A method and system for coordinated operation of pumped storage power station groups

Abstract

This application discloses a collaborative operation method applicable to pumped storage power station groups, relating to the field of power station group collaborative operation. The method includes: collecting regional power grid data and pumped storage unit data; constructing an effective power grid operation matrix and performing full-time-domain sensitivity calculations for the pumped storage power station group to obtain a spatiotemporal sensitivity matrix; constructing a hydropower value model; anchoring the dispatch value of the pumped storage power station group to obtain the power station group value anchor point; prioritizing the pumped storage power stations within the pumped storage power station group to obtain a pumped storage unit priority sequence; selecting target dispatch power stations and backup dispatch power stations from the pumped storage power station group; calculating the stable dispatch power ratio; and combining the stable dispatch power ratio and regional power grid data to generate a collaborative operation strategy for the pumped storage power station group. This application can effectively improve the stability of collaborative operation of pumped storage power station groups.

Description

Technical Field

[0001] This application relates to the field of collaborative operation of power plant groups, and in particular to a collaborative operation method and system applicable to pumped storage power plant groups. Background Technology

[0002] With the continuous and in-depth advancement of the construction of my country's new power system, regional power grids have formed a new operating pattern characterized by a high proportion of new energy penetration and dense ultra-high voltage direct current (UHVDC) transmission. On the one hand, the continuous expansion of wind power and photovoltaic installed capacity, along with their intermittent, fluctuating, and even counter-regulation characteristics, has significantly increased the difficulty of predicting the net load curve of the power grid. On the other hand, the dense inflow and regular operation of multiple UHVDC transmission lines, while enhancing the cross-regional resource allocation capacity, has also made the voltage stability problem at key points increasingly prominent against the backdrop of increased maintenance methods and greater difficulty in coordination and dispatch.

[0003] Against this backdrop, pumped storage power stations, with their unique functional positioning, undertake the fundamental tasks of peak shaving and valley filling and promoting the consumption of new energy sources, while also shouldering the advanced mission of providing stable services such as inertia support and reactive power regulation. However, the current operation of pumped storage power station clusters is still mainly based on the independent operation of individual power stations, lacking overall coordination and collaborative planning. Limited by their own boundaries such as installed capacity, reservoir capacity, and unit characteristics, individual pumped storage power stations often cannot fully play their supporting role in scenarios such as ensuring peak supply and emergency support, resulting in the untapped potential for collaborative regulation and stable support capabilities of pumped storage units within the region. Therefore, there is an urgent need for a collaborative operation method for pumped storage power station clusters that addresses core issues such as single-station constraints, resource misallocation, and insufficient coordination. Summary of the Invention

[0004] This application provides a method and system for the coordinated operation of pumped storage power station groups, which is used to improve the stability of the coordinated operation of pumped storage power station groups.

[0005] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: Firstly, a method for the coordinated operation of pumped storage power station groups is provided, the method comprising: Collect regional power grid data and pumped storage unit data of the pumped storage power station group within the target area; The effective operation matrix of the power grid is constructed iteratively based on the regional power grid data. The full-time sensitivity calculation of the pumped storage power station group is completed by combining the effective operation matrix of the power grid and the pumped storage unit data, and the spatiotemporal sensitivity matrix of the pumped storage power station group is obtained. A hydropower value model for pumped storage power station groups was constructed by combining spatiotemporal sensitivity matrices and regional power grid data. Based on the hydropower value model and using a multi-objective optimization algorithm, the scheduling value anchoring of pumped storage power station groups is completed, and the value anchoring point of the power station group is obtained. Based on the value anchor points of the power station group, the priority of pumped storage power stations within the pumped storage power station group is arranged to obtain the priority sequence of pumped storage units. By combining the spatiotemporal sensitivity matrix and the priority sequence of pumped storage units, target dispatching power stations are selected from the pumped storage power station group, and backup dispatching power stations are selected from the pumped storage power station group based on the target dispatching power stations and pumped storage unit data. The stable dispatch power ratio between the backup power station and the target power station is calculated based on the spatiotemporal sensitivity matrix. The stable dispatch power ratio and regional power grid data are then combined to generate a coordinated operation strategy for the pumped storage power station group.

[0006] Optionally, the regional power grid data includes power grid topology data and power grid operation data. The step of iteratively constructing the effective operation matrix of the power grid based on the regional power grid data, and combining the effective operation matrix of the power grid with the pumped storage unit data to complete the full-time domain sensitivity calculation of the pumped storage power station group, and obtaining the spatiotemporal sensitivity matrix of the pumped storage power station group includes the following steps: Based on the regional power grid data, the power balance equation of the regional power grid in the target area is solved iteratively using the power flow algorithm until the power flow of the regional power grid converges, and the effective operation matrix of the regional power grid is output. For any pumped storage unit in any pumped storage power station within a pumped storage power station group, determine the unit control data and unit electrical parameters based on the pumped storage unit data; Key hub nodes of the regional power grid are selected based on regional power grid data, and node operation parameters of key hub nodes are determined based on power grid operation data. The unit operation elements of the pumped storage unit are extracted by inverse transformation of the grid effective operation matrix, and the steady-state sensitivity index of the pumped storage unit is calculated by combining the unit operation elements and the unit electrical parameters. A baseline time-domain simulation of a preset power grid disturbance scenario is performed by combining node operating parameters and unit control data, and the transient sensitivity index of the pumped storage unit is calculated based on the time-domain simulation results. A spatiotemporal sensitivity matrix for pumped storage power station groups is constructed by combining steady-state and transient sensitivity indices.

[0007] Optionally, identifying key hub nodes of the regional power grid based on regional power grid data includes the following steps: The DC converter stations of the regional power grid are located based on the regional power grid data, and the converter station parameters are determined based on the regional power grid data. The converter station parameters include the converter station reactance parameters and the converter station power parameters. The initial node impedance matrix of the regional power grid is constructed based on the regional power grid data. The initial node impedance matrix is ​​then corrected using converter station parameters to obtain the node impedance matrix of the regional power grid. The impedance magnitude of the AC nodes in the regional power grid is calculated based on the node impedance matrix. The voltage interaction factor between the DC converter station and the AC node of the power grid is calculated based on the impedance modulus, and all voltage interaction factors are integrated to construct a voltage interaction factor matrix. Power weights are assigned to DC converter stations based on their power parameters, and the DC commutation failure factors of all AC nodes in the power grid are calculated by combining the power weights and voltage interaction factors. Based on the DC commutation failure factor, all AC nodes of the power grid were screened out as grid risk nodes, and multiple overlapping commutation risk areas within the regional power grid were delineated based on the grid risk nodes and the power grid topology data. The voltage interaction weighting value of all grid risk nodes in the entire commutation risk overlap area is calculated based on the voltage interaction factor matrix and the entropy weight method. Based on the voltage interaction weighting value, multiple power grid risk nodes were selected from all overlapping commutation risk areas as key hub nodes of the regional power grid.

[0008] Optionally, the pumped-storage unit's operating elements are extracted through inverse transformation of the grid's effective operation matrix, and the steady-state sensitivity index of the pumped-storage unit is calculated by combining the operating elements and the unit's electrical parameters, including the following steps: Perform the inverse matrix transformation of the power grid's effective operation matrix to obtain the inverse effective operation matrix; The pumped storage unit's topological location is determined based on the grid topology data, and the corresponding unit operation element is extracted from the effective operation inverse matrix based on the unit topological location. The unit operation element includes the unit's active power operation element and the unit's reactive power operation element. The node voltage parameters of key hub nodes are determined based on the effective operation inverse matrix, and the voltage sensitivity index of pumped storage units is calculated by combining the active power operation elements, reactive power operation elements and node voltage parameters. The voltage sensitivity index includes active power voltage sensitivity and reactive power voltage sensitivity. The initial frequency sensitivity of the pumped storage unit was calculated based on the unit's electrical parameters. Based on the unit topology of the pumped storage unit, the voltage interaction factor at the corresponding position is extracted from the voltage interaction factor matrix, and the initial frequency sensitivity is spatially corrected using the voltage interaction factor to obtain the frequency sensitivity index. By integrating the voltage sensitivity index and the frequency sensitivity index, the steady-state sensitivity index is obtained.

[0009] Optionally, a baseline time-domain simulation of a preset grid disturbance scenario is performed by combining node operating parameters and unit control data, and the transient sensitivity index of the pumped storage unit is calculated based on the time-domain simulation results, including the following steps: By combining node operating parameters and unit control data, a baseline time-domain simulation of a preset power grid disturbance scenario is performed to obtain the baseline frequency trajectory. Extract the baseline trajectory features of the baseline frequency trajectory, which include the baseline frequency change rate and the baseline frequency trough value; The simulation disturbance step size is determined based on the unit's electrical parameters, and a second time-domain simulation is completed based on the simulation disturbance step size. The disturbance trajectory features of the second time-domain simulation results are extracted, including positive disturbance trajectory features and negative disturbance trajectory features. The rationality of the disturbance trajectory characteristics is verified based on the reference frequency trajectory. If the rationality verification of the disturbance trajectory characteristics passes, the transient sensitivity index of the pumped storage unit is calculated based on the disturbance trajectory characteristics and using the central difference formula.

[0010] Optionally, constructing a hydropower value model for a pumped storage power station group by combining the spatiotemporal sensitivity matrix and regional power grid data includes the following steps: Collect real-time reservoir parameters of the upper and lower reservoirs corresponding to the pumped storage power station group. The real-time reservoir parameters include the real-time water level and the water level difference between the reservoirs. The effective capacity of the reservoir is calculated based on the real-time water level and the pre-acquired water level-capacity curve. The effective capacity of the reservoir includes the effective capacity of the upper reservoir and the effective capacity of the lower reservoir. Calculate the capacity difference between the effective capacity of the upper reservoir and the effective capacity of the lower reservoir, and determine the dispatchable capacity of the upper and lower reservoirs based on the capacity difference. The dispatchable capacity includes the capacity for power generation and the capacity for pumping. The operating conditions of pumped storage power stations in the pumped storage power station group are determined based on power grid operation data, and the water resource penalty factor is calculated by combining the dispatchable reservoir capacity and the operating conditions of pumped storage power stations. The total power grid load of the target area is calculated based on the power grid operation data, and the active power load gap of the target area is calculated based on the total power grid load and the output data of other energy plans obtained in advance. Sensitivity weights are assigned to pumped storage power station groups based on the spatiotemporal sensitivity matrix, and a hydropower value model is constructed by combining the pumped storage power station groups, water resource penalty factors, active power load gap, and dispatchable reservoir capacity.

[0011] Optionally, selecting backup power stations from the pumped-storage power station group based on the target dispatch power station and pumped-storage unit data includes the following steps: The key scheduling attributes of the target power station are extracted from pumped storage unit data and power grid topology data. These key scheduling attributes include power grid zoning attributes and power station basic attributes. Candidate backup power stations are selected from the pumped storage power station group based on the power grid zoning attributes and pumped storage unit priority sequence. Extract key power plant attributes of candidate backup power plants from pumped storage unit data; The overall sensitivity of the target dispatch power station and the candidate backup power station is calculated based on the spatiotemporal sensitivity matrix. After vectorizing the key attributes of scheduling, key attributes of power plants, and the corresponding comprehensive sensitivity of power plants, the vectorized data are input into a clustering algorithm to perform feature clustering, resulting in multiple feature clusters. Filter out the list of power stations within all feature clusters, and mark the list of power stations within the cluster that contains the target dispatch power station as the target power station list; Mark the candidate backup power stations in the target power station list as scheduled backup power stations.

