A distributed power supply supported regional power grid resilience optimization scheduling method and system

Abstract

The application relates to the field of information technology and discloses a distributed power supply supported regional power grid elasticity optimization scheduling method and system. The method comprises the following steps: acquiring integrated distributed power supply real-time output data and operation state parameters, determining an output fluctuation index after data fusion; calculating a gradient contribution degree according to the index, adjusting a power grid power flow deviation, and generating a final weight distribution value; obtaining an instruction individual step through adaptive step optimization; updating an output adjustment instruction, combining the weight and a constraint to judge a power flow balance, and obtaining a preliminary output scheme; comparing the scheme with a real-time power grid, calculating a deviation value to determine a power supply list to be optimized; iteratively calculating a correction gradient to obtain a corrected output adjustment vector; and generating a final output coordination control signal according to the vector. The application solves the problem that the prior art cannot cope with fluctuation and scheduling inaccuracy through dynamic optimization scheduling, and improves power grid stability and scheduling efficiency.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular to a method and system for flexible optimization scheduling of regional power grids supported by distributed power sources. Background Technology

[0002] With the widespread application of renewable energy sources such as photovoltaics, the power grid needs to address the issue of dynamic power source integration, placing higher demands on supply-demand balance and resource optimization. Distributed power sources, due to their decentralized nature, diversity, and environmental dependence, have become a key factor affecting grid stability. However, existing technologies struggle to adapt to complex environments when coordinating their output, easily leading to insufficient grid stability or resource waste.

[0003] Existing methods for coordinating distributed power sources have significant drawbacks: First, they rely on static scheduling models or single control strategies, making it difficult to respond in real time to dynamic changes in the power grid when power output fluctuations occur. For example, traditional methods that predict power output based on historical data cannot cope with sudden increases or decreases in photovoltaic power generation caused by unforeseen weather conditions. Second, they lack a unified framework for processing multi-source heterogeneous data, making it difficult to integrate the operating states of different power sources and affecting the accuracy of scheduling decisions. Third, they cannot accurately quantify and dynamically adjust the output fluctuations of distributed power sources, and due to the heterogeneity of power sources, a unified parameter system cannot be formed. Scheduling instructions are difficult to take into account the actual state of the equipment, which can easily lead to power grid supply and demand imbalances, excessive power flow constraints, and even local power outages during peak electricity consumption.

[0004] This application determines the power output volatility index by acquiring and integrating distributed power source data, calculates and adjusts the gradient contribution to generate weights, then dynamically adjusts the power output step size to generate a preliminary power output plan, and finally generates the final power output coordination control signal through deviation comparison and iterative correction. This aims to solve the problems of existing technologies in dealing with fluctuations and inaccurate scheduling, and improve the stability and scheduling efficiency of the power grid. Summary of the Invention

[0005] This application provides a method and system for resilient optimization scheduling of regional power grids supported by distributed power sources, which is used to solve the problems of existing technologies in dealing with fluctuations and inaccurate scheduling, and to improve power grid stability and scheduling efficiency.

[0006] In a first aspect, this application provides a method for resilient and optimized scheduling of a regional power grid supported by distributed power sources, the method comprising:

[0007] Step S101: Obtain and integrate the real-time output data and operating status parameters of each distributed power source, and determine the output fluctuation index of each distributed power source.

[0008] Step S102: Calculate the gradient contribution of each distributed power source to the grid supply and demand balance target according to the output fluctuation index, obtain grid power flow status data, calculate grid power flow deviation, adjust the gradient contribution based on the grid power flow deviation, and generate the final weight allocation value of each distributed power source based on the adjusted gradient contribution.

[0009] Step S103: Dynamically adjust the output step size parameters of each distributed power source using an adaptive step size optimization method to obtain the personalized instruction step size;

[0010] Step S104: Update the output adjustment instructions of each distributed power source according to the personalized step size of the instruction, and combine the final weight allocation value with the power flow optimization constraints to determine whether the output adjustment instructions meet the power flow balance requirements, and obtain a preliminary output allocation scheme.

[0011] Step S105: Compare the preliminary power output allocation scheme with the real-time grid status, calculate the allocation unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the allocation unevenness deviation value.

[0012] Step S106: Iteratively calculate the correction gradient of the uneven distribution deviation value based on the distributed power source list to obtain the corrected output adjustment vector;

[0013] Step S107: Generate the final output coordination control signal based on the corrected output adjustment vector.

[0014] Optionally, step S101 includes:

[0015] The real-time output data and operating status parameters of the distributed power source are collected from photovoltaic power generation equipment and energy storage equipment. The operating status parameters include at least the photovoltaic output variation rate and the energy storage response delay.

[0016] The real-time output data and the operating status parameters are fused using a multi-source data fusion method to generate a unified set of status parameters.

[0017] Based on the set of state parameters, the output fluctuation index of each distributed power source is obtained by quantifying the difference in output over time and combining it with the weighting coefficients of different operating conditions.

[0018] Optionally, step S102 includes:

[0019] Based on the power output fluctuation index, the gradient contribution of each distributed power source to the grid supply and demand balance target is calculated using the gradient descent algorithm. The gradient contribution characterizes the degree of influence of the power output change of the distributed power source on the grid balance.

[0020] The power flow status data of the real-time power grid is obtained, the difference between the power flow status data and the ideal safe power flow is calculated, and defined as the power flow deviation of the power grid.

[0021] Extract the data fusion accuracy of the multi-source data fusion process, use the power grid power flow deviation and the data fusion accuracy as adjustment criteria, adjust the gradient contribution, and generate the initial weight allocation value corresponding to each distributed power source.

[0022] Verify the grid supply and demand balance effect and resource utilization efficiency index under different initial weight allocation values, adjust the initial weight allocation values ​​of distributed power sources based on the verification results, and generate the final weight allocation values ​​of each distributed power source.

[0023] Optionally, step S103 includes:

[0024] The operating status parameters of each distributed power source are integrated, and the real-time fluctuation amplitude and inter-source heterogeneity influence factor are calculated based on the photovoltaic output variation rate and the energy storage response delay. The real-time fluctuation amplitude characterizes the degree of real-time fluctuation of the output of each distributed power source over time, and the inter-source heterogeneity influence factor characterizes the impact of differences in the operating characteristics of each distributed power source.

[0025] Determine whether the gradient contribution of each distributed power source after adjustment exceeds a preset contribution threshold. If it does, then based on the real-time fluctuation amplitude and the inter-source heterogeneity influence factor, use an adaptive step size optimization method to dynamically adjust the output step size parameter of each distributed power source.

[0026] Based on the adjusted output step size parameter and combined with the adjusted gradient contribution of each distributed power source, a personalized instruction step size is generated for each distributed power source. The personalized instruction step size represents the output adjustment range of each distributed power source.

[0027] Optionally, step S104 includes:

[0028] The output adjustment instructions of each of the distributed power sources are updated according to the personalized step size of the instructions.

[0029] Based on the power grid flow optimization constraints, determine whether the updated output adjustment command meets the power flow balance requirements. The power grid flow optimization constraints include the upper limit of line power and the regional supply and demand power matching requirements.

[0030] If satisfied, the final weight allocation value and collaborative optimization constraints are integrated to generate the preliminary output allocation scheme. The collaborative optimization constraints include the energy storage response delay not exceeding a preset time threshold.

[0031] If the requirements are not met, the gradient descent algorithm is used to adjust the power output adjustment command until the power flow balance requirements are met, thus obtaining the preliminary power output allocation scheme, wherein the gradient descent algorithm takes minimizing the power flow deviation as the objective function.

[0032] Optionally, step S105 includes:

[0033] Extract the actual output demand of distributed generation from real-time power grid flow status data;

[0034] The preset output value of each distributed power source in the preliminary output allocation scheme is compared with the actual output demand of the power source, and the root mean square error method is used to calculate the distribution unevenness deviation value of each distributed power source.

[0035] Distributed power sources whose uneven distribution deviation exceeds a preset deviation threshold are marked as needing optimization, and a list of distributed power sources needing optimization is generated.

[0036] Optionally, step S106 includes:

[0037] For each distributed power source in the distributed power source list, the gradient descent algorithm is used to calculate the correction gradient of the uneven distribution deviation value, and the correction gradient represents the convergence direction of the deviation value;

[0038] Obtain the power grid supply and demand balance target, and iteratively update the output adjustment value of each distributed power source based on the correction gradient and the power grid supply and demand balance target. The iterative update is achieved by subtracting the correction gradient multiplied by the step size in each iteration step.

