Power distribution automation data system compatible with high proportion of distributed power supply and multi-element load access

Abstract

The application discloses a power distribution automation data system compatible with high-proportion distributed power supply and multi-element load access, introduces professional technical resource cooperation of both sides based on basic cumulative data of power grid operation to construct a high-precision intensive data model, has expandability in accordance with power distribution optimization of each sub-item demand in data structure and data algorithm, internally covers source-grid-load multi-oriented data content dimension and data demand dimension based on power distribution automation, and develops data algorithm or data process oriented to various data demands and data optimization on the basis. The data system of the application cooperatively constructs a high-precision intensive data model through multiple scientific research technical resources, realizes expandability of each sub-item demand of power distribution optimization, and provides basic support for source-grid-load multi-oriented new multi-dimensional power grid data system construction.

Description

Technical Field

[0001] This invention relates to the field of intelligent power grid construction and related data processing technologies, and in particular to a distribution automation data system and its application that is compatible with a high proportion of distributed power sources and multiple load access. Background Technology

[0002] With the development of modern technology, the intelligentization, informatization, and datafication of power grids are the trends in technological development. In previous research related to the construction of the Qinghai Province distribution automation system, the applicant's research team focused on the construction of medium-voltage distribution automation and its differentiated models, resulting in corresponding research findings. However, further in-depth research is needed on the technical routes and methods of distribution automation under the background of new power systems. Especially with the construction and development of new power systems and the Qinghai distribution network, the widespread integration of high-proportion distributed power sources, the allocation of demand-side response resources, and the increasing demand for diversified interaction among terminal loads are posing significant challenges to the traditional structure and operation mode of medium- and low-voltage distribution networks, which serve as resource optimization platforms closely aligned with distributed power sources and diversified loads. This also presents inherent technical updates to the matching distribution optimization data structure and data system. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a power distribution automation data system that is compatible with a high proportion of distributed power sources and multiple load access.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows.

[0005] The distribution automation data system is compatible with a high proportion of distributed power sources and diverse load access. Based on the basic accumulated data of power grid operation, it introduces professional technical resources to jointly build a high-precision and intensive data model. In terms of data structure and data algorithm, it has the scalability to meet the needs of various sub-items of distribution optimization. It covers multiple dimensions of data content oriented towards source, grid and load and the data requirement dimension based on distribution automation. On this basis, it can further develop data algorithms or data processes for various data needs and data optimization orientation.

[0006] As a preferred technical solution of the present invention, the data model has external scalability in both data structure and data algorithm. From the initial stage of the data structure, it considers the adaptation and development of external open-source intelligent tensor data flow models, so that it has data consistency and external connection scalability with external data processing platforms, especially open-source intelligent tensor data flow models and their supervised iterative machine optimization tools.

[0007] As a preferred technical solution of the present invention, the underlying initialization structure of the data model is set as follows: the multi-dimensional data structure of the distribution network, including distributed grid-connected power sources, grid nodes, grid branch flow, and multiple loads, is represented as a tensor; based on the large-scale and dynamic source-grid-load-storage multi-directional basic power data provided by the power grid system, the tensor dimension is set to represent orthogonal grid parameters including time, node location, power source type, load type, power parameters, and environmental parameters, etc., and the tensor dimension attributes are extended according to the following two sets of data conditions: ω, the power grid system provides new data parameters or independent power parameters are generated due to the grid access of new types of sensors; ψ, new tensor components or new tensor component layouts are generated for new distribution optimization needs or data processing objectives.

[0008] As a preferred technical solution of the present invention, based on the underlying initialization structure of the data model, we set the entire data system to include at least a data acquisition layer and a data processing layer in terms of data logic and data structure;

[0009] The data acquisition layer connects to a large-scale and dynamic dataset provided by the power grid, containing real-time and cumulative data from distributed power sources and multiple load access points, and performs necessary and basic data processing.

[0010] The main body of the data processing layer completes the tensor structure processing of the power grid basic dataset. At its core, a data tool based on linear allocation is constructed to represent the power grid distribution model as a "distributed source-node parameter network-multi-element load" state space model in a multi-dimensional geometric space. In this state space model, scalar data from the power grid system is used to construct a first-order tensor through linear allocation. The dynamic sequence of the first-order tensor is merged to construct a second-order tensor to represent the dynamic behavior of the distribution network. Furthermore, based on the same linear allocation data process, second-order dynamic tensors from different independent sources such as power parameters, environmental parameters, and efficiency parameters are expanded in orthogonal dimensions, thereby representing the multi-dimensional data parameters of the distribution network using tensors.

