Primary frequency regulation method and apparatus for controllable load resource participating in power system, computer device, computer-readable storage medium, and computer program product

Abstract

The present application relates to the technical field of power system automatic control, and discloses a primary frequency regulation method and apparatus for a controllable load resource participating in a power system, a computer device, a computer-readable storage medium, and a computer program product. The method comprises: acquiring load status data of a power system, and performing multidimensional load clustering on the load status data to obtain a load characteristic time series data set; on the basis of the load characteristic time series data set, using an extended dynamic mode decomposition algorithm to establish a Koopman operator high-dimensional linear model of the power system; using the Koopman operator high-dimensional linear model of the power system to optimize and calculate the load status of the power system to obtain an optimal load control sequence; and using the optimal load control sequence to perform frequency regulation on a controllable load resource participating in the power system. The present application achieves stable prediction of the load status of the power system, and accurate frequency regulation for the controllable load resource participating in the power system, thereby improving the operating efficiency and stability of the power system.

Description

A method, apparatus, computer equipment, computer-readable storage medium, and computer program product for controlling load resources to participate in primary frequency regulation of a power system.

[0001] Cross-references to related applications

[0002] This application claims priority to Chinese Patent Application No. 202411614769.2, filed on November 12, 2024, entitled "A Method and Apparatus for Controllable Load Resources Participating in Primary Frequency Regulation of a Power System", the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application relates to the field of power system automatic control technology, specifically to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for controllable load resources to participate in primary frequency regulation of a power system. Background Technology

[0004] The power grid is a giant inertial system. According to the rotor motion equation, when the active power of the power grid is insufficient, the generator rotor accelerates, and the power grid frequency increases; conversely, the power grid frequency decreases. Therefore, primary frequency regulation is one of the means to dynamically ensure the active power balance of the power grid.

[0005] Power systems rely on generator-side governors to maintain system frequency stability. However, with the large-scale integration of distributed energy resources (DERs), the inertia of the power system decreases, leading to faster acceleration or deceleration of machine rotors during disturbance events. Summary of the Invention

[0006] In view of this, this application provides a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for controlling load resources to participate in the primary frequency regulation of a power system, in order to solve the problems of reduced power system inertia and insufficient frequency stability caused by the large-scale access of distributed energy resources.

[0007] In a first aspect, this application provides a method for controllable load resources to participate in the primary frequency regulation of a power system, the method comprising:

[0008] Obtain load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a time series dataset of load characteristics;

[0009] Based on a time-series dataset of load characteristics, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm with the Koopman operator is established.

[0010] The load state of the power system is optimized by using a high-dimensional linear model of the Koopman operator of the power system to obtain the optimal load control sequence;

[0011] Frequency regulation of controllable load resources participating in the power system is carried out using the optimal load control sequence.

[0012] This embodiment provides a method for controlling load resources to participate in primary frequency regulation of a power system. By performing multidimensional load clustering on load state data, the accurate distribution of power system loads in the load characteristic time series dataset is ensured, guaranteeing load coordination and stability. Furthermore, based on the load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator. This model maps complex nonlinear load dynamics to a high-dimensional linear space, simplifying the prediction complexity of power system load states and achieving global linearization of each load state. Finally, by optimizing the power system load states, the optimal load control sequence is obtained. This optimal load control sequence is then used to regulate the frequency of controllable load resources participating in the power system, achieving stable prediction of power system load states and precise frequency regulation of controllable load resources, thereby improving the operating efficiency and stability of the power system.

[0013] In one optional implementation, multidimensional load clustering is performed on the load state data to obtain a time-series dataset of load characteristics, including:

[0014] Normalize the load status data of the power system;

[0015] The normalized load status data is clustered to obtain the initial load feature groups;

[0016] The initial load feature group is adaptively adjusted to obtain a time series dataset of load features.

[0017] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By normalizing the load state data of the power system, it ensures that each feature data in the load characteristic time series dataset has the same weight in the subsequent model construction process, avoiding the amplification of prediction errors caused by differences in the magnitude of load features. Furthermore, by clustering the normalized load state data, it ensures that the load characteristics within each initial load characteristic group are similar, while the differences between initial load characteristic groups are large, facilitating the distributed optimization and regulation of the power system load in the later stages. Finally, by adaptively adjusting the initial load characteristic groups, it ensures that the loads within the load characteristic time series dataset respond synchronously, guaranteeing that the load management of the power system is always in an optimal state, improving the coordination and stability of the power system load. Optionally, by dynamically adjusting the groups and introducing a consistency control strategy, it ensures that the loads within the group respond synchronously, guaranteeing load coordination and grid stability, and achieving efficient management of the power system load.

[0018] In one optional implementation, based on a load characteristic time-series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established, comprising:

[0019] By embedding the time-series dataset of load characteristics with delayed coordinates, a high-dimensional state vector is obtained;

[0020] A high-dimensional state vector is selected as the basis function, and the least squares method is used to fit the basis function to obtain the Koopman operator;

[0021] Obtain the control input matrix and control variables, and establish a high-dimensional linear model of the power system based on the Koopman operator, control input matrix, and control variables.

