Method, system and equipment for constructing information model of internet of things of transformer substation and medium

Abstract

The invention relates to the technical field of transformer substation Internet of Things information model construction, and discloses a transformer substation Internet of Things information model construction method, system and device and a medium, and the method effectively solves the problems of poor compatibility, insufficient integrity and difficulty in dynamic optimization in the transformer substation Internet of Things information model construction process in the prior art. By constructing the third consistency verification and optimization model, the compatibility, integrity and consistency of the information model in actual communication can be accurately verified, dynamic optimization is realized, and the stability and reliability of the information model in a complex communication environment are greatly improved. According to the invention, the development and application of the Internet of Things technology of the transformer substation are further promoted, and powerful support is provided for intelligent and automatic management of a power system.

Description

Technical Field

[0001] This invention relates to the field of substation Internet of Things (IoT) information model construction technology, and in particular to a method, system, device and medium for constructing a substation IoT information model. Background Technology

[0002] In existing technologies, there are many problems in the construction of IoT information models for substations. On the one hand, due to the complexity of multi-source heterogeneous data, traditional data processing and modeling methods are difficult to cope with effectively, resulting in low data utilization efficiency and insufficient accuracy and reliability of information models.

[0003] For example, the data formats and standards collected by different types of sensors vary greatly, which can easily lead to errors in semantic parsing and classification, affecting the construction and application of subsequent models.

[0004] On the other hand, existing information models lack effective consistency verification and dynamic optimization mechanisms, and cannot adapt in a timely manner to changes in substation business logic and the access of new equipment.

[0005] When the operating environment, voltage level, or maintenance requirements of a substation change, the existing information model may fail to function properly, requiring extensive manual adjustments and optimizations, which consumes significant time and manpower. Furthermore, the lack of automated processes and effective version management during the deployment and updating of the information model can easily lead to inconsistencies between different devices and systems, impacting the overall operational efficiency and reliability of the substation's Internet of Things (IoT). Summary of the Invention

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] Therefore, the present invention provides a method, system, device and medium for constructing an IoT information model for substations, which can solve the problems of multi-source heterogeneous data processing, lack of effective consistency verification and dynamic optimization mechanism, and lack of automation and version management in the construction of IoT information models for substations in the prior art.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] In a first aspect, the present invention provides a method for constructing an Internet of Things (IoT) information model for a substation, comprising:

[0010] Acquire multi-source heterogeneous data from IoT devices in the target substation, and preprocess the multi-source heterogeneous data;

[0011] The multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data;

[0012] A first device semantic recognition model is established based on the preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities.

[0013] Based on the output of the semantic recognition model of the first device, a second information structure mapping model is constructed. The second information structure mapping model is used to map the data structures of different devices into a unified information body format.

[0014] Based on the second information structure mapping model and the business logic rules of the target substation, a third consistency verification and optimization model is constructed.

[0015] The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization.

[0016] By calling the output of the third consistency verification and optimization model through the preset configuration strategy engine, a standardized target substation IoT information model template is generated, and a deployment command is triggered to realize the deployment and update of the model on the edge gateway or platform side.

[0017] As a preferred embodiment of the substation Internet of Things (IoT) information model construction method of the present invention, the establishment of the first device semantic recognition model includes:

[0018] A preset set of device semantic features, which includes several device semantic dimensions;

[0019] Feature vectors related to the semantic feature set of the device are extracted from preprocessed multi-source heterogeneous data;

[0020] Construct a classification model based on a deep neural network, with device semantic features as input and device semantic category labels as output.

[0021] As a preferred embodiment of the substation Internet of Things (IoT) information model construction method of the present invention, the construction of the second information structure mapping model includes:

[0022] A pre-defined standard information body template library is provided, which contains standard data structure definitions applicable to different device types.

[0023] Design an information mapping strategy based on a rule engine to match the semantic category labels of the device output by the first device semantic recognition model with templates in the standard information body template library.

[0024] As a preferred embodiment of the substation Internet of Things (IoT) information model construction method described in this invention, the construction of the third consistency verification and optimization model includes:

[0025] Establish a verification mechanism based on constraint satisfaction problems, and set several constraint conditions;

[0026] The output data of the second information structure mapping model is input into the solver to detect whether there are any violations of constraints.

[0027] Optimization suggestions are generated for the detected inconsistencies, including field renaming, unit conversion, data completion, and structure reorganization.

[0028] This preferred solution effectively ensures the data consistency and accuracy of the substation IoT information model. In actual substation operating environments, data sources are diverse and complex, with differences in data formats and semantics generated by different devices and systems. By constructing a third consistency verification and optimization model, inconsistencies in the data can be detected in a timely manner, avoiding information misunderstandings and decision-making errors caused by data inconsistencies.

[0029] As a preferred embodiment of the substation Internet of Things (IoT) information model construction method of the present invention, the step of dynamic optimization includes:

[0030] An improved genetic algorithm is introduced to perform multi-objective optimization of the information model structure;

[0031] The multiple objectives include minimizing communication overhead and maximizing data readability;

[0032] Output the final information model version after consistency verification and structural optimization.

[0033] As a preferred embodiment of the substation IoT information model construction method of the present invention, the output results of calling the third consistency verification and optimization model through the preset configuration strategy engine include:

[0034] The configuration strategy engine selects a matching information model to generate a strategy based on context information, including the target substation's region, voltage level, and operation and maintenance management requirements.