[0039] Set an iteration number threshold. When the iteration number reaches the iteration number threshold, determine whether the uneven distribution deviation value is lower than a preset deviation threshold. If so, stop the iteration and generate the corrected output adjustment vector.

[0040] Optionally, step S107 includes:

[0041] The modified output adjustment vector is applied to the power grid dispatching system, and the power grid dispatching system is used to achieve data matching between the modified output adjustment vector and the current power flow state of the system.

[0042] The gradient descent algorithm is used to calculate the power grid supply and demand balance deviation after the application of the corrected power output adjustment vector, and it is determined whether the power grid supply and demand balance deviation is greater than a preset balance threshold.

[0043] If so, the adaptive step size optimization method is used to adjust the corrected output adjustment vector until the power grid supply and demand balance deviation is less than or equal to the preset balance threshold, and the final output coordination control signal is generated.

[0044] If not, the corrected output adjustment vector is extracted to generate the final output coordination control signal, which is used to control the output of each distributed power source.

[0045] Secondly, this application provides a regional power grid resilient optimization dispatching system supported by distributed power sources, the system comprising:

[0046] The data fusion module is used to acquire real-time output data of distributed power sources and power grid flow status information, integrate status parameters through multi-source data fusion methods, and calculate the output fluctuation index of each distributed power source.

[0047] The weight allocation module is used to calculate the gradient contribution of each distributed power source to the grid supply and demand balance target based on the power output fluctuation index and the gradient descent algorithm, and to determine the final weight allocation value of each distributed power source accordingly.

[0048] The step size optimization module is used to dynamically adjust the output step size parameters of each distributed power source based on whether the gradient contribution exceeds the preset contribution threshold, and generate personalized instruction step sizes.

[0049] The scheme generation module is used to update the output adjustment instruction according to the personalized step size of the instruction, combine the weight allocation value and the power flow optimization constraints of the power grid, determine whether the power flow balance requirements are met, and generate a preliminary output allocation scheme.

[0050] The deviation calculation module is used to compare the preliminary power distribution plan with the real-time grid status, calculate the distribution unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the deviation value.

[0051] The iterative correction module is used to iteratively calculate the correction gradient of the uneven distribution deviation value for the list of distributed power sources that need to be optimized, and obtain the corrected output adjustment vector.

[0052] The signal generation module is used to generate the final output coordination control signal based on the corrected output adjustment vector, and to realize data matching and dynamic coordination judgment through the power grid dispatching system interface.

[0053] Thirdly, this application provides a regional power grid resilient optimization scheduling device supported by distributed power sources, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the regional power grid resilient optimization scheduling device supported by distributed power sources to execute the aforementioned regional power grid resilient optimization scheduling method supported by distributed power sources.

[0054] This application proposes a regional power grid elastic optimization scheduling method and system supported by distributed generation, applicable to distributed generation coordination control scenarios in new energy power systems. It addresses the problems of existing technologies, such as reliance on static scheduling models, lack of a unified framework for processing multi-source heterogeneous data, and difficulty in accurately quantifying and adjusting the output fluctuations of distributed generation, leading to insufficient grid stability, resource waste, and supply-demand imbalance. Compared with existing technologies, the beneficial effects of this application's technical solution are at least as follows:

[0055] First, this application obtains and integrates real-time output data and operating status parameters of distributed power sources, and dynamically adjusts the output step size parameters by combining an adaptive step size optimization method to generate personalized instruction step sizes. This can adapt in real time to dynamic changes such as sudden increases and decreases in photovoltaic output caused by sudden weather events, thereby improving the grid's real-time response capability to output fluctuations.

[0056] Second, this application adopts a multi-source data fusion method to integrate real-time output data and operating status parameters of photovoltaic, energy storage and other equipment to generate a unified set of status parameters, which provides a unified data foundation for subsequent gradient contribution calculation, weight allocation and scheduling decisions, and improves the accuracy of scheduling decisions.

[0057] Third, this application calculates the power output fluctuation index, combines the gradient descent algorithm to calculate the gradient contribution and adjust the generated weight allocation value, and then obtains the corrected power output adjustment vector through deviation comparison and iterative correction. This can accurately quantify the impact of power output fluctuation, dynamically optimize the power output allocation of each power source, ensure that the power flow constraints of the power grid are met, and reduce the risk of supply and demand imbalance and local power outages.

[0058] Fourth, in the process of weight allocation, step size adjustment and iterative correction, this application incorporates heterogeneous characteristic parameters such as photovoltaic output variation rate and energy storage response delay. The generated output coordination control signal can take into account the operating characteristics of different types of power sources, improve the overall coordination efficiency of the power grid, and avoid the problem of inefficient scheduling caused by differences in equipment characteristics. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a flowchart illustrating the regional power grid resilient optimization scheduling method supported by distributed power sources in this application.

[0061] Figure 2 This is a flowchart illustrating the weight allocation process based on gradient contribution in this application.

[0062] Figure 3 This is a schematic diagram of the regional power grid resilient optimization dispatching system supported by distributed power sources in this application.

[0063] Figure 4 This is a schematic diagram of the structure of the regional power grid flexible optimization scheduling equipment supported by distributed power sources in this application. Detailed Implementation

[0064] This application provides a method and system for resilient optimization scheduling of regional power grids supported by distributed power sources. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0065] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the regional power grid resilient optimization scheduling method supported by distributed power sources in this application includes:

[0066] Step S101: Obtain and integrate the real-time output data and operating status parameters of each distributed power source, and determine the output fluctuation index of each distributed power source.

[0067] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0068] The real-time output data and operating status parameters of the distributed power source are collected from photovoltaic power generation equipment and energy storage equipment. The operating status parameters include at least the photovoltaic output variation rate and the energy storage response delay.

[0069] The real-time output data and the operating status parameters are fused using a multi-source data fusion method to generate a unified set of status parameters.

[0070] Based on the set of state parameters, the output fluctuation index of each distributed power source is obtained by quantifying the difference in output over time and combining it with the weighting coefficients of different operating conditions.

[0071] Specifically, real-time output data and operating status parameters of distributed power sources are collected from photovoltaic (PV) power generation equipment and energy storage equipment. Real-time output data includes the instantaneous power generation of PV power generation equipment and the charging and discharging power of energy storage equipment. Operating status parameters include the PV output variation rate and energy storage response delay. The PV output variation rate reflects the magnitude of change in PV power generation per unit time, while the energy storage response delay reflects the time interval between receiving a dispatch command and actually adjusting the output of the energy storage equipment. In practical applications, if a power grid in a certain area contains 10 PV power generation devices and 5 energy storage devices, the above data needs to be collected for each device separately. The real-time output data of each PV power generation device is collected at a frequency of seconds. For example, if the power generation collected at a certain moment is 80kW, and the power generation collected at an adjacent moment is 72kW, this can initially reflect the real-time changes in the device's output. The charging and discharging power of each energy storage device is collected at a frequency of minutes, and the time from receiving a command to responding is recorded. For example, if after receiving a charging command, actual charging begins after an interval of 2.5 seconds, this time is the energy storage response delay for that instance.

[0072] A multi-source data fusion method is employed to fuse the collected real-time power output data and operating status parameters, generating a unified set of status parameters. During the multi-source data fusion process, different types of data are first standardized, unifying the unit of real-time power output data to kW. The photovoltaic power output variability rate is obtained by calculating the ratio of the difference between the maximum and minimum power output per unit time to the average value, as shown in the formula: Where R is the photovoltaic power output variation rate, The maximum output per unit time. The minimum output per unit time. This represents the average output per unit time. For example, if a photovoltaic power generation device has a maximum output of 100kW, a minimum output of 60kW, and an average output of 80kW within one hour, its photovoltaic output variation rate can be calculated using this formula. For energy storage response delay, the average of multiple measurements is taken as the representative response delay of the energy storage device. For example, if the response delays of a certain energy storage device are 2.3 seconds, 2.5 seconds, and 2.4 seconds respectively, its average response delay is 2.4 seconds. After standardization, the real-time output data of each device, the photovoltaic output variation rate (for photovoltaic devices), and the energy storage response delay (for energy storage devices) are integrated into the same data structure to form the state parameter record corresponding to each device. The state parameter records of all devices together constitute a unified state parameter set.