[0011] As a preferred embodiment of the present invention, the data system further includes a third layer: an optimization control layer; which, based on the tensor data structure of the data processing layer, constructs optimization algorithms and runs data processing processes according to predetermined power distribution requirements and continuously updated power distribution optimization sub-items, and generates an optimization reference data sequence.

[0012] As a preferred technical solution of the present invention, further, based on the above three-layer data architecture, the data layer is set as an open structure. Here, openness includes the extensibility within the layer and the construction of new data layers. The main body of the open structure of the data layer is built according to the following data scenarios: ω, the cumulative update of power grid data, especially the continuous proposal and generation of optimization sub-items in distribution operation; ψ, for the entire data system, in the later stage, it will be artificially constructed with access to external data processing platforms, especially open-source intelligent tensor data flow models and their supervised iterative machine optimization tools.

[0013] As a preferred technical solution of the present invention, the data acquisition layer performs basic data organization on the basic dataset from the power grid, i.e., data preprocessing. This includes data cleaning and / or data standardization and / or data interpolation and / or other data preprocessing based on various distribution optimization sub-items, for electrical parameters such as voltage, current, power, and frequency, as well as environmental parameters such as temperature and humidity, commonly found in distribution network nodes by distributed power sources and multi-loads. Data cleaning removes outliers and noise, using threshold methods or statistical methods to identify and process abnormal data points. Data standardization scales the data to a standard range. Data provided by the power grid system is generally discrete, which is compatible with the underlying initialized tensor data structure. However, various distribution optimization needs usually require the introduction of a complete dynamic data analysis model for differential analysis on the data function (tensor field). Therefore, a general data interpolation preprocessing algorithm is introduced, including two data processing orientations: ω, using linear or polynomial interpolation methods to fill missing data points; and ψ, constructing a continuous data function and tensor field model through interpolation.

[0014] As a preferred embodiment of the present invention, the data preprocessing in the data acquisition layer allows direct use of existing conventional algorithms such as data cleansing and format consistency, while also allowing for the storage of self-built data preprocessing algorithms; and both existing algorithms and self-built extended data preprocessing algorithms are stored in a callable manner in the data acquisition layer; wherein the self-built data preprocessing algorithm includes at least:

[0015] For distributed power sources and multiple loads in the distribution network nodes, the data credibility reference algorithm is set as follows: if the distance between a data point and the mean μ is less than or equal to k times the variance σ, then the data point will be assigned a credibility parameter above the threshold; otherwise, the data point will be assigned a credibility parameter below the threshold. Subsequently, a credibility check data process or a credibility weighting data process is set according to the actual objective situation of the current distribution optimization event. The credibility check data process determines the retention or deletion / replacement of data based on the credibility parameter, and the credibility weighting data process assigns numerical weights to the data credibility within the range of (0-1) based on the credibility parameter and its refined indicators.

[0016] For numericalized electrical parameter sequences, environmental parameter sequences, and other arbitrary numerical parameter sequences or quantifiable parameter sequences of distributed power sources and multiple loads in distribution network nodes, the data normalization scaling algorithm is constructed as follows: ω, each data point is subtracted from the mean and then divided by the standard deviation to transform all data points to a new standard scale; further normalization processing can pre-set the mean of the data sequence to a selected constant such as 0, the standard deviation of the data sequence to a selected constant such as 1, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event; ψ, each data point is subtracted from the minimum value and then divided by the difference between the maximum and minimum values ​​to transform the data point to a new standard scale; further normalization processing can pre-set the maximum value of the data sequence to a selected constant such as 1, the minimum value of the data sequence to a selected constant such as 0, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event.

[0017] As a preferred technical solution of the present invention, the data processing layer represents the power distribution model of the power grid as a "distributed source-node parameter network-multi-element load" state tensor in a multi-dimensional geometric space based on linear allocation. On the one hand, it constructs one or more specific data processes and stores them in the data processing layer as a basic tool library for generating tensor structures that can be called upon. On the other hand, for the real-time updated distribution network node parameters and the continuously expanding distribution optimization sub-objectives, it writes linear allocation data processes that directly correspond to the current distribution optimization sub-objectives as needed, thereby forming new tensor structure generation tools. The new tools formed by any operation are numbered and stored in the data processing layer as an update tool library for tensor structure generation. The basic tool library and the update tool library correspond to the extensible data tool call library group of the data processing layer.

[0018] As a preferred technical solution of the present invention, the tensor data structure constructed by the optimization control layer and the data processing layer is consistent, that is, the power distribution optimization is based on tensor flow and the corresponding data processing constructs optimization objectives and constraints.

