Time period coordinated control method and device for integrated water, light and storage energy network

Through the integrated water-photo-storage energy network time period collaborative control method, the power generation resources are dynamically adjusted, which solves the problem of load demand lag of hydraulic and photovoltaic power generation systems, improves the stability and reliability of the power grid, and reduces resource waste.

CN118630841BActive Publication Date: 2024-11-08SHENZHEN SHENSHUI LONGGANG WATER GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411096346.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-08
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

The existing hydraulic and photovoltaic power generation systems have lag in response to load demand, resulting in insufficient power supply, affecting the stability of the power grid and user power usage experience, and wasting power generation resources, reducing grid reliability.

Method used

Adopt the time period collaborative control method of water, light and storage integrated energy network, obtain historical data through interactive energy networks, build a power generation operation prediction model, pre-construct an ultra-short-term load prediction model, dynamically adjust the allocation of power generation resources, realize real-time monitoring and load balancing, and reduce the power grid Load fluctuations.

Benefits of technology

Real-time monitoring and forecasting of load demand, dynamically adjust power generation resources, ensure the balance between energy supply and demand, improve the stability and reliability of power grid operation, and reduce waste of power generation resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118630841B_ABST
    Figure CN118630841B_ABST
Patent Text Reader

Abstract

The present invention provides a method and device for coordinated control of time periods of an integrated water, photovoltaic and storage energy network, which relates to the field of power distribution control technology. By performing multi-dimensional data collection according to a coordinated control time period window, a real-time load association array, a real-time hydropower association array and a real-time photovoltaic association array are obtained; the real-time load association array is synchronized to an ultra-short-term load forecasting model to obtain real-time load demand; power generation resources are allocated according to the real-time load demand to obtain coordinated power generation operation settings and to perform coordinated control of the energy network. The technical problem that there is a lag in the satisfaction of load demand by power generation resources in the prior art, resulting in waste of power generation resources in the energy network and even reduced reliability of power grid operation is solved. The technical effect of real-time monitoring and prediction of load demand, dynamic adjustment of power generation resources, and reducing power grid load fluctuations while ensuring a balance between energy supply and demand, and improving the stability and reliability of power grid operation is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of power distribution control technology, and in particular to a time period coordinated control method and device for a water, photovoltaic and storage integrated energy network. Background Art

[0002] In existing technologies, hydropower and photovoltaic power generation systems often have a certain lag in responding to load demand. For example, consider an energy network that mainly relies on hydropower and photovoltaic power generation. Due to the inaccurate prediction of changes in sunshine and water flow, when sudden weather changes cause a sharp drop in photovoltaic power generation, or drought causes a drop in hydropower generation, the system may not be able to adjust power generation in time to adapt to the increase in load demand. This lag may lead to insufficient power supply, affecting the stability of the power grid and the user's power experience.

[0003] In summary, the existing technology has the technical problem that there is a lag in meeting the load demand with power generation resources, which leads to waste of power generation resources in the energy network and even reduced reliability of power grid operation. Summary of the invention

[0004] The present invention provides a time period coordinated control method and device for an integrated water, photovoltaic and storage energy network, which is used to solve the technical problem in the prior art that there is a lag in meeting the load demand with power generation resources, resulting in waste of power generation resources in the energy network and even reduced reliability of power grid operation.

[0005] In view of the above problems, the present invention provides a time period coordinated control method and device for an integrated water, photovoltaic and storage energy network.

[0006] According to a first aspect of the present invention, a method for coordinated control of a water-photovoltaic-storage integrated energy network during a certain period of time is provided, the method comprising: obtaining energy allocation data through an interactive energy network, wherein the energy allocation data comprises historical renewable power generation data and historical load demand data; developing a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model comprises a hydraulic prediction branch and a photovoltaic prediction branch; pre-constructing an ultra-short-term load prediction model, wherein the ultra-short-term load prediction model adopts the historical load demand data as construction data; presetting a coordinated control period window, and performing multi-dimensional data collection according to the coordinated control period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array; synchronizing the real-time load association array to the ultra-short-term load prediction model to obtain real-time load demand; allocating power generation resources according to the real-time load demand, and performing allocation evaluation using the real-time hydropower association array, the real-time photovoltaic association array, and the power generation operation prediction model to obtain a coordinated power generation operation setting; using the coordinated control period window as a constraint, and adopting the coordinated power generation operation setting to perform coordinated control of the energy network.

[0007] The second aspect of the present invention provides a time period collaborative control device for an integrated water, photovoltaic and storage energy network, the device comprising: an energy allocation acquisition unit, used to obtain energy allocation data through an interactive energy network, wherein the energy allocation data includes historical renewable power generation data and historical load demand data; a prediction model construction unit, used to develop a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch; a load prediction analysis unit, used to pre-construct an ultra-short-term load prediction model, wherein the ultra-short-term load prediction model uses the historical load demand data as construction data; a data acquisition execution unit, used to preset a collaborative A control time period window is defined, and multi-dimensional data is collected according to the collaborative control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array; a load demand analysis unit is used to synchronize the real-time load association array to the ultra-short-term load forecasting model to obtain real-time load demand; a collaborative control analysis unit is used to allocate power generation resources according to the real-time load demand, and use the real-time hydropower association array, the real-time photovoltaic association array, and the power generation operation forecasting model to perform allocation evaluation to obtain collaborative power generation operation settings; a collaborative control execution unit is used to use the collaborative power generation operation settings to perform collaborative control of the energy network with the collaborative control time period window as a constraint.