[0022] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By embedding delayed coordinates, it transforms the one-dimensional time-series data in the load characteristic time-series dataset into a state representation in a multi-dimensional space. This makes the nonlinear dynamics of the power system more linear in the high-dimensional space, providing a reliable data foundation for subsequent basis function selection and extended dynamic mode decomposition algorithms. Furthermore, by fitting the basis functions using the least squares method, the Koopman operator is obtained, achieving an accurate mapping of the linear relationship between load states. This makes the high-dimensional linear model of the Koopman operator more accurate, realizing a linearized description and dynamic prediction of the power system load state.

[0023] In an optional implementation, based on a load characteristic time-series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator, further comprising:

[0024] The effectiveness of the basis functions is verified, and the basis functions are optimized based on the verification results.

[0025] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By verifying the effectiveness of the basis functions, the linearization effectiveness of the high-dimensional linear model of the Koopman operator of the power system is guaranteed. This enables the optimized basis functions to effectively capture the dynamic behavior of the power system, thereby realizing the linear description and dynamic prediction of the load state of the power system.

[0026] In one optional implementation, the load state of the power system is optimized using a high-dimensional linear model of the power system's Koopman operator to obtain the optimal load control sequence, including:

[0027] Collect the power system load status data at the current moment, input the power system load status data at the current moment into the high-dimensional linear model of the power system's Koopman operator, and obtain the power system load status control sequence at the predicted moment;

[0028] The load state control sequence of the power system at the predicted time is optimized to obtain the optimized load state control sequence;

[0029] With the goal of minimizing frequency deviation and control cost, an optimization objective function is constructed based on the load state optimization control sequence.

[0030] The optimal load control sequence is obtained by solving the objective function.

[0031] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. With the goal of minimizing frequency deviation and control cost, an optimization objective function is constructed based on the load state optimization control sequence. The optimization objective function is then solved to obtain the optimal load control sequence, which enables the power system to minimize frequency deviation and control cost while meeting constraints, thereby achieving optimal load regulation and ensuring the stability of the power system.

[0032] In one optional implementation, the power system load state control sequence at the predicted time is optimized to obtain a load state optimized control sequence, including:

[0033] The power system load state data at the initial time is obtained, and the power system load state data at the initial time is linearly mapped to the prediction time using the Koopman operator to obtain the power system load state data at the prediction time.

[0034] Based on the power system load state data at the prediction time, the prediction confidence interval is determined using a Bayesian estimation algorithm;

[0035] The load state control sequence of the power system at the prediction time is dynamically adjusted based on the prediction confidence interval to obtain the load state optimized control sequence.

[0036] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. Based on the power system load state data at the predicted time, a Bayesian estimation algorithm is used to determine the prediction confidence interval. Then, the power system load state control sequence at the predicted time is dynamically adjusted through the prediction confidence interval. This ensures stable control of the power system load in the face of complex disturbances and uncertainties, enhances the robustness of the power system to errors, measurement noise and external disturbances, and improves the reliability of the power system in load regulation.

[0037] Secondly, this application provides a controllable load resource participating in the primary frequency regulation of a power system, the device comprising:

[0038] The clustering module is used to acquire load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a time series dataset of load characteristics.

[0039] A module is established to build a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm based on a time-series dataset of load characteristics.

[0040] The optimization solution module is used to optimize the load state of the power system using the high-dimensional linear model of the Koopman operator of the power system, and obtain the optimal load control sequence.

[0041] The regulation module is used to regulate the frequency of controllable load resources participating in the power system using the optimal load control sequence.

[0042] Thirdly, this application provides a computer device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the controllable load resource participation in the primary frequency regulation method of the power system described in the first aspect or any corresponding embodiment.

[0043] Fourthly, this application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the method for controlling load resources to participate in the primary frequency regulation of a power system according to the first aspect or any corresponding embodiment described above.

[0044] Fifthly, this application provides a computer program product, including computer instructions for causing a computer to execute the method for controlling load resources to participate in primary frequency regulation of a power system according to the first aspect or any corresponding embodiment described above. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0046] Figure 1 is a flowchart illustrating a method for controllable load resources to participate in primary frequency regulation of a power system according to an embodiment of this application;

[0047] Figure 2 is a flowchart illustrating another method for controllable load resources to participate in primary frequency regulation of a power system according to an embodiment of this application;

[0048] Figure 3 is a flowchart illustrating another method for controllable load resources to participate in primary frequency regulation of a power system according to an embodiment of this application;

[0049] Figure 4 is a structural block diagram of a controllable load resource participating in the primary frequency regulation device of a power system according to an embodiment of this application;

[0050] Figure 5 is a schematic diagram of the hardware structure of a computer device according to an embodiment of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0052] Currently, demand-side control of power grids, especially through intelligent load control to adapt to grid dynamics, has become an important means of frequency regulation. Koopman theory provides a method to transform nonlinear dynamic systems into linear problems, reducing computational complexity and making it suitable for big data-driven prediction and control.

