Power system distributed resource grey box response model parameter setting method

Abstract

The invention discloses a parameter setting method for a distributed resource grey box response model of a power system. The method comprises the following steps: collecting historical data; a model objective function is established, and the model represents the difference between a response curve Pi, pre fitted by using the virtual energy storage model every day and an actual response curve Pi, real; obtaining a solving model of the Pi, pre; setting initial parameters of a Pi, pre solving model; carrying out model solving; error feedback and parameter updating are carried out; carrying out convergence judgment; after the electricity price is determined, the power response curve of each load can be optimized and calculated according to the obtained parameters. According to the method, the internal parameters of the grey box response model are fitted by using the historical data of the user, and the adjusting quantity obtained by optimization approaches the target by adjusting the parameters, so that more accurate and flexible power system scheduling is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power system dispatching and optimization, and in particular to a parameter setting method for a gray box response model of distributed resources in a power system. Background Art

[0002] As the access ratio of renewable energy continues to increase, traditional power system optimization methods face many challenges. In particular, how to accurately dispatch the power system under the influence of the volatility of new energy output and the complexity of demand response has become a key issue. Traditional optimization methods include deterministic optimization methods, scenario-based stochastic optimization, and robust optimization. When faced with uncertainty and complex environments, they lack the ability to dynamically respond to uncertainty, the model assumptions are too idealized, and the computational complexity is high. It is often difficult to provide efficient scheduling solutions. To this end, this patent fits the model parameters through the inverse optimization method to achieve more flexible adaptation and optimization of uncertainty. The parameter setting method based on inverse optimization can effectively overcome these problems. Summary of the invention

[0003] The purpose of the present invention is to provide a parameter setting method for a gray box response model of distributed resources in an electric power system, which utilizes the user's historical data to fit the internal parameters of its gray box response model, and by adjusting these parameters, the optimized regulation amount is close to the target, thereby achieving more accurate and flexible power system scheduling.

[0004] In one aspect of the present invention, the present invention proposes a method for parameter setting of a gray box response model of distributed resources in a power system. According to an embodiment of the present invention, the method comprises the following steps:

[0005] Step 1: Collect historical data, where the types of historical data include historical power curves and historical electricity price curves of distributed resources in the power system;

[0006] Step 2: Establish a model objective function, where the model represents the response curve P fitted by the virtual energy storage model every day. i,pre And the actual response curve P i,real the gap between;

[0007] Step 3: Establish the response curve P obtained by solving the fitting i,pre Optimization model of

[0008] Step 4: Set P i,pre Solve the initial parameters of the model, either by random initialization or by making preliminary estimates using historical data;

[0009] Step 5: Solve the model using the selected solution algorithm. In each iteration, the objective function value is calculated using the current parameter values, and the optimization variables are adjusted according to the constraints.

[0010] Step 6: Convergence judgment: when the change of the objective function is less than a preset threshold and the constraint conditions are met, the iteration is stopped and the optimal solution is output;

[0011] Step 7: After the electricity price is determined, the power response curve of each load can be optimized and calculated according to the optimal solution obtained in step 6.

[0012] In addition, the method for setting parameters of a gray box response model of distributed resources in a power system according to the above embodiment of the present invention may also have the following additional technical features:

[0013] In some embodiments of the present invention, in step 1, the historical power curve of the distributed resources of the power system includes the following data: wind power data, photovoltaic power data, energy storage system power data, and demand response power data.

[0014] In some embodiments of the present invention, the wind power data includes meteorological parameters and the corresponding wind power output; the photovoltaic power data includes meteorological parameters and the corresponding photovoltaic power station power generation; the energy storage system power data includes real-time charging power and discharging power data of the energy storage device, as well as the charging status, battery capacity and efficiency of the energy storage device; the demand response power data includes the user load response curve, the duration and intensity of the demand response event.

[0015] In some embodiments of the present invention, in step 1, the historical electricity price curve includes real-time electricity prices, daytime electricity prices and nighttime electricity prices.

[0016] In some embodiments of the present invention, in step 2, the model objective function established is as follows:

[0017]

[0018] In the formula, N represents the number of days included in the sample, i represents the i-th day, and P i,pre represents the response curve fitted by the virtual energy storage model on the i-th day, P i,real Represents the actual response curve on day i.