[0008] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0009] The method provided by the embodiment of the present invention obtains energy allocation data through an interactive energy network, wherein the energy allocation data includes historical renewable power generation data and historical load demand data; develops a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydropower prediction branch and a photovoltaic prediction branch; pre-constructs an ultra-short-term load prediction model, wherein the ultra-short-term load prediction model uses the historical load demand data as construction data; presets a collaborative control time period window, and performs multi-dimensional data collection according to the collaborative control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array; synchronizes the real-time load association array to the ultra-short-term load prediction model to obtain real-time load demand; allocates power generation resources according to the real-time load demand, and uses the real-time hydropower association array, the real-time photovoltaic association array, and the power generation operation prediction model to perform allocation evaluation to obtain a collaborative power generation operation setting; uses the collaborative control time period window as a constraint, and uses the collaborative power generation operation setting to perform collaborative control of the energy network. It achieves the technical effect of real-time monitoring and forecasting of load demand, dynamically adjusting power generation resources, reducing grid load fluctuations while ensuring a balance between energy supply and demand, and improving the stability and reliability of grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the process flow of the time period coordinated control method of the water-solar-storage integrated energy network provided by the present invention;

[0011] Figure 2 A schematic diagram of a flow chart for constructing a hydraulic prediction branch in the time period coordinated control method for the integrated water, photovoltaic and storage energy network provided by the present invention;

[0012] Figure 3 A schematic diagram of the structure of the time period collaborative control device for the integrated water, photovoltaic and storage energy network provided by the present invention.

[0013] Explanation of the accompanying drawings: energy allocation acquisition unit 11, prediction model construction unit 12, load prediction analysis unit 13, data acquisition execution unit 14, load demand analysis unit 15, collaborative control analysis unit 16, collaborative control execution unit 17. DETAILED DESCRIPTION

[0014] The present invention provides a time period coordinated control method and device for an integrated water, photovoltaic and storage energy network, which is used to solve the technical problem in the prior art that there is a lag in the satisfaction of load demand by power generation resources, resulting in waste of power generation resources in the energy network and even reduced reliability of power grid operation. The method achieves the technical effect of real-time monitoring and prediction of load demand, dynamic adjustment of power generation resources, and reducing power grid load fluctuations while ensuring a balance between energy supply and demand, and improving the stability and reliability of power grid operation.

[0015] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0016] Embodiment 1, as Figure 1 As shown, the present invention provides a time period coordinated control method for a water-photovoltaic-storage integrated energy network, the method comprising:

[0017] A100: The interactive energy network obtains energy allocation data, wherein the energy allocation data includes historical renewable power generation data and historical load demand data.

[0018] Specifically, in this embodiment, the energy network is a complementary energy supply structure formed by integrating hydropower generation, photovoltaic power generation, non-renewable energy generation and electric energy storage systems, based on which real-time monitoring and optimized scheduling of power generation, energy storage and electricity consumption can be achieved.

[0019] At the same time, the energy network in the target power supply area plays the role of storing energy distribution data, recording the historical renewable power generation data of the historical power generation conditions of the hydropower generation equipment and photovoltaic power generation equipment that provide electricity to the target power supply area, and recording the historical load demand data of the historical electricity consumption and power demand conditions in the target power supply area.

[0020] This embodiment will elaborate on the application of the energy allocation data in the coordinated control of power generation in the subsequent description.

[0021] A200: Developing a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch;

[0022] In one embodiment, a power generation operation prediction model is developed based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch, and the method step A200 provided by the present invention further includes:

[0023] A210: Extracting and obtaining a historical hydropower generation log based on the historical renewable power generation data, wherein the historical hydropower generation log includes a plurality of hydropower generation association arrays, a plurality of hydropower generation records, and a plurality of hydropower generation operation settings.

[0024] A220: Construct the hydraulic prediction branch based on a back propagation neural network, and use the historical hydropower generation log as training data to optimize the prediction accuracy of the hydraulic prediction branch.

[0025] A230: Extracting and obtaining a historical photovoltaic log based on the historical renewable power generation data, wherein the historical photovoltaic log includes a plurality of photovoltaic power generation association arrays, a plurality of photovoltaic power generation records, and a plurality of photovoltaic power generation operation settings.

[0026] A240: constructing the photovoltaic prediction branch based on a back propagation neural network, and optimizing the photovoltaic prediction branch using the historical photovoltaic log.

[0027] A250: The hydraulic prediction branch and the photovoltaic prediction branch are connected in parallel to complete the construction of the power generation operation prediction model.

[0028] In one embodiment, each of the plurality of hydroelectric power generation associative arrays includes a reservoir water level, an inflow, and an outflow.

[0029] In one embodiment, Figure 2 As shown, the hydraulic prediction branch is constructed based on a back propagation neural network, and the historical hydropower generation log is used as training data to optimize the prediction accuracy of the hydraulic prediction branch. Step A220 of the method provided by the present invention also includes:

[0030] A221: Time-sequentially sorting the plurality of hydroelectric generation association arrays, the plurality of hydroelectric generation records and the plurality of hydroelectric generation operation settings to obtain an association array sequence, a generation sequence and an operation setting sequence.

[0031] A222: Preset the verification data splitting ratio, and perform interval data extraction on the associative array sequence, power generation sequence and operation setting sequence according to the verification data splitting ratio to obtain a time cross-training set, a time cross-validation set and a time cross-test set.

[0032] A223: Preset a prediction accuracy constraint, and use the time cross-training set, the time cross-validation set and the time cross-test set to optimize the prediction accuracy of the hydraulic prediction branch until the prediction accuracy of the hydraulic prediction branch meets the prediction accuracy constraint.

[0033] Specifically, in this embodiment, based on the historical renewable power generation data in step A100, it can be known that the historical renewable power generation data includes hydropower generation data, based on which, the historical hydropower generation log is directly extracted from the historical renewable power generation data.

[0034] The historical hydropower generation log specifically includes multiple hydropower generation association arrays at multiple historical moments, multiple hydropower generation records, and multiple hydropower generation operation settings. Each of the multiple hydropower generation association arrays includes a reservoir water level, an inflow flow, and an outflow flow. The hydropower generation operation settings are operating parameters of the hydropower generation equipment, including but not limited to head settings, flow settings, guide vane openings, and operating speeds of turbines and generators.

[0035] It should be understood that under the same hydroelectric power generation association array, different hydroelectric power generation operation settings will result in different hydroelectric power generation, and the hydroelectric power generation is affected by the combined effect of the hydroelectric power generation association array and the hydroelectric power generation operation settings.

[0036] Based on this, this embodiment constructs the hydraulic prediction branch based on the back propagation neural network, and the hydraulic prediction branch can reversely determine the hydraulic power generation operation settings of the hydraulic power generation equipment according to the hydropower generation association array and the planned hydropower generation.