[0073] Based on the generated set of state parameters, the output volatility index of each distributed power source is obtained by quantifying the differences in output over time and combining them with weighting coefficients for different operating conditions. When quantifying the differences in output over time, state parameter data within a certain time period is selected, and the standard deviation of the output of each device within that time period is calculated. The larger the standard deviation, the greater the difference in output over time. The weighting coefficients for different operating conditions are set according to the actual operation of the power grid. For example, during peak electricity consumption periods, the power grid has higher requirements for power output stability, so higher weighting coefficients are assigned to parameters related to output stability, set to 0.6; during off-peak electricity consumption periods, the weighting coefficient is set to 0.4. During the fusion calculation, the output standard deviation is multiplied by the weighting coefficient of the corresponding operating condition, and then combined with the influence coefficient of the device's operating state parameters. For example, the influence coefficient of photovoltaic output variation rate for photovoltaic devices is set to 0.3, and the influence coefficient of energy storage response delay for energy storage devices is set to 0.2. Finally, the output volatility index is obtained, and the formula is as follows: Where I is the power output volatility index, Let w be the standard deviation of power output during this time period, w be the weighting coefficient for operating conditions, v be the influence coefficient for operating state parameters, and s be the value of the operating state parameters (photovoltaic power output variation rate or energy storage response delay). Taking a certain photovoltaic power generation equipment as an example, its power output standard deviation during peak electricity consumption periods is... =12kW, operating condition weighting coefficient w=0.6, photovoltaic output variation rate s=0.5, operating state parameter influence coefficient v=0.3, then the output fluctuation index of this equipment is... .

[0074] Step S102: Calculate the gradient contribution of each distributed power source to the grid supply and demand balance target according to the output fluctuation index, obtain grid power flow status data, calculate grid power flow deviation, adjust the gradient contribution based on the grid power flow deviation, and generate the final weight allocation value of each distributed power source based on the adjusted gradient contribution.

[0075] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0076] Based on the power output fluctuation index, the gradient contribution of each distributed power source to the grid supply and demand balance target is calculated using the gradient descent algorithm. The gradient contribution characterizes the degree of influence of the power output change of the distributed power source on the grid balance.

[0077] The power flow status data of the real-time power grid is obtained, the difference between the power flow status data and the ideal safe power flow is calculated, and defined as the power flow deviation of the power grid.

[0078] Extract the data fusion accuracy of the multi-source data fusion process, use the power grid power flow deviation and the data fusion accuracy as adjustment criteria, adjust the gradient contribution, and generate the initial weight allocation value corresponding to each distributed power source.

[0079] Verify the grid supply and demand balance effect and resource utilization efficiency index under different initial weight allocation values, adjust the initial weight allocation values ​​of distributed power sources based on the verification results, and generate the final weight allocation values ​​of each distributed power source.

[0080] Specifically, see Figure 2 When calculating the gradient contribution of each distributed power source to the grid supply-demand balance objective using the gradient descent algorithm based on the power output fluctuation index, it is necessary to first define the grid supply-demand balance objective function. This function focuses on minimizing the difference between the total load demand of the regional grid and the total output of each distributed power source, and its expression is as follows: Where J is the objective function value, and T is the number of time nodes within the scheduling period. Let N be the total load demand of the regional power grid at time t, and N be the number of distributed generation sources. Let be the output value of distributed source i at time t. The gradient contribution is essentially the partial derivative of the objective function with respect to the output of each distributed source, i.e. The sign and magnitude of this value directly reflect the impact of distributed generation output changes on grid balance. A negative partial derivative indicates that increasing the output of that source reduces the objective function value, promoting supply-demand balance; the larger the absolute value, the stronger the impact. In a real-world scenario, assuming a regional grid dispatch cycle has 10 time nodes with total load demands of 200kW, 220kW, ..., 250kW at each node, including 5 distributed generation sources, and the partial derivative of source #3 at a certain moment is -15, it indicates that increasing the output of source #3 has a significant positive impact on grid supply-demand balance, with a gradient contribution of 15 (absolute value representing the degree of impact). This process solves the problem of accurately quantifying the impact of distributed generation output changes on grid balance by transforming abstract influences into calculable gradient contributions through mathematical modeling, providing a quantitative basis for subsequent weight allocation.

[0081] Obtaining real-time power flow data requires collecting parameters such as node voltage and branch power through a power grid monitoring system. For example, a regional power grid may contain 8 nodes and 12 branches. Data collected might include node 1 voltage of 10.2kV, node 3 voltage of 9.8kV, branch 4 power of 80kW, and branch 7 power of 65kW. Ideal safe power flow is determined based on power grid design standards and operating specifications, such as a permissible node voltage range of 9.5kV-10.5kV and a branch power limit of 100kW. Based on this, the ideal voltage for each node and the ideal power for each branch can be determined. Power flow deviation is the difference between the real-time power flow data and the ideal safe power flow. For node voltage deviation, the expression is: (j is the node number, For the real-time voltage of node j, Let the ideal voltage of node j be 10.0kV. If the real-time voltage of node 3 is 9.8kV and the ideal voltage is 10.0kV, then... For branch power deviation, the expression is: (k is the branch number, For the real-time power of branch k, Let k be the ideal power. If the real-time power of branch 4 is 80kW and the ideal power is 90kW, then... By calculating the power flow deviation, we can intuitively grasp the gap between the current power grid operating state and the ideal state, providing a practical basis for adjusting the gradient contribution.

[0082] When extracting the data fusion accuracy during multi-source data fusion, it is necessary to statistically analyze the deviation between the fused data and the actual measured data. For example, for the output data of a distributed power source, if the percentage of samples where the deviation between the fused output value and the actual measured value is within 5% reaches 92%, then the data fusion accuracy of this multi-source data fusion is 92%. When adjusting the gradient contribution based on grid power flow deviation and data fusion accuracy, an adjustment coefficient needs to be set. A larger absolute value of the grid power flow deviation requires a larger adjustment coefficient, indicating a stronger adjustment is needed to correct the deviation. Higher data fusion accuracy requires a smaller adjustment coefficient, as the gradient contribution supported by high-precision data has higher reliability and does not require significant adjustment. The adjustment formula is as follows: ,in, The adjusted gradient contribution. The initial gradient contribution, The influence coefficient of tidal current deviation. The power flow deviation is represented by the average or maximum value of each deviation, which indicates the overall degree of deviation. The data fusion accuracy impact coefficient. For data fusion accuracy (substitute in decimal form). Assume the initial gradient contribution of a distributed power source. Power flow deviation =-0.1 (absolute value 0.1), =2, data fusion accuracy =0.92, =1, then the adjusted gradient contribution The adjusted gradient contributions are normalized to generate initial weight assignments. The normalization formula is as follows: If the adjusted gradient contributions of the five distributed power sources are 4.2, 3.8, 5.1, 3.5, and 4.4 respectively, with a total of 21, then the initial weight allocation value for power source 1 is... =4.2 / 21=0.2, or 20%. This step integrates the fusion accuracy of multi-source data with the power flow deviation of the power grid, so that the weight allocation is not only based on the theoretical gradient contribution, but also combined with the actual data quality and the power grid operation status. This solves the problem of lack of multi-dimensional basis for weight allocation and improves the rationality of the initial weight.

[0083] To verify the grid supply-demand balance effect and resource utilization efficiency under different initial weight allocation values, it is necessary to simulate the grid operation state under different weights. For example, for 5 distributed power sources, 3 initial weight allocation schemes are set: Scheme 1: the weights of each power source are 0.2, 0.19, 0.25, 0.18, and 0.18; Scheme 2: 0.22, 0.2, 0.23, 0.17, and 0.18; Scheme 3: 0.19, 0.21, 0.24, 0.19, and 0.17. For each scheme, the value of the objective function J (supply-demand balance effect index, the smaller the value, the better the effect) and the output utilization rate of each power source (resource utilization efficiency index, utilization rate = actual output / maximum available output, the higher the average value, the better the efficiency) are calculated within the scheduling period. Assuming J=500 for Scheme 1 and an average power utilization rate of 0.82; J=450 for Scheme 2 and an average power utilization rate of 0.85; and J=480 for Scheme 3 and an average power utilization rate of 0.83, based on the verification results, Scheme 2 achieves the best supply-demand balance and resource utilization efficiency. Therefore, using the weights of Scheme 2 as a basis, fine-tuning is performed based on the response of each power source in actual operation. If power source 2 frequently reaches its output limit in the simulation, its weight can be reduced by 0.01 and allocated to power source 1. The final weight allocation values ​​for each distributed power source are 0.23, 0.19, 0.23, 0.17, and 0.18. This verification and adjustment process ensures that the final weight allocation values ​​balance supply and demand and resource utilization efficiency, solving the problem of resource waste caused by scheduling. Through comparison of multiple schemes and actual feedback correction, the optimal allocation of weights is achieved.

