Distributed resource aggregation analysis method and system

Abstract

The invention relates to the technical field of electric power intelligent management, and discloses a distributed resource aggregation analysis method and system, and the system comprises a resource processing module, a data analysis module, an aggregation analysis module and a cooperative regulation and control module. The resource processing module is used for carrying out data acquisition and protocol analysis on the accessed distributed energy supply end; the data analysis module is used for receiving and processing the multi-source heterogeneous data from the resource processing module and constructing a resource portrait; constructing a load model and outputting load prediction data for the I-type energy supply end; aiming at the II-type energy supply end, constructing a random process model of output and response of the II-type energy supply end, and outputting new energy output data; the aggregation analysis module is used for setting an optimal charging and discharging plan of each energy storage unit and a reference scheduling scheme of an adjustable load; and the coordinated regulation and control module is used for generating a scheduling instruction and issuing the scheduling instruction to the corresponding energy supply end. According to the invention, the power supply resources can be efficiently integrated and accurately quantified, and the observable, measurable and controllable level of the distributed power supply resources is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent power management, and in particular to a distributed resource aggregation analysis method and system. Background Art

[0002] Currently, renewable energy sources, such as wind power and photovoltaics, have become a crucial component of the power system. Their distributed and intermittent nature poses a significant challenge to the traditional centralized dispatch model of "source follows load." In regional power grids like Xinjin District, the surge in air conditioning load during summer heatwaves, combined with the peak output of photovoltaic power stations during the afternoon, further widens the peak-to-valley difference. Meanwhile, during low nighttime load periods, high photovoltaic power generation can lead to curtailed power generation. Traditional power grid dispatching methods struggle to accurately perceive and efficiently coordinate massive distributed resources. There is an urgent need to utilize digital technologies, such as virtual power plants, to aggregate and optimize the regulation of dispersed distributed power sources, energy storage systems, and controllable loads. This will enhance the grid's flexibility, reliability, and ability to absorb renewable energy. This is crucial for building a new power system, ensuring regional energy security, and promoting a low-carbon energy transition.

[0003] In the field of virtual power plant resource aggregation technology, a series of research and practices have been carried out both domestically and internationally. Existing technical solutions are usually based on supervisory control and data acquisition (SCADA) systems or energy management systems (EMS) to achieve data collection and monitoring, and use optimization algorithms to dispatch aggregated resources. However, most existing systems still focus on the dispatch of traditional controllable loads and are not adaptable enough to scenarios with a high proportion of renewable energy access. In terms of modeling, traditional methods often use deterministic predictions or simple statistical models to describe renewable energy output, failing to fully characterize its inherent randomness and volatility. In terms of architectural design, existing platforms often adopt a relatively centralized data processing model, which makes it difficult to support the real-time access and efficient computing of massive heterogeneous resources. In addition, existing systems mostly ignore spatiotemporal characteristics such as geographical distribution and meteorological correlation when aggregating resources, resulting in limited reliability and accuracy of the aggregation results.

[0004] Although virtual power plant technology has garnered widespread attention, building an efficient distributed resource aggregation system for scenarios with a high proportion of renewable energy integration still faces numerous technical challenges. First, renewable energy output is highly dependent on meteorological conditions. Its strong randomness and intermittency make accurate prediction and reliable modeling extremely difficult, making traditional deterministic models difficult to apply. Second, distributed resources are diverse and have varying characteristics, including power-side resources such as photovoltaics and energy storage, as well as various flexible loads. Their heterogeneity, dispersion, and varying response characteristics make unified modeling and collaborative optimization extremely complex. Third, the coordination and integration of multi-timescale scheduling across day-ahead, intraday, and real-time scheduling poses significant challenges. Achieving rolling optimization and rapid decision-making in a high-dimensional environment while ensuring robust and cost-effective scheduling plans remains a core challenge in practical applications. Furthermore, the real-time integration of massive devices, high-frequency data collection and processing, and low-latency communication and control also place extremely high demands on the system's infrastructure and computing power. Summary of the Invention

[0005] The present invention aims to provide a distributed resource aggregation analysis method and system, which can efficiently integrate power supply resources and accurately quantify them, effectively improving the observability, measurability and controllability of distributed power supply resources.