[0012] Optionally, the stable dispatch power ratio between the backup power station and the target power station is calculated based on the spatiotemporal sensitivity matrix, and a coordinated operation strategy for the pumped storage power station group is generated by combining the stable dispatch power ratio and regional power grid data, including the following steps: The ratio of the power station sensitivity of the backup power station to the target power station is calculated based on the spatiotemporal sensitivity matrix to obtain the stable dispatch power ratio. The synchronous compensation power of the backup power station is calculated based on the stable dispatch power ratio; The power grid operation characteristics of the regional power grid are extracted based on the power grid operation data, and the power grid operating conditions of the regional power grid are determined based on the power grid operation characteristics. Based on the power grid operation data, the key operating parameters of all critical hub nodes in the regional power grid are verified point by point in time, and the power grid operation risk level of the regional power grid is determined based on the verification results of the key operating parameters. Determine the grid peak-shaving demand of the regional power grid based on the active power load gap; A coordinated operation strategy for pumped storage power stations is generated by combining synchronous compensation power, grid operating conditions, grid operation risk level, and grid peak-shaving demand.

[0013] In a second aspect, this application provides a machine-readable storage medium storing instructions for causing a machine to perform a cooperative operation method applicable to a pumped storage power station group as described in the first aspect.

[0014] Thirdly, this application provides a collaborative operation system suitable for pumped storage power station groups, comprising: The memory is configured to store instructions; and The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the coordinated operation method applicable to pumped storage power plant groups as described in the first aspect.

[0015] Through the aforementioned technical solutions, by completing the full-time-domain sensitivity calculation of pumped-storage power station groups and constructing a spatiotemporal sensitivity matrix for these groups, the real-time operating characteristics of the regional power grid and the quantification of the grid security and peak-shaving regulation efficiency of different generating units are achieved. Simultaneously, transient and steady-state sensitivity indicators are analyzed, overcoming the limitations of traditional pumped-storage dispatching that only focuses on steady-state power flow while neglecting transient security. This achieves a dispatching strategy that simultaneously considers both steady-state grid operation and transient security, avoiding the industry pain point of satisfying peak-shaving needs but triggering grid stability risks. Furthermore, by constructing a hydropower value model for pumped-storage power station groups, the core scarce resource of hydropower is quantitatively valued, strongly binding hydropower value with grid peak-shaving needs, security support needs, and generating unit regulation sensitivity. This enables the allocation of hydropower resources towards maximizing value, fundamentally solving the problem of hydropower resource misallocation in traditional dispatching. By anchoring the dispatching value of pumped-storage power station groups, an optimal value benchmark is provided for subsequent power station priority ranking and target power station selection. Prioritizing pumped storage power stations at the unit level provides a clear selection order for subsequent target pumped storage power station selection. This prioritization is based entirely on quantified power station group value anchors, rather than human experience, ensuring objectivity, fairness, and optimality, and avoiding resource misallocation caused by human intervention. The selection of target dispatch power stations combines priority sequences and spatiotemporal sensitivity matrices to ensure that the selected target dispatch power stations are the most valuable and have the strongest regulation efficiency, achieving the maximum dispatch target with minimal hydropower consumption and significantly improving dispatch efficiency. The quantitative calculation results from all preceding steps are ultimately transformed into an executable power station group collaborative dispatch strategy. This strategy clarifies the output plans, operating condition adjustments, and synchronization methods of target and backup dispatch power stations, ensuring that the generated strategy meets power dispatch management requirements and can be safely and stably implemented. This improves the collaborative operational stability of the pumped storage power station group, thereby ensuring the stable operation of the regional power grid and the stability of power supply.

[0016] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a collaborative operation method for a pumped storage power station group, provided as an embodiment of this application; Figure 2 A flowchart illustrating a spatiotemporal sensitivity matrix construction method provided in this application embodiment; Figure 3 This is a flowchart illustrating a transient sensitivity index calculation method provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0019] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0020] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0021] Figure 1 The illustration schematically shows a process flow diagram of a cooperative operation method for a pumped storage power station group according to an embodiment of this application. Figure 1 As shown in the figure, this application provides a method for the coordinated operation of pumped storage power station groups, which may include the following steps: S101. Collect regional power grid data and pumped storage unit data of the pumped storage power station group within the target area.

[0022] In this embodiment, regional power grid data includes power grid topology data and power grid operation data. Power grid topology data includes the topological connections and voltage level information of all power plants, lines, and busbars within the regional power grid. Power grid operation data includes equipment information of receiving-end converter stations for all high-voltage / ultra-high-voltage direct current transmission projects within the regional power grid. This data can be collected from specialized systems such as the Energy Management System (EMS), Wide Area Phasor Measurement System (WAMS / PMU), New Energy Power Prediction System, and Operational Management System (OMS) of the target region's power grid dispatching agency. Pumped storage unit data refers to the relevant data of all pumped storage units contained in each pumped storage power station within a pumped storage power station group. This includes unit control data, unit electrical parameters, etc., and can be collected from the computer monitoring and control system (SCADA), automatic hydrological monitoring system, unit excitation / speed control system, and power station production management system of each pumped storage power station.

[0023] S102. Based on the regional power grid data, the effective operation matrix of the power grid is iteratively constructed. Combined with the effective operation matrix of the power grid and the pumped storage unit data, the full-time domain sensitivity calculation of the pumped storage power station group is completed, and the spatiotemporal sensitivity matrix of the pumped storage power station group is obtained.

[0024] In this embodiment, power balance equations for each node of the power system in polar coordinates are first established based on grid topology data and grid operation data. Then, an improved Newton-Raphson method is used for iterative solution. After the power flow iteration converges, the core matrix elements that converged in this iteration are extracted based on the converged steady-state operating state of the grid to construct the effective grid operation matrix. Specifically, the extracted content is the Newton-Raphson Fakubi matrix at the convergence moment, which serves as the core of the effective grid operation matrix. The unit control data and electrical parameters of all pumped-storage units are extracted from the pumped-storage unit data. The unit control data includes the speed regulation system droop coefficient, excitation regulation intensity, and speed control pipeline response speed. The unit electrical parameters include the unit electrical parameters. Next, key hub nodes of the regional grid are selected based on regional grid data. These key hub nodes have high voltage cross-weighting values, are sensitive to the risk of multiple DC commutation failures, and play a decisive role in the voltage stability of the regional grid. The node operation parameters of the key hub nodes include the steady-state voltage amplitude, phase angle, and other parameters. The effective operation matrix of the power grid is inverted using the LU decomposition method to obtain the effective operation inverse matrix. The topological location of each pumped-storage unit is determined from the power grid topology data. Based on the unit's topological location, the active and reactive power operating elements corresponding to the pumped-storage unit are extracted from the effective operation inverse matrix. The active and reactive voltage sensitivities of each pumped-storage unit are calculated by combining these elements. Next, the initial frequency sensitivity of the pumped-storage unit is calculated based on its electrical parameters, yielding a frequency sensitivity index. Finally, the steady-state sensitivity index is obtained by integrating the voltage and frequency sensitivity indices.

[0025] Next, transient sensitivity indices were calculated through multiple time-domain simulations. These indices include RoCoF suppression sensitivity and frequency minimum point support sensitivity. RoCoF suppression sensitivity reflects the pumped-storage unit's ability to suppress the rate of frequency change during the initial stage of a fault, while frequency minimum point support sensitivity reflects the pumped-storage unit's ability to improve the depth of frequency drop after a fault. The spatiotemporal sensitivity matrix of the pumped-storage power station group was constructed by integrating the steady-state and transient sensitivity indices.

[0026] S103. A hydropower value model for pumped storage power station groups is constructed by combining the spatiotemporal sensitivity matrix and regional power grid data.

[0027] In this embodiment, sensitivity weights are assigned to the pumped-storage power station group based on the spatiotemporal sensitivity matrix, and a hydropower value model is constructed by combining water resource penalty factors, active power load gaps, and dispatchable reservoir capacity. The hydropower value model uses the stable contribution of pumped-storage units to the power grid as the core weight of hydropower value, while also incorporating water resource scarcity and power grid supply and demand. It realizes the core logic that the higher the stable contribution of pumped-storage hydropower resources, the scarcer they are, and the more needed they are by the power grid, the higher their value. This provides a precise value benchmark for the coordinated operation of pumped-storage power station groups, and can fully tap the stability support potential and hydropower utilization efficiency of the pumped-storage power station group.

[0028] S104. Based on the hydropower value model and using a multi-objective optimization algorithm, the scheduling value anchoring of the pumped storage power station group is completed, and the value anchor point of the power station group is obtained.

[0029] In this embodiment, the core decision variables are first determined. Within the scheduling cycle, which can be set as 24 hours prior to the current day, and 96 15-minute time periods, the active power output of each pumped-storage power station in the power station group in each time period is taken as the continuous core decision variable. At the same time, the operating conditions of the pumped-storage power stations in the corresponding time periods of each power station are taken as discrete auxiliary decision variables. The dimensions of the decision variables are fully matched with the number of time periods and the number of pumped-storage power stations in the scheduling cycle, fully covering the scheduling decision space of the power station group in the entire time domain and for the entire subject. Next, a hierarchical multi-objective optimization function is constructed, with the hydropower value model as the core. The corresponding optimization objectives are set according to the actual application scenario: the first optimization objective is to maximize the total hydropower value of the pumped-storage power station group in the entire time domain, and the second optimization objective is to maximize the renewable energy absorption rate of the regional power grid, that is, to minimize the renewable energy curtailment within the scheduling cycle. This can be achieved by optimizing the timing of the pumping conditions of the pumped-storage power stations to maximize the absorption of surplus electricity during periods of high renewable energy generation and minimize renewable energy curtailment, thereby promoting renewable energy absorption. The third optimization objective is to minimize the risk of power grid operation across the entire time domain. The optimized pumped storage power output plan must meet the power flow convergence requirements of the regional power grid. The voltage of key hub nodes and the power grid frequency must be within the rated safe operating range. The fourth optimization objective is to constrain reservoir capacity. During the scheduling cycle, the water levels of the upper and lower reservoirs of each pumped storage power station must not exceed the safety boundaries between the dead water level and the normal storage water level. The cumulative value of water pumped out in each time period must not exceed the upper limit of the corresponding dispatchable reservoir capacity.

[0030] To address the aforementioned multi-objective optimization problem, a multi-objective optimization algorithm is employed. Genetic optimization or particle swarm optimization can be used. Algorithm parameters are configured based on the pumped storage power station cluster size, the number of scheduling periods, and the dimensions of decision variables. Population size, number of iterations, and convergence threshold are set. The hydropower value model serves as the core of the multi-objective optimization algorithm's fitness calculation. All rigid constraints are used as the algorithm's penalty function boundaries, penalizing individuals that do not meet the constraints and reducing their fitness to ensure the solution process remains within the constraint boundaries. The algorithm iteratively solves the problem, repeatedly executing population initialization, individual fitness calculation, non-dominated sorting, elite retention, and population update operations until the preset maximum number of iterations is reached or the convergence threshold is met. The iteration then terminates, ultimately yielding a Pareto optimal solution set for the pumped storage power station cluster scheduling optimization. Each individual in the solution set corresponds to a full-time-domain scheduling plan for the pumped storage power station cluster that satisfies all constraints.