[0037] The method for training the hydraulic prediction branch to improve the numerical accuracy of the hydraulic power generation operation setting predicted by the hydraulic prediction branch is as follows:

[0038] It should be understood that the multiple hydropower association arrays, multiple hydropower generation records and multiple hydropower operation settings are associated based on data record time mapping. This embodiment performs time sequence sorting on the multiple hydropower association arrays, multiple hydropower generation records and multiple hydropower operation settings to obtain an association array sequence, a generation sequence and an operation setting sequence.

[0039] The verification data split ratio is preset. Exemplarily, the verification data split ratio is training set: verification set: test set = 8:1:1.

[0040] The process of splitting the power generation sequence using the verification data splitting ratio is to determine the total number of data points in the power generation sequence. Exemplarily, the power generation sequence has a total of 1,000 hydropower generation records sorted by data collection time.

[0041] The training set is 80%, so 800 hydropower generation records need to be extracted from 1000 hydropower generation records. In this embodiment, data is extracted by using the same interval, so the extraction interval of the data points is 5, that is, 4 hydropower generation records are randomly extracted for every 5 data points in the power generation sequence.

[0042] Data points are extracted from the power generation sequence according to the interval distance, and 800 hydropower generation records are obtained as the hydropower generation record dimensional data in the time cross-training set.

[0043] By analogy, the same method is used to obtain 100 hydropower generation records as the hydropower generation record dimensional data in the time cross-validation set.

[0044] By analogy, the same method is used to obtain 100 hydropower generation records as the hydropower generation record dimensional data in the time crossover test set.

[0045] Furthermore, according to the dimensional data mapping of the hydropower generation record in the time cross-training set, the data extraction and reorganization of the power generation sequence and the operation setting sequence are performed, and the interval data extraction of the associative array sequence, the power generation sequence and the operation setting sequence is completed to obtain the time cross-training set, the time cross-validation set and the time cross-test set, wherein any one of the data sets in the time cross-training set, the time cross-validation set and the time cross-test set includes multiple groups of hydropower generation associative arrays-hydropower generation records-hydropower generation operation settings.

[0046] A conventional training method of a back-propagation neural network is adopted, the time cross-training set is used to train the prediction accuracy of the hydraulic prediction branch, the time cross-validation set is used to verify the prediction accuracy of the hydraulic prediction branch, and the time cross-test set is used to optimize the hydraulic prediction branch until the prediction accuracy of the hydraulic prediction branch meets the prediction accuracy constraint. Exemplarily, the prediction accuracy constraint is that the prediction accuracy is stably higher than 98%.

[0047] It should be understood that this embodiment obtains a time cross-training set, a time cross-validation set and a time cross-test set based on interval data extraction as training data for the hydraulic prediction branch, thereby achieving the technical effect of avoiding overfitting and improving the prediction credibility of the hydraulic prediction branch.

[0048] A historical photovoltaic log is obtained based on the historical renewable power generation data extraction, wherein the historical photovoltaic log includes multiple photovoltaic power generation association arrays, multiple photovoltaic power generation records and multiple photovoltaic power generation operation settings, wherein each photovoltaic power generation association array includes solar radiation, cloud coverage, temperature, and humidity, and the photovoltaic power generation operation settings are operating parameters of the photovoltaic power generation equipment, including but not limited to the inclination angle of the photovoltaic panel and the azimuth angle of the photovoltaic panel.

[0049] The same training data processing method as that for constructing a hydraulic prediction branch is used to process the historical photovoltaic logs, and the same method for optimizing the prediction accuracy of the hydraulic prediction branch is performed based on the training data obtained through processing. After the photovoltaic prediction branch is constructed based on a back propagation neural network, the processed historical photovoltaic logs are used to perform optimization training for the photovoltaic prediction branch, so that the prediction accuracy of the photovoltaic prediction branch also meets the prediction accuracy constraint.

[0050] The photovoltaic prediction branch can reversely determine the photovoltaic power generation operation settings for achieving the preset photovoltaic power generation after inputting the photovoltaic power generation association array and the preset photovoltaic power generation.

[0051] This embodiment connects the hydraulic prediction branch and the photovoltaic prediction branch in parallel to complete the construction of the power generation operation prediction model. The purpose of constructing the power generation operation prediction model in this embodiment is to achieve a more accurate prediction of the power generation of hydraulic and photovoltaic power generation, and to provide a reliable data source for subsequent power generation resources and improve energy utilization efficiency.

[0052] A300: Pre-constructing an ultra-short-term load forecasting model, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data.

[0053] In one embodiment, an ultra-short-term load forecasting model is pre-constructed, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data, and the method step A300 provided by the present invention further includes:

[0054] A310: Split the historical load demand data to obtain weekday load demand data, weekend load demand data, and holiday load demand data.

[0055] A320: Preset a load-related indicator set, and split the working day load demand data according to the load-related indicator set to obtain multiple working day power loads and multiple working day load association arrays.

[0056] A330: constructing a workday load forecasting branch based on a back propagation neural network, and using the multiple workday power loads and multiple workday load association arrays as training data to optimize the prediction accuracy of the workday load forecasting branch.

[0057] A340: Similarly, a weekend load forecasting branch and a holiday load forecasting branch are constructed based on the weekend load demand data and the holiday load demand data.

[0058] A350: The working day load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch are connected in parallel, and a special event judgment engine is configured at the input ends of the working day load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch to complete the construction of the ultra-short-term load forecasting model.

[0059] Specifically, in this embodiment, the historical load demand data are divided into data sets of weekdays, weekends and holidays according to date types, and the weekday load demand data, weekend load demand data and holiday load demand data are obtained.

[0060] The preset load-related indicator set includes weather temperature, humidity, date and time information (such as day of the week, season), etc. These indicators are directly or indirectly correlated with the power load.

[0061] According to the load-related index set, the working day load demand data is screened, the working day electricity load data related to the specific index and the corresponding load association array are extracted, and the multiple working day electricity loads and the multiple working day load association arrays are obtained.