[0006] To achieve the above objectives, the present invention provides the following basic solutions: Option 1 A distributed resource aggregation analysis system, including a resource processing module, a data analysis module, an aggregation analysis module and a collaborative control module; The resource processing module is used to collect data and parse protocols for the connected distributed energy supply terminals; the distributed energy supply terminals include Class I energy supply terminals whose energy source is controllable loads and Class II energy supply terminals whose energy source is new energy; The data analysis module is used to receive and process multi-source heterogeneous data from the resource processing module to construct a resource profile; and for Class I energy supply terminals, it constructs a load model and outputs load forecast data; and for Class II energy supply terminals, it constructs a random process model of their output and response and outputs new energy output data; The aggregate analysis module is used to receive and process data from the data analysis module and set the optimal charge and discharge plan for each energy storage unit and the benchmark scheduling plan for the adjustable load according to a preset strategy. The preset strategy includes using the MILP algorithm to solve the optimal charge and discharge plan and the benchmark scheduling plan based on load forecast data and renewable energy output data on a day-ahead time scale, and performing periodic updates on an intraday time scale. The collaborative control module is used to generate scheduling instructions based on the optimal charging and discharging plan and the benchmark scheduling scheme and send them to the corresponding energy supply end.

[0007] Option 2 A distributed resource aggregation analysis method, using a distributed resource aggregation analysis system as described in Solution 1 to perform resource aggregation analysis, comprises the following steps: Collect data and parse protocols for connected distributed energy supply terminals, build resource profiles for each energy supply terminal based on historical data, and generate resource profile tags that characterize its operating characteristics; For Class I energy supply terminals, a load model is constructed and load forecast data is output; and for Class II energy supply terminals, a random process model of their output and response is constructed and new energy output data is output; Setting the optimal charge and discharge plan for each energy storage unit and the benchmark scheduling scheme for the adjustable load according to a preset strategy; the preset strategy includes using the MILP algorithm to solve the optimal charge and discharge plan and the benchmark scheduling scheme based on load forecast data and renewable energy output data on a day-ahead time scale; and periodically updating them on an intraday time scale; Based on the optimal charging and discharging plan and the benchmark scheduling scheme, a scheduling instruction is generated and sent to the corresponding energy supply end, and the execution deviation is monitored in real time. When the deviation between the actual output of the new energy and the predicted value exceeds the preset threshold and lasts for a certain period of time, the emergency coordinated stabilization strategy is automatically triggered, and the energy storage unit in the optimal working range is preferentially called for power compensation.

[0008] The working principle and advantages of the present invention are: This invention proposes a distributed resource aggregation analysis method and system that efficiently integrates and accurately quantifies power resources, effectively improving the observability, measurability, and controllability of distributed power resources. Key points: This solution specifically implements differentiated aggregation analysis for scenarios with a high proportion of renewable energy access. The data analysis module specifically models different energy supply types: for Class I energy supply terminals, a load forecasting model is used to capture patterns in human electricity consumption behavior, while for Class II energy supply terminals, a stochastic process model is established to quantify the uncertainties of energy sources such as wind and solar power. This differentiated approach significantly improves the accuracy of resource feature modeling, providing reliable input for subsequent optimized scheduling. Furthermore, compared to existing research that has focused on scheduling at a single time scale, this solution utilizes a mixed integer linear programming (MILP) algorithm to solve the optimal plan at the day-ahead scale, while performing periodic rolling updates at the intraday scale. This ensures both forward-looking planning and the flexibility of real-time adjustments. Furthermore, this solution integrates controllable loads and renewable energy sources into a unified framework for coordinated optimization. By jointly solving energy storage charging and discharging plans with load scheduling schemes, it enables complementary regulation across resource types, resulting in a more flexible and comprehensive scheduling mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a schematic diagram of the system structure of a distributed resource aggregation analysis method and system embodiment of the present invention. DETAILED DESCRIPTION

[0010] The following is a further detailed description through specific implementation methods: The embodiment is basically as shown in the attached Figure 1 Shown: A distributed resource aggregation analysis system, including a resource processing module, a data analysis module, an aggregation analysis module, a collaborative control module, a resource aggregation visualization module and a simulation deduction module.

[0011] The resource processing module is used to collect data and parse protocols for the connected distributed energy supply terminals, which include Class I energy supply terminals whose energy source is controllable loads and Class II energy supply terminals whose energy source is new energy.

[0012] When the resource processing module collects data, the types of data collected include: operating data and environmental data of the distributed energy supply end; the operating data includes real-time active power, reactive power, charge state, energy supply equipment temperature, grid connection point voltage, and historical charge and discharge data of the energy storage unit; the environmental data includes real-time meteorological data within a radius of 5 kilometers centered on the geographical location of each distributed energy supply end, and the real-time meteorological data includes irradiance, ambient temperature, wind speed, and wind direction.