[0019] In some embodiments of the present invention, in step 3, by P i,pre The solution model is as follows:

[0020] sP i,real =minl i P i (2)

[0021]

[0022] In the formula, i represents the i-th day, li represents the electricity price on the i-th day, P i represents the total power on the i-th day, represents the power at time t, P t,max represents the maximum power at time t, P t,min represents the minimum power at time t, E max Indicates the maximum power generated by the generator, E min represents the minimum power generation of the generator, and T represents the total number of moments.

[0023] In some embodiments of the present invention, in step 4, the initial parameter is P t,max ,P t,min ,E max ,E min .

[0024] In some embodiments of the present invention, step 5 is specifically as follows:

[0025] Use equations (7)-(10) to find the derivative of the objective function for each parameter, understand the sensitivity of the objective function to each parameter, and then adjust the parameters according to the value of the derivative in the optimization. Then use the perturbation method to slightly change the value of the parameter and calculate The amount of change;

[0026]

[0027] In the formula, Respectively represent the objective function of the four parameters P t,max ,P t,min ,E max ,E min The partial derivative of , which represents the rate of change of the objective function when the parameter changes; DP t,max , DP t,min ,DE max ,DE min The four parameters P are t,max ,P t,min ,E max ,E min A small disturbance is applied.

[0028] In some embodiments of the present invention, in step 6, the convergence judgment method is as follows:

[0029] Adjust P according to equations (11)-(14) t,max ,P t,min ,E max ,E min The values ​​of these parameters are checked for convergence, and then return to step 3 to continue iteration;

[0030]

[0031] In the formula, They respectively represent the weighting factors of the partial derivatives of the objective function with respect to the four parameters.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention has significant advantages in flexibility, efficiency and accuracy, meets the urgent need for efficient scheduling of modern power systems in the context of increasing proportion of new energy access, and has broad application prospects and practical value. The present invention improves the adaptability and prediction accuracy of the model in complex scenarios by accurately fitting the load response behavior. Using historical data and inverse optimization methods, the model can dynamically adapt to the complex changes of new energy output fluctuations and demand response. The iterative optimization and convergence judgment mechanism are adopted to reduce the computational complexity, save computing resources, and achieve fast and efficient parameter setting. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of a method for setting parameters of a gray box response model of distributed resources in a power system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] A method for setting parameters of a gray box response model of distributed resources in a power system comprises the following steps:

[0037] Step 1: Collect historical data and preprocess

[0038] 1.1 Determine the data source

[0039] First, clarify the types of historical data that need to be collected, including the historical power curve and historical electricity price curve of distributed resources. Specifically, the following types of data are included:

[0040] 1.1.1 Historical power curve of distributed resources:

[0041] (1) Wind power data: including the actual output data of wind farms in different time periods. Usually, it can be obtained through the wind farm monitoring system or the power grid dispatching system. The data content should include meteorological parameters such as wind speed, wind direction, temperature, humidity, and the corresponding wind power output.

[0042] (2) Photovoltaic power data: including the actual power generation data of photovoltaic power stations in different time periods. The data content should include factors such as sunshine intensity, temperature, weather conditions, and the corresponding power generation of photovoltaic power stations.

[0043] (3) Energy storage system power data: including the charging and discharging power of the energy storage system. It is necessary to collect the real-time charging power and discharging power data of the energy storage equipment, as well as the equipment's charging status, battery capacity, efficiency and other parameters.

[0044] (4) Demand response power data: historical regulation data of the demand response system, including user load response curves, duration and intensity of demand response events, etc.

[0045] 1.1.2 Historical electricity price curve:

[0046] Collect electricity market price data in historical time periods, including real-time electricity prices, daytime electricity prices, and nighttime electricity prices. Electricity price data is usually provided by the electricity market trading platform, power companies, or regional power grid dispatching centers. Data should be provided at hourly or shorter time granularity to facilitate subsequent demand response model fitting.

