the system dispatch performance, Table 6 compares the absolute gaps between the dispatch results using real runoff data and using different prediction data as input. The results show that the data generated by XGBoost+GAN is closest to the true performance of the system, and the system dispatch performance gap can be reduced to 1.97%.

[0173] Table 6

[0174]

[0175] Figure 18 shows the optimal dispatch results of three hydropower stations using the runoff data input by XGBoost+GAN. The bar charts represent the monthly hydropower dispatch results of the hydropower stations respectively, and the line charts represent the monthly load shedding in extremely dry years. It can be seen from Figure 18 that even with the support of relatively accurate runoff predictions across quarters, load shedding occurs almost every month except in January, and the load shedding in the dry season is higher than that in the wet season. The trend of the hydropower generation plan coincides with the load curve.

[0176] Figure 19Shows the optimal scheduling scheme of three cascade hydropower stations with a 10MW hydrogen fuel cell. The bar chart represents the monthly hydropower scheduling results of the hydropower station, and the red bar chart represents the output of the hydrogen fuel cell (HFC). The line chart represents the VOLL (Value of Lost Load) for each month in a dry year. With the hydrogen fuel cell equipped, the load shedding losses of HP1 and HP2 approach 0. Although there are still load shedding losses for HP3 in the second half of the year, the total losses have shown a significant reduction. With the help of the fuel cell, the hydropower generation during the dry season is relatively high, and even in a dry year, it can better support the hydropower supply during the dry season.

[0177] Figure 20 Demonstrates the resilience of the hydropower system in an extremely dry year. The blue shaded area represents that compared with the case without hydrogen, the load shedding reduced by 196.92 million kWh provided by the 10 MW capacity hydrogen equipment. Compared with without hydrogen, when using the hydrogen equipment, the load shedding is reduced by 65.09%.

[0178] In this embodiment, a combined scheduling device for runoff prediction and hydropower-hydrogen system is also provided. This device is used to implement the above embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0179] This embodiment provides a combined scheduling device for runoff prediction and hydropower-hydrogen system, as Figure 21 shown, including:

[0180] A data enhancement module 501, configured to obtain historical drought data of a target hydropower station, and perform data enhancement processing on the historical drought data to obtain a runoff enhancement data set of the target hydropower station.

[0181] A model training module 502, configured to train a preset prediction model using the runoff enhancement data set to obtain a trained prediction model.

[0182] A runoff prediction module 503, configured to obtain runoff prediction data of the target hydropower station according to a preset sampling time using the prediction model, where the runoff prediction data includes predicted runoff volumes at each preset sampling time.

[0183] An energy scheduling module 504, configured to perform energy scheduling on the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff volume.

[0184] In some alternative implementation manners, the data enhancement module 501 includes:

[0185] A drought year determination unit for obtaining the hydrological frequency of the historical runoff data of the target hydropower station according to a preset period, and classifying and determining drought years based on the hydrological frequency of the historical runoff data.

[0186] A sample data enhancement unit for using the historical runoff data of drought years as drought sample data, and performing data enhancement processing on the drought sample data based on a preset generative adversarial network to obtain enhanced sample data, and the enhanced sample data and the drought sample data form a runoff enhancement dataset of the target hydropower station.

[0187] In some alternative embodiments, the model training module 502 includes:

[0188] A data partitioning unit for partitioning the runoff enhancement dataset into a training set, a validation set, and a test set according to a preset ratio.

[0189] A model training unit for training multiple preset prediction models using the training set and the validation set to obtain multiple trained prediction models.

[0190] A performance evaluation unit for evaluating the performance of the multiple trained prediction models using the test set, and determining a prediction model for runoff prediction according to the evaluation results.

[0191] In some alternative embodiments, the runoff prediction module 503 includes:

[0192] An input feature acquisition unit for acquiring the input features of the prediction model, where the input features include: the preset historical time runoff of the target hydropower station, the runoff on the same day of the previous year, statistical features, and calendar features.

[0193] A same-day runoff prediction unit for inputting the input features into the prediction model to obtain the same-day runoff prediction data of the target hydropower station.

[0194] A prediction data acquisition unit for using a rolling window method to obtain the runoff prediction data of the target hydropower station at a preset sampling time interval.

[0195] In some alternative embodiments, the energy scheduling module 504 includes:

[0196] An objective function construction unit for constructing an objective function for the annual revenue of the hydropower-hydrogen system:

[0197]

[0198] where represents maximizing the annual revenue, represents the load at time, is the th hydropower station at time, , in addition, the parameter represents the electricity price at a moment, represents the loss of load per unit, is the discrete scheduling time period, is the unit scheduling time window, is the th hydropower station in the basin.

[0199] The energy scheduling unit is used to obtain the output power ratio of the hydrogen fuel cell and hydropower when the objective function is maximized, and use the output power ratio as the optimal scheduling result for energy scheduling. The loss of load of the th hydropower station at the moment is determined by the output power ratio of the hydrogen fuel cell and hydropower at the moment.

[0200] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding above-mentioned embodiments, and will not be elaborated here.

[0201] The runoff prediction and combined dispatching device of the hydropower-hydrogen system in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0202] The embodiment of the present invention also provides a computer device having the Figure 21 shown runoff prediction and combined dispatching device of the hydropower-hydrogen system.

[0203] Please refer to Figure 22 , Figure 22 is a schematic structural diagram of a computer device provided by an optional embodiment of the present invention. As Figure 22 shown, the computer device includes: one or more processors 10, a memory 20, and an interface for connecting each component, including a high-speed interface and a low-speed interface. Each component communicates with each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a set of blade servers, or a multi-processor system).Figure 22 Take a processor 10 as an example.