[0031] Next, the entropy weight method is used to screen the Pareto optimal solution set. Specifically, the numerical values ​​of each optimization objective for each individual in the Pareto optimal solution set are normalized to eliminate the dimensional differences between different objectives. Benefit-type objectives are positively normalized, while cost-type objectives are negatively normalized, ensuring all indicators fall within a comparable range of 0 to 1. Then, the objective weights of each optimization objective are calculated based on the entropy weight method. By calculating the information entropy of each optimization objective, the dispersion and effective information content of the objective are quantified. Objectives with higher dispersion carry more effective information and are assigned higher weights. Finally, based on the weighted standardized objective matrix, the positive and negative ideal solutions for each optimization objective are determined. The Euclidean distance from each individual in the solution set to the positive and negative ideal solutions is calculated as the relative proximity of each individual. The individual with the highest relative proximity is selected as the global optimal compromise solution for the pumped storage power station group scheduling optimization.

[0032] The value anchoring of pumped-storage power station group scheduling is completed based on the globally optimal compromise solution. The global value anchoring point and the individual power station scheduling value anchoring point are quantified separately, forming the power station group value anchoring point. First, the global value anchoring point of the power station group is determined by extracting three core benchmark values ​​from the globally optimal compromise solution. The first core benchmark is the optimal benchmark value of the total hydropower value of the pumped-storage power station group across the entire time domain, i.e., the maximum value of the total hydropower value of the power station group within the scheduling period corresponding to the optimal solution, serving as the global anchoring benchmark for the overall scheduling value of the power station group. The second core benchmark is the time-series distribution benchmark of the hydropower value across the entire time domain, i.e., the optimal value of the total hydropower value of the power station group in each scheduling period of the optimal solution, serving as the time-series anchoring benchmark for the scheduling value in different periods. The third core benchmark is the optimal benchmark of the comprehensive efficiency of the power station group, i.e., the optimal combination value of the new energy absorption rate, grid operation risk, and unit operation loss corresponding to the optimal solution, serving as the efficiency boundary anchoring point for the scheduling value of the power station group. Then, the above three core benchmark values ​​are normalized and weighted summed to obtain the global value anchoring point of the power station group. The weights can be determined using the entropy weight method or the multi-level analysis method. Next, based on the hydropower value model, the hydropower value of each pumped-storage power station is accumulated across different time periods to obtain the total hydropower value of the pumped-storage power station. Then, the ratio of the total hydropower value of the pumped-storage power station to the optimal benchmark value of the total hydropower value over the entire time domain is calculated to obtain the value proportion of a single power station. Next, the dispatchable reservoir capacity contribution of the pumped-storage power station is calculated: Dispatchable reservoir capacity contribution = Power generation capacity + Pumped reservoir capacity / Total dispatchable reservoir capacity of the power station group. The overall sensitivity of the pumped-storage power station and the value proportion of the single power station are weighted and summed to obtain the dispatch value anchor point of the single power station. The weights can be determined using the entropy weight method. Integrating the dispatch value anchor point of the single power station and the global value anchor point of the power station group, the value anchor point of the power station group is obtained. This lays the data foundation for the subsequent priority ranking of pumped-storage power stations within the pumped-storage power station group.

[0033] S105. Based on the value anchor point of the power station group, complete the priority arrangement of pumped storage power stations within the pumped storage power station group to obtain the priority sequence of pumped storage units.

[0034] In this embodiment, pumped storage power stations are first sorted in descending order based on the single power station scheduling value anchor point to obtain a priority sequence for pumped storage power stations. Then, the pumped storage units within each pumped storage power station are prioritized, with the order determined by a combination of the rated capacity and ramp rate of each unit. The rated capacity refers to the maximum active power output / input capability that, under specified design and manufacturing conditions and conforming to rated operating conditions, can operate safely and stably for a long period; it is a benchmark parameter for measuring the core operating capability of the unit. After normalizing the rated capacity and ramp rate, a weighted sum is calculated, with each weight potentially being 0.5, to obtain the comprehensive performance parameters of the unit. Based on these comprehensive performance parameters, the pumped storage units within the pumped storage power station are sorted in descending order to obtain a single unit priority sequence. This single unit priority sequence is then inserted into the pumped storage power station priority sequence to obtain the pumped storage unit priority sequence.

[0035] S106. Combine the spatiotemporal sensitivity matrix and the pumped storage unit priority sequence to select the target dispatching power station from the pumped storage power station group, and select the dispatch backup power station from the pumped storage power station group based on the target dispatching power station and pumped storage unit data.

[0036] In this embodiment, steady-state and transient sensitivity indices for all key hub nodes are extracted from the spatiotemporal sensitivity matrix, and their arithmetic averages are calculated to obtain the overall steady-state and transient sensitivity of a single power station. The weighting ratio is determined based on the current operating conditions of the power grid: 60% for steady-state comprehensive sensitivity and 40% for transient comprehensive sensitivity under normal operating conditions; and 30% for steady-state comprehensive sensitivity and 70% for transient comprehensive sensitivity under warning or emergency operating conditions. The overall regulation efficiency score of the power station is obtained by weighted summation of the steady-state and transient comprehensive sensitivities based on these weights, quantifying the actual support efficiency of the power station for grid stability. Pumped storage power stations with an overall regulation efficiency score greater than a preset score threshold are selected as candidate dispatch power stations. The candidate dispatch power station ranked first is selected as the target dispatch power station based on the priority sequence of pumped storage units. The score threshold can be determined based on the average overall regulation efficiency score of all pumped storage power stations.

[0037] The entire pumped storage power station group is traversed, and those in the same electrical zone of the power grid as the target dispatching power station and ranking in the top 60% of the pumped storage power station priority sequence are prioritized as candidate backup power stations. Next, key power station attributes of the candidate backup power stations are extracted from the pumped storage unit data. Then, the comprehensive power station sensitivity of the target dispatching power station and the candidate backup power stations is calculated based on the spatiotemporal sensitivity matrix. The comprehensive power station sensitivity and key dispatching attributes of the target dispatching power station are vectorized and concatenated along the same rules to form the target power station vector. Similarly, the comprehensive sensitivity and key power station attributes of the candidate backup power stations are vectorized and concatenated along the same rules to form the candidate power station vector. The target power station vector and the candidate backup power station vector are input into a clustering algorithm for iterative clustering. The algorithm iteratively executes initial cluster center selection, sample Euclidean distance calculation, sample cluster allocation, cluster center update, and convergence judgment operations until the maximum number of iterations is reached or the convergence judgment threshold is met, at which point the iteration terminates, and the initial clustering results are output, resulting in multiple feature clusters. Filter out the list of power stations within all feature clusters, and mark the list of power stations within the cluster that contains the target scheduling power station as the target power station list, and mark the candidate backup power stations in the target power station list as scheduling backup power stations.

[0038] S107. Calculate the stable dispatch power ratio between the backup power station and the target power station based on the spatiotemporal sensitivity matrix, and generate a power station collaborative operation strategy for the pumped storage power station group by combining the stable dispatch power ratio and regional power grid data.

[0039] In this embodiment, the comprehensive sensitivity of the backup power station and the target power station is first calculated based on the spatiotemporal sensitivity matrix. Then, the ratio of their comprehensive sensitivities is calculated to obtain the comprehensive power station sensitivity. The planned regulation power of each target power station is determined based on the active power load gap; it is positive during power generation and negative during pumping operation. Next, the product of the ratio of the planned regulation power to the stable dispatch power is calculated to obtain the synchronous compensation power of the backup power station. Then, the grid operation characteristics of the regional grid are extracted based on grid operation data, and the grid operating conditions are determined based on these characteristics. These operating conditions include normal, early warning, and emergency conditions. Finally, based on the comprehensive power station sensitivity and synchronous compensation power, a coordinated operation strategy for the pumped storage power station group under different grid operating conditions is generated for the backup power station and the target power station.

[0040] In one embodiment, reference is made to Figure 2 The regional power grid data includes power grid topology data and power grid operation data. The step of iteratively constructing the effective operation matrix of the power grid based on the regional power grid data, and combining the effective operation matrix of the power grid with the pumped storage unit data to complete the full-time domain sensitivity calculation of the pumped storage power station group, and obtaining the spatiotemporal sensitivity matrix of the pumped storage power station group includes the following steps: S201. Based on the regional power grid data and using the power flow algorithm, iteratively solve the power balance equation of the regional power grid in the target area until the power flow of the regional power grid converges, and output the effective operation matrix of the regional power grid. S202. For any pumped storage unit in any pumped storage power station within a pumped storage power station group, determine the unit control data and unit electrical parameters of the pumped storage unit based on the pumped storage unit data. S203. Select key hub nodes of the regional power grid based on regional power grid data, and determine the node operation parameters of the key hub nodes based on power grid operation data. S204. Extract the unit operation elements of the pumped storage unit through the inverse transformation of the grid effective operation matrix, and calculate the steady-state sensitivity index of the pumped storage unit by combining the unit operation elements and the unit electrical parameters. S205. Combine node operating parameters and unit control data to perform a baseline time-domain simulation of the preset power grid disturbance scenario, and calculate the transient sensitivity index of the pumped storage unit based on the time-domain simulation results. S206. The spatiotemporal sensitivity matrix of the pumped storage power station group is constructed by combining steady-state sensitivity index and transient sensitivity index.

[0041] In this embodiment, the power balance equations for each node of the power system in polar coordinates are first established based on the power grid topology data and power grid operation data:

[0042] in, , The active and reactive power injected into AC node i of the power grid are positive for power generation and negative for load or pumping. and Let i be the voltage amplitude at AC node i of the power grid. , and Let be the voltage phase angle difference between AC node i and AC node j of the power grid. , The real and imaginary parts of the node admittance matrix are given by the real part (conductivity) and the imaginary part (susceptance), which are determined by the real-time topology of the power grid.

[0043] The power balance equation is a strongly nonlinear equation and cannot be solved analytically directly. Therefore, an improved Newton-Raphson method is used for iterative solution. This method applies the power balance equation to the current real-time operating point of the power grid. , A Taylor series expansion is performed at the point, and the power imbalance at each AC node of the power grid is calculated in each iteration. The Jacobian matrix used for iteration is updated synchronously. If the maximum value of the power imbalance is less than a pre-set convergence threshold, such as 10, the process is considered complete. -1 If the power flow convergence is determined, the iteration terminates; if convergence is not achieved and the maximum number of iterations has not been reached, the node voltages and phase angles are updated before proceeding to the next iteration. Once the power flow iteration converges, based on the converged steady-state operation of the power grid, the core matrix elements of the final converged iteration are extracted to construct the effective operation matrix of the power grid. Specifically, the extracted content is the Newton-Raphson-Fajokian matrix at the convergence point, which serves as the core of the effective operation matrix. This matrix accurately reflects the linear mapping relationship between the power injection changes at each AC node in the regional power grid and the node voltage amplitude and phase angle under the steady-state operation point of power flow convergence. It is the core mathematical model for subsequent steady-state sensitivity partial derivative calculations.

[0044] The unit control data and electrical parameters of all pumped-storage units were extracted from the pumped-storage unit data. The unit control data included the speed control system droop coefficient, excitation regulation intensity, and speed control pipeline response speed. The unit electrical parameters were also extracted. Next, key hub nodes of the regional power grid were selected based on regional power grid data. These key hub nodes are core AC nodes with high voltage cross-weighting values, high sensitivity to the risk of multiple DC commutation failures, and a decisive role in the voltage stability of the regional power grid. The node operating parameters of the key hub nodes included the steady-state voltage amplitude and phase angle. The effective operating matrix of the power grid was inverted using the LU decomposition method to obtain the effective operating inverse matrix. The topological location of each pumped-storage unit was determined from the power grid topology data. Based on the unit topological location, the active power and reactive power operating elements corresponding to the pumped-storage units in the effective operating inverse matrix were extracted. The active power voltage sensitivity and reactive power voltage sensitivity of each pumped-storage unit were calculated by combining the active power and reactive power operating elements. Next, the initial frequency sensitivity of the pumped-storage unit is calculated based on the unit's electrical parameters, thus obtaining the frequency sensitivity index. Integrating the voltage sensitivity index and the frequency sensitivity index, the steady-state sensitivity index is obtained.