[0013] Specifically, in this embodiment, data is collected from Supervisory Control and Data Acquisition (SCADA), Energy Management System (EMS), Remote Terminal Units (RTUs) at each plant or station, or Internet of Things (IoT) gateways. For Class I energy supply terminals (controllable loads, including industrial adjustable equipment and flexible residential loads), macro-level power consumption data such as total active power, total reactive power, switch status, and key circuit currents are collected on a minute-by-minute basis (e.g., 5-15 minute intervals). For Class II energy supply terminals (renewable energy sources, including photovoltaic power stations and distributed photovoltaic / wind power), detailed operating status data is collected on a high frequency basis of seconds (e.g., 1-5 seconds), including but not limited to real-time active / reactive power, DC-side voltage / current, AC-side voltage / current, internal temperature, alarm information, grid connection point frequency, and energy storage device-specific state of charge (SOC), state of health (SOH), and charge and discharge cycles.

[0014] Furthermore, the resource processing module has a built-in protocol adaptation engine for performing protocol parsing functions; it supports establishing connections with Profinet protocol and MPI / PPI protocol commonly used for communication with Class I energy supply terminals (such as industrial programmable logic controllers PLCs), as well as IEC 61850, Modbus TCP, MQTT, DNP3 and other protocols commonly used for communication with Class II energy supply terminals (photovoltaic inverters, energy storage converters PCS, wind turbine controllers).

[0015] The resource processing module is also provided with a pre-processing unit; the pre-processing unit is used to perform data cleaning and quality verification. Specifically, in this embodiment, preset data validity rules are applied for real-time cleaning and verification, and the data validity rules include: range verification - checking whether the data is within a reasonable physical range (such as the SOC value is between 0-100%); jump verification - identifying sudden changes between adjacent data points (such as the power change rate exceeds 20% of the rated value) and smoothing or marking them; correlation verification - for Class II energy supply terminals, cross-validating meteorological data (such as when the irradiance is zero at night, the photovoltaic output should be zero).

[0016] The data analysis module is used to receive and process multi-source heterogeneous data from the resource processing module to construct a resource profile. For Class I energy supply terminals, it builds a load model and outputs load forecast data. For Class II energy supply terminals, it builds a stochastic process model of their output and response and outputs new energy output data.

[0017] Specifically, building a resource profile involves the following steps: Based on the data from various energy supply terminals, a resource identifier is assigned to each energy supply terminal. These resource identifiers include Class I resource tags—interruptible load, shiftable load, maximum curtailment power, minimum response time, and comfort constraints; and Class II resource tags—PV / storage type, rated capacity, current adjustable capacity, weather dependency, volatility level, and SOH health status.

[0018] When constructing the random process model, the data analysis module includes the following steps: Stochastic differential equations are used to model the output fluctuation of the Class II energy supply terminal. For the photovoltaic Class II energy supply terminal where the new energy source is solar energy, the photovoltaic output The dynamics of is described by the Jacobi process: ; in, is the change in output over time; mean reversion rate , conditional mean , volatility coefficient It is generated through dynamic mapping of a pre-trained gradient boosting decision tree model based on real-time irradiance and ambient temperature. Specifically, the real-time collected meteorological data such as total radiation irradiance, diffuse irradiance, cloud cover, ambient temperature, wind speed, humidity, etc. are used to form a feature vector M(t). A pre-trained gradient boosting decision tree (GBDT) regression model is then used, taking M(t) and its historical window data as input, and outputting the Jacobi process parameters at the current moment in real time online. The GBDT model excels at capturing complex nonlinear relationships between features and can effectively learn how meteorological conditions affect the fluctuating characteristics of output.

[0019] By using the Jacobi process to replace the traditional single deterministic prediction point, we can use its inherent ability to constrain random variables to a fixed interval. characteristics, which can perfectly match the normalized photovoltaic output The physical boundaries of the photovoltaic output can simultaneously describe the fluctuation range of photovoltaic output (limited by 0 and the maximum available output) and its internal mean reversion and random diffusion characteristics, thereby achieving accurate evaluation of photovoltaic output.

[0020] The data analysis module is provided with an energy storage health status assessment unit, which is used to calculate the cumulative throughput power using the ampere-hour integration method based on the collected historical charge and discharge data of the energy storage unit (including cycle start / end SOC, total charge capacity, total discharge capacity, average charge and discharge power, duration, average temperature, etc.), and combined with its rated cycle life according to the formula: , dynamically estimates its health status and automatically reduces the adjustable capacity upper limit of energy storage units with SOH values ​​below 80% to 90% of their nominal value.