[0053] Therefore, this application provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By integrating advanced Koopman model predictive control technology with an intelligent load management system, the method effectively regulates the primary frequency of controllable load resources participating in the power system, thereby improving the frequency stability and response speed of the power system.

[0054] This application provides a method for controllable load resources to participate in the primary frequency regulation of a power system. It should be noted that the executing entity of this method can be a device for controlling load resources to participate in the primary frequency regulation of the power system. This device can be implemented as part or all of an electronic device through software, hardware, or a combination of both. The electronic device can be a server or a terminal. In this application embodiment, the server can be a single server or a server cluster composed of multiple servers. The terminal in this application embodiment can be a smartphone, personal computer, tablet computer, wearable device, or other intelligent hardware device such as an intelligent robot. The following method embodiments all use an electronic device as the executing entity for illustration.

[0055] According to an embodiment of this application, a method for controlling load resources to participate in primary frequency regulation of a power system is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0056] This embodiment provides a method for controllable load resources to participate in the primary frequency regulation of a power system, which can be used in the aforementioned electronic equipment. Figure 1 is a flowchart of a method for controllable load resources to participate in the primary frequency regulation of a power system according to an embodiment of this application. As shown in Figure 1, the process includes the following steps:

[0057] Step S101: Obtain the load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a load feature time series dataset.

[0058] Specifically, by deploying PMUs (Phasor Measurement Units) and RTUs (Remote Terminal Units) in the power system, load status data of the power system can be monitored in real time. The load status data includes the load's voltage, current, power, and frequency. Among them, the PMU can provide high-precision phasor measurement data, while the RTU is responsible for collecting and transmitting data from remote devices, ensuring the stability and reliability of the power system.

[0059] Optionally, a central controller is deployed in the power system. The central controller receives monitoring data from the PMU and RTU (i.e., load status data of the power system) and analyzes the real-time load status of the power system. Through communication with distributed actuators at the load points, each actuator integrates an intelligent controller, which can dynamically adjust the load according to the instructions of the central controller to cope with load changes in the power grid and improve the operating efficiency and stability of the power system.

[0060] Optionally, to effectively manage the load of the power system, a multidimensional load clustering method is introduced, and a load characteristic time series dataset is constructed based on load characteristics (such as geographical location, load type and historical response characteristics), meteorological data (such as temperature, wind speed, etc.) and real-time load status (such as fluctuation frequency, fluctuation amplitude, etc.).

[0061] Step S102: Based on the load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator.

[0062] Step S103: The load state of the power system is optimized and solved using the high-dimensional linear model of the Koopman operator of the power system to obtain the optimal load control sequence.

[0063] Step S104: Frequency regulation of controllable load resources participating in the power system is performed using the optimal load control sequence.

[0064] Specifically, the load is adjusted according to the optimal load control sequence, and the prediction and control are continuously updated through a real-time feedback mechanism to form a closed-loop control.

[0065] Optionally, when a frequency disturbance occurs in the power system, the loads of each group are adjusted. The power system quickly identifies and predicts local frequency changes. Based on the prediction results, the MPC (Model Predictive Control) strategy is used to dynamically adjust the power of each group of loads and provide upward / downward reserve power, thereby quickly restoring the system frequency to the rated value.

[0066] This embodiment provides a method for controlling load resources to participate in primary frequency regulation of a power system. By performing multidimensional load clustering on load state data, the accurate distribution of power system loads in the load characteristic time series dataset is ensured, guaranteeing load coordination and stability. Furthermore, based on the load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator. This model maps complex nonlinear load dynamics to a high-dimensional linear space, simplifying the prediction complexity of power system load states and achieving global linearization of each load state. Finally, by optimizing the power system load states, the optimal load control sequence is obtained. This optimal load control sequence is then used to regulate the frequency of controllable load resources participating in the power system, achieving stable prediction of power system load states and precise frequency regulation of controllable load resources, thereby improving the operating efficiency and stability of the power system.

[0067] This embodiment provides a method for controllable load resources to participate in the primary frequency regulation of a power system, which can be used in the aforementioned electronic equipment. Figure 2 is a flowchart of a method for controllable load resources to participate in the primary frequency regulation of a power system according to an embodiment of this application. As shown in Figure 2, the process includes the following steps:

[0068] Step S201: Obtain the load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a load feature time series dataset.

[0069] Specifically, step S201 includes:

[0070] Step S2011: Normalize the load status data of the power system.