[0045] Next, transient sensitivity indices were calculated through multiple time-domain simulations. These indices include RoCoF suppression sensitivity and frequency minimum point support sensitivity. RoCoF suppression sensitivity reflects the pumped-storage unit's ability to suppress the rate of frequency change during the initial stage of a fault, while frequency minimum point support sensitivity reflects the pumped-storage unit's ability to improve the depth of frequency drop after a fault. The transient sensitivity indices calculated through these steps can serve as sensitivity weighting factors in the subsequent hydropower value model construction process. Units with higher sensitivity have higher hydropower value, providing a core transient dimension quantitative basis for the weight allocation in the hydropower value model. The spatiotemporal sensitivity matrix of the pumped-storage power station group was constructed by integrating the steady-state and transient sensitivity indices.

[0046] In one embodiment, identifying key hub nodes of the regional power grid based on regional power grid data includes the following steps: The DC converter stations of the regional power grid are located based on the regional power grid data, and the converter station parameters are determined based on the regional power grid data. The converter station parameters include the converter station reactance parameters and the converter station power parameters. The initial node impedance matrix of the regional power grid is constructed based on the regional power grid data. The initial node impedance matrix is ​​then corrected using converter station parameters to obtain the node impedance matrix of the regional power grid. The impedance magnitude of the AC nodes in the regional power grid is calculated based on the node impedance matrix. The voltage interaction factor between the DC converter station and the AC node of the power grid is calculated based on the impedance modulus, and all voltage interaction factors are integrated to construct a voltage interaction factor matrix. Power weights are assigned to DC converter stations based on their power parameters, and the DC commutation failure factors of all AC nodes in the power grid are calculated by combining the power weights and voltage interaction factors. Based on the DC commutation failure factor, all AC nodes of the power grid were screened out as grid risk nodes, and multiple overlapping commutation risk areas within the regional power grid were delineated based on the grid risk nodes and the power grid topology data. The voltage interaction weighting value of all grid risk nodes in the entire commutation risk overlap area is calculated based on the voltage interaction factor matrix and the entropy weight method. Based on the voltage interaction weighting value, multiple power grid risk nodes were selected from all overlapping commutation risk areas as key hub nodes of the regional power grid.

[0047] In this embodiment, the regional power grid data includes power grid topology data and power grid operation data. The power grid topology data includes the topological connections and voltage level information of all substations, lines, and busbars within the regional power grid. The power grid operation data includes the equipment information of receiving-end converter stations for all high-voltage / ultra-high-voltage DC transmission projects within the regional power grid. Based on the aforementioned regional power grid data, all converter stations with voltage levels ≥500kV within the regional power grid are traversed, and all DC receiving-end converter stations currently in operation are selected, referred to as DC converter stations. Converter station parameters are extracted from the power grid operation data. These parameters include converter station reactance parameters and converter station power parameters. The converter station reactance parameters include the short-circuit reactance and commutation reactance of the converter transformer within the DC converter station. The commutation reactance, which is the sum of the leakage reactance of the converter transformer and the reactor of the converter valve, is a core parameter determining the voltage characteristics during the DC commutation process. The power parameters of the converter station include the rated transmission power, real-time operating transmission power, and current power factor of the DC converter station. The real-time operating transmission power is the active power of DC transmission at the current moment of the DC converter station, and the rated transmission power is the maximum continuous transmission power designed for the DC converter station.

[0048] The extracted converter station parameters were validated to remove DC converter stations that were out of service or whose real-time transmission power was less than 10% of the rated power, thus avoiding interference from invalid data in subsequent risk quantification. At the same time, the validated converter station parameters were normalized by selecting the unified reference capacity and reference voltage of the regional power grid as the rated voltage of the AC bus of the converter station, and completing the per-unit conversion of reactance parameters and power parameters to provide a unified dimension for subsequent matrix construction and numerical calculation.

[0049] From the regional power grid topology data, all AC bus nodes within the regional power grid are extracted. These AC nodes are then filtered to remove isolated nodes, out-of-operation nodes, and low-voltage distribution network nodes (those with voltage levels below 220kV). Only AC nodes with voltage levels of 220kV and above electrically connected to the DC converter station are retained as the main AC nodes. Next, from the power grid topology data, the resistance, reactance, and ground admittance parameters of all AC lines within the regional power grid, the short-circuit reactance and turns ratio parameters of all transformers, and the reference voltage and ground capacitance parameters of all AC nodes are extracted. Based on the power system node voltage method and following the standard node admittance matrix construction rules, an initial node admittance matrix for the regional power grid is formed. The matrix dimension is M×M, where M is the total number of AC nodes in the regional power grid. The initial node admittance matrix is ​​then inverted to obtain the initial node impedance matrix of the regional power grid. This matrix fully reflects the electrical coupling characteristics between any two nodes in an AC power grid. Note that the nodes here include AC nodes of the power grid and AC bus nodes of DC converter stations. The diagonal elements of the matrix are the self-impedance parameters of the corresponding nodes, and the off-diagonal elements are the mutual impedance parameters between the two nodes.

[0050] Iterate through all AC nodes in the regional power grid, extracting two types of core impedance parameters from the initial node impedance matrix: the self-impedance parameter of each AC node and the mutual impedance parameter between each AC node and the AC bus nodes of each DC converter station. Add the commutation reactance of each DC converter station as the series reactance to ground of the AC bus node to the corresponding diagonal elements of the initial node impedance matrix to complete the matrix correction. The corrected self-impedance parameter is: ,in, The imaginary unit, This refers to the AC bus node number of the DC converter station. This refers to the commutation reactance of the DC converter station. The mutual impedance parameters in the initial node impedance matrix are used as the basis for the consistency update of the remaining off-diagonal elements, which are then updated in sync with the correction of the diagonal elements. Finally, the node impedance matrix of the regional power grid with integrated DC commutation characteristics is obtained.

[0051] The self-impedance parameters of all AC nodes in the power grid and the mutual impedance parameters between all AC nodes and the AC bus nodes of each DC converter station are extracted from the node impedance matrix. Then, the self-impedance magnitude and the mutual impedance magnitude are extracted separately, and the self-impedance magnitude and the mutual impedance magnitude are integrated into the impedance magnitude. The formula for calculating the impedance magnitude is: Where A is the real part of the self-impedance parameter or mutual impedance parameter, and B is the imaginary part of the self-impedance parameter or mutual impedance parameter.

[0052] The ratio of mutual impedance magnitude to self-impedance magnitude is calculated to obtain the voltage interaction factor between the DC converter station and the AC node of the power grid. This process is repeated for all DC converter stations and power grid AC nodes, calculating the voltage interaction factor for each power grid AC node on each DC converter station. Using the power grid AC nodes as rows and the DC converter stations as columns, all calculated voltage interaction factors are normalized to construct a voltage interaction factor matrix. This matrix comprehensively reflects the voltage impact characteristics of all power grid AC nodes on all DC converter stations within the regional power grid, serving as the core data foundation for subsequent multi-DC coupling risk quantification.

[0053] Power weights are assigned to each DC converter station based on its real-time operating power. First, the sum of the real-time operating power of all DC converter stations is calculated. Then, the ratio of each DC converter station's real-time operating power to the sum of its real-time operating power is calculated. This ratio is normalized and used as the power weight for each DC converter station. This is because the higher the real-time operating power of a DC converter station, the higher its power proportion in the regional power grid, and the more severe the impact on the grid in the event of a commutation failure. Therefore, it should be given a higher weight in the comprehensive risk quantification.

[0054] The DC commutation failure factor for each AC node in the power grid is calculated by combining power weight and voltage interaction factor. The calculation formula is as follows: ,in, The voltage interaction factor between AC node j of the power grid and DC converter station i. Let represent the power weight of DC converter station i, and N represent the total number of DC converter stations. The larger the DC commutation failure factor, the stronger the comprehensive impact of voltage fluctuations at the AC nodes of the power grid on the multi-DC commutation process of the regional power grid, and the more likely it is to cause single or multiple DC commutation failures.

[0055] The mean and standard deviation of the DC commutation failure factor for all AC nodes in the power grid are calculated. The sum of these two values ​​is then used as the failure factor threshold. AC nodes with a DC commutation failure factor greater than this threshold are marked as risk nodes. Based on the power grid's electrical partitioning and topology connections, the regional power grid is divided into multiple electrically coupled zones. For example, AC nodes with the same voltage level sequence, the same dispatch electrical partition, and direct physical line connections of the same voltage level between two nodes, without intermediate step-up / step-down transformers or other intermediate bus nodes, can be classified into the same electrically coupled zone. For each electrically coupled zone, the number of risk nodes and the number of DC converter stations covering each risk node are counted. If a risk node in a zone corresponds to a high-voltage interaction factor of two or more DC converter stations (i.e., the voltage interaction factor is greater than or equal to a preset threshold, such as 0.5), then the zone is marked as a commutation risk overlap area. Adjacent overlapping commutation risk areas are merged to form multiple overlapping commutation risk areas within the regional power grid. Each overlapping commutation risk area corresponds to a set of coupling risks of DC converter stations and is the core electrical area in the regional power grid most prone to triggering multi-DC cascading commutation failures. Alternatively, community detection algorithms, such as the Louvain algorithm, can be used. This algorithm constructs the power grid electrical topology network based on the node impedance matrix and uses the mutual impedance magnitude between nodes as edge weights to divide the risky nodes into communities, with each community representing a commutation risk overlapping area.

[0056] For each overlapping commutation risk zone, a voltage interaction factor sub-matrix corresponding to all grid risk nodes within that zone is extracted; this sub-matrix consists of the DC commutation failure factors between the grid risk node and all DC converter stations. Next, the information entropy of each DC converter station is calculated using the information entropy formula. Based on the calculated information entropy, the difference coefficient of each DC converter station is calculated: difference coefficient = 1 - information entropy. Then, the sum of the difference coefficients of all DC converter stations is calculated, and the ratio of each DC converter station's difference coefficient to the sum of the difference coefficients is used as the voltage interaction weighting value for each DC converter station. Each overlapping commutation risk zone is sorted in descending order of voltage interaction weighting value. The grid risk nodes in the top 50% of overlapping commutation risk zones are designated as key hub nodes of the regional power grid. Key hub nodes are core AC nodes with high voltage interaction weighting values, high sensitivity to multiple DC commutation failure risks, and a decisive role in the voltage stability of the regional power grid.

[0057] In one embodiment, the pumped-storage unit's operating elements are extracted through inverse transformation of the grid effective operation matrix, and the steady-state sensitivity index of the pumped-storage unit is calculated by combining the operating elements and the unit's electrical parameters, including the following steps: Perform the inverse matrix transformation of the power grid's effective operation matrix to obtain the inverse effective operation matrix; The pumped storage unit's topological location is determined based on the grid topology data, and the corresponding unit operation element is extracted from the effective operation inverse matrix based on the unit topological location. The unit operation element includes the unit's active power operation element and the unit's reactive power operation element. The node voltage parameters of key hub nodes are determined based on the effective operation inverse matrix, and the voltage sensitivity index of pumped storage units is calculated by combining the active power operation elements, reactive power operation elements and node voltage parameters. The voltage sensitivity index includes active power voltage sensitivity and reactive power voltage sensitivity. The initial frequency sensitivity of the pumped storage unit was calculated based on the unit's electrical parameters. Based on the unit topology of the pumped storage unit, the voltage interaction factor at the corresponding position is extracted from the voltage interaction factor matrix, and the initial frequency sensitivity is spatially corrected using the voltage interaction factor to obtain the frequency sensitivity index. By integrating the voltage sensitivity index and the frequency sensitivity index, the steady-state sensitivity index is obtained.

