Industrial park multi-energy dynamic complementary scheduling method and system based on multi-source data

Abstract

The invention discloses an industrial park multi-energy dynamic complementary scheduling method and system based on multi-source data, and belongs to the technical field of energy scheduling. The method comprises the following steps: an acquisition module, a first determination module, a second determination module, a constraint determination module and a scheduling scheme determination module. According to the method, the power change rate is predicted in real time and quantified into the dynamic stability margin, then the dynamic stability margin is used as the rigid constraint to be embedded into the economic dispatching optimization model, the standby capacity is reserved in advance in the dispatching decision-making stage to prevent risks, the balance of economy and safety is converted into economic optimization within the safety boundary, and the economic optimization efficiency is improved. According to the method, the safety requirement is improved from a soft target to a hard constraint, active beforehand defense of an electric energy quality problem is realized, economic optimal scheduling is realized while the system stability is guaranteed, and the problem that voltage sag and frequency fluctuation are difficult to effectively suppress in renewable energy fluctuation processing in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy scheduling, in particular to an industrial park multi-energy dynamic complementary scheduling method and system based on multi-source data. BACKGROUND

[0002] The penetration rate of sustainable power resources such as wind power and photovoltaic power in industrial park multi-energy systems is continuously increasing, and has become an important part of new power systems. However, the output of sustainable power sources is highly dependent on environmental factors, and its inherent intermittency and severe volatility pose a serious challenge to the power quality of the park power grid, especially frequent voltage sag, frequency deviation and other problems, which have become a key constraint to high-quality power supply in the park.

[0003] In related technologies, the processing of sustainable power fluctuations mainly relies on two types of solutions: one is to use an output prediction model based on historical weather data to arrange traditional unit output adjustment in advance to balance fluctuations; the second is to configure energy storage systems (ESS), static var generators (SVG) and other devices on the grid side to compensate for fluctuations afterwards.

[0004] However, the above solutions have significant limitations: first, the prediction model accuracy decreases dramatically when encountering extreme weather or rapid weather changes, and scheduling instructions based on incorrect predictions can even amplify system risks; second, the response of energy storage and SVG devices is based on local measurements, which is a "after-the-fact" adjustment and has a millisecond delay, which cannot completely avoid the impact of voltage sag on sensitive loads. Therefore, the existing technology cannot achieve proactive suppression and coordinated control of power quality problems, and cannot meet the dual needs of safe and economic operation of the park power grid with a high proportion of renewable energy access. SUMMARY

[0005] The embodiments of the present application provide an industrial park multi-energy dynamic complementary scheduling method and system based on multi-source data, which solves the problem that the processing of renewable energy fluctuations in the prior art cannot effectively suppress voltage sag and frequency fluctuations, and achieves proactive prevention of power quality problems, while ensuring system stability and achieving optimal economic scheduling.

[0006] The embodiments of the present application provide an industrial park multi-energy dynamic complementary scheduling method based on multi-source data, which is applied to an industrial park multi-energy dynamic complementary scheduling system based on multi-source data, and includes the following steps: Step 1, real-time acquisition and synchronous processing of multi-source data of the industrial park, wherein the multi-source data includes historical output and weather data, real-time output and weather data, and historical and real-time power grid operation state data; Step 2, based on historical output and meteorological data, offline training time series prediction model, input real-time output and meteorological data, predict power value sequence in future prediction window, and then determine the expected maximum power change rate by calculating the maximum power difference between adjacent time points in the power value sequence; Step 3, based on historical grid operation state data, by statistical disturbance event, the calibrated grid inertia response time constant is calculated, according to the expected maximum power change rate, through the preset margin quantization function, the dynamic stability margin value is calculated; Step 4, based on real-time grid operation state data, an economic dispatch optimization model is constructed to minimize the total operation cost of the park, and the dynamic stability margin value is set as the lower limit of the total available standby capacity constraint in the economic dispatch optimization model; Step 5, the economic dispatch optimization model is solved by using the optimization solver, and the optimal dispatch scheme under the total available standby capacity constraint and other operation constraints is obtained, and the optimal dispatch scheme is the optimal dispatch instruction set for each controllable energy unit in the park, and is executed.