[0071] Specifically, the load status data of the power system is normalized to eliminate the influence of dimensions and ensure that data with different characteristics are within the same numerical range. The Z-score (standard score) normalization method is used for normalization, and the calculation process is as follows:

[0072] Where, x norm This represents the normalized load state data, where x represents the load state data, μ represents the mean, and σ represents the standard deviation.

[0073] Optionally, a time-series data window length can be set for the normalized load status data. Based on the periodic or non-periodic changes of different load characteristics, the model can be ensured to capture short-term and long-term dynamic behaviors, thereby improving prediction accuracy. In the data preprocessing, an appropriate time-series data window length is selected, and a multi-window mechanism is introduced. By combining short-term and long-term load fluctuations, time-series data with different window lengths are used, enabling the model to simultaneously capture both rapid short-term changes and long-term trends.

[0074] Step S2012: Cluster the normalized load status data to obtain the initial load feature group.

[0075] Specifically, the K-means++ algorithm (an improvement on the traditional K-means algorithm) is used to determine the initial cluster centers, avoiding the sensitivity of the K-means algorithm (an unsupervised learning algorithm for cluster analysis) to initial points. The clustering objective is to minimize the sum of squared distances between load samples (i.e., normalized load status data) and their cluster centers, ensuring similar load characteristics within clusters and large differences between clusters, thus achieving more reasonable load grouping. The expression for the K-means++ algorithm is as follows:

[0076] Where N represents the total number of samples of load status data after normalization, q represents the number of clusters, and c j Let x represent the j-th cluster center. i This represents the i-th load status data.

[0077] Step S2013: Adaptively adjust the initial load feature group to obtain the load feature time series dataset.

[0078] Specifically, clustering makes the load characteristics distribution of the power system clearer. However, load characteristics are not static. Therefore, to cope with load changes over time, an adaptive online clustering update mechanism is used to ensure that the power system is always in an optimal state. Since the real-time changes in grid load require dynamic adjustment of group configuration, an online clustering update algorithm is introduced to recalculate cluster centers based on real-time load behavior and adaptively adjust groups to ensure that the load management of the power system is always in an optimal state.

[0079] Optionally, a gradual update mechanism is adopted, adjusting cluster centers step by step with small increments as new load status data arrives to avoid large abrupt changes. An online clustering algorithm is introduced to dynamically adjust group configurations based on real-time load behavior, maintaining optimal load characteristic groups. Simultaneously, a collaborative control strategy, such as a consensus algorithm, is employed within the load characteristic groups to ensure synchronized load response and reduce individual biases. The calculation formula for cluster center updates is shown below:

[0080] in, C represents the updated cluster centers. j This indicates the initial load characteristic group.

[0081] Step S202: Based on the load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator. For details, please refer to step S101 of the embodiment shown in Figure 1, which will not be repeated here.

[0082] Step S203 involves optimizing the load state of the power system using a high-dimensional linear model with the Koopman operator to obtain the optimal load control sequence. For details, please refer to step S103 of the embodiment shown in Figure 1; it will not be repeated here.

[0083] Step S204 involves using the optimal load control sequence to regulate the frequency of controllable load resources participating in the power system. For details, please refer to step S104 of the embodiment shown in Figure 1, which will not be repeated here.

[0084] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By normalizing the load state data of the power system, it ensures that each feature data in the load characteristic time series dataset has the same weight in the subsequent model construction process, avoiding the amplification of prediction errors caused by differences in the magnitude of load features. Furthermore, by clustering the normalized load state data, it ensures that the load characteristics within each initial load characteristic group are similar, while the differences between initial load characteristic groups are large, facilitating the distributed optimization and regulation of the power system load in the later stages. Finally, by adaptively adjusting the initial load characteristic groups, it ensures that the loads within the load characteristic time series dataset respond synchronously, guaranteeing that the load management of the power system is always in an optimal state, improving the coordination and stability of the power system load. Optionally, by dynamically adjusting the groups and introducing a consistency control strategy, it ensures that the loads within the group respond synchronously, guaranteeing load coordination and grid stability, and achieving efficient management of the power system load.

[0085] This embodiment provides a method for controllable load resources to participate in the primary frequency regulation of a power system, which can be used in the aforementioned electronic equipment. Figure 3 is a flowchart of a method for controllable load resources to participate in the primary frequency regulation of a power system according to an embodiment of this application. As shown in Figure 3, the process includes the following steps:

[0086] Step S301: Obtain the load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a load feature time series dataset. For details, please refer to step S201 of the embodiment shown in Figure 2, which will not be repeated here.

[0087] Step S302: Based on the load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator.

[0088] Specifically, step S302 includes:

[0089] Step S3021: Delay coordinate embedding is performed on the load feature time series dataset to obtain a high-dimensional state vector.

[0090] Specifically, the goal of time-delayed coordinate embedding is to incorporate temporal information into the state representation, making the system dynamics more linear in a higher-dimensional space. Therefore, by combining the states of data in the load characteristic time series dataset at different time points into a high-dimensional vector, the load state space of the power system is reconstructed to capture the nonlinear characteristics of load dynamics.

