Multi-level quota switching optimization method for new energy scenario uncertainty
By constructing a distributed robust optimization model and a multi-core K-means clustering algorithm, the difficulty in evaluating the section limit of the power system caused by the uncertainty of new energy scenarios is solved, efficient and accurate section limit evaluation is achieved, and the safety and computational efficiency of the power system are improved.
Patent Information
- Application Number
- CN202510071386.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-16
- Publication Date
- 2025-09-30
- Estimated Expiration
- 2045-01-16
AI Technical Summary
Existing technologies are unable to effectively cope with the uncertainty of new energy scenarios, resulting in difficulty in balancing the accuracy and computational efficiency of power system section limit assessments. Traditional methods are computationally overburdened when faced with a high proportion of renewable energy access and are difficult to adapt to diverse operating scenarios.
A distributed robust optimization model is constructed, combined with a multi-core K-means clustering algorithm, and a multi-level limit switching optimization method is used to improve the robustness and accuracy of section limit assessment. The distributed robust optimization model and the multi-core K-means clustering algorithm are used to identify diversified operating modes, construct multi-level section limit rules, and realize flexible switching of section limits.
It improves the robustness and computational efficiency of section limit assessment, can adapt to the diverse changes in new energy scenarios, and enhances the ability to search for safe operation boundaries of the power system.
Smart Images
Figure CN119990425B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power systems, and in particular to a multi-level quota switching optimization method for uncertainties in new energy scenarios. Background Art
[0002] With the deepening reform of the global energy structure, the integration of a high proportion of renewable energy into the power system has become the core path to promote energy transformation. However, the inherent intermittent and random nature of renewable energy generation significantly increases the uncertainty of power system operation, posing a severe challenge to the assessment of transmission network section limits. Section limits are an important indicator for measuring the anti-disturbance capability and stability of the power grid. Especially for the weak points of the hub power grid, their accurate assessment is the key to ensuring the safe operation of the system. However, traditional section assessment methods mostly rely on the total transfer capacity (TTC) of typical operating modes and are based on historical operating conditions. They are difficult to cope with the complex nonlinear characteristics and scenario diversity brought about by the increase in the penetration rate of renewable energy. In addition, the setting of TTC is a computationally intensive task with high computational requirements. Its computational burden hinders the accurate search for safe operation boundaries based on TTC assessment safety margin methods.
[0003] Artificial intelligence algorithms have been widely studied and applied to complex calculations in power systems due to their powerful nonlinear fitting capabilities and high computational efficiency. Existing artificial intelligence algorithms (deep neural networks, reinforcement learning, etc.) have demonstrated outstanding computational performance in TTC assessment and control. However, the "common problem" of deep learning - poor interpretability - remains a key barrier to its practical application in the industry. There are two key issues that need to be addressed in accurate and highly reliable section limit assessment and robust optimization scheduling: First, the uncertainty characterization of new energy scenarios needs to take into account the random occurrence of extreme scenarios, and the probability distribution of new energy uncertainty should meet a certain uncertainty, that is, the new energy scenarios used for learning can support robust and generalizable learning; second, the section limit assessment calculation should adapt to a variety of operating scenarios, that is, there must be high requirements for the accuracy of the limit, while at the same time ensuring computational efficiency and engineering practicality. Summary of the Invention
[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a multi-level limit switching optimization method for new energy scenarios with uncertainties.