[0204] The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above-mentioned hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above-mentioned programmable logic device can be a complex programmable logic device, a field-programmable gate array, a generic array logic, or any combination thereof.

[0205] Among them, the memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0206] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely provided with respect to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and a combination thereof.

[0207] The memory 20 can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memories. The computer device further includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

[0208] Embodiments of the present invention also provide a computer-readable storage medium. The methods according to the embodiments of the present invention can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the methods described herein can be stored as such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the methods shown in the above embodiments are implemented.

[0209] A part of the present invention can be applied as a computer program product, for example, computer program instructions, which when executed by a computer, can call or provide the methods and / or technical solutions according to the present invention through the operation of the computer. The forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Herein, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to the computer.

[0210] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for runoff prediction and joint scheduling of hydropower-hydrogen system, characterized in that: The method comprises: Acquire historical drought data of a target hydropower station, and perform data enhancement processing on the historical drought data to obtain a runoff enhancement data set of the target hydropower station, wherein the historical drought data is historical runoff data in drought years; Using the runoff enhancement data set to train a preset prediction model to obtain a trained prediction model; Using the prediction model to obtain runoff prediction data of the target hydropower station according to preset sampling times, the runoff prediction data including predicted runoff volumes at each preset sampling time; Performing energy dispatch of the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff; According to the energy dispatch results, the first sensitivity of different hydropower proportions to load loss, the second sensitivity of different hydrogen equipment capacities to load loss under different hydropower proportions, and the third sensitivity of runoff to load loss under different hydropower proportions are analyzed; The runoff prediction data of the target hydropower station is obtained, and the hydropower ratio and hydrogen equipment capacity of the target hydropower station are adjusted according to the first sensitivity, the second sensitivity, and the third sensitivity.

2. The method according to claim 1, characterized in that: The historical drought data of the target hydropower station is obtained, and the historical drought data is subjected to data enhancement processing to obtain a runoff enhancement data set of the target hydropower station, including: Obtain the hydrological frequency of historical runoff data of the target hydropower station according to a preset period, and classify the drought years according to the hydrological frequency of the historical runoff data; The historical runoff data of the drought year is used as drought sample data, and data enhancement processing is performed on the drought sample data based on a preset generative adversarial network to obtain enhanced sample data. The enhanced sample data and the drought sample data constitute a runoff enhancement data set of the target hydropower station.

3. The method according to claim 1, characterized in that The preset prediction model is trained using the runoff enhancement data set to obtain a trained prediction model, including: Dividing the runoff enhancement dataset into a training set, a trial set and a test set according to a preset ratio; Using the training set and the test set to train multiple preset prediction models to obtain multiple trained prediction models; The test set is used to perform performance evaluation on the trained multiple prediction models, and a prediction model for runoff prediction is determined based on the evaluation results.

4. The method according to claim 1, characterized in that The prediction model is used to obtain the runoff prediction data of the target hydropower station according to the preset sampling time, including: Acquiring input features of the prediction model, the input features including: runoff of a target hydropower station at a preset historical time, runoff on the same day of the previous year, statistical features, and calendar features; Inputting the input features into the prediction model to obtain the runoff prediction data of the target hydropower station on the same day; The rolling window method is used to obtain the runoff prediction data of the target hydropower station according to the preset sampling time interval.

5. The method according to claim 1, characterized in that: Energy dispatching of the hydropower-hydrogen system is performed based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff, including: Construct the objective function of the annual revenue of the hydropower-hydrogen system: Among them, Maximize means maximizing annual income, express The load of time, For the Hydropower stations in The amount of load loss at the moment, In addition, the parameter express The electricity price at the time, It represents the unit load loss, is a discrete scheduling time period, Scheduling time windows for units, The first hydroelectric power stations; When the objective function is the largest, the output ratio of hydrogen fuel cells to hydropower is obtained, and the output ratio is used as the optimal scheduling result for energy scheduling. Hydropower stations in The load loss at the moment is given by The output ratio of hydrogen fuel cells and hydropower at that moment is determined.

6. A runoff prediction and joint dispatching device for hydropower-hydrogen system, characterized in that: The device comprises: A data enhancement module is used to obtain historical drought data of a target hydropower station and perform data enhancement processing on the historical drought data to obtain a runoff enhancement data set of the target hydropower station, wherein the historical drought data is historical runoff data in drought years; A model training module, used to train a preset prediction model using the runoff enhancement data set to obtain a trained prediction model; A runoff prediction module, used to obtain the runoff prediction data of the target hydropower station according to the preset sampling time by using the prediction model, wherein the runoff prediction data includes the predicted runoff volume at each preset sampling time; An energy dispatching module, used for dispatching energy of the hydropower-hydrogen system based on the annual revenue demand of the hydropower-hydrogen system and the predicted runoff; The result analysis module is used to analyze the first sensitivity of different hydropower ratios to load loss, the second sensitivity of different hydrogen equipment capacities to load loss under different hydropower ratios, and the third sensitivity of runoff to load loss under different hydropower ratios according to the energy dispatch results; The capacity adjustment module is used to obtain the runoff prediction data of the target hydropower station, and adjust the hydropower ratio and hydrogen equipment capacity of the target hydropower station according to the first sensitivity, the second sensitivity, and the third sensitivity.

7. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 5 by executing the computer instructions.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 5.

9. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the method according to any one of claims 1 to 5.

Images