[0091] Optionally, the steps of time-delay coordinate embedding include: determining the optimal embedding dimension m and the time delay interval τ, ensuring that the embedded state vector can effectively capture the dynamic characteristics of the system, and constructing the power system load state space based on the optimal embedding dimension and the time delay interval τ to capture the dynamic characteristics. The power system load state space can be represented as: ζ(x)=(x(t),x(t-τ),...,x(t-(m-1)τ)) (4)

[0092] Where ζ(x) represents the state vector after time delay coordinate embedding, i.e., the high-dimensional state vector, t-τ represents the load state after a time delay of τ (the observation data at different time points are combined into a vector by time delay coordinate embedding to reconstruct the state space), and x(t) represents the load state at the current time point t in the load feature time series dataset.

[0093] Optionally, the delay interval τ can be determined using the autocorrelation function method or the mutual information method, and the optimal embedding dimension m can be determined using the pseudo nearest neighbor method or the saturation dimension method.

[0094] In step S3022, a high-dimensional state vector is selected as the basis function, and the basis function is fitted using the least squares method to obtain the Koopman operator.

[0095] Specifically, the high-dimensional state vector after embedding the delayed coordinates is used as the input to the basis function selection and EDMD (Extended Dynamic Mode Decomposition) algorithm. Using the high-dimensional state vector, the EDMD algorithm extracts linear characteristics and constructs the Koopman operator by selecting appropriate basis functions.

[0096] Optionally, the basis functions are used to describe the linear representation of the load state of the power system in the state space. The basis functions can be polynomial functions, trigonometric functions, or radial basis functions, etc. The specific choice depends on the complexity of the power system and the application scenario. For example, if the dynamics of the power system have obvious periodicity, such as the periodic fluctuations in data such as wind speed, power generation, and load, trigonometric basis functions are used.

[0097] Alternatively, the basis functions can be expressed as: φ(x)=[φ1(ζ(x)), φ2(ζ(x)), ..., φ n (ζ(x))] (5)

[0098] Where φ(x) represents the set of basis functions, n represents the number of basis functions, and φ n (ζ(x)) represents the basis function.

[0099] Optionally, EDMD constructs a Koopman operator by performing least-squares fitting on the basis functions. The Koopman operator is used to represent the linear relationship of the power system evolving from one load state to another. The formula for calculating the matrix form of the Koopman operator K is shown below:

[0100] Where, x t This represents the high-dimensional state vector at time t.

[0101] Optionally, the validity of the basis functions can be verified, and the basis functions can be optimized based on the verification results.

[0102] Optionally, the verification process for the effectiveness of the basis functions includes verification of the basis functions' reconstruction capability and verification of the Koopman operator; wherein, the reconstruction effect of the basis functions is evaluated using the reconstruction error, which is the mean squared error (MSE). The formula for calculating the mean squared error (MSE) is as follows:

[0103] Where u∈n, if the reconstructed mean square error (MSE) is less than the preset error, it means that the selected basis function can effectively capture the dynamic characteristics of the system; if the reconstructed mean square error (MSE) is greater than or equal to the preset error, the basis function is reselected and the Koopman operator is recalculated.

[0104] Optionally, by calculating the reconstruction error, the accuracy of the basis functions in reconstructing the load state of the power system can be quantified, and it can be determined whether the basis functions are sufficient to express the nonlinear characteristics. If the reconstruction error is small, it means that the set of basis functions is powerful enough to effectively describe the complex dynamic behavior of the system. Conversely, a large error may mean that the basis functions are insufficient to describe all the characteristics of the system, and the basis functions need to be adjusted or replaced.

[0105] Optionally, the Koopman operator can be verified by calculating its eigenvalues ​​and modes, and then by analyzing its spectral distribution. After obtaining the Koopman operator K, the calculation of eigenvalues ​​and modes can be performed through the following steps:

[0106] (1) Eigenvalue decomposition: The eigenvalues ​​and eigenvectors of the Koopman operator K are calculated using the following formula: Kz = λz (8)

[0107] Where λ represents the eigenvalue and z represents the eigenvector.

[0108] (2) Modal calculation: Koopman mode refers to the system response on a specific eigenvector. The formula for calculating the Koopman mode vector is as follows: Φ=φ(x)V (9)

[0109] Where V represents the eigenvector matrix and Φ represents the Koopman mode, which describes the influence of specific eigenvalues ​​(i.e., frequency and growth / decrease characteristics) on each high-dimensional state vector.