[0084] Step S103: Dynamically adjust the output step size parameters of each distributed power source using an adaptive step size optimization method to obtain the personalized instruction step size.

[0085] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0086] The operating status parameters of each distributed power source are integrated, and the real-time fluctuation amplitude and inter-source heterogeneity influence factor are calculated based on the photovoltaic output variation rate and the energy storage response delay. The real-time fluctuation amplitude characterizes the degree of real-time fluctuation of the output of each distributed power source over time, and the inter-source heterogeneity influence factor characterizes the impact of differences in the operating characteristics of each distributed power source.

[0087] Determine whether the gradient contribution of each distributed power source after adjustment exceeds a preset contribution threshold. If it does, then based on the real-time fluctuation amplitude and the inter-source heterogeneity influence factor, use an adaptive step size optimization method to dynamically adjust the output step size parameter of each distributed power source.

[0088] Based on the adjusted output step size parameter and combined with the adjusted gradient contribution of each distributed power source, a personalized instruction step size is generated for each distributed power source. The personalized instruction step size represents the output adjustment range of each distributed power source.

[0089] Specifically, when integrating the operating status parameters of various distributed power sources, it is necessary to first collect two core parameters: the photovoltaic output variation rate of photovoltaic power generation equipment and the energy storage response delay of energy storage equipment. The photovoltaic output variation rate is calculated by statistically analyzing the change in photovoltaic output per unit time, using the following formula: ,in For the photovoltaic output variability rate, Let be the output value of the photovoltaic equipment at time t. These represent the maximum and minimum photovoltaic output per unit time, respectively. This represents the average photovoltaic output per unit time. For example, if a photovoltaic power generation device has a maximum output of 120kW, a minimum output of 60kW, and an average output of 90kW within one hour, substituting these values ​​into the formula yields... The energy storage response delay is obtained by recording the time interval between receiving the output adjustment command and the actual output change of the energy storage device, and the average of multiple measurements is taken as the energy storage response delay of the device. For example, if the response delays of a certain energy storage device are 2.1s, 2.3s, and 2.2s respectively in three responses, its... .

[0090] When calculating the real-time fluctuation amplitude and inter-source heterogeneity impact factor based on the photovoltaic output variability rate and energy storage response delay, the calculation of the real-time fluctuation amplitude A needs to distinguish between power source types: for photovoltaic equipment, That is, the product of the photovoltaic power output variation rate and the average power output, as in the example above. The real-time fluctuation range of the photovoltaic equipment's output per hour is approximately 60.3 kW; for energy storage equipment, ,in, This refers to the maximum adjustable output difference of the energy storage device. For example, if the maximum charge / discharge difference of a certain energy storage device is 80kW, then this value represents the maximum adjustable output difference. =2.2s, then This indicates that the adjustable output amplitude of the energy storage device is approximately 36.36 kW per second. The inter-source heterogeneity impact factor K is used to quantify the impact of differences in the operating characteristics of different types of power sources; the calculation formula is... If the aforementioned photovoltaic equipment is operated in conjunction with energy storage equipment, substituting the data yields... , K The smaller the value, the smaller the impact of the differences in operating characteristics between the two types of power sources on grid dispatch. This process solves the problem of not being able to quantify the impact of heterogeneous power source characteristics on dispatch. By converting the operating parameters of photovoltaic and energy storage into calculable real-time fluctuation amplitudes and inter-source heterogeneity impact factors, it provides a quantitative basis for subsequent step size adjustments.

[0091] When determining whether the adjusted gradient contribution of each distributed power source exceeds the preset contribution threshold, it is necessary to first clarify the adjusted gradient contribution G' (i.e., the adjusted gradient contribution when generating the initial weight allocation value in step S102) and the preset contribution threshold. Based on the grid capacity and dispatch requirements, for example, if the total grid capacity of a certain region is 1000kW, then... Set to 50. If a distributed power source is adjusted... ,Exceed If the value is 50, then the output step size parameter needs to be dynamically adjusted using an adaptive step size optimization method based on the real-time fluctuation amplitude and the heterogeneous influence factor between sources. The core of the adaptive step size optimization method is to dynamically adjust the step size according to the real-time fluctuation of the power supply and its heterogeneous characteristics. The calculation formula is as follows: ,in, The baseline output step size is set by the power grid dispatching system based on historical operating data. For example, the baseline output step size for a certain regional power grid is 10kW. In the example above, =10kW, K=0.4, =50, G'=65, substituting these values ​​into the formula yields... That is, the output step size parameter of the distributed power source is adjusted from the base of 10kW to 4.62kW. If a power source G'=40, it does not exceed... If the value is 50, then the output step size parameter is kept at the baseline value of 10kW. This dynamic adjustment method solves the problem that the static step size cannot adapt to the power output fluctuation. By combining the gradient contribution and the heterogeneous characteristics of the power source to adjust the step size, it avoids the grid power flow fluctuation caused by an excessively large step size, or the dispatch response speed being affected by an excessively small step size.

[0092] When generating personalized instruction step sizes based on the adjusted output step size parameters and the adjusted gradient contributions of each distributed power source, a correlation needs to be established between the two through weight allocation. Personalized Instruction Step Size The calculation formula is ,in, For the first i The output step size parameter after power supply adjustment For the first i Gradient contribution of each power supply after adjustment This is the sum of the gradient contributions of all power sources after adjustment. Assume a regional power grid contains three distributed power sources with adjusted output step sizes of 4.62kW, 10kW, and 8.5kW, and adjusted gradient contributions of 65, 40, and 42, respectively. =65+40+42=147. Therefore, the first power source... The second power supply The third power source These personalized step sizes represent the output adjustment range of the three distributed power sources. For example, the first power source needs to adjust by approximately 1.89kW per dispatch, the second power source by 2.5kW, and the third power source by 2.89kW. This process solves the problem that dispatch instructions cannot take into account the actual state of the equipment. By generating personalized step sizes for different power sources, it ensures that the dispatch instructions match the output capacity of the power sources and their impact on grid balance, avoiding supply and demand imbalances or equipment overloads caused by uniform instructions.

[0093] Step S104: Update the output adjustment instructions of each distributed power source according to the personalized step size of the instruction, and combine the final weight allocation value with the power flow optimization constraints to determine whether the output adjustment instructions meet the power flow balance requirements, and obtain a preliminary output allocation scheme.

[0094] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0095] The output adjustment instructions of each of the distributed power sources are updated according to the personalized step size of the instructions.

[0096] Based on the power grid flow optimization constraints, determine whether the updated output adjustment command meets the power flow balance requirements. The power grid flow optimization constraints include the upper limit of line power and the regional supply and demand power matching requirements.

[0097] If satisfied, the final weight allocation value and collaborative optimization constraints are integrated to generate the preliminary output allocation scheme. The collaborative optimization constraints include the energy storage response delay not exceeding a preset time threshold.

[0098] If the requirements are not met, the gradient descent algorithm is used to adjust the power output adjustment command until the power flow balance requirements are met, thus obtaining the preliminary power output allocation scheme, wherein the gradient descent algorithm takes minimizing the power flow deviation as the objective function.

[0099] Specifically, when updating the output adjustment commands of each distributed power source according to the personalized command step size, it is necessary to first determine the current output reference value of each power source, and then determine the adjustment direction and magnitude based on the personalized command step size. The personalized command step size represents the output adjustment magnitude of each power source. If the current output reference value of a certain distributed power source is... The personalized step size of the instruction is S. When the power grid needs to increase its total output, the output adjustment instruction is as follows: When the power grid needs to reduce its total output, the output adjustment command is as follows: For example, a photovoltaic power generation device has a current base output of 80kW and a personalized instruction step size of 1.89kW. If the grid load increases and the device needs to increase its output, its output adjustment instruction will be 80 + 1.89 = 81.89kW. Similarly, an energy storage device has a current base output of 50kW (discharge state) and a personalized instruction step size of 2.5kW. If the grid load decreases and the device needs to reduce its discharge, its output adjustment instruction will be 50 - 2.5 = 47.5kW. This process transforms abstract step size parameters into specific output adjustment instructions, providing an operational basis for subsequent power flow balance judgments, solving the problem of lack of specificity in dispatch instructions, and ensuring that the adjustment instructions for each power source match its own characteristics.