[0007] Further, the determination of the expected maximum power change rate comprises: input real-time output and meteorological data into the trained time series prediction model to obtain the predicted power value sequence of the future N prediction window steps ; calculate the absolute value of the power difference between all adjacent time points in the sequence, that is , ; Finally, select the maximum value , and divide by the prediction window step , to obtain the expected maximum power change rate .

[0008] Further, the specific form of the margin quantization function is: ; wherein, is the dynamic stability margin value; is the absolute value of the expected maximum power change rate; is the grid inertia response time constant; is the fixed minimum standby capacity of the system.

[0009] Further, the dynamic stability margin value is set as the lower limit of the total available standby capacity constraint in the economic dispatch optimization model, that is, the constraint condition of the economic dispatch optimization model includes the following inequality: ; wherein, is the available reserve capacity that the jth controllable energy unit in the park can provide at time t; M is the total number of controllable energy units; is the dynamic stability margin value.

[0010] Further, the controllable energy unit includes: a gas turbine, an energy storage system, and an interruptible industrial load that has signed a demand response agreement; The optimal dispatching instruction is issued to the local controller of the corresponding unit through an industrial communication protocol for execution.

[0011] Further, the historical and real-time power grid operating state data includes total park load, power grid frequency, and bus voltage; The historical power grid operating state data is used for calibration of the power grid inertia response time constant; The real-time power grid operating state data is used for power balance constraints of the economic dispatching optimization model.

[0012] Further, the economic dispatching optimization model is a mixed integer linear programming model; The optimization solver solves the economic dispatching optimization model, and an output solution vector contains optimal active power set values and start-stop states of each controllable energy unit, which is directly converted into the optimal dispatching instruction set.

[0013] Further, the total available reserve capacity constraint is a constraint condition that requires that the total system reserve capacity is not less than the dynamic stability margin value.

[0014] Further, the other operating constraints include: power balance constraints for generation and power consumption balance, unit operating constraints for limiting the output range of each power generation device, energy storage system constraints for regulating the charging and discharging power and energy state of the energy storage system, and load constraints for limiting the interruption range of the load.

[0015] Embodiments of the present application provide an industrial park multi-energy dynamic complementary dispatching system based on multi-source data, for implementing an industrial park multi-energy dynamic complementary dispatching method based on multi-source data, which includes: historical output and meteorological data, real-time output and meteorological data, and historical and real-time power grid operating state data; The first determination module is configured to train a time series prediction model offline based on the historical output and meteorological data, input real-time output and meteorological data, predict a power value sequence in a future prediction window, and then determine an expected maximum power change rate by calculating a maximum power difference value between adjacent time points in the power value sequence. The second determination module is configured to calculate the dynamic stability margin value by a preset margin quantization function according to the expected maximum power change rate, based on the historical power grid operating state data, by counting the calibrated power grid inertia response time constant in the disturbance event. The constraint determination module is used to construct an economic dispatch optimization model based on real-time power grid operation status data with the goal of minimizing the total operating cost of the park, and to set the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic dispatch optimization model. The scheduling scheme determination module is used to solve the economic scheduling optimization model using an optimization solver to obtain the optimal scheduling scheme that satisfies the total available reserve capacity constraint and other operational constraints. The optimal scheduling scheme is a set of optimized scheduling instructions for each controllable energy unit in the park, and is issued for execution.