[0005] The purpose of the present invention is achieved through the following technical solution: a multi-level quota switching optimization method for new energy scenario uncertainty, comprising the following steps:
[0006] Step S1: construct a distributed robust optimization model that considers the uncertainty of new energy extreme scenarios and section limit constraints;
[0007] The distribution robust optimization model is expressed as:
[0008]
[0009] in, and Represent scheduling time zone, section set, scenario set and probability distribution set respectively, p sc represents the probability of occurrence of scenario sc; H(·) and G(·) are the equality constraint and inequality constraint in the model respectively; are the unit adjustment price coefficients in the day-ahead unit combination model and the new energy multi-scenario; P G (t), P D (t), P RE (t) and u G (t) are respectively the active power output of synchronous generator, active power of load, active power of renewable energy grid connection and start / stop status of synchronous generator; They are respectively the positive deviation, negative deviation, positive deviation state and negative deviation state of the active output of the synchronous generator in the scenario sc; is the predicted value of new energy under scenario sc; is the power flow transfer factor matrix; P f (t),P f,sc (t) is the transmission power of section f in the initial prediction scenario and the uncertainty scenario sc, respectively; A is the parameter matrix of feature adjustment in the new energy scenario, reflecting the changes in the day-ahead dispatch results in each scenario; f(·) is a multi-level limit switcher based on the distance criterion; is the multi-section dynamic limit constraint set; ξ sc is the random prediction error of new energy;
[0010] Step S2: Based on the SimpleMKKM model, a multi-section dynamic limit constraint set is obtained; and an operation mode cluster with classification labels is obtained. It contains multiple sub-operation clusters Each sub-cluster reflects a typical operating mode;
[0011] Step S3: Perform multi-level quota switching based on the operating mode and the minimum square norm of the cluster center as a criterion.
[0012] Specifically, the random prediction error ξ sc and the probability of occurrence of scenario sc p sc The following mathematical relationship is satisfied:
[0013]
[0014] Among them, N is the total number of samples; interval The uncertainty range of the mean probability distribution of new energy scenarios is defined. Specifically, the distributed robust optimization model is reconstructed into a main problem and A two-stage model of the subproblems is solved by column and generation constraints.
[0015] Specifically, the multi-core K-means clustering model is expressed as:
[0016]
[0017] Where, and are the basic kernel weight set and cluster partition matrix set, is a set of sampled operation modes, including all historical operations and their search extreme operation scenarios, x i is the q-dimensional operation mode feature vector obtained after feature extraction, K γ is a set of precomputed kernel matrices; represents the pth kernel function, m and k are the number of kernel functions and clusters respectively, γ p and H represent the weight of the p-th basic kernel and the weight of the cluster partition matrix, respectively.
[0018] Specifically, the multi-core K-means clustering model is reconstructed as a minimization problem, specifically:
[0019]
[0020] Specifically, a simplified gradient descent algorithm is used to iteratively update γ:
[0021]
[0022] Specifically, in step S3, multi-level quota switching is performed using the following formula:
[0023]
[0024] Among them, j is the category index.
[0025] The present invention has the following advantages:
[0026] The present invention sets interval constraints on the mean value of new energy distribution, constructs a distributionally robust optimization model that takes into account extreme scenarios of new energy and section limit constraints, improves the robustness of the section limit switcher in a high-proportion new energy system, applies a multi-core clustering algorithm based on multi-core spatial mapping to identify diverse operating modes, and constructs multi-level section limit rules to achieve robust switching of multi-level section limits. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 Schematic diagram of the handover optimization method of the present invention. DETAILED DESCRIPTION
[0028] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for the purpose of explaining the present invention and are not intended to limit the present invention. That is, the embodiments described herein are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and illustrated in the drawings herein can be arranged and designed in various different configurations.
[0029] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but is merely intended to represent selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.
[0030] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.