[0019] The optimization control layer is set to be highly open, as it has completely different data processes based on different single power distribution optimization events;

[0020] The optimization control layer also has a modular and extensible tool library. On the one hand, it modularizes the data algorithms and data processing processes of historical power distribution optimization events into optimization control tool suites. On the other hand, the tools that the optimization control layer can call are not directly set as the aforementioned optimization control tool suites. Instead, any optimization control tool suite is divided into blocks and marked as callable and stored in the library according to its data processing steps (and allows the data processing steps to be subdivided downwards or merged upwards according to the data processing objectives, thereby realizing block processing according to the data processing effectiveness). In this way, no matter how much the historical optimization control tool suites differ from the current power distribution events, as long as some of the data optimization steps are corresponding, the sub-modules of the optimization control tool suites stored in the library can be called in a specific data processing process.

[0021] The beneficial effects of adopting the above technical solution are as follows: This invention, through the collaborative efforts of multiple research parties, has constructed a dedicated distribution automation data model and data system highly adapted to the current new power grid. Based on this, it has made initial attempts to develop simulation data examples for real distribution optimization environments and preliminary development of executable code. The data system of this invention has strong technical advantages. By introducing professional technical resources and collaboratively constructing a high-precision and intensive data model, it achieves scalability for the needs of various sub-items of distribution optimization, providing fundamental support for the construction of a new multi-dimensional power grid data system with multiple orientations of source-grid-load.

[0022] Overall, the data system of this invention has outward scalability in terms of data structure and algorithm, enabling it to achieve data consistency and external connectivity with external data processing platforms, especially open-source intelligent tensor data flow models and their supervised iterative machine optimization tools. Its underlying initialization structure represents the multidimensional data structure of the distribution network as tensors, constructs a data tool based on linear allocation, and represents the distribution model as a "distributed source-node parameter network-multi-element load" state space model in multidimensional geometric space. This enables large-scale and dynamic processing of real-time and cumulative data of the power grid, and by setting infinitely expandable tensor dimension attributes, it achieves rapid adaptation and processing of new data parameters and distribution optimization requirements.

[0023] The construction of the underlying initialization structure and data processing layer of the data model is a core aspect of our technology development. This novel data structure model is specifically designed for the high proportion of distributed sources and the developing diverse load characteristics of the Qinghai power grid, exhibiting high adaptability, scalability, and wide applicability. It has significant advantages in handling complex and dynamically changing distribution optimization data systems. The highly structured data and tensor representation method, which characterizes the multidimensional data structure of the distribution network as tensors, provides a highly structured and unified way to represent power grid data. This representation not only helps to process and analyze data more effectively but also makes it easier to integrate different types of data. Simultaneously, the design of the initialization structure allows for the unlimited scalability of power grid data, meaning that as new data is added to the power grid system... With the addition of data parameters or the integration of new types of sensors, the model can easily adapt to and integrate this new information; this flexibility is crucial for coping with rapidly changing power grid environments and evolving technological demands. Furthermore, this underlying data structure is directly compatible with a high proportion of distributed power sources and diverse load integration; in fact, its data dimensions and data filling space are infinitely expandable, thus allowing for the subsequent addition of more distribution-related factors. Moreover, it can more easily adapt to new distribution optimization needs; as the challenges and demands facing the power grid continue to change, this data system can quickly adapt to new optimization requirements, whether by generating new tensor components or adjusting the layout of existing components. Furthermore, this tensor data structure has direct link compatibility with machine learning and deep learning technologies. Specifically, regarding the data logic architecture of the data processing layer, we know that for a multi-dimensional data system of a smart grid containing numerous distributed power sources and various renewable energy loads, we pioneered a method to represent the power grid's distribution model as a "distributed source-node parameter network-multi-load" state-space model in a multi-dimensional geometric space. By constructing a first-order tensor from the scalar data of the power grid system through linear allocation, we can capture the basic attributes and states of each node in the power grid. The dynamic sequences of these first-order tensors are merged to construct a second-order tensor, thereby representing the dynamic behavior of the distribution network. This method allows us to extract the dynamic characteristics of the power grid from time-series data, facilitating further analysis. This provides a foundation for optimization. After constructing the basic second-order dynamic tensor, we further expand it in the orthogonal dimension to integrate data from different independent sources such as power parameters, environmental parameters, and efficiency parameters into a unified tensor representation. By representing the multidimensional data parameters of the distribution network using tensors, we can use tools such as PCA and matrix sections to extract key features and patterns from the data. Overall, this method of constructing a linearly distributed data tool and a multidimensional state-space model provides a powerful framework for distribution optimization and data processing in the context of local new power grids, and has significant advantages in terms of data processing efficiency, accuracy, depth, data breadth, and intensification.