[0058] In this embodiment, an inverse matrix transformation is performed on the effective operation matrix of the power grid. The LU decomposition method is used to complete the matrix inversion operation, yielding the effective operation inverse matrix. From the power grid topology data, the AC node connected to the high-voltage side of the step-up transformer for each pumped-storage unit in the pumped-storage power station is extracted. The unique number, voltage level, and row and column index positions of these AC nodes in the effective operation inverse matrix are determined; these positions represent the unit topology positions of the pumped-storage units. The effective operation inverse matrix is ​​then divided into four corresponding blocks.

[0059] in, The inverse sub-block represents the partial derivative of the voltage phase angle with respect to active power. The inverse sub-block represents the partial derivative of the voltage phase angle with respect to reactive power. The inverse subblock represents the partial derivative of voltage amplitude with respect to active power. This represents the inverse sub-block of the partial derivative of voltage amplitude with respect to reactive power. It is divided into blocks from the inverse matrix based on the unit topology location. Extract the row and column elements corresponding to the AC node of the unit connected to the power grid to form the active power operation element of the unit. The physical meaning of this sub-block is the change in voltage amplitude of each AC node in the entire power grid caused by the change in the unit active power output of the pumped storage unit. It is the core data for calculating active power voltage sensitivity. The sub-block is divided from the inverse matrix according to the unit topology location. Extract the row and column elements corresponding to the AC node of the unit connected to the power grid to form the reactive power operation element of the unit; the physical meaning of this sub-block is: the change in the voltage amplitude of each AC node of the entire power grid caused by the change in the unit reactive power output of the pumped storage unit, which is the core data for calculating reactive voltage sensitivity.

[0060] Extract the node voltage parameters of all key hub nodes from the effective operating inverse matrix. Calculate the product of the node voltage parameter of each key hub node and the active power operating element of each pumped-storage unit. Calculate the ratio of this product to the grid power reference value to obtain the active power voltage sensitivity. The grid power reference value is fixed at 100 MVA. Calculate the product of the node voltage parameter of each key hub node and the reactive power operating element of each pumped-storage unit. Calculate the ratio of this product to the grid power reference value to obtain the reactive power voltage sensitivity.

[0061] The initial frequency sensitivity of the pumped-storage unit is calculated based on its electrical parameters. Initial frequency sensitivity refers to the magnitude of system frequency change caused by a change in the unit's active power output. The rated capacity of the pumped-storage unit is extracted from its electrical parameters. Then, the static droop coefficient of the pumped-storage unit's speed regulation system is determined according to industry standards, typically 4%~5%, with a per-unit value of 0.04~0.05. The reciprocal of the product of the static droop coefficients is calculated to obtain the initial frequency sensitivity. Next, based on the unit's topological location, the voltage interaction factors corresponding to the location are extracted from the voltage interaction factor matrix. These voltage interaction factors are then used to spatially correct the initial frequency sensitivity; that is, after normalizing the voltage interaction factors, all normalized voltage interaction factors are integrated into an interaction factor set. The information entropy corresponding to each voltage interaction factor is calculated to quantify the effective information carried by each factor. The difference coefficient of each voltage interaction factor is calculated based on the information entropy: difference coefficient = 1 - information entropy. The ratio of the difference coefficient of a single voltage interaction factor to the sum of the difference coefficients is calculated to obtain the single-factor weight. The spatial correction coefficient of the pumped-storage unit is obtained by weighted summation of the interaction factor set within the interaction factor set. The product of the spatial correction coefficient and the initial frequency sensitivity is calculated to obtain the frequency sensitivity index. The steady-state sensitivity index is obtained by integrating the voltage sensitivity index and the frequency sensitivity index.

[0062] In one embodiment, reference is made to Figure 3 The following steps are taken to perform a baseline time-domain simulation of a preset power grid disturbance scenario, combining node operating parameters and unit control data, and to calculate the transient sensitivity index of the pumped storage unit based on the time-domain simulation results: S301. Combine node operating parameters and unit control data to perform a baseline time-domain simulation of the preset power grid disturbance scenario to obtain the baseline frequency trajectory; S302. Extract the baseline trajectory features of the baseline frequency trajectory, wherein the baseline trajectory features include the baseline frequency change rate and the baseline frequency trough value. S303. Determine the simulation disturbance step size based on the unit's electrical parameters, complete the secondary time-domain simulation based on the simulation disturbance step size, and extract the disturbance trajectory features of the secondary time-domain simulation results. The disturbance trajectory features include positive disturbance trajectory features and negative disturbance trajectory features. S304. Verify the rationality of the disturbance trajectory characteristics based on the reference frequency trajectory; S305. If the rationality verification of the disturbance trajectory characteristics passes, the transient sensitivity index of the pumped storage unit is calculated based on the disturbance trajectory characteristics and using the central difference formula.

[0063] In this embodiment, the preset grid disturbance scenarios can be DC system faults and renewable energy grid disconnection faults. DC system faults refer to single-circuit UHVDC bipolar blocking faults. When performing time-domain simulation, it is necessary to specify the fault occurrence time, such as simulation duration of 0s, and the fault clearing time, such as simulation duration of 0.1s. The fault location is selected as the AC bus of the DC converter station with the highest power proportion in the regional power grid. Renewable energy grid disconnection faults refer to the largest capacity renewable energy power station in the target area, such as wind power or photovoltaic centralized grid disconnection faults, with a disconnection capacity of not less than 30% of the total installed capacity of renewable energy in the region. The fault occurrence time can be set to 0s, without fault clearing steps, to simulate instantaneous full grid disconnection.

[0064] Operating parameters are extracted from regional power grid data, including steady-state voltage amplitude and phase angle of key hub nodes, line impedance and transformer parameters from power grid topology data, converter station parameters of DC converter stations such as commutation reactance and real-time transmission power, and real-time output data of renewable energy power plants. These are then integrated to construct a regional power grid simulation model. Next, pumped-storage unit control data is extracted from pumped-storage unit data, including speed regulation system droop coefficient, excitation regulation intensity, speed control pipeline response speed, and unit electrical parameters such as inertia constant, rated capacity, and short-circuit reactance. The pumped-storage units are then connected to the corresponding AC nodes in the regional power grid simulation model according to their current actual operating conditions, such as power generation, pumping, or no-load operation, ensuring consistency between unit control logic and field conditions. The total simulation duration is set to 10-15 seconds, covering the entire process of fault occurrence, frequency drop, and natural recovery. The simulation step size can be set to 0.01 seconds, consistent with conventional electromechanical transient simulations, balancing computational accuracy and efficiency. The simulation algorithm can employ implicit trapezoidal integral method to ensure simulation stability.

[0065] After completing the above parameter configuration steps, start the simulation software, such as PSASP / BPA / PSS / E / PSCAD, etc., and execute the preset power grid disturbance scenario simulation. Do not apply any additional power disturbance to the pumped storage units, and only record the natural response process of the power grid. After the simulation is completed, extract the frequency response data of key hub nodes to form a reference frequency trajectory. The reference frequency trajectory contains the frequency value corresponding to every 0.01 seconds, which fully reflects the dynamic evolution process of the power grid frequency after the disturbance.

[0066] Next, the baseline trajectory features of the baseline frequency trajectory are extracted. These features include the baseline frequency change rate and the baseline frequency trough. The baseline frequency change rate refers to the maximum rate of change of the grid frequency after a fault occurs, reflecting the steepness of the frequency drop / rise, and is a core indicator for measuring the transient frequency stability of the power grid. Starting from the fault occurrence time, such as 0s, a baseline frequency trajectory segment within a subsequent 0.5-second window is extracted, such as 0~0.5s. Then, the frequency change rate at each instant within this segment is calculated using numerical differentiation, and the value with the largest absolute value is taken as the baseline frequency change rate. The baseline frequency trough refers to the lowest value reached by the grid frequency after a fault occurs, reflecting the depth of the grid frequency drop. It directly determines whether the low-frequency load shedding device is triggered and is a key threshold for the transient frequency safety of the power grid. A trajectory segment from the fault occurrence until the frequency recovers to more than 95% of the rated frequency can be extracted, usually 0~5 seconds. The minimum frequency value within this segment is selected as the baseline frequency trough.

[0067] The simulation disturbance step size is determined based on the rated capacity in the unit's electrical parameters. The simulation disturbance step size can be selected as 1% to 5% of the rated capacity. For example, if the rated capacity of a single unit is 300MW, the disturbance step size can be 3 to 15MW. Next, step size verification is performed. One pumped-storage unit is randomly selected, and simulations are conducted using three candidate step sizes: 1%, 2%, and 5%. The frequency characteristic value change trends under different step sizes are compared, and the smallest step size at which the characteristic value change enters the stable range is selected. Alternatively, an intermediate value can be directly selected. Based on the above regional power grid simulation model, a positive power disturbance is applied to all pumped-storage units, i.e., their active power output ΔP is uniformly increased. The specific implementation of the positive disturbance is as follows: when the pumped-storage unit is in pumping mode, the pumping power ΔP is reduced; when the pumped-storage unit is in generating mode, the active power output ΔP is increased. Next, using the same simulation step size, total duration, and integration algorithm as the aforementioned baseline time-domain simulation, a second time-domain simulation is performed to obtain the positive disturbance frequency trajectory. Then, using the same method as extracting the baseline trajectory features, the positive disturbance frequency change rate and the positive disturbance frequency trough value are extracted from the positive disturbance frequency trajectory and integrated into the positive disturbance trajectory features.

[0068] Next, a negative disturbance simulation is performed. A negative disturbance refers to a disturbance that reduces the net active power output of the power grid, forming a symmetrical perturbation with the positive disturbance. This is used to offset calculation errors caused by system nonlinearity. This step also strictly adheres to the single-variable control principle. The negative power disturbance is based on the regional power grid simulation model, with all other parameters completely consistent with the regional power grid simulation model. The specific implementation of the negative disturbance is as follows: when the pumped-storage unit is in generating mode, the active power output ΔP is reduced; when the pumped-storage unit is in pumping mode, the pumping power ΔP is increased. Both operations reduce the net active power output of the power grid, which exacerbates the frequency drop, forming a completely symmetrical perturbation direction with the positive disturbance, satisfying the calculation requirements of the central difference method. Then, using the exact same simulation settings as the baseline time-domain simulation, a second time-domain simulation is performed to obtain the negative disturbance frequency trajectory. Finally, the same method is used to extract the negative disturbance frequency change rate and the negative disturbance frequency trough value from the negative disturbance frequency trajectory. The negative disturbance trajectory features are obtained by integrating the negative disturbance frequency change rate and the negative disturbance frequency trough value, and the disturbance trajectory features are obtained by integrating the positive disturbance trajectory features and the negative disturbance trajectory features.