[0110] (3) Spectral analysis: By analyzing eigenvalues ​​and Koopman modes, the frequency and decay / growth modes of the power system are obtained. The specific analysis methods are as follows: Eigenvalue analysis: The modulus of the eigenvalue reflects the dynamic characteristics of the system. An eigenvalue with a modulus less than 1 indicates that the corresponding mode will decay, that is, the power system is stable. An eigenvalue with a modulus greater than 1 indicates that the mode will grow, indicating that the power system is unstable. Mode analysis: Koopman modes are associated with eigenvalues ​​and describe how the system responds to different disturbances. For example, a mode with an eigenvalue modulus close to 1 indicates that the mode of the power system changes slowly, while a mode with a modulus much greater than 1 indicates that the power system may experience violent fluctuations or instability.

[0111] If the modulus of the eigenvalues ​​is found to be too large during the analysis, it indicates that the power system may have high instability. The stability of the power system can then be improved by adjusting the basis functions. For example, more suitable basis functions can be selected (such as increasing the order of the polynomial basis functions or changing the type of basis functions) to improve the stability of the Koopman operator and accurately capture the dynamic characteristics of the system.

[0112] Step S3023: Obtain the control input matrix and control variables, and establish a high-dimensional linear model of the power system based on the Koopman operator, the control input matrix, and the control variables.

[0113] Specifically, a high-dimensional linear model of the power system based on the Koopman operator is established, and then multi-step prediction is performed using a recursive formula; the expression of the high-dimensional linear model of the Koopman operator is shown below:

[0114] in, This represents the power system load state at time k+1. Let B represent the power system load state at time k, and let u represent the control input matrix. k Indicates a control variable.

[0115] Optionally, the high-dimensional linear model of the power system using the Koopman operator can predict the load state at multiple forecast times, ensuring that the load regulation system can respond to possible grid changes in advance.

[0116] Step S303: The load state of the power system is optimized and solved using the high-dimensional linear model of the Koopman operator of the power system to obtain the optimal load control sequence.

[0117] Specifically, step S303 includes:

[0118] Step S3031: Collect the power system load state data at the current moment, input the power system load state data at the current moment into the high-dimensional linear model of the power system's Koopman operator, and obtain the power system load state control sequence at the predicted moment.

[0119] Step S3032: Optimize the power system load state control sequence at the predicted time to obtain the optimized load state control sequence.

[0120] In some optional implementations, step S3032 above includes:

[0121] Step a1: Obtain the power system load state data at the initial time, and use the Koopman operator to linearly map the power system load state data at the initial time to the prediction time to obtain the power system load state data at the prediction time.

[0122] Specifically, the power system load state data at the prediction time is predicted using the Koopman operator, and multi-step prediction is performed using the following recursive formula:

[0123] Optionally, the initial power system load state data may include load state information such as voltage, power, and frequency. A high-dimensional state vector is formed by time-delayed coordinate embedding. By recursively applying the constructed Koopman operator to the high-dimensional state vector, the state of the power system load at multiple future times can be predicted, thereby providing forward-looking information for dynamic regulation of the power grid.

[0124] Step a2: Based on the power system load state data at the prediction time, determine the prediction confidence interval using a Bayesian estimation algorithm.

[0125] Specifically, to address power system errors, measurement noise, and external disturbances, a Bayesian estimation algorithm is used to quantify the uncertainty of future states. The Bayesian estimation algorithm combines the power system load state data (i.e., the power system load state data at the initial moment) and the power system load state data at the prediction moment to derive the confidence interval of the future load state, i.e., the prediction confidence interval. The prediction confidence interval can indicate the possible deviation range of the prediction result, so that load regulation can be dynamically adjusted according to the uncertainty of the prediction. By introducing the prediction confidence interval, the prediction model can estimate the upper and lower limits of the state at multiple future moments, i.e., the possible fluctuation range at each prediction moment.

[0126] Optionally, the Bayesian estimation algorithm includes the following steps: Prior distribution: Based on the power system load state data at the initial time, a prior distribution of the power system load state is set to provide a preliminary estimate of the future state; Likelihood function: The likelihood function that matches the state is calculated using the power system load state data at the prediction time, representing the degree of matching between the observed data and the actual state; Posterior distribution: Combining the prior distribution and the likelihood function, the posterior distribution is updated to obtain an uncertainty prediction of the future load state.

[0127] Step a3: Dynamically adjust the power system load state control sequence at the prediction time based on the prediction confidence interval to obtain the load state optimized control sequence.

[0128] Specifically, the power system load state control sequence at the predicted time is dynamically adjusted to ensure that the power system load state control sequence at the predicted time conforms to the prediction confidence interval, and to ensure that the power system can make robust load regulation decisions even in the presence of disturbances or noise.

[0129] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. Based on the power system load state data at the predicted time, a Bayesian estimation algorithm is used to determine the prediction confidence interval. Then, the power system load state control sequence at the predicted time is dynamically adjusted through the prediction confidence interval. This ensures stable control of the power system load in the face of complex disturbances and uncertainties, enhances the robustness of the power system to errors, measurement noise and external disturbances, and improves the reliability of the power system in load regulation.

[0130] Step S3033: With the goal of minimizing frequency deviation and control cost, construct an optimization objective function based on the load state optimization control sequence.