[0024] The technical advantages of the present invention are detailed in the embodiments described below. Detailed Implementation

[0025] The following embodiments detail the present invention. In the description of the following embodiments, specific details such as particular system structures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail. It should be understood that, as used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. It should also be understood that, as used in this specification and the appended claims, the term "and / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases “if determined” or “if [the described condition or event] is detected” can be interpreted, depending on the context, as meaning “once determined” or “in response to determined” or “once [the described condition or event] is detected” or “in response to the detection of [the described condition or event]”.

[0026] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. References to "one embodiment" or "some embodiments" in this application mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0027] Example 1

[0028] For new power grids that include a high proportion of distributed power sources, diverse new energy sources, and traditional energy loads (as well as distributed energy storage systems), existing data models are fundamentally incompatible. The construction of a new data system is based on the basic accumulated data of power grid operation, while introducing professional technical resources. Both parties collaborate to build a highly accurate and intensive data model. In terms of data structure and data algorithms, it has the scalability to meet the needs of various sub-items of distribution optimization. It covers multiple dimensions of data content oriented towards source, grid, and load, as well as the data requirement dimension based on distribution automation. On this basis, it expands and develops data algorithms or data processes oriented towards various data needs and data optimization.

[0029] In terms of overall architecture, on the one hand, we set the entire data system as including a data acquisition layer, a data processing layer, and an optimization control layer; on the other hand, the data layer is set as an scalable structure. Specifically, the data acquisition layer connects to the large-scale and dynamic dataset provided by the power grid, containing real-time and cumulative data from distributed power sources and multiple load access points, and performs necessary and basic data processing; the data processing layer mainly completes the tensor structure processing of the power grid's basic dataset; the optimization control layer, based on the tensor data structure of the data processing layer, constructs optimization algorithms and runs data processing processes according to predetermined power distribution requirements and continuously updated power distribution optimization sub-items, generating optimization reference data sequences. Regarding the construction of the scalable architecture, based on the above three-layer data architecture, the data layer is set as an open structure. This openness includes the scalability within the layer and the construction of new data layers; the open structure of the data layer is mainly built according to the following data scenarios: ω, the cumulative updates of power grid data, especially the continuous proposal and generation of optimization sub-items in power distribution operation; ψ, for the entire data system, in the later stages, it will be connected to external data processing platforms, especially open-source intelligent tensor data flow models and their supervised iterative machine optimization tools, for artificial intelligence construction.

[0030] The data processing layer is one of the core components of data model construction. To address this issue, we fundamentally considered the objective form of the new distributed source-load context, and thus set the underlying initialization structure of the distribution optimization data model as follows: The multi-dimensional data structure of the distribution network, including distributed grid-connected power sources, grid nodes, grid branch flow, and diverse loads, is represented as a tensor. Based on the large-scale and dynamic multi-directional basic power data of the source-grid-load-storage system provided by the grid system, the tensor dimension is set to represent orthogonal grid parameters including time, node location, power source type, load type, power parameters, and environmental parameters. The tensor dimension attributes are extended according to the following two sets of data conditions: ω, the grid system provides new data parameters or due to the grid connecting new types of sensors. It generates independent power parameters; generates new tensor components or new tensor component layouts to meet new distribution optimization needs or data processing objectives; based on this, the data processing layer constructs a data tool based on linear allocation to represent the power grid distribution model as a "distributed source-node parameter network-multi-element load" state space model in a multi-dimensional geometric space. In our state space model, scalar data from the power grid system is used to construct a first-order tensor through linear allocation. The dynamic sequence of the first-order tensor is merged to construct a second-order tensor to represent the dynamic behavior of the distribution network. Furthermore, based on the same linear allocation data process, the second-order dynamic tensors from different independent sources such as power parameters, environmental parameters, and efficiency parameters are expanded in orthogonal dimensions, thereby representing the multi-dimensional data parameters of the distribution network using tensors.

[0031] In addition, from the initial construction of the underlying architecture logic of the data structure, we considered the adaptation and development of external open-source intelligent tensor data flow models. This makes the data model we built have data consistency and external connectivity and scalability with external data processing platforms, especially open-source intelligent tensor data flow models (and their supervised iterative machine optimization tools); thus, we can carry out system optimization based on artificial intelligence in the later stage.

[0032] Example 2

[0033] The data acquisition layer also performs basic data processing on the basic dataset from the power grid, i.e., data preprocessing. For distributed power sources and multiple loads in the distribution network nodes, common electrical parameters such as voltage, current, power, and frequency, as well as environmental parameters such as temperature and humidity, and other types of data, data cleaning and / or data standardization and / or data interpolation and / or other data preprocessing based on various distribution optimization sub-items are performed.