[0100] When determining whether updated output adjustment commands meet power flow balance requirements based on power flow optimization constraints, these constraints include line power limits and regional supply-demand power matching requirements. Line power limits are determined by the physical carrying capacity of the power grid lines. For example, in a regional power grid, the power limit for line 1 is 100kW, and the power limit for line 2 is 80kW. Regional supply-demand power matching requirements mean that the sum of the total output adjustment commands of all distributed power sources within the region must be consistent with the change in regional load demand. The formula is as follows: ,in, For the first i The difference between the power output adjustment command and the reference value (i.e.) ), This represents the change in regional load demand. Assume a regional power grid includes three distributed generation sources. The changes are +1.89kW, -2.5kW, and +2.89kW respectively, totaling 1.89 - 2.5 + 2.89 = 2.28kW. If the regional load demand changes... =2.3kW, then the power supply and demand matching error is |2.28-2.3|=0.02kW, which can be considered to meet the regional power supply and demand matching requirements. Simultaneously, power flow calculations are needed to simulate the power values ​​of each line under the output adjustment command. For example, under the aforementioned adjustment command, if the simulated power of line 1 is 95kW (less than the upper limit of 100kW) and the simulated power of line 2 is 78kW (less than the upper limit of 80kW), then the updated output adjustment command is determined to meet the power flow balance requirements. If the simulated power of a line exceeds the upper limit, or the power supply and demand matching error exceeds the preset range (e.g., 0.1kW), then the power flow balance requirements are not met. This judgment process, through quantitative constraints, transforms power flow balance from a qualitative requirement into a quantitative judgment, solving the problem of a lack of standards in power flow constraint judgment and ensuring that dispatching decisions comply with the power grid safety operation specifications.

[0101] If the power flow balance requirements are met, when generating a preliminary power allocation scheme by integrating the final weight allocation values ​​and collaborative optimization constraints, the final weight allocation values ​​reflect the contribution priority of each power source in the grid. Collaborative optimization constraints include ensuring that the energy storage response delay does not exceed a preset time threshold (e.g., 2.5s). The preliminary power allocation scheme needs to clearly define the final output command of each power source, its corresponding weight percentage, and whether it meets the collaborative optimization constraints. For example, in a certain scheme, the final output commands of three power sources are 81.89kW (weight 0.23), 47.5kW (weight 0.19), and 62.89kW (weight 0.24), respectively, and the response delay of all energy storage devices is 2.2s (less than 2.5s). This information is then organized into a structured scheme, including fields such as power source number, device type, final output command, weight percentage, and response delay. This integration process combines weights, constraints, and specific output commands to form an executable scheduling scheme, solving the problem of incomplete scheduling schemes and ensuring that the scheme takes into account both contribution priority and equipment operating limitations.

[0102] If the power flow balance requirements are not met, the gradient descent algorithm is used to adjust the power output adjustment command until the requirements are met. The gradient descent algorithm uses minimizing the power flow deviation as its objective function. ,in, Let be the simulated power of the k-th line. Let m be the power limit for the k-th line, and m be the number of lines. During the algorithm iteration, the step size for each adjustment is... (e.g., 0.01), the adjustment amount is ,in, For the objective function on the th i The partial derivative of the power output adjustment of a power source. For example, if the simulated power of a certain line is 102kW (exceeding the upper limit of 100kW) and the power supply-demand matching error is 0.5kW, by calculating the partial derivative, it can be seen that the power output adjustment of the power source connected to this line needs to be reduced. Assuming that a certain power source... =20, then the adjusted =1.89 - 0.01 × 20 = 1.69 kW. Repeated iterative adjustments and power flow calculations are performed until the simulated power of the lines does not exceed the upper limit and the power supply-demand matching error is less than the preset value. The set of power output adjustment commands at this point constitutes the preliminary power output allocation scheme. This optimization process dynamically corrects commands through mathematical algorithms, solving the problem that static scheduling cannot cope with power flow imbalances and ensuring that the scheme always meets the constraints of safe grid operation.

[0103] Step S105: Compare the preliminary power output allocation scheme with the real-time grid status, calculate the allocation unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the allocation unevenness deviation value.

[0104] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0105] Extract the actual output demand of distributed generation from real-time power grid flow status data;

[0106] The preset output value of each distributed power source in the preliminary output allocation scheme is compared with the actual output demand of the power source, and the root mean square error method is used to calculate the distribution unevenness deviation value of each distributed power source.

[0107] Distributed power sources whose uneven distribution deviation exceeds a preset deviation threshold are marked as needing optimization, and a list of distributed power sources needing optimization is generated.

[0108] Specifically, when extracting the actual output demand of distributed generation sources from real-time power flow data, it is necessary to first collect real-time data such as node voltage, branch power, and total regional load through the power grid monitoring system, and then infer the actual output demand of each distributed generation source based on the power flow calculation model. Real-time power flow data includes the voltage amplitude and phase angle of each node, and the active and reactive power of each branch. For example, in a regional power grid, node 1 has a voltage of 10.1kV and a phase angle of 0°, branch 1 has an active power of 85kW, and the total regional load is 320kW. The power flow calculation model takes the total regional load and line parameters (resistance, reactance) as input, and solves the node power balance equations using the Newton-Raphson method to obtain the injected power of each node. The actual output demand of a distributed generation source is equal to the injected power of its connected node. Assuming a photovoltaic (PV) power generation device is connected to node 2, and the power injected into node 2 is calculated to be 92kW, then the actual output demand of the PV device is 92kW. Similarly, if an energy storage device is connected to node 3, and the power injected into node 3 is 68kW, then the actual output demand of the energy storage device is 68kW. This process addresses the disconnect between the dispatch scheme and the actual grid demand by using real-time power flow data to infer the actual demand, providing a true benchmark for subsequent deviation comparisons.

[0109] The preset output values ​​of each distributed power source in the initial power allocation plan are compared with the actual power output demand. When calculating the distribution unevenness deviation value of each distributed power source using the root mean square error (RMSE) method, the preset output value is the output command specified for each power source in the initial plan. For example, the preset output value of the photovoltaic equipment is 81.89 kW, and the preset output value of the energy storage equipment is 62.89 kW. The RMSE calculation formula is as follows: Where T is the number of samplings within the calculation period (e.g., taking 60 sampling points within 10 minutes). The preset output value is given at time t. Let t be the actual power demand at time t. Assume that within 10 minutes, the differences between the preset power output and the actual power demand at each sampling point of a distributed power source are -10.11kW, -9.8kW, ..., -10.3kW, respectively. Summing these differences gives... Substituting into the formula, we can obtain This value represents the uneven distribution deviation of the power supply, characterizing the degree of deviation between the preset output and the actual demand. This calculation process quantifies the deviation, transforming a qualitative description into a quantitative indicator through a mathematical model.

[0110] Distributed power sources whose uneven distribution deviation exceeds a preset deviation threshold are marked as requiring optimization. When generating a list of distributed power sources requiring optimization, the preset deviation threshold is set according to the grid operation accuracy requirements; for example, a regional grid might set the threshold to 8kW. If the uneven distribution deviation of a photovoltaic device is 10.06kW (exceeding 8kW), and the uneven distribution deviation of an energy storage device is 5.2kW (less than 8kW), then the photovoltaic device is marked as requiring optimization. The generated list must include information such as power source number, device type, preset output value, actual output demand, and uneven distribution deviation value. For example, an entry in the list might be "Power source number 1, type: photovoltaic, preset output: 81.89kW, actual demand: 92kW, deviation value: 10.06kW". This list clearly identifies the key targets for adjustment, resolving the problem of unclear optimization objectives, providing precise guidance for subsequent iterative corrections, and avoiding resource waste or grid fluctuations caused by blind adjustments.

[0111] Step S106: Iteratively calculate the correction gradient of the uneven distribution deviation value based on the distributed power source list to obtain the corrected output adjustment vector.