[0031] The present invention will be further described below in conjunction with the accompanying drawings, but the scope of protection of the present invention is not limited to the following. Figure 1 As shown in FIG, the multi-level quota switching optimization method for new energy scenario uncertainty includes the following steps:
[0032] Step S1: Construct a distributionally robust optimization model that considers the uncertainty of extreme scenarios of new energy and the constraints of cross-sectional limits. To enhance the robustness of the model, the present invention incorporates the uncertainty of renewable energy forecasts and establishes a scenario-oriented distributionally robust optimization model (SDRO), which integrates multiple random scenarios of new energy. By effectively avoiding complex mathematical transformations, the model has better engineering practicality. The distributionally robust optimization model is expressed as:
[0033]
[0034] in, and Represent scheduling time zone, section set, scenario set and probability distribution set respectively, p sc represents the probability of occurrence of scenario sc; H(·) and G(·) are the equality constraint and inequality constraint in the model respectively; are the unit adjustment price coefficients in the day-ahead unit combination model and the new energy multi-scenario; P G (t), P D (t), P RE (t) and u G (t) are respectively the active power output of synchronous generator, active power of load, active power of renewable energy grid connection and start / stop status of synchronous generator; They are respectively the positive deviation, negative deviation, positive deviation state and negative deviation state of the active output of the synchronous generator in the scenario sc; is the predicted value of new energy under scenario sc; P f,sc (t) is the transmission power of section f in scenario sc; A is the parameter matrix of feature adjustment in the new energy scenario, reflecting the changes in the day-ahead scheduling results in each scenario; f(·) is a multi-level limit switch based on the distance criterion; is the multi-section dynamic limit constraint set; ξ sc is the random prediction error of new energy;
[0035] The random prediction error ξ sc and the probability of occurrence of scenario sc p sc The following mathematical relationship is satisfied:
[0036]
[0037] The probability distribution of the new energy scenario is constrained by the Wasserstein radius, which limits its deviation from the probability distribution of the sampled operating mode dataset. The uncertainty range of the mean probability distribution of new energy scenarios is defined; by adjusting μ, the robustness of the model in extreme scenarios can be enhanced. (1) and (2) A robust unit commitment model is constructed based on dynamic section limit switching. The model can be reconstructed into a main problem and A two-stage model for each subproblem, which can be solved through column-and-constraint generation (C&CG);
[0038] Step S2: Based on the SimpleMKKM model, a multi-section dynamic limit constraint set is obtained; and an operation mode cluster with classification labels is obtained. It contains multiple sub-operation clusters Each sub-cluster reflects a typical operating mode;
[0039] Traditional section limits only consider a single conservative limit value. The reason is that the setting of the limit only considers the single feature of the section flow and the stability verification result. The insufficient characterization of the operating mode leads to the generation of a "one-size-fits-all" conservative limit. Considering a wider range of operating characteristics can characterize more accurate limit boundaries, but after introducing more features, there is a linear inseparability problem between operating modes, especially in large-scale power systems. In order to solve this problem, the present invention adopts multi-kernel unsupervised learning, such as multi-kernel K-means clustering (MKKM), to project the operating mode into the multi-kernel Hilbert space, thereby enhancing the ability to capture the complex relationship between the modes and being able to identify different operating modes. Specifically, a global optimal algorithm SimpleMKKM is adopted, which has been proven to have excellent generalization error performance and is used to generate representative operating modes;
[0040] The SimpleMKKM model is expressed as:
[0041]
[0042] Where, and are the basic kernel weight set and cluster partition matrix set, is a set of sampled operation modes, including all historical operations and their search extreme operation scenarios, x i is the q-dimensional operation mode feature vector obtained after feature extraction, K γ is a set of precomputed kernel matrices; represents the pth kernel function, m and k are the number of kernel functions and clusters respectively, γ p and H represent the weight of the p-th basic kernel and the weight of the cluster partition matrix, respectively.
[0043] Specifically, the multi-core K-means clustering model is reconstructed as a minimization problem, where It is proved to be differentiable, specifically:
[0044]
[0045] A simplified gradient descent algorithm is used to iteratively update γ to ensure global accelerated convergence of the model:
[0046]
[0047] Through (6), the descending direction of γ is obtained as d = [d1,…,d m ] Τ .
[0048] The SimpleMKKM model is optimized using the following algorithm:
[0049] 1: Input: t=1,flag=1.
[0050] 2:while flag do:
[0051] 3: By solving (4b) Calculate H
[0052] 4: Calculate d p (p=1,…,m).
[0053] 5: Update γ (t+1) =γ (t) +αd
[0054] 6:if max|γ (t+1) -γ (t) |<1e-5then
[0055] 7: flag = 0.