[0034] Considering the discretized nature of the basic data provided by the power grid system, which is compatible with the underlying initial tensor data structure, various power distribution optimization needs usually require the introduction of a complete dynamic data analysis model to perform differential analysis on the data function (tensor field). To this end, we have generally constructed a data interpolation preprocessing algorithm, which includes two data processing orientations: ω, using linear interpolation or polynomial interpolation methods to reasonably fill missing data points; ψ, constructing a continuous data function and tensor field model through interpolation processing.

[0035] The data preprocessing in the data acquisition layer allows for the direct use of existing conventional algorithms such as data clarification and format consistency, while also allowing for the storage of self-built data preprocessing algorithms. Furthermore, both existing algorithms and self-built extended data preprocessing algorithms are stored in the data acquisition layer and can be called upon.

[0036] Here, we will focus on introducing our self-built data preprocessing algorithm. ① For distributed power sources and multiple loads in the distribution network nodes, the data credibility reference algorithm is set as follows: if the distance between a data point and the mean μ is less than or equal to k times the variance σ, then the data point will be assigned a credibility parameter above the threshold; otherwise, the data point will be assigned a credibility parameter below the threshold. Subsequently, based on the actual objective situation of the current distribution optimization event, a credibility check data process or a credibility weighting data process is set. The credibility check data process determines the retention or deletion / replacement of data based on the credibility parameter, while the credibility weighting data process assigns numerical weights to the data credibility within the range of (0-1) based on the credibility parameter and its refined indicators. ② For numericalized electrical parameter sequences, environmental parameter sequences, and other arbitrary numerical parameter sequences or quantifiable parameter sequences of distributed power sources and multiple loads in the distribution network nodes, the data normalization scaling algorithm is constructed as follows: ω, each data point is subtracted from the mean and then divided by the standard deviation to transform all data points to a new standard scale; further normalization processing can pre-set the mean of the data sequence to a selected constant such as 0, the standard deviation of the data sequence to a selected constant such as 1, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event; ψ, each data point is subtracted from the minimum value and then divided by the difference between the maximum and minimum values ​​to transform the data point to a new standard scale; further normalization processing can pre-set the maximum value of the data sequence to a selected constant such as 1, the minimum value of the data sequence to a selected constant such as 0, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event.

[0037] Example 3

[0038] Based on the underlying initialization data structure and core data logic of the data processing layer in Example 1, a tensor structure tool consisting of a basic tool library and an update tool library can be constructed. Specifically, based on linear allocation, the power grid distribution model is represented as a "distributed source-node parameter network-multi-element load" state tensor in a multi-dimensional geometric space. On the one hand, one or more specific data processes are constructed and stored in the data processing layer as the basic tool library for generating tensor structures that can be called upon. On the other hand, for the real-time updated distribution network node parameters and the continuously expanding distribution optimization sub-objectives, linear allocation data processes directly corresponding to the current distribution optimization sub-objectives are written as needed, thereby forming new tensor structure generation tools. The new tools formed by any operation are numbered and stored in the data processing layer as the update tool library for tensor structure generation. The basic tool library and the update tool library correspond to the extensible data tool call library group of the data processing layer. The following are some examples of tools stored in the data processing layer.

[0039] ① It has an internal double linear and internally computable multilinear data entry tool. It represents distributed power sources (including energy storage discharge and discharge from new energy vehicles that may be connected to the grid in the future, and may or may not include traditional generators), currently interested nodes, transformers, lines (flow carriers), new loads such as new energy charging, and traditional loads) in the power network as block-based second-order tensors (matrixes). Loads and distributed power sources are independently represented as vectors. As a simplified model, assuming the power system has n nodes, m lines, p generators, q loads, and r distributed power sources, we can represent the system model as an n×n nodeized matrix Y, an m×n line parameter matrix B, a p×n generator output vector Pg, a q×n load power vector Pd, and an r×n distributed power source output vector Pr. The node-based matrix Y is established based on the equivalent admittance of system components. The matrix order is the same as the number of nodes, and the diagonal elements are the sum of the admittances connected to that node and the admittance to ground. The off-diagonal elements are the negatives of the admittances of the lines between nodes i and j. Its diagonal element Yij (i=j) is the node's self-admittance, equal to the sum of the admittances of all branches connected to that node. For simplification, it can be set as a sparse matrix, where the mutual admittance between two non-adjacent nodes in the power network is 0. This allows the introduction of the mature Tideflow algorithm for matrix data processing. The line parameter matrix B represents the line parameters, such as resistance and reactance; the vector Pg represents the generator output; the load power vector Pd represents the load power demand; and the distributed generation output vector Pr represents the output of the distributed generation. As for the optimization objective of the current power distribution optimization event, in this simplified model, for example, if our optimization objective is to minimize system loss, then we can represent the system loss as a scalar function f(Y,B,Pg,Pd,Pr), where f is a function of Y,B,Pg,Pd,Pr.For example, we introduce tideflow to calculate system losses, for which we express f as f(Y,B,Pg,Pd,Pr)=\sum{i=1}^{n}\sum{j=1}^{n}(Y{ij}×Vi-Vj)^2; where Vi and Vj represent the voltages at nodes i and j, respectively; further introducing constraints, considering the general case, we can express it as a system of inequalities g(Y,B,Pg,Pd,Pr)\leq0, where g is a function of Y,B,Pg,Pd,Pr. In this way, we can set some constraints based on the current power distribution optimization task, such as node voltage limits, line capacity limits, generator output limits, load power limits, and distributed power output limits. For example, the node voltage limit can be expressed as: g{ij}(Y,B,Pg,Pd,Pr)=Vi^2-V{ij}^2\leq0; where V{ij} represents the voltage difference between node i and node j. Thus, the current power distribution optimization task (system loss optimization) can be globally represented on this simplified model. Based on the construction of the above multidimensional linear representation model, mature solution algorithms can be easily introduced for optimization, such as using linear programming or nonlinear programming algorithms to solve the optimization problem, obtain the optimal solution (Y^*,B^*,Pg^*,Pd^*,Pr^*), and adjust the system according to the optimal solution (such as power generation allocation, node transformer taps, line capacity or flow tidal control, etc.).