[0047] 1.2 Channels and methods of data collection:

[0048] 1.2.1 Distributed resource power data: Obtain real-time power data through the monitoring systems of the power grid dispatching center, wind farms and photovoltaic power stations. For wind power and photovoltaic power, data collection should ensure that the power generation curves under different weather conditions are covered to avoid abnormal data interference caused by special weather conditions.

[0049] 1.2.2 Electricity price data: Obtain historical electricity price data of the electricity market from the electricity market trading platform or the power grid dispatch center. Electricity price data is usually provided by the market regulatory department or power company, and can be obtained through API interface, regular reports or historical database.

[0050] 1.3 Data Preprocessing

[0051] 1.3.1 Data cleaning:

[0052] (1) Remove outliers: Filter out sudden changes in wind power, photovoltaic or demand response curves.

[0053] (2) Data completion: Use interpolation or statistical methods to complete missing electricity price or power data.

[0054] 1.3.2 Normalization processing:

[0055] Normalize historical data (such as power and price) to the interval [0,1] to reduce the impact of dimension on optimization.

[0056] 1.3.3 Time Alignment:

[0057] Ensure that power data and electricity price data are aligned in time series and handle data sources with different time spans.

[0058] 1.3.4 Denoising:

[0059] A filtering method (such as the sliding average method) is used to smooth the historical power curve and eliminate random noise.

[0060] 1.3.5 Feature extraction:

[0061] Key variables (such as wind speed and light intensity) are extracted from meteorological data as influencing factors.

[0062] Step 2: Use formula (1) to establish the model objective function. The model represents the response curve P fitted by the virtual energy storage model every day. i,pre And the actual response curve P i,real The gap between:

[0063]

[0064] In the formula, N represents the number of days included in the sample, i represents the i-th day, and P i,pre represents the response curve fitted by the virtual energy storage model on the i-th day, P i,real Represents the actual response curve on day i.

[0065] This step improves the model's ability to fit the load response behavior by minimizing the error function, thereby enhancing the model's adaptability in complex scenarios.

[0066] Step 3: Establish an optimization model to solve the response curve fitted by the type. Using equations (2)-(6) to obtain P i,pre The solution model, P i,pre It is the result of solving an optimization model and requires subsequent steps to perform iterative calculations, which explicitly express the mathematical expression of the optimization problem.

[0067] sP i,real =minl i P i (2)

[0068]

[0069] In the formula, l i represents the electricity price on the i-th day, P i represents the total power on the i-th day, represents the power at time t, P t,max represents the maximum power at time t, P t,min represents the minimum power at time t, E maxIndicates the maximum power generation of the generator, E min represents the minimum power generation of the generator, and T represents the total number of moments. Since the model of each load is unknown, the subsequent steps need to solve P t,max ,P t,min , E max ,E min optimal values ​​for these parameters.

[0070] Step 4: Set the initial parameters. Set the initial parameters P t,max ,P t,min ,E max ,E min The value of can be initialized randomly or preliminarily estimated through historical data to provide a starting point for the iterative optimization in step 5.

[0071] Step 5: Use equations (7)-(10) to calculate the derivative of the objective function F in step 2 for each parameter. The purpose is to understand the sensitivity of the objective function to each parameter, so that the parameters can be adjusted according to the value of the derivative during optimization. Then use the perturbation method to slightly change the value of the parameter and calculate The feedback mechanism is used to optimize the parameters so that the error is gradually reduced. The error is calculated based on the optimization result, and the parameter value is adjusted to provide updated parameters for the next iteration.

[0072]

[0073] In the formula, Respectively represent the objective function of the four parameters P t,max ,P t,min ,E max ,E min The partial derivative of , which represents the rate of change of the objective function when the parameter changes; DP t,max , DP t,min ,DE max ,DE min The four parameters P are t,max ,P t,min ,E max ,E min A small perturbation is applied to slightly change the value of the parameter.

[0074] Step 6: Perform convergence judgment. When the change of the objective function is less than a preset threshold and the constraints are met, stop the iteration and output the optimal solution.

[0075] Specifically, adjust P according to equations (11)-(14) t,max ,P t,min ,E max ,E minThe values ​​of these parameters are checked for convergence, and whether the error meets the preset threshold, and whether to stop the optimization. If optimization needs to be continued, return to step 3 to continue the iteration.