[0056] 8:end if
[0057] 9:t=t+1
[0058] 10:end while
[0059] 11:
[0060] 12:Output:y
[0061] Through the above-mentioned pattern recognition algorithm based on SimpleMKKM, the operation mode clusters with classification labels are obtained Contains multiple sub-operation clusters The modes in each sub-cluster of operating modes have the highest degree of feature similarity, and each sub-cluster can reflect a typical operating mode, where Through this method, the present invention alleviates the strong nonlinear separability of the safety margin caused by characterizing the operating mode only by cross-section power transmission, and uses a conservative setting method in each sub-cluster:
[0062] The maximum cross-sectional current in the stable mode and the minimum cross-sectional current in the unstable mode are selected within the cluster, and the minimum of the two is taken;
[0063] Step S3: Using the operation mode and the least squares norm of the cluster center as the criterion to perform multi-level quota switching:
[0064]
[0065] Among them, j is the category index.
[0066] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make many possible changes and modifications to the technical solution of the present invention using the above technical content, or modify it into an equivalent embodiment with equivalent changes. Therefore, any changes, modifications, equivalent changes, and modifications made to the above embodiments based on the technology of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the present technical solution.
Claims
1. A multi-level quota switching optimization method for new energy scenario uncertainty, characterized by: The following steps are involved: Step S1: construct a distributed robust optimization model that considers the uncertainty of new energy extreme scenarios and section limit constraints; The distribution robust optimization model is expressed as: in, and Represent scheduling time zone, section set, scenario set and probability distribution set respectively, p sc represents the probability of occurrence of scenario sc; H(·) and G(·) are the equality constraint and inequality constraint in the model respectively; are the unit adjustment price coefficients in the day-ahead unit combination model and the new energy multi-scenario; P G (t), P D (t), P RE (t) and u G (t) are respectively the active power output of synchronous generator, active power of load, active power of renewable energy grid connection and start / stop status of synchronous generator; They are respectively the positive deviation, negative deviation, positive deviation state and negative deviation state of the active output of the synchronous generator in the scenario sc; is the predicted value of new energy under scenario sc; is the power flow transfer factor matrix; P f (t), P f，sc (t) is the transmission power of section f in the initial prediction scenario and the uncertainty scenario sc, respectively; A is the parameter matrix of feature adjustment in the new energy scenario, reflecting the changes in the day-ahead dispatch results in each scenario; f(·) is a multi-level limit switcher based on the distance criterion; is the multi-section dynamic limit constraint set; ξ sc is the random prediction error of new energy; Step S2: Based on the SimpleMKKM model, a multi-section dynamic limit constraint set is obtained; and an operation mode cluster with classification labels is obtained. It contains multiple sub-operation clusters Each sub-cluster reflects a typical operating mode; Step S3: Perform multi-level quota switching based on the operating mode and the minimum square norm of the cluster center as a criterion.
2. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 1 is characterized by: The random prediction error ξ sc and the probability of occurrence of scenario sc p sc The following mathematical relationship is satisfied: Among them, N is the total number of samples; interval The uncertainty range of the mean probability distribution of new energy scenarios is defined.
3. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 2 is characterized by: The distributed robust optimization model is reformulated as a master problem and A two-stage model of the subproblems is solved by column and generation constraints.
4. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 1 is characterized by: The SimpleMKKM model in step S2 is expressed as: Where, is the field of real numbers, and They are the basic kernel weight set and cluster partition matrix set, E is the sampling operation mode set, including all historical operations and their search extreme operation scenarios, x i is the q-dimensional operation mode feature vector obtained after feature extraction, K γ is a set of pre-calculated kernel matrices, the kernel matrix K calculated by multiple independent kernel functions P Composition, γ is the weight coefficient vector of each kernel matrix; represents the pth kernel function, m and k are the number of kernel functions and clusters respectively, γ p and H represent the weight of the p-th basic kernel and the weight of the cluster partition matrix, respectively.
5. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 4 is characterized by: The multi-core K-means clustering model is restructured as a minimization problem, specifically:
6. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 5 is characterized by: Use a simplified gradient descent algorithm to iteratively update γ:
7. The multi-level quota switching optimization method for new energy scenario uncertainty according to claim 6 is characterized by: In step S3, multi-level quota switching is performed by the following formula: Among them, j is the category index.