[0021] The energy storage health assessment unit regularly calculates the latest SOH value (e.g., daily). If the SOH value falls below 80%, a capacity correction factor (0.9) is automatically generated and sent to the aggregation analysis module and the coordinated control module. When formulating scheduling plans, these downstream modules multiply the maximum charge and discharge power and maximum available capacity of the energy storage by the capacity correction factor. This effectively accounts for capacity degradation caused by equipment aging during control, avoids issuing instructions that exceed the equipment's actual capabilities, and helps extend its service life.

[0022] The data analysis module is also provided with an output characteristic index calculation unit for calculating the volatility index and ramp risk index of photovoltaic output. When calculating the volatility, the photovoltaic output is calculated within a rolling time window (such as 15 minutes). The standard deviation of the first-order differences (differences between consecutive data points) is used as a quantitative indicator of short-term volatility, namely, a volatility index. When calculating ramp risk, the maximum increase and decrease rates of PV output per unit time (e.g., 1 minute, 5 minutes) under different weather types are first calculated based on historical PV output data and corresponding environmental data to form a ramp event probability distribution. In this embodiment, an extreme ramp event is identified when the increase rate exceeds a certain maximum value (e.g., 50 kW / min) or the absolute value of the decrease rate exceeds a certain maximum value (e.g., -60 kW / min, with an absolute value of 60 kW / min). Based on the ramp event probability distribution obtained above, the probability of an extreme ramp event is calculated as the ramp risk indicator. For example, the probability of an extreme ramp event under a certain weather type and unit time = the number of samples with an increase rate exceeding the threshold under that weather type and unit time / the total number of samples of that type.

[0023] The aggregation analysis module is used to receive and process data from the data analysis module, and set the optimal charging and discharging plan of each energy storage unit and the benchmark scheduling plan of the adjustable load according to the preset strategy.

[0024] The preset strategy includes, on a day-ahead time scale, using the MILP algorithm to solve the optimal charging and discharging plan and benchmark scheduling scheme based on load forecast data and renewable energy output data; and performing periodic updates on an intraday time scale.

[0025] When using the MILP algorithm to solve the optimal charging and discharging plan and the benchmark scheduling scheme, a MILP model is established with the goal of minimizing the total operating cost of the virtual power plant. The objective function of the MILP model is: ; in, is the cost of purchasing electricity from the main grid during period t, It is the online electricity price. is the purchased power; is the revenue from selling electricity to the main grid during period t, It is the on-grid electricity price. is the power sold, and the negative sign represents the revenue; is the energy storage loss cost, is the energy storage depreciation coefficient, It is the absolute sum of the charging and discharging power of i energy storage units.

[0026] The constraints of the MILP model include: system power balance constraint, energy storage SOC dynamic update constraint and its operation boundary constraint, new energy output uncertainty constraint, curtailment constraint and duration constraint.

[0027] In this embodiment, the system power balance constraint is specifically expressed as: .

[0028] in, Indicates the power purchased from the main grid during period t, represents the total output of all photovoltaic units (i) in period t, represents the sum of the charging and discharging power of all energy storage units (i) in time period t (charging is negative and discharging is positive); represents the load demand during period t, represents the load reduction during period t.

[0029] The energy storage SOC dynamic update constraint is specifically expressed as: ; in, Indicates the state of charge of the i-th energy storage unit at time t (the value range is 0-100%); represents the charge and discharge power of the i-th energy storage unit during period t; represents the time step, Represents the rated capacity of the i-th energy storage unit.

[0030] The operational boundary constraint of the energy storage SOC is specifically expressed as: .

[0031] In this embodiment, The value can be 10%~20%. The value can be 90%~95%.

[0032] The uncertainty constraint of new energy output is specifically expressed as: .

[0033] in, is the maximum charging power of the i-th energy storage unit, and the negative sign indicates that the power is negative during charging; is the maximum discharge power of the i-th energy storage unit.

[0034] The reducible amount constraint is specifically expressed as: .in, represents the power reduction of the jth controllable load in period t, It represents the maximum allowable power reduction of the j-th controllable load in time period t.

[0035] The duration constraint is specifically expressed as follows: a load can be called at most N times in a day, or the continuous calling time cannot exceed M periods. The values ​​of M and N are set according to the actual power supply demand.