[0016] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: By shifting the core of prediction from the absolute value of power generation to the rate of power change, which directly reflects the degree of grid impact, and quantifying it into a dynamic stability margin constraint that must be followed based on a clear physical mapping relationship, rather than a soft cost term that can be weighed in the objective function, this scheme forces the economic dispatch algorithm to perform economic optimization under the premise of satisfying this rigid security constraint. This enables the system to reserve precise buffer resources in advance for impending power surges, transforming passive compensation into active defense. As a result, it fundamentally suppresses power quality anomalies such as voltage dips and frequency deviations caused by renewable energy power fluctuations, significantly improving power supply reliability and the resilience of the park's power grid. Attached Figure Description

[0017] Figure 1 A flowchart of a multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data, provided in this application embodiment;

[0018] Figure 2 This is a schematic diagram of the structure of a multi-energy dynamic complementary scheduling system for industrial parks based on multi-source data, provided in an embodiment of this application. Detailed Implementation

[0019] This application provides a method and system for dynamic complementary scheduling of multiple energy sources in industrial parks based on multi-source data. It solves the problem that the existing technology is unable to effectively suppress voltage sags and frequency fluctuations when dealing with renewable energy fluctuations. By predicting the power change rate in real time and quantifying it into a dynamic stability margin, and then embedding it as a rigid constraint into the economic scheduling optimization model, it realizes proactive defense against power quality problems and achieves optimal economic scheduling while ensuring system stability.

[0020] In related technologies, handling renewable energy fluctuations mainly relies on predicting the absolute value of its output or configuring equipment for ex-post compensation. However, absolute power prediction has significant errors during sudden weather changes, and its prediction results are not directly linearly correlated with the actual impact intensity experienced by the grid. Ex-post compensation schemes have inherent delays and cannot completely avoid power quality events. If traditional methods are used, either inaccurate predictions will lead to misallocation of dispatch resources, or response delays will result in insufficient compensation, both of which will make it difficult to effectively suppress voltage sags and frequency fluctuations, thereby affecting the normal operation of production equipment in the industrial park.

[0021] Based on the aforementioned technical issues, this application focuses the prediction on the rate of power change, a physical quantity that directly characterizes the intensity of power grid impacts. It then transforms this rate of change into a concrete dynamic stability margin requirement through a margin quantification function. This dynamic stability margin value is embedded as a time-varying hard constraint into the economic dispatch optimization model, thereby driving the system to proactively and accurately reserve resources to cope with future impacts, achieving pre-emptive prevention.

[0022] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0023] like Figure 1 As shown, this application provides a method for dynamic complementary scheduling of multiple energy sources in industrial parks based on multi-source data. This method is applied to a dynamic complementary scheduling system for multiple energy sources in industrial parks based on multi-source data, and includes: Step 1: Collect and process multi-source data from the industrial park in real time. The multi-source data includes historical power output and meteorological data for model training, real-time power output and meteorological data for real-time prediction, and historical and real-time power grid operation status data for parameter calibration and economic dispatch.

[0024] In one specific embodiment, for a power industrial park that includes photovoltaic power plants, gas turbines, and energy storage systems, the system collects second-level photovoltaic power output data and second-level irradiance data from the on-site weather station over the past year. Simultaneously, it collects historical grid frequency data and reports of major disturbance events recorded by the grid dispatch center.

[0025] The power grid frequency refers to the alternating current frequency of the power system; major disturbance events refer to events that suddenly disrupt the power balance of the power grid and cause significant fluctuations in frequency or voltage. These mainly include the sudden switching of large-capacity loads, the sudden disconnection of generation units within the area from the grid, and sudden power changes in the interconnection lines with the main grid; reports of major disturbance events must include the event time, event type, event scale, key indicators before and after the event, and the actions of the control system.

[0026] Step 2: Based on historical power output and meteorological data, train the time series prediction model offline, and use the real-time power output and meteorological data as model input to predict the power value sequence within the future prediction window. Then, by calculating the maximum power difference between adjacent times in the power value sequence, determine the expected maximum power change rate.

[0027] In one specific embodiment, an LSTM neural network is trained using historical photovoltaic (PV) output and meteorological data. The model learns to predict the PV output curve for the next 5 minutes based on the changing trends of current meteorological characteristics such as irradiance and cloud cover.