[0040] ② Basic data entry tools under a simple (without internal secondary structure) multidimensional linear structure. This basic structure can serve as a fundamental modular tool. Generally, it cannot be directly applied to the data processing layer for organizing basic power data, but it can be used as a called module for other data entry tools. Typically, the initial tensor of the simple multidimensional linear architecture should be set according to the current power distribution optimization task. Here, in a simplified case, we assume that we first need to construct a fourth-order tensor to represent the multidimensional data structure of the power grid. Let this fourth-order tensor be T, and let T(i,j,k,l) ​​= data, where i represents the time dimension, j represents the spatial location, k represents the power source type, l represents the load type, and data represents the specific power grid data; each dimension corresponds to an attribute of the data structure. Based on this, some tensor processing algorithms can be constructed to represent the features within the data dimensions, such as building a tensor slice tool to view the power source and load situation at a specific time and location, and building a tensor product to analyze the relationship between different power source and load types, etc. Based on this fundamental model, the dynamic behavior of the power grid is characterized using linear operations. In continuous-time states, the following computational tools can be constructed: dx / dt = Ax(t) + Bu(t); y(t) = Cx(t) + Du(t); or in discrete-time states: x(t+1) = Ax(t) + Bu(t); y(t) = Cx(t) + Du(t); where x(t) is the initial state, u(t) is the control input, y(t) is the output, and A, B, C, and D are system matrices. In this way, parameters such as voltage, current, power, and impedance (distributed across multiple dimensions of the tensor T and discretely expanded) can be extracted, feature-analyzed, and optimized within the basic multidimensional linear structure based on the current power distribution optimization task (connecting to the optimization control layer).

[0041] ③ An ingestion tool (infrastructure) with a hierarchical structure directly generated based on the initial data logic. For the initial linear layer, for each unique source (which can be a distributed production device or load, or a selected indicator in the current power distribution optimization task), we can define a first-order tensor, where each element represents a specific state at a certain point in time, such as power output, voltage, etc. Assuming we have N sources and T time points, we can define a first-order tensor Xi∈R^T for each source i: Xi=[xi(1),xi(2),...,xi(T)]; here, xi(t) contains all relevant states of source i at time point t. For the intermediate linear layer, once we obtain the initial linear layer containing all sources, we can superimpose these initial linear layers to form an intermediate linear layer to represent the power distribution network state at all time points, expressed as: X∈R^(N x T):X=[X1;X2;...;XN];In the relative structure of the initial and intermediate layers, we construct the intermediate layers by stacking the initial layers vertically; the construction of the tensor representation layer, here, for intermediate layers from different categories (such as power parameters, environmental parameters, and efficiency parameters) (for example, intermediate layers represented as second-order tensors), we can stack these second-order tensors to form a multi-dimensional tensor to comprehensively represent the state of the distribution network. For example, assuming we have C categories of parameters, we can define a C-dimensional tensor X∈R^(N x T x C):X=[X1,X2,...,XC];Here, it is different from the construction of the intermediate layers. Generally, we can consider stacking the second-order tensors according to the third dimension (category dimension). Thus, by operating layer by layer, we can realize the comprehensive and dynamic modeling and representation of the distribution network using this multi-dimensional tensor (corresponding to the current distribution optimization task when selecting parameters).