[0036] Through the above constraints, the feasible domain of the MILP model can be jointly defined, allowing the MILP model to optimize the scheduling strategy while satisfying the physical rules.

[0037] In the preset strategy, an update is performed once every 15 minutes on a daily time scale. Based on the latest actual SOC data and new energy output data, with the goal of minimizing the adjustment amount, the MPC framework (model predictive control framework) is used to correct the optimal charging and discharging plan and benchmark scheduling plan for the previous day, and output the equivalent adjustable power upper limit, equivalent adjustable power lower limit and climbing rate curve of the aggregate.

[0038] In the preset strategy, on the intraday time scale, the objective function constructed with the goal of minimizing the adjustment amount is: ; in, is the power of energy storage planned in the past few days, is the adjusted power.

[0039] The collaborative control module is used to generate scheduling instructions based on the optimal charging and discharging plan and the benchmark scheduling scheme and send them to the corresponding energy supply end. Specifically, in this embodiment, the collaborative control module receives the optimal charging and discharging plan and the benchmark scheduling scheme (JSON format) from the aggregation analysis module through a message queue (such as Kafka), and extracts the scheduling instructions of each energy supply end at each time point (for example: for energy storage units, the scheduling instructions include the power setting value, expected state of charge and operation mode at each time point; for controllable loads, the scheduling instructions include the planned power reduction, baseline load forecast value and instruction status at each time point), and then sends them to the device gateway or controller corresponding to each energy supply end. All instructions are accompanied by a unique instruction ID, a generation timestamp and an execution validity period (if there is no response within 15 minutes, it will automatically expire), and are sent through a standardized protocol (such as MQTT-SN) to adapt to the communication capabilities of different energy supply ends.

[0040] The resource aggregation visualization module is used to perform the following operations: on a web-based GIS map, different resource types, including photovoltaics and energy storage, are represented by circular markers of different colors. The size of the marker represents its rated capacity, and the depth of the marker color represents the proportion of its current adjustable capacity to the rated capacity. For example, dark red represents an adjustable capacity close to 100%, and light pink represents an adjustable capacity of less than 20%.

[0041] The simulation module is used to allow users to customize meteorological scenarios (such as customizing meteorological parameters through sliders, drop-down boxes, table entry, etc., including setting light intensity, temperature, wind speed, cloud cover coefficient, radiation attenuation rate, etc.), and drive the data analysis module and the aggregation analysis module to perform simulation calculations, and output an assessment report on the total output range, expected benefits and expected scheduling risks (such as load loss risk, that is, the probability of load shedding due to insufficient output of new energy and depletion of energy storage power) of the aggregate in the next 24 hours under the customized scenario.

[0042] During simulation calculations, the environmental data sources of the data analysis module and the aggregation analysis module are switched to user-defined meteorological scenarios. Specifically, when a simulation is triggered, the simulation module intercepts the regular environmental data acquisition requests of the data analysis module and the aggregation analysis module (such as obtaining data from a real-time meteorological monitoring system or a historical database) through middleware (such as a message queue) and redirects them to the simulation scenario data cache. The simulation scenario data cache generates simulated environmental data according to the user-defined meteorological scenario and in a time series (with a time step of 15 minutes or 1 hour), including photovoltaic irradiation prediction series, temperature series, etc., and injects them into the data analysis module according to the original data format and frequency, ensuring that the module logic can run based on the simulation data without modification.

[0043] This embodiment also provides a distributed resource aggregation analysis method, which uses the above-mentioned distributed resource aggregation analysis system to perform resource aggregation analysis; the method includes the following steps: Data collection and protocol analysis are performed on the connected distributed energy supply terminals, and resource portraits are built for each energy supply terminal based on historical data, and resource portrait labels are generated to characterize its operating characteristics.

[0044] For Class I energy supply terminals, a load model is constructed and load forecast data is output; and for Class II energy supply terminals, a random process model of their output and response is constructed and new energy output data is output.

[0045] The optimal charging and discharging plan for each energy storage unit and the benchmark scheduling scheme for the adjustable load are set according to a preset strategy. The preset strategy includes, on a day-ahead time scale, using the MILP algorithm to solve the optimal charging and discharging plan and the benchmark scheduling scheme based on load forecast data and renewable energy output data; and periodically updating them on an intraday time scale.