[0028] Step 3: Based on the historical power grid operating status data, the dynamic stability margin value is calculated by using the power grid inertial response time constant calibrated in the statistical disturbance events and the expected maximum power change rate, through a preset margin quantification function. The dynamic stability margin is used to reflect the system adjustment capability reserved to resist expected power fluctuations.

[0029] In one specific embodiment, historical disturbance reports were analyzed to identify several events where large equipment tripped, causing a sudden drop in power. For example, one event recorded a 2MW power shortfall within 0.2 seconds, with the grid frequency dropping by 0.1Hz before backup was activated. Based on the grid's frequency response characteristics, this 0.1Hz frequency difference is equivalent to the system inertia automatically compensating for approximately 1MW of power. Through the analysis of multiple such events, the system's inertial response time constant was calculated to be approximately 0.1 minutes.

[0030] The specific method for obtaining the power grid inertial response time constant is as follows: Selected from historical database The effective perturbation event, for the first The first event is to record its power deficit. (Unit: MW) and by analyzing the frequency waveform curve, the initial frequency change rate (unit: Hz / s) is calculated. Then, based on the principle that the instantaneous power compensation provided by the system's inertial response is approximately equal to the power deficit at the initial moment of the disturbance, the estimated value of the time constant represented by this event is calculated. ,in Duration of the disturbance, in seconds. For example, a 2MW shock lasting 0.2 seconds. That's 0.2 seconds, or 0.2 / 60 ≈ 0.0033 minutes. Ultimately, through... This incident The arithmetic mean of the values ​​and their conversion to minutes yields the system's grid inertial response time constant. (Unit: min) is the average time span during which the system relies on its own inertia to withstand power surges.

[0031] Step 4: Based on real-time power grid operation status data, construct an economic dispatch optimization model with the goal of minimizing the total operating cost of the park, and set the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic dispatch optimization model.

[0032] Real-time power grid operation status data includes the park's real-time total load, key bus voltage, and system frequency.

[0033] Step 5: Use an optimization solver to solve the economic dispatch optimization model to obtain the optimal dispatch scheme that satisfies the total available reserve capacity constraint and other operational constraints. The optimal dispatch scheme is a set of optimized dispatch instructions for each controllable energy unit in the park, which is then issued and executed to actively suppress power quality anomalies caused by fluctuations in sustainable power resources.

[0034] Total available reserve capacity constraint refers to the constraint that the total reserve capacity of the system is not less than the dynamic stability margin value; other operational constraints include power balance constraints to ensure the balance between power generation and consumption, unit operation constraints to limit the output range of each power generation device, energy storage system constraints to specify the charging and discharging power and energy state of the energy storage system, and load constraints to limit the range of load interruption. These constraints together constitute the complete boundary conditions to ensure the safe and stable operation of the system.

[0035] Furthermore, determining the expected maximum power change rate includes: Offline training of time series prediction models was conducted based on historical power output data and historical meteorological data of sustainable power resources. The offline training of the time series prediction model specifically involves first collecting historical data as a training set, with input features being historical sustainable power output sequences and historical meteorological data sequences, labeled with corresponding future power value sequences; then preprocessing the data; next, building a time series prediction neural network; during training, the input sequence is input into the encoder to generate a context vector, which the decoder uses to gradually predict future sequences; the difference between the predicted and actual values ​​is calculated using a loss function, and the network weights are adjusted through error backpropagation using the Adam optimizer; this process is iterated until the model accuracy meets the requirements, and finally the optimal model is saved for real-time system use.

[0036] By inputting real-time power output and meteorological data into a trained time series prediction model, a sequence of predicted power values ​​for the next N prediction window steps is obtained. ; Calculate the absolute value of the power difference between all adjacent time points in the sequence, i.e. , ; Finally, select the maximum value. and divide by the prediction window step size The expected maximum power change rate is obtained. .