[0112] In one specific embodiment, the process of performing step S106 may specifically include the following steps:

[0113] For each distributed power source in the distributed power source list, the gradient descent algorithm is used to calculate the correction gradient of the uneven distribution deviation value, and the correction gradient represents the convergence direction of the deviation value;

[0114] Obtain the power grid supply and demand balance target, and iteratively update the output adjustment value of each distributed power source based on the correction gradient and the power grid supply and demand balance target. The iterative update is achieved by subtracting the correction gradient multiplied by the step size in each iteration step.

[0115] Set an iteration number threshold. When the iteration number reaches the iteration number threshold, determine whether the uneven distribution deviation value is lower than a preset deviation threshold. If so, stop the iteration and generate the corrected output adjustment vector.

[0116] Specifically, when calculating the correction gradient for the uneven distribution deviation value using the gradient descent algorithm for each distributed power source in the distributed power source list, the functional expression for the uneven distribution deviation value needs to be clearly defined first. The uneven distribution deviation value is represented by the root mean square error (RMSE). To simplify the gradient calculation, the square of the RMSE is taken as the objective function. ,in, The preset output value is given at time t. Let be the actual output demand at time t, and T be the number of samples within the calculation period. The corrected gradient is the objective function modulo the preset output value. The partial derivatives, i.e. The sign of this value indicates the direction of convergence of the deviation value—if A positive value indicates an increase. This can decrease F (i.e., reduce RMSE), shifting the deviation value towards the convergence direction; if If it is negative, then it needs to be reduced. Achieve deviation convergence. For example, in a list of distributed power sources, the photovoltaic devices at a certain moment within a sampling period T=60. =81.89kW, =92kW, substituting into the formula, we get This indicates that the difference between the preset output value and the actual demand needs to be reduced (i.e., increased). This process promotes deviation convergence. It solves the problem of not being able to determine the direction of deviation correction by transforming deviation correction from blind adjustment to directional optimization through gradient calculation, providing clear guidance for subsequent output updates.

[0117] When obtaining the power grid supply and demand balance target, the core objective is to minimize the difference between the total regional output and the total load, expressed as follows: ,in, The total output value is preset for all distributed power sources. For the total regional load, the requirements are as follows: The output adjustment value of each distributed power source is less than the preset balance threshold (e.g., 5kW). When iteratively updating the output adjustment value of each distributed power source based on the correction gradient and the grid supply and demand balance target, the iterative update formula is: ,in, For the first iThe preset output value of the power source in the k-th iteration. This is the preset output value for the (k+1)th iteration. The iteration step size (set according to the grid response sensitivity, such as...) =0.5), For the first i The corrected gradient of each power source. Assume that the photovoltaic device in its first iteration... =81.89kW, , =0.5, substituting into the formula, we get Simultaneously, it is necessary to verify whether the updated total output meets the supply-demand balance target. If the updated total output of all power sources is 322kW, the total regional load... If the output is less than 5kW, continue iterating; if the total output deviates from the load demand, the step size needs to be fine-tuned during iteration. If the total output is too high, reduce This avoids exacerbating the supply-demand imbalance. This iterative process solves the problem that static adjustments cannot dynamically adapt to demand. By updating output in real time in conjunction with supply-demand balance targets, it ensures that deviation corrections do not deviate from the overall operation requirements of the power grid.

[0118] When setting the iteration threshold, this threshold is determined based on the grid dispatch response speed requirements; for example, a regional grid might set the threshold to 20 iterations. Once the iteration count reaches the threshold, the RMSE (Ratio of Uneven Distribution Score) of each distributed power source needs to be recalculated and compared with a preset deviation threshold (e.g., 8kW). For instance, after 20 iterations, a distributed power source's... During the sampling period of T=60, Substituting into the RMSE formula, we get If the output adjustment value is exactly equal to the preset deviation threshold, the convergence requirement is met, and the iteration stops. At this point, the difference between the final output adjustment value obtained by each distributed power source and the initial preset output value needs to be integrated to generate a corrected output adjustment vector. The output adjustment vector uses the output adjustment amount of each power source in the list as a component. For example, if the list contains 3 power sources with final output adjustment amounts of 90.5-81.89=8.61kW, 67.8-62.89=4.91kW, and 75.2-70.3=4.9kW respectively, then the corrected output adjustment vector is [8.61, 4.91, 4.9] (unit: kW). Each component in this vector corresponds to the output adjustment range of one power source in the list. If the RMSE of a power source is 8.5kW (exceeding 8kW) after the number of iterations reaches the threshold, the iteration needs to continue until the RMSE is lower than the threshold or the number of iterations reaches a newly set threshold again (such as 30 times). This termination and vector generation process solves the problem of optimization without a clear termination condition. By using dual judgment (number of iterations and deviation threshold), it ensures that the output adjustment vector meets the accuracy requirements while avoiding scheduling delays caused by excessive iteration.

[0119] Step S107: Generate the final output coordination control signal based on the corrected output adjustment vector.

[0120] In one specific embodiment, the process of executing step S107 may specifically include the following steps:

[0121] The modified output adjustment vector is applied to the power grid dispatching system, and the power grid dispatching system is used to achieve data matching between the modified output adjustment vector and the current power flow state of the system.

[0122] The gradient descent algorithm is used to calculate the power grid supply and demand balance deviation after the application of the corrected power output adjustment vector, and it is determined whether the power grid supply and demand balance deviation is greater than a preset balance threshold.

[0123] If so, the adaptive step size optimization method is used to adjust the corrected output adjustment vector until the power grid supply and demand balance deviation is less than or equal to the preset balance threshold, and the final output coordination control signal is generated.

[0124] If not, the corrected output adjustment vector is extracted to generate the final output coordination control signal, which is used to control the output of each distributed power source.

[0125] Specifically, when applying the modified output adjustment vector to the power grid dispatching system, the structure of the output adjustment vector must first be defined. Each component corresponds to the output adjustment amount of each power source in the distributed generation list. For example, the vector V=[8.61,4.91,4.9] (unit: kW) corresponds to the output increase of three distributed power sources. The power grid dispatching system needs to extract the current power flow status data, including the voltage of each node, the power of each branch, and the current actual output of each power source. For example, node 1 voltage is 10.1kV, branch 1 power is 85kW, power source 1 current actual output is 81.89kW, power source 2 current actual output is 62.89kW, and power source 3 current actual output is 70.3kW. The data matching process requires associating the vector components with the current actual output of the corresponding power source and calculating the target output value of each power source after applying the vector. The formula is as follows: ,in, For the first i The current actual output of the power source. For vector number 1 i Each component. Substituting the data, we can obtain the target output of power supply 1. Power supply 2 target output Power supply 3 target output Simultaneously, the dispatching system needs to verify whether the power flow state corresponding to the target output is compatible with the current system power flow. For example, by simulating the power of each branch under the target output through power flow calculation, if the target output of power source 1 is 90.5kW, and the simulated power value of its connected branch is 88kW (less than the branch power limit of 100kW), then the data matching is considered successful. This process solves the problem of the disconnect between dispatching instructions and system status, ensuring that vector applications conform to the current operating conditions of the power grid through data matching, and avoiding power flow disturbances caused by blind adjustments.

[0126] When calculating the power grid supply-demand balance deviation after applying the corrected output adjustment vector using the gradient descent algorithm, a power grid supply-demand balance deviation function must first be defined. This deviation is characterized by the absolute value of the difference between the total target output and the total load of the region, and the formula is as follows: ,in, The sum of the outputs of all power sources. Let be the total load of the region. The objective function of the gradient descent algorithm is: The degree of supply-demand balance is reflected by calculating the objective function value. Assume the total regional load... =320kW, the total target output of the three power sources is 90.5 + 67.8 + 75.2 = 233.5kW. Substituting into the deviation formula, we can get... The preset balance threshold is set according to the grid stability requirements. For example, if the threshold for a certain area's grid is set to 5kW, then... The supply and demand balance requirement is not met; if the total target output is 318kW in another scenario, If the condition is met, then the requirement is deemed satisfied. This calculation process quantifies the supply and demand balance, solving the problem of a lack of standards for judging supply and demand balance, and transforming the balance state from a qualitative description into a quantitative indicator through a mathematical model.