[0046] Based on the optimal charging and discharging plan and the benchmark scheduling scheme, a scheduling instruction is generated and sent to the corresponding energy supply end, and the execution deviation is monitored in real time. When the deviation between the actual output of renewable energy and the predicted value exceeds a preset threshold (such as 15%, which can be temporarily relaxed to 20% when the photovoltaic prediction error is large on cloudy days) and lasts for a certain period of time (such as 3 minutes to 5 minutes), the emergency coordinated smoothing strategy is automatically triggered, and energy storage units in the optimal working range are preferentially called for power compensation.

[0047] In this embodiment, the optimal working range is defined as: SOC is between 40% and 80% (taking into account both charge and discharge flexibility and life protection), (in good health), (low energy loss) and no fault alarm within 30 minutes.

[0048] When performing power compensation, the initial compensation amount is set equal to the actual output deviation value of the new energy (if the actual output is lower than the predicted value, energy storage discharge is required to compensate; otherwise, energy storage charging is required to absorb it). It is then allocated to the selected energy storage units according to the proportion of energy storage capacity (for example, if the total compensation demand is 500kW, energy storage capacity A is 2MWh and energy storage capacity B is 3MWh, then A will bear 200kW and B will bear 300kW). At the same time, the compensation power of a single energy storage unit is limited to no more than 80% of its rated power to avoid overload.

[0049] During execution, the compensation effect is updated every 30 seconds: if the deviation falls within the threshold, the current compensation amount is maintained; if the deviation still exceeds the threshold, the secondary energy storage (SOC 30%-40% or 80%-90%) is activated for additional compensation; if the deviation does not converge after three consecutive updates, the Class I energy supply is coordinated for adjustment (such as temporarily reducing power by invoking interruptible loads). If the deviation remains below the threshold for two minutes, the strategy automatically exits, and the energy storage gradually returns to the original charging and discharging plan (the recovery rate does not exceed 50% of its maximum ramp rate to avoid impacting the grid). At the same time, emergency process data (trigger time, compensation amount, energy storage response speed, etc.) is uploaded to the aggregation analysis module to optimize the charging and discharging plan for the next day.

[0050] The distributed resource aggregation analysis method and system provided in this embodiment can efficiently integrate power supply resources and accurately quantify them, effectively improving the observability, measurability, and controllability of distributed power supply resources.

[0051] At the resource access and perception level, a protocol adaptation engine achieves broad compatibility with heterogeneous modeling devices, ensuring comprehensive and real-time data collection. It also deeply integrates meteorological data with equipment operation data, providing a reliable data foundation for subsequent accurate modeling and forecasting. At the modeling level, this solution abandons traditional deterministic methods and specifically employs stochastic differential equation theory, specifically the Jacobi process, to characterize the non-Gaussian and non-stationary fluctuation characteristics of photovoltaic output. The model parameters are dynamically generated from real-time meteorological data through a machine learning model. This ensures that the model not only has a solid mathematical and physical foundation but also can dynamically adapt to changes in the external environment, significantly improving the accuracy of quantifying the uncertainty of renewable energy output.

[0052] At the aggregation and scheduling level, this plan sets up an optimization framework that closely connects multiple time scales of day-ahead and intraday. Day-ahead optimization formulates a preliminary plan with the best economic efficiency, while intraday rolling optimization corrects deviations in real time based on ultra-short-term forecasts. Through a hierarchical and progressive optimization mechanism, it effectively balances the economy and robustness of the scheduling plan, which helps to improve the power grid's ability to absorb distributed renewable energy and its operational safety.

[0053] To validate the effectiveness of this solution, a simulation analysis was conducted on power dispatch during peak summer hours at an industrial park in Xinjin District. The scenario was set for a week in August, with the weather forecast indicating sunny to cloudy weather with scattered afternoon showers. The power grid issued a demand response order from 2:00 PM to 3:00 PM the following day, requiring the area to reduce its load by 1.5 MW.

[0054] Aggregated resources include: Class II energy supply (new energy) - park rooftop photovoltaic power station (total capacity 4MW), distributed energy storage station (total capacity 2MW / 4MWh); Class I energy supply (controllable load) - Factory A (production line can be interrupted, maximum reduction of 800kW), Technology Company B (central air conditioning can be adjusted, maximum reduction of 400kW), Data Center C (cooling system can be adjusted, maximum reduction of 300kW).