[0069] Next, the rationality verification of the disturbance trajectory characteristics is completed using the reference frequency trajectory. Positive disturbances, to increase the net active power output of the grid, will inevitably suppress frequency drops and slow down the rate of frequency change. Therefore, two conditions must be met simultaneously: first, the frequency trough of the positive disturbance must be greater than or equal to the reference frequency trough; second, the absolute value of the rate of change of the positive disturbance frequency must be less than or equal to the absolute value of the rate of change of the reference frequency. Negative disturbances, to reduce the net active power output of the grid, will inevitably exacerbate frequency drops and accelerate the rate of frequency change. Therefore, two conditions must be met simultaneously: first, the frequency trough of the negative disturbance must be less than or equal to the reference frequency trough; second, the absolute value of the rate of change of the negative disturbance frequency must be greater than or equal to the absolute value of the rate of change of the reference frequency. If either condition is not met, the rationality verification of the disturbance trajectory characteristics fails, and the disturbance simulation needs to be repeated. If both conditions are met, the rationality verification of the disturbance trajectory characteristics passes, and subsequent steps are performed. Based on the disturbance trajectory characteristics and using the central difference formula, the transient sensitivity index of the pumped-storage unit is calculated.

[0070] Transient sensitivity indices include RoCoF suppression sensitivity and frequency minimum support sensitivity. RoCoF suppression sensitivity reflects the pumped-storage unit's ability to suppress the rate of frequency change during the initial stage of a fault, while frequency minimum support sensitivity reflects the pumped-storage unit's ability to improve the depth of frequency drop after a fault. The center-difference method has second-order calculation accuracy and can offset the first-order nonlinear error of the system through symmetrical perturbation. Compared with the first-order accuracy of forward-difference and backward-difference methods, the calculation results are more accurate and better reflect the actual characteristics of the power grid. The RoCoF suppression sensitivity is obtained by subtracting the rate of frequency change of the negative disturbance from the rate of frequency change of the positive disturbance and dividing by 2ΔP, where ΔP is the simulation disturbance step size. The frequency minimum support sensitivity is obtained by subtracting the frequency minimum of the negative disturbance from the frequency minimum of the positive disturbance and dividing by 2ΔP.

[0071] The transient sensitivity index calculated through the above steps can be used as a sensitivity weighting factor in the subsequent hydropower value model construction steps to participate in the hydropower value calculation. The higher the sensitivity of the unit, the higher the hydropower value, which provides a core transient dimension quantitative basis for the weight allocation of the hydropower value model.

[0072] In one embodiment, constructing a hydropower value model for a pumped storage power station group by combining a spatiotemporal sensitivity matrix and regional power grid data includes the following steps: Collect real-time reservoir parameters of the upper and lower reservoirs corresponding to the pumped storage power station group. The real-time reservoir parameters include the real-time water level and the water level difference between the reservoirs. The effective capacity of the reservoir is calculated based on the real-time water level and the pre-acquired water level-capacity curve. The effective capacity of the reservoir includes the effective capacity of the upper reservoir and the effective capacity of the lower reservoir. Calculate the capacity difference between the effective capacity of the upper reservoir and the effective capacity of the lower reservoir, and determine the dispatchable capacity of the upper and lower reservoirs based on the capacity difference. The dispatchable capacity includes the capacity for power generation and the capacity for pumping. The operating conditions of pumped storage power stations in the pumped storage power station group are determined based on power grid operation data, and the water resource penalty factor is calculated by combining the dispatchable reservoir capacity and the operating conditions of pumped storage power stations. The total power grid load of the target area is calculated based on the power grid operation data, and the active power load gap of the target area is calculated based on the total power grid load and the output data of other energy plans obtained in advance. Sensitivity weights are assigned to pumped storage power station groups based on the spatiotemporal sensitivity matrix, and a hydropower value model is constructed by combining the pumped storage power station groups, water resource penalty factors, active power load gap, and dispatchable reservoir capacity.

[0073] In this embodiment, real-time reservoir parameters for the upper and lower reservoirs corresponding to the pumped-storage power station group are first collected. These parameters include real-time reservoir water levels and the water level difference. Real-time water levels refer to the measured real-time water level upstream of the upper reservoir dam and the measured real-time water level of the tailrace of the lower reservoir, i.e., the real-time water levels of the upper and lower reservoirs. The water level difference is the difference between the real-time water levels of the upper and lower reservoirs, i.e., the real-time operating head of the pumped-storage unit. This parameter is simultaneously used for subsequent unit efficiency correction and hydropower value adjustment. Outliers are removed from the collected raw data, and moving average filtering is used to process jump data caused by wind, waves, and water level fluctuations, ultimately outputting stable and effective real-time reservoir parameters.

[0074] For each pumped-storage power station's upper and lower reservoirs, pre-existing, field-calibrated water level-capacity curves in the dispatch system are invoked. If reservoir siltation measurements and curve updates are completed during power station operation, the latest calibrated curve is used to ensure the accuracy of the water level-capacity mapping relationship. The real-time water level of the upper reservoir is substituted into the upper reservoir's water level-capacity curve, and interpolation is performed to calculate the current corresponding capacity of the upper reservoir. The current corresponding capacity is then subtracted from the dead water level's corresponding dead capacity to obtain the effective capacity of the upper reservoir. The effective capacity of the upper reservoir refers to the maximum amount of water that can be discharged downwards from the upper reservoir for power generation at the current water level; it must not be lower than 0. If the calculated value is negative, it indicates that the current water level is below the dead water level, and the effective capacity of the upper reservoir is recorded as 0. Substitute the real-time water level of the lower reservoir into the water level-capacity curve of the lower reservoir, and interpolate to calculate the current corresponding capacity of the lower reservoir. Subtract the current corresponding capacity of the lower reservoir from the maximum capacity corresponding to the highest water level of the lower reservoir to obtain the effective capacity of the lower reservoir. The effective capacity of the lower reservoir refers to the maximum amount of tailwater discharged from the upper reservoir for power generation that the lower reservoir can hold at the current water level. It must not be lower than 0. If the calculated value is negative, it means that the current water level is higher than the highest water level, and the effective capacity is recorded as 0.

[0075] The absolute value of the difference between the effective capacity of the upper and lower reservoirs is calculated to obtain the capacity difference. A smaller capacity difference indicates a higher matching degree between the upper and lower reservoir capacities, and a more relaxed constraint on the dispatchable capacity. A larger difference indicates that one of the reservoirs has become a bottleneck, and the dispatchable capacity is limited by the effective capacity of the bottleneck reservoir. The power-generating capacity is constrained by both the water release capacity of the upper reservoir and the water capacity of the lower reservoir. The calculation formula is as follows: ,in, This refers to the effective capacity of the upper reservoir. This refers to the effective capacity of the reservoir. This refers to the water conversion coefficient, used to correct for reservoir evaporation, leakage, and generator water intake losses. The engineering value range is 0.97~0.99. The generating capacity refers to the maximum amount of water that the power station can continuously generate electricity and release water under the current conditions without exceeding the water level constraints of the upper and lower reservoirs. The pumpable capacity is constrained by both the pumpable capacity of the lower reservoir and the storage capacity of the upper reservoir. First, the difference between the current corresponding capacity of the lower reservoir and its dead capacity is calculated to obtain the effective pumpable capacity. Then, the difference between the maximum capacity corresponding to the normal storage level of the upper reservoir and its current corresponding capacity is calculated to obtain the effective storage capacity of the upper reservoir. Finally, the pumpable capacity is calculated using the pumpable capacity formula: ,in, This refers to the effective water storage capacity of the upper reservoir. This refers to the effective pumping capacity of a reservoir. This refers to the water conversion coefficient.

[0076] The operating conditions of pumped storage power stations in a pumped storage power station group are determined based on grid operation data. These operating conditions are categorized into four types: generation, pumping, shutdown, and standby. First, the real-time active power value, power direction indicator, and remote control status of the unit's grid connection main switch at the grid connection port are extracted from the grid operation data. The unit's grid connection port is the physical boundary point connecting the pumped storage unit to the regional grid's high-voltage transmission line. A preliminary operating condition judgment is made based on the remote control status of the unit's grid connection main switch. If the unit's grid connection main switch is open, then... If the pumped storage power station is determined to be in shutdown condition, and the main switch for grid connection is closed, then the next step is to determine the power direction. If the unit outputs active power to the grid, i.e., the real-time active power value at the grid connection port is positive, then the pumped storage power station is determined to be in power generation condition. If the grid inputs active power to the unit, i.e., the real-time active power value at the grid connection port is negative, then the pumped storage power station is determined to be in pumping condition. If the grid-connected active power is 0 or close to 0, and the unit is in standby mode under grid dispatch, then it is determined to be in standby condition.

[0077] When a pumped-storage power station is in generating or standby mode, the ratio of the generating capacity to the maximum dispatchable capacity is calculated to obtain the capacity availability rate. The maximum dispatchable capacity is a rated inherent parameter determined during the design phase of the pumped-storage power station, representing the theoretical maximum dispatchable water volume under full operating conditions. This value can be extracted from the technical documents of the pumped-storage power station. When the pumped-storage power station is in pumping mode, the ratio of the pumpable capacity to the pumpable water capacity is calculated to obtain the capacity availability rate. Then, the water resource penalty factor is calculated based on the capacity availability rate, using the following formula:

[0078] in, The penalty intensity coefficient is set to 3-5 in engineering applications to control the exponential growth rate. Under power grid fault warning conditions, the upper limit of 5 can be used to amplify the scarcity penalty and prevent formula singularities when reservoir availability is 0%. This is the saturation coefficient, with an engineering value ranging from 0.01 to 0.03. It is used to limit the maximum value of the penalty factor and prevent the value from amplifying indefinitely. For warehouse capacity availability, It refers to the dispatchable reservoir capacity of the m-th pumped storage power station, which includes the reservoir capacity for power generation and the reservoir capacity for pumping water.

[0079] Based on grid operation data, real-time active power load data for all load nodes within the target area are determined. These are then summed to obtain the total grid load within the dispatch cycle. Next, planned output data for other energy sources within the target area, excluding pumped storage power stations, is acquired. This includes ultra-short-term forecasts for new energy sources such as wind and solar power, and daily planned outputs for conventional synchronous power sources such as thermal power, conventional hydropower, and nuclear power. It may also include planned exchange power from inter-provincial power transmission lines, with inputs being positive and outputs negative. The difference between the total grid load and the ultra-short-term forecast output is calculated to obtain the net load that the grid needs to fill with pumped storage power stations and other adjustable power sources, i.e., the grid net load. Subtracting the daily planned output and planned exchange power from the grid net load, and adding the required spinning reserve capacity for grid dispatch, yields the active power load gap for the target area. When the active power load gap is greater than 0, it indicates that the regional grid needs pumped storage power stations to fill the gap. When the active power load gap is less than 0, it indicates that the regional grid has an active power surplus, requiring pumped storage power stations to pump water for absorption. The larger the absolute value of the active power load gap, the stronger the regional grid's demand for pumped storage regulation.

[0080] Sensitivity weights are assigned to pumped-storage power station groups based on the spatiotemporal sensitivity matrix, and a hydropower value model is constructed by combining the pumped-storage power station groups, water resource penalty factors, active power load gaps, and dispatchable reservoir capacity. Specifically, the comprehensive sensitivity weight of each pumped-storage power station to key nodes of the power grid is extracted from the spatiotemporal sensitivity matrix. This comprehensive sensitivity weight is a weighted synthesis of steady-state voltage sensitivity, steady-state frequency sensitivity, and transient frequency sensitivity. The weight allocation matches the current security state of the power grid: under normal grid operation, the steady-state sensitivity weight accounts for 60%, and the transient sensitivity weight accounts for 40%; under grid fault warning or emergency conditions, the transient sensitivity weight increases to 70%, strengthening the weight of stability contribution. The hydropower value model is as follows:

[0081] in, This refers to the dispatchable reservoir capacity of the m-th pumped storage power station, which includes both power generation capacity and pumping capacity. and This is the dynamic adjustment coefficient; the sum of the two is 1, under normal grid operation conditions. It can be 0.4. It can be 0.6, balancing stability contribution and hydropower utilization efficiency; under power grid fault early warning conditions. It can be 0.7. It can be 0.3, prioritizing the contribution to ensuring grid stability. Let be the hydropower value of the m-th pumped storage power station during the t-th dispatch period. A higher value indicates a higher comprehensive value per unit volume of water. The active power load gap in the target area, It is a water resource penalty factor.