[0127] If the power grid supply-demand balance deviation exceeds a preset balance threshold, when adjusting the corrected output adjustment vector using the adaptive step-size optimization method, the step-size adjustment coefficient must first be determined. The adaptive step-size optimization method aims to minimize... For the target, the step size adjustment factor θ The formula is dynamically set based on the magnitude of the deviation. ,in, This is a preset balance threshold. Substituting the data yields... The adjusted vector This is used to increase the output adjustment of each power source to narrow the supply-demand gap. Adjusted vector (Unit: kW). The total target output is recalculated as 90.5 + 9.11 + 67.8 + 5.20 + 75.2 + 5.18 = 252.99 kW. If the power is still greater than 5kW, the adjustment process needs to be repeated until... When the vector is adjusted to V''=[35.2,20.5,20.3] (unit: kW), the total target output is 81.89+35.2+62.89+20.5+70.3+20.3=310.98kW. Continue to adjust the step size coefficient final vector (Unit: kW), the total target output is 81.89 + 54.56 + 62.89 + 31.78 + 70.3 + 31.47 = 332.9 kW. (This example illustrates the adjustment process; in practice, multiple iterations are required until the deviation meets the target.) Once the target is met, the final vector is converted into a power output coordination control signal. This signal includes instructions such as the target output value for each power source and the adjustment duration (e.g., completing the adjustment within 5 minutes). If the supply-demand balance deviation is less than or equal to a preset threshold, the corrected power output adjustment vector is directly extracted and converted into a control signal. For example, after applying vector V=[2.1,1.8,1.1] (unit: kW), the total target output is 319kW. The control signal instructs power supply 1 to increase from 81.89kW to 83.99kW, power supply 2 from 62.89kW to 64.7kW, and power supply 3 from 70.3kW to 71.4kW. This control signal generation process solves the problem of the lack of dynamic adaptability in dispatching instructions. Through adaptive adjustment, it ensures that the instructions always meet the supply and demand balance requirements, while avoiding excessive adjustments that could cause grid fluctuations.

[0128] The above describes the regional power grid resilient optimization scheduling method supported by distributed power sources in the embodiments of this application. Please refer to [link / reference]. Figure 3 The following describes the regional power grid resilient optimization scheduling system 300 supported by distributed power sources in the embodiments of this application. The regional power grid resilient optimization scheduling system 300 supported by distributed power sources in this application includes:

[0129] The data fusion module 301 is used to acquire real-time output data of distributed power sources and power grid flow status information, integrate status parameters through multi-source data fusion method, and calculate the output fluctuation index of each distributed power source.

[0130] The weight allocation module 302 is used to calculate the gradient contribution of each distributed power source to the grid supply and demand balance target based on the power output fluctuation index and the gradient descent algorithm, and determine the final weight allocation value of each distributed power source accordingly.

[0131] The step size optimization module 303 is used to dynamically adjust the output step size parameters of each distributed power source according to whether the gradient contribution exceeds the preset contribution threshold, and generate personalized instruction step size.

[0132] The scheme generation module 304 is used to update the output adjustment instruction according to the personalized step size of the instruction, combine the weight allocation value and the power flow optimization constraints of the power grid, determine whether the power flow balance requirements are met, and generate a preliminary output allocation scheme.

[0133] The deviation calculation module 305 is used to compare the preliminary power distribution scheme with the real-time grid status, calculate the distribution unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the deviation value.

[0134] The iterative correction module 306 is used to iteratively calculate the correction gradient of the uneven distribution deviation value for the list of distributed power sources that need to be optimized, and obtain the corrected output adjustment vector.

[0135] The signal generation module 307 is used to generate the final output coordination control signal based on the corrected output adjustment vector, and to realize data matching and dynamic coordination judgment through the power grid dispatching system interface.

[0136] Through the collaborative efforts of the aforementioned components, the system constructs a distributed power generation regional power grid elastic dispatching system based on "data-driven, dynamic optimization, and closed-loop control." This system achieves end-to-end management, from the integration of multi-source heterogeneous data from distributed power sources to precise regulation of power grid supply and demand balance.

[0137] The data fusion module 301 collects real-time output data from photovoltaic power generation equipment and operating status parameters from energy storage equipment. It generates a unified set of status parameters using a preset data fusion algorithm and calculates the output volatility index, providing a standardized data foundation for subsequent gradient contribution calculations. The weight allocation module 302 uses the output volatility index as input and employs a gradient descent algorithm to calculate the gradient contribution of each power source to the grid supply-demand balance target. It then adjusts the gradient contribution based on grid power flow deviation and data fusion accuracy, generating initial weight allocation values. After verifying the supply-demand balance effect and resource utilization efficiency under different initial weights, it finally determines the weight allocation values ​​for each power source, clarifies scheduling priorities, and avoids the shortcomings of static weights that cannot adapt to dynamic grid changes. The step size optimization module 303 integrates operating status parameters, calculates real-time fluctuation amplitude and inter-source heterogeneous influence factors, and determines whether the gradient contribution exceeds a preset threshold. If it exceeds the threshold, an adaptive step size optimization method is used to dynamically adjust the output step size parameter. Combined with the adjusted gradient contribution, it generates a personalized step size command, matching the step size with power source characteristics and grid fluctuations, thus solving the problem of response lag or over-adjustment caused by a uniform step size. The scheme generation module 304 updates the output adjustment instructions according to the personalized step size of the instructions. It then combines the final weight allocation value with the power flow optimization constraints (line power limit, regional supply and demand power matching requirements) to determine the power flow balance. If the balance is satisfied, it integrates the collaborative optimization constraints (energy storage response delay not exceeding a preset time threshold) to generate a preliminary output allocation scheme. If the balance is not satisfied, it adjusts the instructions using a gradient descent algorithm (with minimizing power flow deviation as the objective function) until the target is met, ensuring that the preliminary scheme meets the requirements for safe grid operation. The deviation calculation module 305 extracts the actual output demand from real-time power flow status data, calculates the root mean square error between the preset output value of the preliminary scheme and the actual demand, obtains the distribution unevenness deviation value, marks the power sources with deviations exceeding the threshold to generate an optimization list, and avoids blind adjustments that waste resources. The iterative correction module 306 calculates the correction gradient (representing the direction of deviation convergence) for the deviation value using the gradient descent algorithm for the optimization list. Combined with the power grid supply-demand balance target, it iteratively updates the output adjustment value by subtracting the correction gradient multiplied by the step size in each iteration step. It sets an iteration number threshold and verifies whether the deviation meets the target. Once the target is met, a corrected output adjustment vector is generated, achieving directional convergence of the deviation and improving the accuracy of output allocation. The signal generation module 307 applies the corrected output adjustment vector to the power grid dispatching system, completing data matching with the current power flow state. It calculates the supply-demand balance deviation using the gradient descent algorithm. If the deviation exceeds a preset balance threshold, an adaptive step-size optimization method is used to adjust the vector until the target is met. Finally, an output coordination control signal is generated, forming a closed loop from data acquisition and scheme generation to command execution. This solves the problems of existing technologies struggling to cope with output fluctuations and inaccurate dispatching, enabling flexible dispatching of the regional power grid.

[0138] above Figure 3The distributed power source-supported regional power grid elastic optimization scheduling system in this application embodiment is described in detail from the perspective of modular functional entities. The distributed power source-supported regional power grid elastic optimization scheduling equipment in this application embodiment is described in detail from the perspective of hardware processing.