[0055] Day-ahead phase—Developing an economically optimal plan: The resource processing module collects the status of all resources and indicates an average SOC of 65% for energy storage. Based on the weather forecast (sunny to partly cloudy), the data analysis module generates a probabilistic forecast of PV output using a Jacobi process model. The forecast indicates an expected output of 3.2 MW at 2:00 PM the next day, but with high uncertainty (a 70% probability of between 2.8 MW and 3.5 MW). The aggregation analysis module performs MILP day-ahead optimization. The calculations show that the most economical solution to meet the 1.5 MW reduction target is to utilize Class I resources, with Factory A reducing power by 800 kW, Company B by 400 kW, and Data Center C by 300 kW (a total of 1.5 MW). This solution offers the lowest cost and avoids depreciation losses by eliminating the need for energy storage. This baseline dispatch plan is then generated and submitted to the grid dispatch center.

[0056] Intraday Phase—Real-Time Adjustment: At 1:45 PM the next day, the weather suddenly changed, bringing early rainfall. Real-time monitoring by the resource processing module indicated that actual PV output had plummeted to 1.8 MW, 1.4 MW below the previous day's forecast. If only 1.5 MW of load were cut as originally planned, the park's total load would have been short by 2.9 MW: 1.4 MW (PV shortfall) + 1.5 MW (planned reduction), a potential drop in the park's internal grid frequency or even a blackout.

[0057] In this case, the data analysis module immediately triggered an ultra-short-term forecast update, predicting that PV output would remain low for the next hour. The aggregation analysis module triggered rolling MPC optimization. Under the new constraints (significantly reduced PV output), the calculation was recalculated with the goal of minimizing adjustments and avoiding power outages. A new decision was then output: immediately terminating the curtailment order for Data Center C (due to its high priority, power was restored to ensure server safety) and initiating a coordinated energy storage stabilization strategy. The coordinated control module instructed the energy storage station to discharge at a maximum power of 1 MW. It also instructed Factory A and Company B to maintain the original curtailment plan (800 kW + 400 kW). The total compensation power was now 1 MW (storage) + 1.2 MW (load reduction) = 2.2 MW, making up for most of the shortfall.

[0058] This demonstrates that the system can rapidly compensate for power shortfalls and intelligently avoid grid frequency fluctuations and potential power outages. In the absence of disturbances, the system selects the lowest-cost load response solution through day-ahead optimization, saving energy storage depreciation costs. In extreme situations, rapid adjustments prevent the substantial economic losses that could result from power outages, safeguarding production safety. Furthermore, through precise stochastic prediction and multi-timescale optimization, it maximizes the utilization of photovoltaic power generation, converting it into stable output through energy storage even under fluctuating conditions, significantly reducing curtailment.

[0059] The above is only an embodiment of the present invention. Common knowledge such as the specific structure and characteristics of the scheme is not described in detail here. Ordinary technicians in the relevant field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the guidance of this application. Some typical well-known structures or well-known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent.

Claims

1. A distributed resource aggregation analysis system, characterized in that: It includes resource processing module, data analysis module, aggregation analysis module and collaborative control module; The resource processing module is used to collect data and parse protocols for the connected distributed energy supply terminals; the distributed energy supply terminals include Class I energy supply terminals whose energy source is controllable loads and Class II energy supply terminals whose energy source is new energy; The data analysis module is used to receive and process multi-source heterogeneous data from the resource processing module to construct a resource profile; and for Class I energy supply terminals, it constructs a load model and outputs load forecast data; For the Class II energy supply end, a random process model of its output and response is constructed and the new energy output data is output; The aggregate analysis module is used to receive and process data from the data analysis module and set the optimal charge and discharge plan for each energy storage unit and the benchmark scheduling plan for the adjustable load according to a preset strategy. The preset strategy includes using the MILP algorithm to solve the optimal charge and discharge plan and the benchmark scheduling plan based on load forecast data and renewable energy output data on a day-ahead time scale. And it is updated periodically on the intraday time scale; The collaborative control module is used to generate scheduling instructions based on the optimal charging and discharging plan and the benchmark scheduling scheme and send them to the corresponding energy supply end.

2. A distributed resource aggregation analysis system according to claim 1, characterized in that: When the resource processing module collects data, the types of data collected include: operating data and environmental data of the distributed energy supply end; the operating data includes real-time active power, reactive power, charge state, energy supply equipment temperature, grid connection point voltage, and historical charge and discharge data of the energy storage unit; the environmental data includes real-time meteorological data within a radius of 5 kilometers centered on the geographical location of each distributed energy supply end, and the real-time meteorological data includes irradiance, ambient temperature, wind speed, and wind direction.