[0082] The hydropower value model takes the stable contribution of pumped storage units to the power grid as the core weight of hydropower value. It also integrates the scarcity of water resources and the supply and demand of the power grid, realizing the core logic that the higher the stable contribution of pumped storage hydropower resources, the scarcer they are, and the more the power grid needs them, the higher their value. This provides an accurate value benchmark for the coordinated operation of pumped storage power station groups and can fully tap the stability support potential and hydropower utilization efficiency of pumped storage power station groups.

[0083] In one embodiment, selecting backup power stations from the pumped storage power station group based on target dispatch power station and pumped storage unit data includes the following steps: The key scheduling attributes of the target power station are extracted from pumped storage unit data and power grid topology data. These key scheduling attributes include power grid zoning attributes and power station basic attributes. Candidate backup power stations are selected from the pumped storage power station group based on the power grid zoning attributes and pumped storage unit priority sequence. Extract key power plant attributes of candidate backup power plants from pumped storage unit data; The overall sensitivity of the target dispatch power station and the candidate backup power station is calculated based on the spatiotemporal sensitivity matrix. After vectorizing the key attributes of scheduling, key attributes of power plants, and the corresponding comprehensive sensitivity of power plants, the vectorized data are input into a clustering algorithm to perform feature clustering, resulting in multiple feature clusters. Filter out the list of power stations within all feature clusters, and mark the list of power stations within the cluster that contains the target dispatch power station as the target power station list; Mark the candidate backup power stations in the target power station list as scheduled backup power stations.

[0084] In this embodiment, based on grid topology data, the AC bus nodes of the grid connected to the high-voltage side of the step-up transformer of each unit within the target dispatching power station are identified. The unique number, voltage level, and row and column index position of each node in the grid node admittance matrix are determined, thus completing the grid topology spatial positioning of the target dispatching power station. Next, according to the electrical zoning rules and combined with the topology node positioning results of the target dispatching power station, the grid electrical zoning number, zoning name, zoning voltage level sequence, and DC converter station landing point information within the zoning are extracted to form standardized grid zoning attributes. From the pumped-storage unit data, the basic attributes of the target dispatching power station are extracted, including the total rated installed capacity, rated capacity of a single unit, number of units, unit design ramp rate, unit inertia constant, water hammer effect time constant, and control parameters of the unit speed control system and excitation system. These parameters are normalized to obtain the key attributes of the power station. The entire pumped storage power station group is traversed, and pumped storage power stations located in the same electrical zone of the power grid as the target dispatching power station and ranking in the top 60% of the pumped storage power station priority sequence are selected as candidate backup power stations. Next, the key attributes of the candidate backup power stations are extracted from the pumped storage unit data. The attribute dimensions of the key attributes of the power stations are completely consistent with the key attributes of the dispatching, including parameters such as power grid zone name, total rated installed capacity of the power station, rated capacity of a single unit, number of units, and unit design ramp rate.

[0085] The comprehensive sensitivity of the target scheduling power station and candidate backup power stations is calculated based on the spatiotemporal sensitivity matrix. This comprehensive sensitivity can be obtained by weighted summation of steady-state and transient sensitivities, with weights determined using entropy weighting or expert scoring. Next, the comprehensive sensitivity and key scheduling attributes of the target scheduling power station are vectorized and concatenated along the same dimensional rules to form the target power station vector. Similarly, the comprehensive sensitivity and key attributes of the candidate backup power stations are vectorized and concatenated along the same dimensional rules to form the candidate power station vector. The K-means++ clustering algorithm can be used as the core clustering algorithm. For the sample size of the pumped-storage power station group, an initial cluster number k is set, where k is one-third of the number of candidate backup power stations. The maximum number of iterations is set to 500, the convergence threshold is 1e-6, and Euclidean distance is used to quantify the similarity between vectors. The candidate power station vectors and target power station vectors are input into a clustering algorithm for iterative clustering. The algorithm iteratively executes initial cluster center selection, sample Euclidean distance calculation, sample cluster allocation, cluster center update, and convergence determination operations until the maximum number of iterations is reached or the convergence threshold is met. The iteration then terminates, and the initial clustering results are output, resulting in multiple feature clusters. A list of power stations within each feature cluster is then selected, and the list of power stations within each cluster that includes the target scheduling power station is marked as the target power station list. Candidate backup power stations within the target power station list are marked as scheduling backup power stations.

[0086] In one embodiment, the stable dispatch power ratio between the backup power station and the target power station is calculated based on the spatiotemporal sensitivity matrix, and a coordinated operation strategy for the pumped storage power station group is generated by combining the stable dispatch power ratio and regional power grid data, including the following steps: The ratio of the power station sensitivity of the backup power station to the target power station is calculated based on the spatiotemporal sensitivity matrix to obtain the stable dispatch power ratio. The synchronous compensation power of the backup power station is calculated based on the stable dispatch power ratio; The power grid operation characteristics of the regional power grid are extracted based on the power grid operation data, and the power grid operating conditions of the regional power grid are determined based on the power grid operation characteristics. Based on the power grid operation data, the key operating parameters of all critical hub nodes in the regional power grid are verified point by point in time, and the power grid operation risk level of the regional power grid is determined based on the verification results of the key operating parameters. Determine the grid peak-shaving demand of the regional power grid based on the active power load gap; A coordinated operation strategy for pumped storage power stations is generated by combining synchronous compensation power, grid operating conditions, grid operation risk level, and grid peak-shaving demand.

[0087] In this embodiment, the comprehensive sensitivity of the backup power station and the target power station is first calculated based on the spatiotemporal sensitivity matrix. Then, the ratio of their comprehensive sensitivities is calculated to obtain the comprehensive power station sensitivity. The planned regulation power of each target power station is determined based on the active power load gap; it is positive during power generation and negative during pumping operation. Next, the product of the planned regulation power and the ratio of the stable dispatch power is calculated to obtain the synchronous compensation power of the backup power station. Then, the regional power grid operation characteristics are extracted based on the power grid operation data. These characteristics include the normalized results of parameters such as the difference between the system frequency and the rated 50Hz, the voltage deviation rate of key hub nodes, the line power flow utilization rate, the power fluctuation coefficient of the DC converter station, and the power output fluctuation rate of new energy sources. When all the characteristic values ​​in the above power grid operation characteristics are within the safety threshold, the power grid operation is considered normal. If one or two characteristic values ​​exceed the safety threshold but do not reach the emergency threshold, it is considered a warning condition. If three or more characteristic values ​​exceed the safety threshold or any one characteristic reaches the emergency threshold, the power grid operation is considered an emergency condition. Safety thresholds and emergency thresholds can be determined by consulting national power industry standards and power grid enterprise dispatching procedures, extracting the safe operating range and extreme tolerance range specified in the standards. For example, the national standard basic range of system frequency is: normal operation 49.5~50.5Hz, extreme tolerance 49.0~50.8Hz. Therefore, the basic range of the frequency safety threshold is 49.5~50.5Hz, and the basic range of the emergency threshold is 49.0~50.8Hz.

[0088] The verification time point is set according to the smallest time unit of the scheduling period, such as 5 minutes. The key operating parameters of critical hub nodes are defined as follows: node voltage amplitude, node frequency, power flow of connected lines, and voltage interaction factor with DC converter stations. All verification time points are traversed, and the actual values ​​of the key operating parameters of all critical hub nodes in the regional power grid are collected at each time point. The actual values ​​are compared with the node's rated operating parameters and safety thresholds, and the parameter exceedance is recorded, including: the node exceeding the limit, the type of parameter exceeding the limit, the amplitude of the exceedance, and the duration of the exceedance. Parameters that do not exceed the limit are marked as normal. The parameter exceedance of critical hub nodes is used as the core input, and weights are assigned to different exceedance types. For example, the weight of voltage / frequency exceedance is higher than that of power flow exceedance, and the weight of exceedance of adjacent nodes of DC converter stations is higher than that of other nodes. The weighted summation method is used to calculate the quantified value of the power grid operation risk at each verification time point to obtain the power grid operation risk level. The higher the power grid operation risk level, the greater the risk. Risk levels will be classified according to the degree of risk in power grid operation. For example, 0-0.3 is low risk, 0.3-0.7 is medium risk, and 0.7-1 is high risk. The risk level will be marked at each time point.

[0089] Extract the active power load gap dataset for the target region at each time point within the current dispatch cycle. A positive gap value indicates a generation gap in the regional power grid, requiring pumped storage power stations to compensate for peak loads. A negative gap value indicates an active power surplus gap in the regional power grid, requiring pumped storage power stations to pump water to absorb the load and fill the valley. This forms an active power load gap time series curve. Based on the direction of the active power load gap, the peak-shaving demand of the regional power grid is divided into generation peak-shaving demand and pumped water absorption demand. Using the absolute value of the active power load gap at each time point as a basis, and combining the spinning reserve capacity coefficient of the power grid dispatch (usually taken as 5%-10% of the load), calculate the peak-shaving demand power scale at each time point. The peak-shaving demand power scale is equal to the product of the absolute value of the active power load gap and (1 - spinning reserve capacity coefficient).

[0090] Next, a coordinated operation strategy for pumped storage power stations is generated by combining the peak-shaving demand power scale, synchronous compensation power, grid operating conditions, and grid operation risk. Specifically, when the grid operating conditions are normal, the target dispatch power station is the core execution entity, and daily peak shaving and valley filling are performed according to the grid's peak-shaving demand: when the load is at its peak and the active power load gap is positive, power generation is carried out according to plan, with output matching 70%-80% of the peak load gap; when the load is at its valley and the active power load gap is negative, pumping is carried out according to plan to absorb the grid's surplus electricity and new energy output, with pumping power matching the core proportion of the valley surplus electricity. Throughout the process, pumped storage units that are at the top of the priority sequence are prioritized, and the output of pumped storage units does not exceed the upper limit of the dispatchable reservoir capacity. When the power grid is in a warning condition, the target dispatch power station prioritizes ensuring grid stability and simultaneously meets peak-shaving needs. It prioritizes calling upon the backup power station with the highest overall unit sensitivity to perform full synchronous compensation power, adjusting output synchronously with the main dispatch power station according to the stable dispatch power ratio. Together with the target dispatch power station, they form a coordinated support for the regional power grid. If the active power load gap widens, generation output is increased synchronously; if the grid active power surplus widens, pumping power is increased synchronously. If the power grid is in an emergency condition, the target dispatch power station immediately executes emergency support actions, adjusting output according to the maximum ramp rate of the pumped storage units. It prioritizes calling upon the highest priority pumped storage units with the highest transient sensitivity to quickly fill the power gap and suppress frequency drops. If a fault results in an active power surplus, it immediately pumps water at full power to quickly absorb the excess electricity and suppress frequency rise. It also simultaneously outputs full reactive power to support and stabilize the voltage of key hub nodes. Simultaneously, all dispatched backup power stations will simultaneously execute full emergency coordinated output, reaching the maximum synchronous compensation power according to the stable dispatch power ratio. They will also synchronize with the main dispatch power station to complete the full-load switching, forming the full regulation capacity output of the power station group. This maximizes the inertia support and power regulation capabilities of the pumped storage power station group, quickly mitigating the impact of grid faults. During the fault recovery phase, the target dispatched power station and dispatched backup power stations will synchronously and smoothly reduce their output according to the grid frequency or voltage recovery rhythm, avoiding secondary risks caused by large power fluctuations, until the grid returns to steady-state operation. Simultaneously, the reservoir capacity status of each power station will be checked to ensure that the water level safety boundary is not exceeded.