[0062] It should be understood that the multiple working day power loads and multiple working day load association arrays are associated and mapped, each working day power load is an instantaneous power load, and each working day load association array includes instantaneous weather temperature, humidity, and non-instantaneous date and time information (such as day of the week, season).

[0063] The same technology as that for constructing the hydraulic prediction branch is adopted to construct the working day load prediction branch based on the back propagation neural network, and the plurality of working day electricity loads and the plurality of working day load association arrays are used as training data, and the conventional training method of the back propagation neural network is adopted to optimize the prediction accuracy of the working day load prediction branch.

[0064] By analogy, the same method of constructing and training the weekday load forecasting branch is adopted to construct and generate a weekend load forecasting branch and a holiday load forecasting branch based on the weekend load demand data and the holiday load demand data.

[0065] The weekday load forecasting branch, weekend load forecasting branch and holiday load forecasting branch are connected in parallel to obtain a comprehensive ultra-short-term load forecasting model. At the same time, a special event judgment engine is integrated at the input end of the ultra-short-term load forecasting model. The function of the special event judgment engine is to judge to which category of time (weekdays, weekends or holidays) the collection time of the input data of the ultra-short-term load forecasting model belongs, and then select the corresponding load forecasting branch in the weekday load forecasting branch, weekend load forecasting branch and holiday load forecasting branch according to the judgment result.

[0066] The ultra-short-term load forecasting model constructed in this embodiment can take into account the load characteristics and patterns in different time periods, improve the accuracy of load forecasting, and provide powerful decision-making support for the operation and scheduling of the power grid and energy management.

[0067] A400: Preset a collaborative control time period window, and perform multi-dimensional data collection according to the collaborative control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array.

[0068] Specifically, in this embodiment, the collaborative control time period window is essentially the interval time period for regulating hydropower and photovoltaic power generation in the energy network. The collaborative control time period window is used to perform historical data of the time interval corresponding to the past collaborative control time period window as reference data for the next collaborative control time period window to regulate hydropower and photovoltaic power generation.

[0069] This embodiment performs multi-dimensional data collection according to the collaborative control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array, wherein the data of the real-time load association array is composed of instantaneous weather temperature and humidity, and non-instantaneous date and time information (such as day of the week, season), the data of the real-time hydropower association array is composed of reservoir water level, inflow, and outflow, and the data of the real-time photovoltaic association array is composed of solar radiation, cloud coverage, temperature, and humidity.

[0070] A500: Synchronize the real-time load association array to the ultra-short-term load forecasting model to obtain real-time load demand.

[0071] The real-time load association array is synchronized to the ultra-short-term load forecasting model. First, the special event judgment engine at the input end of the model selects the correct load forecasting branch according to the collection date of the real-time load association array, and then synchronizes the real-time load association array to the corresponding load forecasting branch to perform the real-time load demand for the entire time interval corresponding to the collaborative control period window.

[0072] A600: Allocate power generation resources according to the real-time load demand, and use the real-time hydropower association array, real-time photovoltaic association array and power generation operation prediction model to perform allocation evaluation to obtain coordinated power generation operation settings.

[0073] In one embodiment, power generation resources are allocated according to the real-time load demand, and the real-time hydropower association array, the real-time photovoltaic association array and the power generation operation prediction model are used to perform allocation evaluation to obtain a coordinated power generation operation setting. The method step A600 provided by the present invention also includes:

[0074] A610: Taking 1 / 2 of the real-time load demand as the optimization starting point for the allocation of power generation resources, and taking 1 / K of the real-time load demand as the adjustment step for the allocation of power generation resources.

[0075] A620: After combining the optimization starting point with the real-time hydropower associated array and the real-time photovoltaic associated array respectively, the optimization starting point is mapped and synchronized to the hydraulic prediction branch and the photovoltaic prediction branch of the power generation operation prediction model to obtain the first hydraulic operation setting and the first photovoltaic operation setting.

[0076] A630: Interacting with the energy network to obtain a standard hydraulic operation setting, and calculating a first deviation between the first hydraulic operation setting and the standard hydraulic operation setting.

[0077] A640: Interacting with the energy network to obtain a standard photovoltaic operation setting, and calculating a second deviation between the first photovoltaic operation setting and the standard photovoltaic operation setting.

[0078] A650: Preset a first deviation threshold and a second deviation threshold. If the first deviation and the second deviation both satisfy the first deviation threshold and the second deviation threshold, use the first hydraulic operation setting and the first photovoltaic operation setting as the collaborative power generation operation setting.

[0079] A660: If the first deviation does not satisfy the first deviation threshold and / or the second deviation does not satisfy the second deviation threshold, update the optimization starting point according to the adjustment step to obtain a first updated resource allocation.

[0080] A670: Use the power generation operation prediction model, the first deviation threshold and the second deviation threshold to perform allocation evaluation of the first updated resource allocation until the collaborative power generation operation setting is obtained.

[0081] Specifically, it should be understood that, in this embodiment, the allocation of power generation resources is essentially to allocate the power load demand of the target power supply area to different power generation resources, such as hydropower and photovoltaic power generation.

[0082] The method for allocating power generation resources in this embodiment is as follows: 1 / 2 of the real-time load demand is used as the optimization starting point for allocating power generation resources to ensure that at least half of the load demand is met. 1 / K (K is a positive integer, usually greater than 50) of the real-time load demand is used as the adjustment step of allocating power generation resources, and the specific value of K depends on the need for fine adjustment.

[0083] The optimization starting point (1 / 2 of the real-time load demand) is combined with the real-time hydropower association array and then synchronized to the hydraulic prediction branch of the power generation operation prediction model to analyze and predict the operating parameters of the hydropower equipment to obtain the first hydraulic operation setting. It should be understood that under the hydropower generation state of the real-time hydropower association array, the first hydraulic operation setting is used to control the hydropower generation process, so as to obtain the real-time power generation that meets 1 / 2 of the real-time load demand.

[0084] Similarly, after combining the data of the optimization starting point and the real-time photovoltaic associated array, the data is synchronized to the photovoltaic prediction branch of the power generation operation prediction model to obtain the first photovoltaic operation setting. Under the lighting environment of the real-time photovoltaic associated array, the actual power generation of the first photovoltaic operation setting meets 1 / 2 of the real-time load demand.