[0139] Reference Figure 4 This application also provides a distributed power source-supported regional power grid resilient optimization scheduling device 400, which can be a server, and its internal structure can be as follows: Figure 4 As shown. The distributed power generation-supported regional power grid resilient optimization scheduling device includes a processor 402, a memory 403, a display screen 404, an input device 405, a network interface 406, and a database 407 connected via a system bus 401. The processor 402, designed as a computer, provides computing and control capabilities. The memory 403 of the distributed power generation-supported regional power grid resilient optimization scheduling device includes a non-volatile storage medium 4031 and internal memory 4032. The non-volatile storage medium 4031 stores the operating system and computer programs. The internal memory 4032 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database 407 of the distributed power generation-supported regional power grid resilient optimization scheduling device stores the corresponding data in this embodiment. The network interface 406 of the distributed power generation-supported regional power grid resilient optimization scheduling device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0140] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the solution of this application, and does not constitute a limitation on the regional power grid resilient optimization scheduling equipment supported by distributed power sources applied to it. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the methods, systems, and equipment described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0141] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for resilient and optimized scheduling of a regional power grid supported by distributed power sources, characterized in that, The method includes: Step S101: Obtain and integrate the real-time output data and operating status parameters of each distributed power source, and determine the output fluctuation index of each distributed power source. Step S102: Based on the output fluctuation index, calculate the gradient contribution of each distributed power source to the grid supply and demand balance target, obtain grid power flow status data, calculate grid power flow deviation, adjust the gradient contribution based on the grid power flow deviation, and generate the final weight allocation value of each distributed power source based on the adjusted gradient contribution. The gradient contribution characterizes the degree of influence of distributed power source output changes on grid supply and demand balance, and is calculated by combining the gradient descent algorithm with the output fluctuation index. Step S103: Dynamically adjust the output step size parameters of each distributed power source using an adaptive step size optimization method to obtain the personalized instruction step size; Step S104: Update the output adjustment instructions of each distributed power source according to the personalized step size of the instruction, and combine the final weight allocation value with the power flow optimization constraints to determine whether the output adjustment instructions meet the power flow balance requirements, and obtain a preliminary output allocation scheme. Step S105: Compare the preliminary power output allocation scheme with the real-time grid status, calculate the allocation unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the allocation unevenness deviation value. Step S106: Iteratively calculate the correction gradient of the uneven distribution deviation value based on the distributed power source list to obtain the corrected output adjustment vector; Step S107: Generate the final output coordination control signal based on the corrected output adjustment vector.

2. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 1, characterized in that, Step S101 includes: The real-time output data and operating status parameters of the distributed power source are collected from photovoltaic power generation equipment and energy storage equipment. The operating status parameters include at least the photovoltaic output variation rate and the energy storage response delay. The real-time output data and the operating status parameters are fused using a multi-source data fusion method to generate a unified set of status parameters. Based on the set of state parameters, the output fluctuation index of each distributed power source is obtained by quantifying the difference in output over time and combining it with the weight coefficients of different operating conditions.

3. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 2, characterized in that, Step S102 includes: Based on the power output fluctuation index, the gradient contribution of each distributed power source to the grid supply and demand balance target is calculated using the gradient descent algorithm. The gradient contribution characterizes the degree of influence of the power output change of the distributed power source on the grid balance. The power flow status data of the real-time power grid is obtained, the difference between the power flow status data and the ideal safe power flow is calculated, and defined as the power flow deviation of the power grid. Extract the data fusion accuracy of the multi-source data fusion process, use the power grid power flow deviation and the data fusion accuracy as adjustment criteria, adjust the gradient contribution, and generate the initial weight allocation value corresponding to each distributed power source. Verify the grid supply and demand balance effect and resource utilization efficiency index under different initial weight allocation values, adjust the initial weight allocation values ​​of distributed power sources based on the verification results, and generate the final weight allocation values ​​of each distributed power source.

4. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 2, characterized in that, Step S103 includes: The operating status parameters of each distributed power source are integrated, and the real-time fluctuation amplitude and inter-source heterogeneity influence factor are calculated based on the photovoltaic output variation rate and the energy storage response delay. The real-time fluctuation amplitude characterizes the degree of real-time fluctuation of the output of each distributed power source over time, and the inter-source heterogeneity influence factor characterizes the impact of differences in the operating characteristics of each distributed power source. Determine whether the gradient contribution of each distributed power source after adjustment exceeds a preset contribution threshold. If it does, then based on the real-time fluctuation amplitude and the inter-source heterogeneity influence factor, use an adaptive step size optimization method to dynamically adjust the output step size parameter of each distributed power source. Based on the adjusted output step size parameter and combined with the adjusted gradient contribution of each distributed power source, a personalized instruction step size is generated for each distributed power source. The personalized instruction step size represents the output adjustment range of each distributed power source.

5. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 2, characterized in that, Step S104 includes: The output adjustment instructions of each of the distributed power sources are updated according to the personalized step size of the instructions. Based on the power grid flow optimization constraints, determine whether the updated output adjustment command meets the power flow balance requirements. The power grid flow optimization constraints include the upper limit of line power and the regional supply and demand power matching requirements. If satisfied, the final weight allocation value and collaborative optimization constraints are integrated to generate the preliminary output allocation scheme. The collaborative optimization constraints include the energy storage response delay not exceeding a preset time threshold. If the requirements are not met, the gradient descent algorithm is used to adjust the power output adjustment command until the power flow balance requirements are met, thus obtaining the preliminary power output allocation scheme, wherein the gradient descent algorithm takes minimizing the power flow deviation as the objective function.

6. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 1, characterized in that, Step S105 includes: Extract the actual output demand of distributed generation from real-time power grid flow status data; The preset output value of each distributed power source in the preliminary output allocation scheme is compared with the actual output demand of the power source, and the root mean square error method is used to calculate the distribution unevenness deviation value of each distributed power source. Distributed power sources whose uneven distribution deviation exceeds a preset deviation threshold are marked as needing optimization, and a list of distributed power sources needing optimization is generated.

7. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 1, characterized in that, Step S106 includes: For each distributed power source in the distributed power source list, the gradient descent algorithm is used to calculate the correction gradient of the uneven distribution deviation value, and the correction gradient represents the convergence direction of the deviation value; Obtain the power grid supply and demand balance target, and iteratively update the output adjustment value of each distributed power source based on the correction gradient and the power grid supply and demand balance target. The iterative update is achieved by subtracting the correction gradient multiplied by the step size in each iteration step. Set an iteration number threshold. When the iteration number reaches the iteration number threshold, determine whether the uneven distribution deviation value is lower than a preset deviation threshold. If so, stop the iteration and generate the corrected output adjustment vector.

8. The regional power grid resilient optimization scheduling method supported by distributed power sources according to claim 1, characterized in that, Step S107 includes: The modified output adjustment vector is applied to the power grid dispatching system, and the power grid dispatching system is used to achieve data matching between the modified output adjustment vector and the current power flow state of the system. The gradient descent algorithm is used to calculate the power grid supply and demand balance deviation after the application of the corrected power output adjustment vector, and it is determined whether the power grid supply and demand balance deviation is greater than a preset balance threshold. If so, the adaptive step size optimization method is used to adjust the corrected output adjustment vector until the power grid supply and demand balance deviation is less than or equal to the preset balance threshold, and the final output coordination control signal is generated. If not, the corrected output adjustment vector is extracted to generate the final output coordination control signal, which is used to control the output of each distributed power source.

9. A regional power grid resilient optimization dispatching system supported by distributed power sources, characterized in that, For implementing the distributed power generation-supported regional power grid resilient optimization scheduling method as described in any one of claims 1 to 8, the distributed power generation-supported regional power grid resilient optimization scheduling system comprises: The data fusion module is used to acquire real-time output data of distributed power sources and power grid flow status information, integrate status parameters through multi-source data fusion methods, and calculate the output fluctuation index of each distributed power source. The weight allocation module is used to calculate the gradient contribution of each distributed power source to the grid supply and demand balance target based on the power output fluctuation index and the gradient descent algorithm, and determine the final weight allocation value of each distributed power source accordingly. The gradient contribution represents the degree of influence of the output change of the distributed power source on the grid supply and demand balance. The step size optimization module is used to dynamically adjust the output step size parameters of each distributed power source based on whether the gradient contribution exceeds the preset contribution threshold, and generate personalized instruction step sizes. The scheme generation module is used to update the output adjustment instruction according to the personalized step size of the instruction, combine the weight allocation value and the power flow optimization constraints of the power grid, determine whether the power flow balance requirements are met, and generate a preliminary output allocation scheme. The deviation calculation module is used to compare the preliminary power distribution plan with the real-time grid status, calculate the distribution unevenness deviation value, and determine the list of distributed power sources that need to be optimized based on the deviation value. The iterative correction module is used to iteratively calculate the correction gradient of the uneven distribution deviation value for the list of distributed power sources that need to be optimized, and obtain the corrected output adjustment vector. The signal generation module is used to generate the final output coordination control signal based on the corrected output adjustment vector, and to realize data matching and dynamic coordination judgment through the power grid dispatching system interface.

10. A regional power grid resilient optimization scheduling device supported by distributed power sources, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the regional power grid resilient optimization scheduling method supported by distributed power sources as described in any one of claims 1 to 8.

Images