[0091] This application also provides a machine-readable storage medium storing instructions for causing a machine to execute a cooperative operation method applicable to a pumped storage power station group according to any one of the preceding embodiments.

[0092] This application also provides a collaborative operation system suitable for pumped storage power station groups, including: The memory is configured to store instructions; and The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the cooperative operation method applicable to a pumped storage power station group according to any one of the preceding statements.

[0093] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0094] The memory can be an internal storage unit of a computer device, such as a hard disk or RAM, or an external storage device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) provided on the computer device. Furthermore, the memory can be a combination of internal storage units and external storage devices of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0095] This application also provides a machine-readable storage medium storing instructions that cause a machine to execute the above-described cooperative operation method applicable to pumped storage power station groups.

[0096] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0098] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0099] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0100] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0101] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0102] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0103] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0104] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for coordinated operation of pumped storage power station groups, characterized in that, The method includes the following steps: Collect regional power grid data and pumped storage unit data of the pumped storage power station group within the target area; The effective operation matrix of the power grid is constructed iteratively based on the regional power grid data. The full-time sensitivity calculation of the pumped storage power station group is completed by combining the effective operation matrix of the power grid and the pumped storage unit data, and the spatiotemporal sensitivity matrix of the pumped storage power station group is obtained. A hydropower value model for pumped storage power station groups was constructed by combining spatiotemporal sensitivity matrices and regional power grid data. Based on the hydropower value model and using a multi-objective optimization algorithm, the scheduling value anchoring of pumped storage power station groups is completed, and the value anchoring point of the power station group is obtained. Based on the value anchor points of the power station group, the priority of pumped storage power stations within the pumped storage power station group is arranged to obtain the priority sequence of pumped storage units. By combining the spatiotemporal sensitivity matrix and the priority sequence of pumped storage units, target dispatching power stations are selected from the pumped storage power station group, and backup dispatching power stations are selected from the pumped storage power station group based on the target dispatching power stations and pumped storage unit data. The stable dispatch power ratio between the backup power station and the target power station is calculated based on the spatiotemporal sensitivity matrix. The stable dispatch power ratio and regional power grid data are then combined to generate a coordinated operation strategy for the pumped storage power station group.

2. The method according to claim 1, characterized in that, The regional power grid data includes power grid topology data and power grid operation data. The step of iteratively constructing the effective operation matrix of the power grid based on the regional power grid data, and combining the effective operation matrix of the power grid with the pumped storage unit data to complete the full-time domain sensitivity calculation of the pumped storage power station group, and obtaining the spatiotemporal sensitivity matrix of the pumped storage power station group includes the following steps: Based on the regional power grid data, the power balance equation of the regional power grid in the target area is solved iteratively using the power flow algorithm until the power flow of the regional power grid converges, and the effective operation matrix of the regional power grid is output. For any pumped storage unit in any pumped storage power station within a pumped storage power station group, determine the unit control data and unit electrical parameters based on the pumped storage unit data; Key hub nodes of the regional power grid are selected based on regional power grid data, and node operation parameters of key hub nodes are determined based on power grid operation data. The unit operation elements of the pumped storage unit are extracted by inverse transformation of the grid effective operation matrix, and the steady-state sensitivity index of the pumped storage unit is calculated by combining the unit operation elements and the unit electrical parameters. A baseline time-domain simulation of a preset power grid disturbance scenario is performed by combining node operating parameters and unit control data, and the transient sensitivity index of the pumped storage unit is calculated based on the time-domain simulation results. A spatiotemporal sensitivity matrix for pumped storage power station groups is constructed by combining steady-state and transient sensitivity indices.

3. The method according to claim 2, characterized in that, The process of selecting key hub nodes of the regional power grid based on regional power grid data includes the following steps: The DC converter stations of the regional power grid are located based on the regional power grid data, and the converter station parameters are determined based on the regional power grid data. The converter station parameters include the converter station reactance parameters and the converter station power parameters. The initial node impedance matrix of the regional power grid is constructed based on the regional power grid data. The initial node impedance matrix is ​​then corrected using converter station parameters to obtain the node impedance matrix of the regional power grid. The impedance magnitude of the AC nodes in the regional power grid is calculated based on the node impedance matrix. The voltage interaction factor between the DC converter station and the AC node of the power grid is calculated based on the impedance modulus, and all voltage interaction factors are integrated to construct a voltage interaction factor matrix. Power weights are assigned to DC converter stations based on their power parameters, and the DC commutation failure factors of all AC nodes in the power grid are calculated by combining the power weights and voltage interaction factors. Based on the DC commutation failure factor, all AC nodes of the power grid were screened out as grid risk nodes, and multiple overlapping commutation risk areas within the regional power grid were delineated based on the grid risk nodes and the power grid topology data. The voltage interaction weighting value of all grid risk nodes in the entire commutation risk overlap area is calculated based on the voltage interaction factor matrix and the entropy weight method. Based on the voltage interaction weighting value, multiple power grid risk nodes were selected from all overlapping commutation risk areas as key hub nodes of the regional power grid.

4. The method according to claim 3, characterized in that, The process of extracting the unit operating elements of the pumped-storage unit through inverse transformation of the grid effective operation matrix, and calculating the steady-state sensitivity index of the pumped-storage unit by combining the unit operating elements and the unit electrical parameters, includes the following steps: Perform the inverse matrix transformation of the power grid's effective operation matrix to obtain the inverse effective operation matrix; The pumped storage unit's topological location is determined based on the grid topology data, and the corresponding unit operation element is extracted from the effective operation inverse matrix based on the unit topological location. The unit operation element includes the unit's active power operation element and the unit's reactive power operation element. The node voltage parameters of key hub nodes are determined based on the effective operation inverse matrix, and the voltage sensitivity index of pumped storage units is calculated by combining the active power operation elements, reactive power operation elements and node voltage parameters. The voltage sensitivity index includes active power voltage sensitivity and reactive power voltage sensitivity. The initial frequency sensitivity of the pumped storage unit was calculated based on the unit's electrical parameters. Based on the unit topology of the pumped storage unit, the voltage interaction factor at the corresponding position is extracted from the voltage interaction factor matrix, and the initial frequency sensitivity is spatially corrected using the voltage interaction factor to obtain the frequency sensitivity index. By integrating the voltage sensitivity index and the frequency sensitivity index, the steady-state sensitivity index is obtained.

5. The method according to claim 2, characterized in that, The process of combining node operating parameters and unit control data to perform a baseline time-domain simulation of a preset power grid disturbance scenario, and calculating the transient sensitivity index of the pumped-storage unit based on the time-domain simulation results, includes the following steps: By combining node operating parameters and unit control data, a baseline time-domain simulation of a preset power grid disturbance scenario is performed to obtain the baseline frequency trajectory. Extract the baseline trajectory features of the baseline frequency trajectory, which include the baseline frequency change rate and the baseline frequency trough value; The simulation disturbance step size is determined based on the unit's electrical parameters, and a second time-domain simulation is completed based on the simulation disturbance step size. The disturbance trajectory features of the second time-domain simulation results are extracted, including positive disturbance trajectory features and negative disturbance trajectory features. The rationality of the disturbance trajectory characteristics is verified based on the reference frequency trajectory. If the rationality verification of the disturbance trajectory characteristics passes, the transient sensitivity index of the pumped storage unit is calculated based on the disturbance trajectory characteristics and using the central difference formula.

6. The method according to claim 1, characterized in that, The process of constructing a hydropower value model for pumped storage power station groups by combining spatiotemporal sensitivity matrices and regional power grid data includes the following steps: Collect real-time reservoir parameters of the upper and lower reservoirs corresponding to the pumped storage power station group. The real-time reservoir parameters include the real-time water level and the water level difference between the reservoirs. The effective capacity of the reservoir is calculated based on the real-time water level and the pre-acquired water level-capacity curve. The effective capacity of the reservoir includes the effective capacity of the upper reservoir and the effective capacity of the lower reservoir. Calculate the capacity difference between the effective capacity of the upper reservoir and the effective capacity of the lower reservoir, and determine the dispatchable capacity of the upper and lower reservoirs based on the capacity difference. The dispatchable capacity includes the capacity for power generation and the capacity for pumping. The operating conditions of pumped storage power stations in the pumped storage power station group are determined based on power grid operation data, and the water resource penalty factor is calculated by combining the dispatchable reservoir capacity and the operating conditions of pumped storage power stations. The total power grid load of the target area is calculated based on the power grid operation data, and the active power load gap of the target area is calculated based on the total power grid load and the output data of other energy plans obtained in advance. Sensitivity weights are assigned to pumped storage power station groups based on the spatiotemporal sensitivity matrix, and a hydropower value model is constructed by combining the pumped storage power station groups, water resource penalty factors, active power load gap, and dispatchable reservoir capacity.

7. The method according to claim 6, characterized in that, The step of selecting backup power stations from the pumped storage power station group based on the target dispatch power station and pumped storage unit data includes the following steps: The key scheduling attributes of the target power station are extracted from pumped storage unit data and power grid topology data. These key scheduling attributes include power grid zoning attributes and power station basic attributes. Candidate backup power stations are selected from the pumped storage power station group based on the power grid zoning attributes and pumped storage unit priority sequence. Extract key power plant attributes of candidate backup power plants from pumped storage unit data; The overall sensitivity of the target dispatch power station and the candidate backup power station is calculated based on the spatiotemporal sensitivity matrix. After vectorizing the key attributes of scheduling, key attributes of power plants, and the corresponding comprehensive sensitivity of power plants, the vectorized data are input into a clustering algorithm to perform feature clustering, resulting in multiple feature clusters. Filter out the list of power stations within all feature clusters, and mark the list of power stations within the cluster that contains the target dispatch power station as the target power station list; Mark the candidate backup power stations in the target power station list as scheduled backup power stations.

8. The method according to claim 7, characterized in that, The process of calculating the stable dispatch power ratio between the backup power station and the target power station based on the spatiotemporal sensitivity matrix, and generating a coordinated operation strategy for the pumped storage power station group by combining the stable dispatch power ratio and regional power grid data, includes the following steps: The ratio of the power station sensitivity of the backup power station to the target power station is calculated based on the spatiotemporal sensitivity matrix to obtain the stable dispatch power ratio. The synchronous compensation power of the backup power station is calculated based on the stable dispatch power ratio; The power grid operation characteristics of the regional power grid are extracted based on the power grid operation data, and the power grid operating conditions of the regional power grid are determined based on the power grid operation characteristics. Based on the power grid operation data, the key operating parameters of all critical hub nodes in the regional power grid are verified point by point in time, and the power grid operation risk level of the regional power grid is determined based on the verification results of the key operating parameters. Determine the grid peak-shaving demand of the regional power grid based on the active power load gap; A coordinated operation strategy for pumped storage power stations is generated by combining synchronous compensation power, grid operating conditions, grid operation risk level, and grid peak-shaving demand.

9. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform a cooperative operation method applicable to a pumped storage power station group according to any one of claims 1 to 8.

10. A collaborative operation system suitable for pumped storage power station groups, characterized in that, include: The memory is configured to store instructions; as well as The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the cooperative operation method applicable to a pumped storage power station group according to any one of claims 1 to 8.

Images