[0085] Next, the energy network is interacted to obtain a standard hydraulic operation setting, and a first deviation between the first hydraulic operation setting and the standard hydraulic operation setting is calculated based on the Euclidean distance. The energy network is interacted to obtain a standard photovoltaic operation setting, and a second deviation between the first photovoltaic operation setting and the standard photovoltaic operation setting is calculated.

[0086] A first deviation threshold and a second deviation threshold are preset. The first deviation threshold is a safety range set for hydropower generation, used to determine whether the first hydraulic operation setting is within an acceptable operating range. The second deviation threshold is a safety range set for photovoltaic power generation, used to determine whether the first photovoltaic operation setting is within an acceptable operating range.

[0087] If the first deviation and the second deviation are both within their respective thresholds, it means that the predicted operation settings both meet the real-time load demand and do not exceed the normal operating range of the equipment.

[0088] If the first deviation and the second deviation both satisfy the first deviation threshold and the second deviation threshold, the first hydraulic operation setting and the first photovoltaic operation setting are used as the cooperative power generation operation setting.

[0089] If the first deviation does not satisfy the first deviation threshold and / or the second deviation does not satisfy the second deviation threshold, this means that the current operating settings may have an adverse effect on the device.

[0090] Based on this, the optimization starting point is updated according to the adjustment step to obtain a first updated resource allocation. Exemplarily, the first updated resource allocation is that hydropower generation occupies 55% of the real-time load demand and thermal power generation occupies 45% of the real-time load demand.

[0091] The updated resource allocation is evaluated by using the power generation operation prediction model in combination with real-time data and the adjusted optimization starting point to obtain the second photovoltaic operation setting and the second hydraulic operation setting. Then, the second photovoltaic operation setting and the second hydraulic operation setting are evaluated by using the standard photovoltaic operation setting, the standard hydraulic operation setting, the first deviation threshold and the second deviation threshold.

[0092] Until the updated photovoltaic operation setting and the updated hydraulic operation setting that meet the threshold requirement are obtained, the updated photovoltaic operation setting and the updated hydraulic operation setting are used as the coordinated power generation operation setting.

[0093] This embodiment achieves the optimal utilization of hydropower and photovoltaic power generation resources in the energy network while ensuring the safety of power generation equipment, enables the energy network to flexibly adapt to changing load demands and environmental conditions, and ensures power generation efficiency and grid stability.

[0094] A700: Taking the collaborative control time period window as a constraint, adopt the collaborative power generation operation setting to perform collaborative control of the energy network.

[0095] Specifically, in this embodiment, the equipment operation parameters of the hydroelectric power generation equipment and the photovoltaic power generation equipment in the energy network are adjusted according to the collaborative power generation operation setting, and the duration of the collaborative power generation operation setting is the corresponding time span of the collaborative control period window.

[0096] This embodiment achieves the technical effect of optimizing the utilization of power generation resources, improving energy efficiency, and reducing operating costs while ensuring power supply safety and reliability based on the energy network through short-term collaborative control of the energy network.

[0097] In one embodiment, the method steps provided by the present invention further include:

[0098] A810: Interact with the energy network to obtain historical electricity price data.

[0099] A820: Split the historical electricity price data to obtain weekday electricity price data, weekend electricity price data and holiday electricity price data.

[0100] A830: Fixing the time intervals for updating the electricity price timestamps of the working day electricity price data, weekend electricity price data and holiday electricity price data, respectively, to obtain multiple groups of working day electricity price fluctuation sequences, multiple groups of weekend electricity price fluctuation sequences and multiple groups of holiday electricity price fluctuation sequences.

[0101] A840: Analyze the electricity price distribution characteristics of the multiple groups of weekday electricity price fluctuation sequences, the multiple groups of weekend electricity price fluctuation sequences, and the multiple groups of holiday electricity price fluctuation sequences, respectively, to obtain a table of period numbers of weekday electricity prices, a table of period numbers of weekend electricity prices, and a table of period numbers of holiday electricity prices.

[0102] A850: Generate a tiered electricity price period table by combining the weekday electricity price period table, weekend electricity price period table and holiday electricity price period table.

[0103] In one embodiment, the method steps provided by the present invention further include:

[0104] A851: interactively obtain the execution time of the collaborative power generation operation setting, and use the execution time to traverse the ladder electricity price time period table to obtain the real-time electricity price.

[0105] A852: Preset the electric energy storage price threshold, and determine whether the real-time electricity price meets the electric energy storage price threshold.

[0106] A853: If the real-time electricity price meets the electricity storage price threshold, a thermal power generation energy storage instruction is generated.

[0107] Specifically, in this embodiment, the energy network is interacted with to collect historical electricity price data within a period of time, and the collected historical electricity price data is classified according to the date type (weekdays, weekends, holidays) to obtain weekday electricity price data, weekend electricity price data, and holiday electricity price data;

[0108] It should be understood that the electricity price data usually carries a timestamp, indicating the specific time point at which the electricity price data is recorded. Based on this, this embodiment fixes the time interval for updating the electricity price timestamps for the working day electricity price data, weekend electricity price data, and holiday electricity price data (for example, every 15 minutes, every hour, etc.), and obtains multiple groups of working day electricity price fluctuation sequences, multiple groups of weekend electricity price fluctuation sequences, and multiple groups of holiday electricity price fluctuation sequences. For example, if the time interval is set to every hour, an electricity value is extracted from the working day electricity price data every hour to form a working day electricity price fluctuation sequence.

[0109] The electricity price distribution of the multiple groups of weekday electricity price fluctuation sequences is analyzed to determine the electricity price levels in different time periods of the day and form a weekday electricity price period table. Similarly, the electricity price distribution characteristics of the multiple groups of weekend electricity price fluctuation sequences and the multiple groups of holiday electricity price fluctuation sequences are analyzed to obtain a weekday electricity price period table, a weekend electricity price period table and a holiday electricity price period table. The weekday electricity price period table, the weekend electricity price period table and the holiday electricity price period table are combined to generate a ladder electricity price period table.