[0037] In one specific implementation, at the current scheduling moment, the latest photovoltaic power output and meteorological data are collected and input into a pre-trained LSTM model. The model outputs a power prediction sequence for the next 5 minutes (300 data points per second). The algorithm iterates through this sequence and finds that the largest predicted power drop per second is 0.3MW. Therefore, the expected maximum power change rate is calculated. .

[0038] The power slope prediction model is used to reflect the drastic changes in sustainable power output over a prediction step time. Its output is a sequence of expected maximum power change rates, which characterizes the maximum change in sustainable power output per unit time.

[0039] Furthermore, the specific form of the margin quantification function is as follows: ; in, This represents the dynamic stability margin value. This is the absolute value of the expected maximum power change rate; This is the power grid inertial response time constant, whose value is obtained by analyzing historical power grid disturbance events; This is the system's fixed minimum standby capacity.

[0040] Fixed minimum reserve capacity This is a constant set based on operational experience, which can be set to 2% to 5% of the park's maximum load to cope with unforeseen minor fluctuations and ensure reliability; the total available reserve capacity is a real-time calculated value, referring to the sum of the upward and downward adjustments of reserve capacity that all controllable energy units in the park can make under the current state, and its value is determined by the real-time operating status and physical limits of each unit; the total available reserve capacity constraint is an inequality constraint embedded in the economic dispatch optimization model in step 4, specifically... This mandates that the system's real-time backup capacity must be greater than or equal to the calculated dynamic stability margin; other operational constraints include power balance constraints, unit output upper and lower limit constraints, unit ramp rate constraints, and energy storage SOC constraints.

[0041] In one specific embodiment, the park sets a fixed minimum standby capacity. The value is 0.5MW. The dynamic stability margin is calculated using the formula: .

[0042] Furthermore, setting the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic scheduling optimization model means that the constraints of the economic scheduling optimization model include the following inequalities: ; in, Let M be the available backup capacity that the j-th controllable energy unit in the park can provide at time t; M is the total number of controllable energy units. This represents the dynamic stability margin value.

[0043] Specifically, the dynamic stability margin, a physical quantity reflecting the system's ability to withstand disturbances, is transformed into a quantifiable reserve capacity constraint and directly embedded into the economic dispatch optimization model, achieving two important transformations: First, it shifts from the traditional "post-event compensation" model to an "exponential prevention" model, reserving reserve capacity in advance during the dispatch decision-making stage to mitigate risks; second, it shifts from "economic and security trade-offs" to "economic optimization within the security boundary," elevating security requirements from soft targets to hard constraints, effectively solving the problem of power quality management under a high proportion of renewable energy access.

[0044] The constraints of the economic dispatch optimization model specifically include: 1) power balance constraints, which are used to ensure that the power generation and power consumption are equal in real time; 2) unit operation constraints, which are used to limit the output range of each power generation device; 3) energy storage system constraints, including charging and discharging power limits and energy state limits; 4) interruptible load constraints, which are used to specify the allowable range of load interruption; 5) dynamic reserve constraints, which require that the total reserve capacity of the system is not less than the dynamic stability margin value.

[0045] The aforementioned economic dispatch optimization model is input into an optimization solver (such as CPLEX or Gurobi) for solution. The solution process employs a deterministic optimization algorithm, searching for the optimal dispatch scheme that minimizes the total operating cost within the feasible solution space that satisfies all constraints. The solver output is directly converted into executable instructions for each controllable energy unit, including the gas turbine output setpoint, the energy storage system's charging and discharging power instructions, and the interruptible load reduction instructions.

[0046] Furthermore, the controllable energy unit includes: a gas turbine, an energy storage system, and interruptible industrial loads with a demand response agreement; The optimized scheduling instructions are sent to the local controller of the corresponding unit for execution via industrial communication protocols.

[0047] Among them, gas turbine commands are executed through the governor, energy storage system commands are executed through the power conversion system, and interruptible load commands are executed through the load management terminal.