[0042] Example 4

[0043] The optimization control layer is set to be highly open, and it is consistent with the tensor data structure constructed by the data processing layer. That is, power distribution optimization is based on tensor flow and corresponding data processing constructs optimization objectives and constraints.

[0044] In the specific data processing process, the data processes differ significantly depending on the specific power distribution optimization event. The optimization control layer also features a modular, extensible tool library. On one hand, it modularizes the data algorithms and processing processes built from historical power distribution optimization events into optimization control tool suites. On the other hand, the tools that the optimization control layer can call are not directly set as these optimization control tool suites. Instead, any optimization control tool suite is divided into blocks based on its data processing steps (allowing for further subdivision or merging of these steps according to the data processing objectives, thus achieving block processing based on data processing effectiveness). These blocks are then marked and made callable in the library. In this way, regardless of the significant differences between historical optimization control tool suites and current power distribution events, as long as some data optimization steps are corresponding, the stored optimization control tool suite sub-modules can be called in a specific data processing process. Several examples of the optimization control layer's data processing process are shown below. The core technology of the optimization control layer's data processing process is filed for patent application separately.

[0045] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. In each embodiment, the hardware implementation of the technology can directly utilize existing smart devices, including but not limited to industrial control computers, PCs, smartphones, handheld devices, and floor-standing devices. The input device is preferably an on-screen keyboard, the data storage and computing modules utilize existing memory, calculators, and controllers, the internal communication module utilizes existing communication ports and protocols, and the remote communication utilizes existing GPRS networks, the World Wide Web, etc.

[0046] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. The functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in a computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0047] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0048] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A distribution automation data system compatible with a high proportion of distributed power sources and diverse load access, characterized in that: Based on the basic accumulated data of power grid operation, a high-precision and intensive data model is constructed by introducing algorithm technology resources. In terms of data structure and data algorithm, it has the scalability to meet the needs of various sub-items of power distribution optimization. It covers the data content dimensions of multiple orientations of source-grid-load and the data requirement dimensions based on power distribution automation. On this basis, it can further develop data algorithms or data processes for various data needs and data optimization orientation. The underlying initialization structure of the data model is set as follows: the multi-dimensional data structure of the distribution network, which includes distributed grid-connected power sources, grid nodes, grid branch flows, and multiple loads, is represented as a tensor data structure; based on the large-scale and dynamic source-grid-load-storage multi-directional basic power data provided by the power grid system, the tensor dimension is set to represent orthogonal grid parameters including time, node location, power source type, load type, power parameters, and environmental parameters. The tensor dimension attributes are extended according to the following two sets of data conditions: ω, the power grid system provides new data parameters or independent power parameters are generated by new sensors connected to the grid; ψ, new tensor components or new tensor component layouts are generated for new distribution optimization needs or data processing objectives. Based on the underlying initialization structure of the data model, we set the entire data system to include at least a data acquisition layer and a data processing layer in terms of data logic and data structure. Among them, the data acquisition layer connects to the large-scale and dynamic dataset provided by the power grid, which includes real-time and cumulative data of distributed power sources and multiple load access points, and performs necessary and basic data processing. The data processing layer primarily handles the tensor structure processing of the power grid's basic dataset. The core component involves constructing a linearly distributed data tool to represent the power grid's distribution model as a "distributed source-node parameter network-multi-load" state-space model in a multi-dimensional geometric space. In this state-space model, scalar data from the power grid system is linearly distributed to construct a first-order tensor. The dynamic sequences of these first-order tensors are then merged to construct a second-order tensor, representing the dynamic behavior of the distribution network. Furthermore, based on the same linearly distributed data process, second-order dynamic tensors from different independent sources—power parameters, environmental parameters, and efficiency parameters—are expanded in orthogonal dimensions, thereby representing the multi-dimensional data parameters of the distribution network using tensors. The data processing layer, based on linear allocation, represents the power grid distribution model as a "distributed source-node parameter network-multi-element load" state tensor in a multi-dimensional geometric space. On one hand, it constructs one or more specific data processes and stores them in the data processing layer as a basic tool library for generating tensor structures that can be called upon. On the other hand, for the real-time updated distribution network node parameters and the continuously expanding distribution optimization sub-objectives, it writes linear allocation data processes that directly correspond to the current distribution optimization sub-objectives as needed, thereby forming new tensor structure generation tools. The new tools formed by any operation are numbered and stored in the data processing layer as an update tool library for tensor structure generation. The basic tool library and the update tool library are merged together to correspond to the extensible data tool call library group of the data processing layer.

2. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 1, characterized in that: The data system also includes a third layer: the optimization and control layer; Its tensor data structure based on the data processing layer constructs optimization algorithms and runs data processing processes according to the established power distribution requirements and continuously updated power distribution optimization sub-items, and generates optimization reference data sequences.

3. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 2, characterized in that: Based on the above three-layer data architecture, the following open structure is further set up. The openness here includes the scalability within the layers and the construction of new data layers: ω, the cumulative update of power grid data, that is, the continuous proposal and generation of optimization sub-items in power distribution operation; ψ, for the entire data system, the later access to external data processing platform, namely the open-source intelligent tensor data flow model and its supervised iterative machine optimization tools, to carry out artificial intelligence construction.

4. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 1, characterized in that: The data acquisition layer performs basic data processing on the basic dataset from the power grid, i.e., data preprocessing. For distributed power sources and multiple loads in the distribution network nodes, it performs data cleaning and / or data standardization and / or data interpolation and / or other data preprocessing based on various distribution optimization sub-items. Data cleaning removes outliers and noise to clean the data. Thresholding or statistical methods are used to identify and process outlier data points, while data standardization scales the data to a standard range. Meanwhile, the data provided by the power grid system is discretized, which is compatible with the underlying initial tensor data structure. However, various power distribution optimization needs typically require the introduction of a complete dynamic data analysis model for differential analysis on the data function. Therefore, a data interpolation preprocessing algorithm is introduced, including two data processing orientations: ω, using linear or polynomial interpolation to fill in missing data points; and ψ, constructing a continuous data function and tensor field model through interpolation.

5. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 4, characterized in that: The data preprocessing in the data acquisition layer allows for the direct use of existing data cleaning and format standardization algorithms, while also allowing for the expansion and storage of self-built data preprocessing algorithms; Furthermore, both existing algorithms and self-built extended data preprocessing algorithms are stored in a callable manner at the data acquisition layer; The self-built data preprocessing algorithm mentioned above includes at least: For distributed power sources and multiple loads in the distribution network nodes, the data credibility reference algorithm is set as follows: if the distance between a data point and the mean μ is less than or equal to k times the variance σ, then the data point will be assigned a credibility parameter above the threshold; otherwise, the data point will be assigned a credibility parameter below the threshold. Subsequently, a credibility check data process or a credibility weighting data process is set according to the actual objective situation of the current distribution optimization event. The credibility check data process determines the retention or deletion / replacement of data based on the credibility parameter, and the credibility weighting data process assigns numerical weights to the data credibility within the range of (0-1) based on the credibility parameter and its refined indicators. For numericalized electrical parameter sequences, environmental parameter sequences, and other arbitrary numerical parameter sequences or quantifiable parameter sequences of distributed power sources and multiple loads in distribution network nodes, the data normalization scaling algorithm is constructed as follows: ω, each data point is subtracted from the mean and then divided by the standard deviation to transform all data points to a new standard scale; further normalization processing pre-sets the mean of the data sequence to a selected constant, the standard deviation of the data sequence to a selected constant, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event; ψ, each data point is subtracted from the minimum value and then divided by the difference between the maximum and minimum values ​​to transform the data point to a new standard scale; further normalization processing pre-sets the maximum value of the data sequence to a selected constant, the minimum value of the data sequence to a selected constant, or other fixed integer values ​​or their percentage values ​​that match the current distribution optimization event.

6. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 2, characterized in that: The tensor data structure constructed by the optimization control layer and the data processing layer is consistent, that is, the power distribution optimization is based on tensor flow and the corresponding data processing constructs the optimization objectives and constraints. The optimization control layer is set to be highly open, as it has completely different data processes based on different single power distribution optimization events; The optimization control layer also features a modular and extensible tool library. On one hand, it modularizes the data algorithms and data processing processes used to construct historical power distribution optimization events into optimization control tool suites. On the other hand, the tools that the optimization control layer can call are not directly set as the aforementioned optimization control tool suites. Instead, any optimization control tool suite is configured according to its data processing steps, and the data processing steps can be further subdivided or merged according to the data processing objectives. This enables block processing based on data processing effectiveness, with block marking and callable storage. In this way, no matter how different the historical optimization control tool suite is from the current power distribution event, as long as some data optimization steps are corresponding, the stored optimization control tool suite sub-modules can be called in a specific data processing process.

7. The distribution automation data system compatible with high proportion of distributed power sources and multiple load access as described in claim 1, characterized in that: The data model is externally compatible and extensible in both data structure and data algorithm. From the initial stage of the data structure, it is designed to be adapted to external open-source intelligent tensor data flow models, so that it has data consistency and external connectivity and extensibility with external data processing platforms, namely open-source intelligent tensor data flow models and their supervised iterative machine optimization tools.