[0110] Interact with the energy network to obtain the specific execution time of the collaborative power generation operation setting, determine whether the execution time is after a weekday / weekend / holiday, and use the execution time to traverse the corresponding table in the tiered electricity price period table to obtain the real-time electricity price, which is the predicted value.

[0111] A preset electric energy storage price threshold is determined, and whether the real-time electricity price satisfies the electric energy storage price threshold. If the real-time electricity price satisfies the electric energy storage price threshold, the current period is a low electricity price period. If there are thermal power generation equipment in the energy network at this time and the thermal power generation equipment is in operation, a thermal power generation storage instruction is generated to prompt the storage of the current electricity generation, thereby improving the economic benefits of electricity production.

[0112] This embodiment achieves the technical effect of optimizing the allocation and use of power resources, achieving cost savings and improving economic benefits by enabling the energy network to flexibly respond to electricity price fluctuations.

[0113] Embodiment 2, based on the same inventive concept as the water-photovoltaic-storage integrated energy network time period coordinated control method in the above embodiment, as Figure 3 As shown, the present invention provides a time period coordinated control device for a water-photovoltaic-storage integrated energy network, wherein the device comprises:

[0114] The energy allocation obtaining unit 11 is used to obtain energy allocation data through an interactive energy network, wherein the energy allocation data includes historical renewable power generation data and historical load demand data.

[0115] The prediction model building unit 12 is used to develop a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch.

[0116] The load forecasting analysis unit 13 is used to pre-build an ultra-short-term load forecasting model, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data.

[0117] The data collection execution unit 14 is used to preset a collaborative control time period window, and perform multi-dimensional data collection according to the collaborative control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array.

[0118] The load demand analysis unit 15 is used to synchronize the real-time load association array to the ultra-short-term load forecasting model to obtain the real-time load demand.

[0119] The collaborative control analysis unit 16 is used to allocate power generation resources according to the real-time load demand, and use the real-time hydropower association array, real-time photovoltaic association array and power generation operation prediction model to perform allocation evaluation to obtain collaborative power generation operation settings.

[0120] The collaborative control execution unit 17 is used to perform collaborative control of the energy network using the collaborative power generation operation setting with the collaborative control time period window as a constraint.

[0121] In one embodiment, the apparatus further comprises:

[0122] The energy network is interacted to obtain historical electricity price data; the historical electricity price data is split to obtain weekday electricity price data, weekend electricity price data and holiday electricity price data; the time intervals for electricity price update timestamps of the weekday electricity price data, weekend electricity price data and holiday electricity price data are fixed to obtain multiple groups of weekday electricity price fluctuation sequences, multiple groups of weekend electricity price fluctuation sequences and multiple groups of holiday electricity price fluctuation sequences; electricity price distribution characteristics are analyzed on the multiple groups of weekday electricity price fluctuation sequences, multiple groups of weekend electricity price fluctuation sequences and multiple groups of holiday electricity price fluctuation sequences to obtain weekday electricity price period tables, weekend electricity price period tables and holiday electricity price period tables; the weekday electricity price period tables, weekend electricity price period tables and holiday electricity price period tables are combined to generate a ladder electricity price period table.

[0123] In one embodiment, the apparatus further comprises:

[0124] Interactively obtain the execution time of the collaborative power generation operation setting, and use the execution time to traverse the ladder electricity price time period table to obtain the real-time electricity price; preset the electric energy storage price threshold, and judge whether the real-time electricity price meets the electric energy storage price threshold; if the real-time electricity price meets the electric energy storage price threshold, generate a thermal power generation energy storage instruction.

[0125] In one embodiment, the prediction model building unit 12 further includes:

[0126] A historical hydroelectric power generation log is obtained based on the historical renewable power generation data, wherein the historical hydroelectric power generation log includes multiple hydroelectric power generation association arrays, multiple hydroelectric power generation records and multiple hydroelectric power generation operation settings; the hydraulic prediction branch is constructed based on a back propagation neural network, and the historical hydroelectric power generation log is used as training data to optimize the prediction accuracy of the hydraulic prediction branch; a historical photovoltaic log is obtained based on the historical renewable power generation data, wherein the historical photovoltaic log includes multiple photovoltaic power generation association arrays, multiple photovoltaic power generation records and multiple photovoltaic power generation operation settings; the photovoltaic prediction branch is constructed based on a back propagation neural network, and the photovoltaic prediction branch is optimized using the historical photovoltaic log; the hydraulic prediction branch and the photovoltaic prediction branch are connected in parallel to complete the construction of the power generation operation prediction model.

[0127] In one embodiment, the prediction model building unit 12 further includes:

[0128] Each of the plurality of hydroelectric power generation association arrays includes a reservoir water level, an inflow flow, and an outflow flow.

[0129] In one embodiment, the prediction model building unit 12 further includes:

[0130] The multiple hydropower generation association arrays, multiple hydropower generation records and multiple hydropower generation operation settings are time-series sorted to obtain an association array sequence, a generation sequence and an operation setting sequence; a verification data splitting ratio is preset, and interval data is extracted from the association array sequence, the generation sequence and the operation setting sequence according to the verification data splitting ratio to obtain a time cross-training set, a time cross-validation set and a time cross-test set; a prediction accuracy constraint is preset, and the time cross-training set, the time cross-validation set and the time cross-test set are used to optimize the prediction accuracy of the hydraulic prediction branch until the prediction accuracy of the hydraulic prediction branch meets the prediction accuracy constraint.

[0131] In one embodiment, the collaborative control analysis unit 16 further includes:

[0132] The historical load demand data is split to obtain workday load demand data, weekend load demand data and holiday load demand data; a load-related index set is preset, and the workday load demand data is split according to the load-related index set to obtain multiple workday power loads and multiple workday load association arrays; a workday load forecasting branch is constructed based on a back propagation neural network, and the multiple workday power loads and multiple workday load association arrays are used as training data to optimize the prediction accuracy of the workday load forecasting branch; by analogy, a weekend load forecasting branch and a holiday load forecasting branch are constructed based on the weekend load demand data and the holiday load demand data; the workday load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch are connected in parallel, and a special event judgment engine is configured at the input end of the workday load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch to complete the construction of the ultra-short-term load forecasting model.