[0076]

[0077] In the formula They respectively represent the weighting factors of the partial derivatives of the objective function with respect to the four parameters.

[0078] Step 7: After the electricity price is determined, the P t,max ,P t,min ,E max ,E min These parameters are optimized to calculate the power response curve for each load.

[0079] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, they shall all fall within the protection scope of the present invention.

Claims

1. A method for parameter setting of a gray box response model of distributed resources in a power system, characterized in that: The following steps are involved: Step 1: Collect historical data, where the types of historical data include historical power curves and historical electricity price curves of distributed resources in the power system; Step 2: Establish a model objective function, where the model represents the response curve P fitted by the virtual energy storage model every day. i,pre And the actual response curve P i,real the gap between; Step 3: Establish the response curve P obtained by solving the fitting i,pre Optimization model of Step 4: Set P i,pre Solve the initial parameters of the model, either by random initialization or by making preliminary estimates using historical data; Step 5: Solve the model using the selected solution algorithm. In each iteration, the objective function value is calculated using the current parameter values, and the optimization variables are adjusted according to the constraints. Step 6: Convergence judgment: when the change of the objective function is less than a preset threshold and the constraint conditions are met, the iteration is stopped and the optimal solution is output; Step 7: After the electricity price is determined, the power response curve of each load can be optimized and calculated according to the optimal solution obtained in step 6.

2. A method for parameter setting of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: In step 1, the historical power curve of the distributed resources of the power system includes the following data: wind power data, photovoltaic power data, energy storage system power data, and demand response power data.

3. A method for setting parameters of a gray box response model of distributed resources in a power system according to claim 2, characterized in that: The wind power data includes meteorological parameters and the corresponding wind power output; the photovoltaic power data includes meteorological parameters and the corresponding photovoltaic power station power generation; the energy storage system power data includes the real-time charging power and discharging power data of the energy storage device, as well as the charging status, battery capacity and efficiency of the energy storage device; The demand response power data includes user load response curve, duration and intensity of demand response events.

4. A method for parameter setting of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: In step 1, the historical electricity price curve includes real-time electricity price, daytime electricity price and nighttime electricity price.

5. A method for parameter setting of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: In step 2, the model objective function established is as follows: In the formula, N represents the number of days included in the sample, i represents the i-th day, and P i,pre represents the response curve fitted by the virtual energy storage model on the i-th day, P i,real Represents the actual response curve on day i.

6. A method for setting parameters of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: In step 3, P i,pre The solution model is as follows: Wed Mon i,real =minl i P i (2) In the formula, l i represents the electricity price on the i-th day, i represents the i-th day, P i represents the total power on the i-th day, represents the power at time t on the i-th day, P t,max represents the maximum power at time t, P t,min represents the minimum power at time t, E max Indicates the maximum power generated by the generator, E min represents the minimum power generation of the generator, and T represents the total number of moments.

7. A method for setting parameters of a gray box response model of distributed resources in a power system according to claim 6, characterized in that: In step 4, the initial parameter is P t,max ,P t,min ,E max ,E min .

8. A method for parameter setting of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: The step 5 is specifically as follows: Use equations (7)-(10) to find the derivative of the objective function for each parameter. In the optimization, adjust the parameter according to the value of the derivative. Then use the perturbation method to slightly change the value of the parameter and calculate The amount of change; In the formula, Respectively represent the objective function of the four parameters P t,max ,P t,min ,E max ,E min The partial derivative of , which represents the rate of change of the objective function when the parameter changes; DP t,max , DP t,min ,DE max ,DE min The four parameters P are t,max ,P t,min ,E max ,E min A small disturbance is applied.

9. A method for parameter setting of a gray box response model of distributed resources in a power system according to claim 1, characterized in that: In step 6, the convergence judgment method is as follows: Adjust P according to equations (11)-(14) t,max ,P t,min ,E max ,E min The values ​​of these parameters are checked for convergence, and then return to step 3 to continue the iteration; In the formula, They respectively represent the weighting factors of the partial derivatives of the objective function with respect to the four parameters.

Images