[0048] In this embodiment, the system adopts a rolling optimization mechanism, performing a complete scheduling calculation every 5 minutes and updating the scheduling instructions based on the latest system status and forecast data. This ensures that the scheduling scheme can respond to changes in system status in a timely manner and achieve proactive defense against power quality problems.

[0049] Furthermore, the historical and real-time power grid operation status data includes the total load of the park, power grid frequency, and bus voltage; The historical power grid operating status data is used for calibration of the power grid inertial response time constant; The real-time power grid operation status data is used for power balance constraints in the economic dispatch optimization model.

[0050] Furthermore, the economic scheduling optimization model is a mixed-integer linear programming model; The optimization solver solves the economic scheduling optimization model, and the output solution vector contains the optimal active power setpoint and start / stop state of each controllable energy unit. This solution vector is directly converted into the optimized scheduling instruction set.

[0051] In this embodiment, the output results directly correspond to the control setpoints of each unit, ensuring the executability of the scheduling scheme. The entire scheme forms a complete technical closed loop from data acquisition to instruction issuance, with clear data flow and logical connections between each link. The solution vector output by the optimization solver is directly converted into an optimized scheduling instruction set through the instruction conversion module built into the park energy management system (EMS). This module automatically maps the power values ​​of each unit in the mathematical solution vector to the corresponding equipment control instructions, and encapsulates them according to standard industrial communication protocols such as Modbus TCP and IEC 104. Finally, the instructions are sent to the local controllers of each controllable unit for execution through the real-time communication interface of the scheduling system.

[0052] like Figure 2 As shown, this application provides an industrial park multi-energy dynamic complementary scheduling system based on multi-source data, used to implement the industrial park multi-energy dynamic complementary scheduling method based on multi-source data, including: an acquisition module, a first determination module, a second determination module, a constraint determination module, and a scheduling scheme determination module; The acquisition module is used to collect and process multi-source data from the industrial park in real time. The multi-source data includes historical power output and meteorological data, real-time power output and meteorological data, and historical and real-time power grid operation status data. The first determining module is used to train a time series prediction model offline based on historical power output and meteorological data, input real-time power output and meteorological data, predict the power value sequence within the future prediction window, and then determine the expected maximum power change rate by calculating the maximum power difference between adjacent times in the power value sequence. The second determining module is used to calculate the dynamic stability margin value based on historical power grid operating status data, by statistically analyzing the calibrated power grid inertial response time constant in disturbance events, and by using a preset margin quantification function according to the expected maximum power change rate. The constraint determination module is used to construct an economic scheduling optimization model with the goal of minimizing the total operating cost of the park, and to set the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic scheduling optimization model. The scheduling scheme determination module is used to solve the economic scheduling optimization model based on real-time power grid operation status data using an optimization solver to obtain the optimal scheduling scheme that satisfies the total available reserve capacity constraint and other operational constraints. The optimal scheduling scheme is a set of optimized scheduling instructions for each controllable energy unit in the park, and is issued for execution.

[0053] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0054] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0055] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0056] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0057] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0058] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data, characterized in that, Includes the following steps: Step 1: Collect and process multi-source data from the industrial park in real time. The multi-source data includes historical power output and meteorological data, real-time power output and meteorological data, and historical and real-time power grid operation status data. Step 2: Based on historical power output and meteorological data, train the time series prediction model offline, input real-time power output and meteorological data, predict the power value sequence within the future prediction window, and then determine the expected maximum power change rate by calculating the maximum power difference between adjacent times in the power value sequence. Step 3: Based on historical power grid operation data, the dynamic stability margin value is calculated by statistically analyzing the calibrated power grid inertial response time constant during disturbance events and using a preset margin quantification function according to the expected maximum power change rate. Step 4: Based on real-time power grid operation status data, construct an economic dispatch optimization model with the goal of minimizing the total operating cost of the park, and set the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic dispatch optimization model; Step 5: Use an optimization solver to solve the economic scheduling optimization model to obtain the optimal scheduling scheme that satisfies the total available reserve capacity constraint and other operational constraints. The optimal scheduling scheme is a set of optimized scheduling instructions for each controllable energy unit in the park, and then issues them for execution.

2. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, Determining the expected maximum power change rate includes: By inputting real-time power output and meteorological data into a trained time series prediction model, a sequence of predicted power values ​​for the next N prediction window steps is obtained. ; Calculate the absolute value of the power difference between all adjacent time points in the sequence, i.e. , ; Finally, select the maximum value. and divide by the prediction window step size The expected maximum power change rate is obtained. .

3. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The specific form of the margin quantification function is as follows: ； in, This represents the dynamic stability margin value. This is the absolute value of the expected maximum power change rate; The inertial response time constant of the power grid; This is the system's fixed minimum standby capacity.

4. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, Setting the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic scheduling optimization model means that the constraints of the economic scheduling optimization model include the following inequalities: ； in, Let M be the available backup capacity that the j-th controllable energy unit in the park can provide at time t; M is the total number of controllable energy units. This represents the dynamic stability margin value.

5. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The controllable energy unit includes: a gas turbine, an energy storage system, and interruptible industrial loads with a demand response agreement; The optimized scheduling instructions are sent to the local controller of the corresponding unit for execution via industrial communication protocols.

6. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The historical and real-time power grid operation status data includes the park's total load, power grid frequency, and bus voltage. The historical power grid operating status data is used for calibration of the power grid inertial response time constant; The real-time power grid operation status data is used for power balance constraints in the economic dispatch optimization model.

7. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The economic scheduling optimization model is a mixed integer linear programming model; The optimization solver solves the economic scheduling optimization model, and the output solution vector contains the optimal active power setpoint and start / stop state of each controllable energy unit. This solution vector is directly converted into the optimized scheduling instruction set.

8. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The total available reserve capacity constraint refers to the constraint that requires the total reserve capacity of the system to be no less than the dynamic stability margin value.

9. The multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in claim 1, characterized in that, The other operational constraints include: power balance constraints to balance power generation and consumption, unit operation constraints to limit the output range of each power generation device, energy storage system constraints to specify the charging and discharging power and energy state of the energy storage system, and load constraints to limit the range of load interruption.

10. A multi-energy dynamic complementary scheduling system for industrial parks based on multi-source data, used to implement the multi-energy dynamic complementary scheduling method for industrial parks based on multi-source data as described in any one of claims 1-9, characterized in that, include: Acquisition module, first determination module, second determination module, constraint determination module, scheduling scheme determination module; The acquisition module is used to collect and process multi-source data from the industrial park in real time. The multi-source data includes historical power output and meteorological data, real-time power output and meteorological data, and historical and real-time power grid operation status data. The first determining module is used to train a time series prediction model offline based on historical power output and meteorological data, input real-time power output and meteorological data, predict the power value sequence within the future prediction window, and then determine the expected maximum power change rate by calculating the maximum power difference between adjacent times in the power value sequence. The second determining module is used to calculate the dynamic stability margin value based on historical power grid operating status data, by statistically analyzing the calibrated power grid inertial response time constant in disturbance events, and by using a preset margin quantification function according to the expected maximum power change rate. The constraint determination module is used to construct an economic dispatch optimization model based on real-time power grid operation status data with the goal of minimizing the total operating cost of the park, and to set the dynamic stability margin value as the lower limit of the total available reserve capacity constraint in the economic dispatch optimization model. The scheduling scheme determination module is used to solve the economic scheduling optimization model using an optimization solver to obtain the optimal scheduling scheme that satisfies the total available reserve capacity constraint and other operational constraints. The optimal scheduling scheme is a set of optimized scheduling instructions for each controllable energy unit in the park, and is issued for execution.

Images