[0133] In one embodiment, the collaborative control analysis unit 16 further includes:

[0134] 1 / 2 of the real-time load demand is used as the optimization starting point of power generation resource allocation, and 1 / K of the real-time load demand is used as the adjustment step of power generation resource allocation; after combining the optimization starting point with the real-time hydropower association array and the real-time photovoltaic association array, the optimization starting point is mapped and synchronized to the hydraulic prediction branch and the photovoltaic prediction branch of the power generation operation prediction model to obtain a first hydraulic operation setting and a first photovoltaic operation setting; the energy network is interacted to obtain a standard hydraulic operation setting, and a first deviation between the first hydraulic operation setting and the standard hydraulic operation setting is calculated; the energy network is interacted to obtain a standard photovoltaic operation setting, and a second deviation between the first photovoltaic operation setting and the standard photovoltaic operation setting is calculated; a first deviation threshold and a second deviation threshold are preset, and if the first deviation and the second deviation both meet the first deviation threshold and the second deviation threshold, the first hydraulic operation setting and the first photovoltaic operation setting are used as the coordinated power generation operation setting; if the first deviation does not meet the first deviation threshold and / or the second deviation does not meet the second deviation threshold, the optimization starting point is updated according to the adjustment step to obtain a first updated resource allocation; the power generation operation prediction model, the first deviation threshold and the second deviation threshold are used to perform allocation evaluation of the first updated resource allocation until the coordinated power generation operation setting is obtained.

[0135] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0136] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the patent protection scope of the present invention.

Claims

1. A time period coordinated control method for an integrated water, light and storage energy network, characterized in that: The method comprises: The interactive energy network obtains energy allocation data, wherein the energy allocation data includes historical renewable power generation data and historical load demand data; Developing a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch; Pre-constructing an ultra-short-term load forecasting model, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data; Preset a collaborative control time window, and perform multi-dimensional data collection according to the collaborative control time window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array; Synchronizing the real-time load association array to the ultra-short-term load forecasting model to obtain real-time load demand; Allocating power generation resources according to the real-time load demand, and using the real-time hydropower association array, the real-time photovoltaic association array and the power generation operation prediction model to perform allocation evaluation to obtain a coordinated power generation operation setting; Taking the collaborative control time window as a constraint, adopting the collaborative power generation operation setting to perform collaborative control of the energy network; Pre-constructing an ultra-short-term load forecasting model, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data, including: Splitting the historical load demand data to obtain workday load demand data, weekend load demand data and holiday load demand data; Preset a load-related indicator set, and split the working day load demand data according to the load-related indicator set to obtain multiple working day power loads and multiple working day load association arrays; Constructing a workday load forecasting branch based on a back propagation neural network, and optimizing the forecasting accuracy of the workday load forecasting branch by using the multiple workday power loads and multiple workday load association arrays as training data; By analogy, a weekend load forecasting branch and a holiday load forecasting branch are constructed based on the weekend load demand data and the holiday load demand data; The working day load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch are connected in parallel, and a special event judgment engine is configured at the input ends of the working day load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch to complete the construction of the ultra-short-term load forecasting model; The power generation resources are allocated according to the real-time load demand, and the real-time hydropower association array, the real-time photovoltaic association array and the power generation operation prediction model are used to perform allocation evaluation to obtain the coordinated power generation operation setting, including: Taking 1 / 2 of the real-time load demand as the optimization starting point for power generation resource allocation, and taking 1 / K of the real-time load demand as the adjustment step for power generation resource allocation; After combining the optimization starting point with the real-time hydropower association array and the real-time photovoltaic association array, the optimization starting point is mapped and synchronized to the hydraulic prediction branch and the photovoltaic prediction branch of the power generation operation prediction model to obtain the first hydraulic operation setting and the first photovoltaic operation setting; Interacting with the energy network to obtain a standard hydraulic operation setting, and calculating a first deviation between the first hydraulic operation setting and the standard hydraulic operation setting; Interacting with the energy network to obtain a standard photovoltaic operation setting, and calculating a second deviation between the first photovoltaic operation setting and the standard photovoltaic operation setting; Preset a first deviation threshold and a second deviation threshold, and if the first deviation and the second deviation both meet the first deviation threshold and the second deviation threshold, use the first hydraulic operation setting and the first photovoltaic operation setting as the coordinated power generation operation setting; If the first deviation does not satisfy the first deviation threshold and / or the second deviation does not satisfy the second deviation threshold, updating the optimization starting point according to the adjustment step size to obtain a first updated resource allocation; The power generation operation prediction model, the first deviation threshold and the second deviation threshold are used to perform allocation evaluation of the first updated resource allocation until the coordinated power generation operation setting is obtained.

2. The time period coordinated control method of the integrated water, photovoltaic and storage energy network according to claim 1 is characterized in that: The method further comprises: Interacting with the energy network to obtain historical electricity price data; Splitting the historical electricity price data to obtain weekday electricity price data, weekend electricity price data and holiday electricity price data; The time intervals for updating the electricity price timestamps of the working day electricity price data, the weekend electricity price data and the holiday electricity price data are fixed to obtain multiple groups of working day electricity price fluctuation sequences, multiple groups of weekend electricity price fluctuation sequences and multiple groups of holiday electricity price fluctuation sequences; Performing electricity price distribution characteristic analysis on the multiple groups of weekday electricity price fluctuation sequences, the multiple groups of weekend electricity price fluctuation sequences, and the multiple groups of holiday electricity price fluctuation sequences, respectively, to obtain a table of weekday electricity price time periods, a table of weekend electricity price time periods, and a table of holiday electricity price time periods; The weekday electricity price period table, weekend electricity price period table and holiday electricity price period table are combined to generate a tiered electricity price period table.

3. The time period coordinated control method of the integrated water, photovoltaic and storage energy network according to claim 2 is characterized in that: The method further comprises: Interactively obtain the execution time of the collaborative power generation operation setting, and use the execution time to traverse the ladder electricity price time period table to obtain the real-time electricity price; Preset an electric energy storage price threshold, and determine whether the real-time electricity price meets the electric energy storage price threshold; If the real-time electricity price meets the electric energy storage price threshold, a thermal power generation energy storage instruction is generated.