[0131] Specifically, the core task of MPC is to achieve precise load regulation by optimizing the control sequence. In this process, considering the actual constraints of the power system (such as power upper and lower limits, frequency deviation tolerance) and control costs, the MPC problem can be expressed as an optimization objective function, the expression of which is shown below:

[0132] Among them, J * Let u0:N represent the objective function to be optimized. p -1 indicates the number of prediction steps N from the current time. p The control sequence, l(x) k ,u k This indicates that the power system load state and control actions at time k are taken into account. k Stage cost, V f (xN p This represents the control cost of the final load state, used to guide the power system to the desired state.

[0133] Step S3034: Solve the objective function to obtain the optimal load control sequence.

[0134] Specifically, by solving the objective function, frequency deviation and control cost can be minimized while ensuring that the constraints are met, thereby achieving optimal load regulation.

[0135] Optionally, the interior point method or gradient descent method can be used to solve the objective function to obtain the optimal load control sequence.

[0136] Step S304 involves using the optimal load control sequence to regulate the frequency of controllable load resources participating in the power system. For details, please refer to step S204 of the embodiment shown in Figure 2, which will not be repeated here.

[0137] This embodiment provides a method for controlling load resources to participate in the primary frequency regulation of a power system. By embedding delayed coordinates, it transforms the one-dimensional time-series data in the load characteristic time-series dataset into a state representation in a multi-dimensional space. This makes the nonlinear dynamics of the power system more linear in the high-dimensional space, providing a reliable data foundation for subsequent basis function selection and extended dynamic mode decomposition algorithms. Furthermore, by fitting the basis functions using the least squares method, the Koopman operator is obtained, achieving an accurate mapping of the linear relationship between load states. This makes the high-dimensional linear model of the Koopman operator more accurate, realizing a linear description and dynamic prediction of the power system load state. Finally, with the goal of minimizing frequency deviation and control cost, an optimization objective function is constructed based on the load state optimized control sequence. The optimal load control sequence is then solved to obtain the optimal load control sequence. This allows the power system to minimize frequency deviation and control cost while meeting constraints, thereby achieving optimal load regulation and ensuring the stability of the power system.

[0138] This embodiment also provides a controllable load resource participation device for primary frequency regulation in a power system. This device is used to implement the above embodiments and optional implementation methods, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0139] This embodiment provides a controllable load resource participating in the primary frequency regulation of a power system, as shown in Figure 4, including:

[0140] Clustering module 401 is used to acquire load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a time series dataset of load characteristics.

[0141] Module 402 is established to build a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm based on a time-series dataset of load characteristics.

[0142] The optimization solution module 403 is used to optimize the load state of the power system using the high-dimensional linear model of the Koopman operator of the power system to obtain the optimal load control sequence.

[0143] The regulation module 404 is used to regulate the frequency of controllable load resources participating in the power system using the optimal load control sequence.

[0144] In some alternative implementations, clustering module 401 includes:

[0145] The normalization processing unit is used to normalize the load status data of the power system.

[0146] Clustering units are used to cluster the normalized load status data to obtain initial load feature groups;

[0147] The adjustment unit is used to adaptively adjust the initial load feature group to obtain a time series dataset of load features.

[0148] In some alternative implementations, the establishment module 402 includes:

[0149] The embedding unit is used to embed the time-series dataset of load features with delayed coordinates to obtain a high-dimensional state vector.

[0150] The fitting unit is used to select a high-dimensional state vector as a basis function and fit the basis function using the least squares method to obtain the Koopman operator.

[0151] A unit is established to obtain the control input matrix and control variables. Based on the Koopman operator, the control input matrix, and the control variables, a high-dimensional linear model of the power system using the Koopman operator is established.

[0152] In some alternative implementations, the establishment module 402 further includes:

[0153] The verification unit is used to verify the effectiveness of the basis functions and optimize them based on the verification results.

[0154] In some alternative implementations, the optimization solution module 403 includes:

[0155] The acquisition unit is used to acquire the power system load status data at the current moment, input the power system load status data at the current moment into the high-dimensional linear model of the power system's Koopman operator, and obtain the power system load status control sequence at the predicted moment.

[0156] The optimization unit is used to optimize the power system load state control sequence at the predicted time to obtain the optimized load state control sequence.

[0157] The construction unit aims to minimize frequency deviation and control cost, and constructs an optimization objective function based on the load state optimization control sequence.

[0158] The solver unit is used to solve the objective function to obtain the optimal load control sequence.

[0159] In some optional implementations, the optimization unit includes:

[0160] The mapping subunit is used to obtain the power system load state data at the initial time and use the Koopman operator to linearly map the power system load state data at the initial time to the prediction time, so as to obtain the power system load state data at the prediction time.

[0161] The sub-unit is determined, and the prediction confidence interval is determined using the Bayesian estimation algorithm based on the power system load state data at the prediction time.