3. A distributed resource aggregation analysis system according to claim 2, characterized in that: When constructing the random process model, the data analysis module includes the following steps: Stochastic differential equations are used to model the output fluctuation of the Class II energy supply terminal. For the photovoltaic Class II energy supply terminal where the new energy source is solar energy, the photovoltaic output The dynamics of is described by the Jacobi process: ； in, is the change in output over time; mean reversion rate , conditional mean , volatility coefficient It is generated through dynamic mapping of the gradient boosting decision tree model pre-trained by real-time irradiance and ambient temperature.

4. A distributed resource aggregation analysis system according to claim 2, characterized in that: The data analysis module is provided with an energy storage health status assessment unit, which is used to calculate the cumulative throughput power based on the collected historical charge and discharge data of the energy storage unit using the ampere-hour integration method, and combined with its rated cycle life according to the formula: , dynamically estimates its health status and automatically reduces the adjustable capacity upper limit of energy storage units with SOH values ​​below 80% to 90% of their nominal value.

5. A distributed resource aggregation analysis system according to claim 2, characterized in that: When using the MILP algorithm to solve the optimal charging and discharging plan and the benchmark scheduling scheme, a MILP model is established with the goal of minimizing the total operating cost of the virtual power plant. The objective function of the MILP model is: ； in, is the cost of purchasing electricity from the main grid during period t, It is the online electricity price. is the purchased power; is the revenue from selling electricity to the main grid during period t, It is the on-grid electricity price. is the power sold, and the negative sign represents the revenue; is the energy storage loss cost, is the energy storage depreciation coefficient, is the absolute sum of the charging and discharging power of i energy storage units; The constraints of the MILP model include: system power balance constraints, energy storage SOC dynamic update constraints and its operation boundary constraints, and new energy output uncertainty constraints.

6. A distributed resource aggregation analysis system according to claim 1, characterized in that: In the preset strategy, an update is performed once every 15 minutes on a daily time scale. Based on the latest actual SOC data and new energy output data, with the goal of minimizing the adjustment amount, the MPC framework is used to correct the optimal charging and discharging plan and benchmark scheduling plan for the day before, and the equivalent adjustable power upper limit, equivalent adjustable power lower limit and climbing rate curve of the aggregate are output.

7. A distributed resource aggregation analysis system according to claim 6, characterized in that: In the preset strategy, on the intraday time scale, the objective function constructed with the goal of minimizing the adjustment amount is: ； in, is the power of energy storage planned in the past few days, is the adjusted power.

8. A distributed resource aggregation analysis system according to claim 1, characterized in that: It also includes a resource aggregation visualization module; the resource aggregation visualization module is used to perform the following operations: on a web-based GIS map, different resource types, including photovoltaics and energy storage, are represented by circular marks of different colors. The size of the mark represents its rated capacity, and the depth of the mark represents the proportion of its current adjustable capacity to the rated capacity.

9. A distributed resource aggregation analysis system according to claim 1, characterized in that: It also includes a simulation and deduction module, which is used for users to customize meteorological scenarios, and drive the data analysis module and the aggregation analysis module to perform simulation calculations, and output an evaluation report on the total output range, expected benefits and expected scheduling risks of the aggregate in the next 24 hours under the customized scenario.

10. A distributed resource aggregation analysis method, characterized in that: Applying a distributed resource aggregation analysis system according to any one of claims 1 to 9 to perform resource aggregation analysis comprises the following steps: Collect data and parse protocols for connected distributed energy supply terminals, build resource profiles for each energy supply terminal based on historical data, and generate resource profile tags that characterize its operating characteristics; For Class I energy supply terminals, a load model is constructed and load forecast data is output; and for Class II energy supply terminals, a random process model of their output and response is constructed and new energy output data is output; Setting the optimal charge and discharge plan for each energy storage unit and the benchmark scheduling scheme for the adjustable load according to a preset strategy; the preset strategy includes using the MILP algorithm to solve the optimal charge and discharge plan and the benchmark scheduling scheme based on load forecast data and renewable energy output data on a day-ahead time scale; and periodically updating them on an intraday time scale; Based on the optimal charging and discharging plan and the benchmark scheduling scheme, a scheduling instruction is generated and sent to the corresponding energy supply end, and the execution deviation is monitored in real time. When the deviation between the actual output of the new energy and the predicted value exceeds the preset threshold and lasts for a certain period of time, the emergency coordinated stabilization strategy is automatically triggered, and the energy storage unit in the optimal working range is preferentially called for power compensation.

Images