4. The time period coordinated control method of the integrated water, photovoltaic and storage energy network according to claim 1 is characterized in that: Developing a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch, and the method further includes: Extracting and obtaining a historical hydroelectric power generation log based on the historical renewable power generation data, wherein the historical hydroelectric power generation log includes a plurality of hydroelectric power generation association arrays, a plurality of hydroelectric power generation records, and a plurality of hydroelectric power generation operation settings; Constructing the hydraulic prediction branch based on a back propagation neural network, and optimizing the prediction accuracy of the hydraulic prediction branch by using the historical hydropower generation log as training data; Extracting and obtaining a historical photovoltaic log based on the historical renewable power generation data, wherein the historical photovoltaic log includes a plurality of photovoltaic power generation association arrays, a plurality of photovoltaic power generation records, and a plurality of photovoltaic power generation operation settings; Constructing the photovoltaic prediction branch based on a back propagation neural network, and optimizing the photovoltaic prediction branch using the historical photovoltaic log; The hydraulic prediction branch and the photovoltaic prediction branch are connected in parallel to complete the construction of the power generation operation prediction model.

5. The time period coordinated control method of the integrated water, photovoltaic and storage energy network according to claim 4 is characterized in that: Each of the plurality of hydroelectric power generation association arrays includes a reservoir water level, an inflow flow, and an outflow flow.

6. The time period coordinated control method of the integrated water, photovoltaic and storage energy network according to claim 5 is characterized in that: The hydraulic prediction branch is constructed based on a back propagation neural network, and the historical hydropower generation log is used as training data to optimize the prediction accuracy of the hydraulic prediction branch. The method further includes: Sequentially sorting the plurality of hydroelectric generation association arrays, the plurality of hydroelectric generation records, and the plurality of hydroelectric generation operation settings to obtain an association array sequence, a generation sequence, and an operation setting sequence; Preset the verification data split ratio, and perform interval data extraction on the associative array sequence, the power generation sequence and the operation setting sequence according to the verification data split ratio to obtain a time crossover training set, a time crossover validation set and a time crossover test set; A prediction accuracy constraint is preset, and the time cross-training set, the time cross-validation set and the time cross-test set are used to optimize the prediction accuracy of the hydraulic prediction branch until the prediction accuracy of the hydraulic prediction branch meets the prediction accuracy constraint.

7. A time period coordinated control device for an integrated water, light and storage energy network, characterized in that: The device is used to implement the time period coordinated control method of the integrated water, photovoltaic and storage energy network according to any one of claims 1 to 6, and comprises: An energy allocation acquisition unit, configured to acquire energy allocation data through an interactive energy network, wherein the energy allocation data includes historical renewable power generation data and historical load demand data; A prediction model building unit, used to develop a power generation operation prediction model based on the historical renewable power generation data, wherein the power generation operation prediction model includes a hydraulic prediction branch and a photovoltaic prediction branch; A load forecasting analysis unit, used for pre-building an ultra-short-term load forecasting model, wherein the ultra-short-term load forecasting model uses the historical load demand data as construction data; A data acquisition execution unit, used for presetting a coordinated control time period window, and performing multi-dimensional data acquisition according to the coordinated control time period window to obtain a real-time load association array, a real-time hydropower association array, and a real-time photovoltaic association array; A load demand analysis unit, used for synchronizing the real-time load association array to the ultra-short-term load forecasting model to obtain real-time load demand; A collaborative control analysis unit, configured to allocate power generation resources according to the real-time load demand, and use the real-time hydropower association array, the real-time photovoltaic association array and the power generation operation prediction model to perform allocation evaluation to obtain collaborative power generation operation settings; A collaborative control execution unit, configured to perform collaborative control of the energy network using the collaborative power generation operation setting with the collaborative control time window as a constraint; The load forecasting and analysis unit includes: splitting the historical load demand data to obtain workday load demand data, weekend load demand data and holiday load demand data; presetting a load-related index set, and splitting the workday load demand data according to the load-related index set to obtain multiple workday power loads and multiple workday load association arrays; constructing a workday load forecasting branch based on a back propagation neural network, and using the multiple workday power loads and multiple workday load association arrays as training data to optimize the prediction accuracy of the workday load forecasting branch; and so on, constructing and generating a weekend load forecasting branch and a holiday load forecasting branch based on the weekend load demand data and the holiday load demand data; connecting the workday load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch in parallel, and configuring a special event judgment engine at the input end of the workday load forecasting branch, the weekend load forecasting branch and the holiday load forecasting branch to complete the construction of the ultra-short-term load forecasting model; The collaborative control analysis unit includes: taking 1 / 2 of the real-time load demand as the optimization starting point for power generation resource allocation, and taking 1 / K of the real-time load demand as the adjustment step for power generation resource allocation; after combining the optimization starting point with the real-time hydropower association array and the real-time photovoltaic association array, mapping and synchronizing to the hydraulic prediction branch and the photovoltaic prediction branch of the power generation operation prediction model, obtaining the first hydraulic operation setting and the first photovoltaic operation setting; interacting with the energy network to obtain the standard hydraulic operation setting, and calculating the first deviation between the first hydraulic operation setting and the standard hydraulic operation setting; interacting with the energy network to obtain the standard photovoltaic operation setting, and calculating a second deviation between the first photovoltaic operation setting and the standard photovoltaic operation setting; a first deviation threshold and a second deviation threshold are preset, and if the first deviation and the second deviation both satisfy the first deviation threshold and the second deviation threshold, the first hydraulic operation setting and the first photovoltaic operation setting are used as the collaborative power generation operation setting; if the first deviation does not satisfy the first deviation threshold and / or the second deviation does not satisfy the second deviation threshold, the optimization starting point is updated according to the adjustment step to obtain a first updated resource allocation; the power generation operation prediction model, the first deviation threshold and the second deviation threshold are used to perform an allocation evaluation of the first updated resource allocation until the collaborative power generation operation setting is obtained.