[0162] The adjustment subunit is used to dynamically adjust the power system load state control sequence at the prediction time based on the prediction confidence interval, so as to obtain the load state optimized control sequence.

[0163] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0164] In this embodiment, a controllable load resource participating in the primary frequency regulation device of the power system is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0165] This application also provides a computer device having a primary frequency regulation device for controllable load resources participating in the power system, as shown in FIG4 above.

[0166] Please refer to Figure 5, which is a schematic diagram of the structure of a computer device provided in an optional embodiment of this application. As shown in Figure 5, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components communicate with each other using different buses and can be installed on a common motherboard or otherwise as needed. The processor can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, multiple processors and / or multiple buses can be used with multiple memories, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 5 uses one processor 10 as an example.

[0167] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Optionally, processor 10 may also include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GPRS), or any combination thereof.

[0168] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0169] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0170] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0171] The computer device also includes an input device 30 and an output device 40. The processor 10, memory 20, input device 30 and output device 40 can be connected via a bus or other means; Figure 5 shows an example of a connection via a bus.

[0172] Input device 30 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the computer device, such as a touchscreen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 40 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The aforementioned display devices include, but are not limited to, liquid crystal displays, light-emitting diodes, displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.

[0173] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; optionally, the storage medium may also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0174] A portion of this application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0175] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and all such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A method for a controllable load resource to participate in primary frequency regulation of a power system, the method comprising: The method includes: Obtain load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a time series dataset of load features; Based on the aforementioned load characteristic time series dataset, a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm is established using the Koopman operator. The load state of the power system is optimized and solved using the high-dimensional linear model of the Koopman operator of the power system to obtain the optimal load control sequence; The optimal load control sequence is used to regulate the frequency of controllable load resources participating in the power system.

2. The method of claim 1, wherein, The process of performing multidimensional load clustering on the load status data to obtain a time-series dataset of load features includes: The load status data of the power system is normalized. The normalized load status data is clustered to obtain the initial load feature groups; The initial load feature group is adaptively adjusted to obtain the load feature time series dataset.

3. The method of claim 1, wherein, The process of establishing a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm based on the load characteristic time-series dataset includes: The time-series dataset of load features is embedded with delayed coordinates to obtain a high-dimensional state vector; The high-dimensional state vector is selected as the basis function, and the basis function is fitted using the least squares method to obtain the Koopman operator; Obtain the control input matrix and control variables, and establish a high-dimensional linear model of the power system based on the Koopman operator, the control input matrix, and the control variables.

4. The method of claim 3, wherein, The step of establishing a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm based on the load characteristic time-series dataset also includes: The effectiveness of the basis functions is verified, and the basis functions are optimized based on the verification results.

5. The method of claim 3, wherein, The optimization solution for the load state of the power system using the high-dimensional linear model of the Koopman operator is used to obtain the optimal load control sequence, including: Collect the power system load status data at the current moment, input the power system load status data at the current moment into the high-dimensional linear model of the power system's Koopman operator, and obtain the power system load status control sequence at the predicted moment; The power system load state control sequence at the predicted time is optimized to obtain the optimized load state control sequence. With the goal of minimizing frequency deviation and control cost, an optimization objective function is constructed based on the load state optimization control sequence. The optimal load control sequence is obtained by solving the objective function.

6. The method of claim 5, wherein, The optimization of the power system load state control sequence at the predicted time to obtain the load state optimized control sequence includes: The power system load state data at the initial time is obtained, and the power system load state data at the initial time is linearly mapped to the prediction time using the Koopman operator to obtain the power system load state data at the prediction time. Based on the power system load state data at the predicted time, the prediction confidence interval is determined using a Bayesian estimation algorithm; The load state control sequence of the power system at the predicted time is dynamically adjusted based on the predicted confidence interval to obtain the optimized load state control sequence.

7. A device for a controllable load resource to participate in primary frequency modulation of a power system, characterized in that, The device includes: The clustering module is used to acquire load status data of the power system, perform multidimensional load clustering on the load status data, and obtain a time-series dataset of load features. A module is established to build a high-dimensional linear model of the power system using the extended dynamic mode decomposition algorithm based on the aforementioned load characteristic time-series dataset. The optimization solution module is used to optimize the load state of the power system using the high-dimensional linear model of the Koopman operator of the power system, and obtain the optimal load control sequence. The regulation module is used to regulate the frequency of controllable load resources participating in the power system using the optimal load control sequence.

8. A computer device, comprising: include: A memory and a processor are interconnected, the memory stores computer instructions, and the processor executes the computer instructions to perform the primary frequency regulation method for controllable load resources participating in the power system as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the method for controlling load resources to participate in the primary frequency regulation of the power system as described in any one of claims 1 to 6.

10. A computer program product, characterised in that, Includes computer instructions, which are used to cause a computer to execute the method for controlling load resources to participate in the primary frequency regulation of the power system as described in any one of claims 1 to 6.

Images