[0035] Receive the final information model version from the third consistency verification and optimization model;

[0036] Automatically generate model deployment configuration files suitable for edge gateways, main station systems, or cloud platforms;

[0037] Trigger the automated deployment process to push the information model to relevant devices and systems, and record version change logs.

[0038] As a preferred embodiment of the substation Internet of Things (IoT) information model construction method described in this invention, it further includes:

[0039] Establish an information model operation monitoring mechanism to continuously monitor the performance data of the final information model version in actual communication. The performance data includes at least parsing success rate, transmission delay and error rate.

[0040] The monitoring data is fed back to the semantic recognition model of the first device and the mapping model of the second information structure for iterative updates of the model;

[0041] When a new device is detected or a change in business requirements is detected, the information model reconstruction process is automatically initiated.

[0042] Secondly, this invention provides a substation Internet of Things (IoT) information model construction system, comprising:

[0043] The data acquisition and processing module is used to acquire multi-source heterogeneous data from IoT devices in the target substation and to preprocess the multi-source heterogeneous data.

[0044] The multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data;

[0045] The first model building module is used to build a first device semantic recognition model based on preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities.

[0046] The second model building module is used to construct a second information structure mapping model based on the output results of the first device semantic recognition model. The second information structure mapping model is used to map the data structures of different devices into a unified information body format.

[0047] The third model building module is used to construct a third consistency verification and optimization model based on the second information structure mapping model and the business logic rules of the target substation.

[0048] The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization.

[0049] The template generation module is used to call the output results of the third consistency verification and optimization model through the preset configuration strategy engine, generate a standardized target substation IoT information model template, and trigger deployment instructions to realize the deployment and update of the model on the edge gateway or platform side.

[0050] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0051] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0052] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a method for constructing an information model for a substation Internet of Things (IoT). This method effectively solves the problems of poor compatibility, insufficient completeness, and difficulty in dynamic optimization during the construction of substation IoT information models in existing technologies. By constructing a third consistency verification and optimization model, the compatibility, completeness, and consistency of the information model in actual communication can be accurately verified, and dynamic optimization can be achieved, greatly improving the stability and reliability of the information model in complex communication environments. This invention further promotes the development and application of substation IoT technology and provides strong support for the intelligent and automated management of power systems. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 A flowchart illustrating a method for constructing an Internet of Things (IoT) information model for a substation, as provided in one embodiment of the present invention.

[0055] Figure 2 This is an internal structure diagram of an electronic device for a method of constructing an Internet of Things (IoT) information model for a substation, as provided in one embodiment of the present invention. Detailed Implementation

[0056] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0057] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for constructing an Internet of Things (IoT) information model for a substation, comprising:

[0058] Existing technologies suffer from several problems. For example, insufficient data integration and processing capabilities make it difficult to efficiently fuse and analyze massive amounts of multi-source, heterogeneous data, resulting in information models that cannot comprehensively and accurately reflect the actual operation of substations. Furthermore, some models lack effective tracking and feedback mechanisms for real-time equipment status, failing to promptly detect changes and make corresponding adjustments, leading to poor timeliness and adaptability of the information models. Additionally, the lack of unified standards and specifications in the construction of information models hinders information exchange and sharing between different manufacturers and systems, impacting the overall collaborative operation efficiency of the substation IoT.

[0059] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to implement the method for constructing the Internet of Things information model of the substation using multiple embodiments.

[0060] Figure 1 A flowchart illustrating a method for constructing an Internet of Things (IoT) information model for a substation is shown, including:

[0061] S101, acquire multi-source heterogeneous data from IoT devices in the target substation, and preprocess the multi-source heterogeneous data, including:

[0062] It should be noted that to build an IoT information model for a substation, it is necessary to start with the relevant data from the target substation equipment. This requires comprehensively acquiring multi-source heterogeneous data generated by the IoT devices in the target substation. These data sources are wide-ranging. For example, sensor data reflects various operating parameters of the equipment, control command data reflects the operation of the equipment, environmental monitoring data helps to understand the external environmental conditions of the substation, and equipment operation log data records the historical operating information of the equipment.

[0063] In this embodiment of the invention, the multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data;

[0064] In some specific implementations, data cleaning algorithms can be used to remove noise and outliers from sensor data, and data normalization methods can be used to unify data collected by different sensors to the same scale range, thereby improving the accuracy and efficiency of subsequent processing.

[0065] In some specific implementations, control command data should be standardized in format to ensure that control commands of different types and sources have a unified expression form, which facilitates system recognition and processing.

[0066] In some specific implementations, environmental monitoring data needs to be classified and feature extracted. Data on different environmental factors are classified and organized, and key features are extracted in order to better analyze the impact of the environment on substation equipment.

[0067] In some specific implementations, data mining techniques can be used to extract potential patterns and valuable information from equipment operation log data, such as equipment failure modes and operating trends.

[0068] It should be noted that when acquiring multi-source heterogeneous data, the wide range and diverse types of data, along with significant differences in data format, semantics, and structure, pose a significant challenge to data integration and processing. Multi-source heterogeneous data may originate from equipment manufactured by different companies, employing different communication protocols and data standards, making direct data fusion and analysis difficult. Furthermore, with the continuous development and upgrading of substations, new equipment and systems are constantly being integrated, further increasing the complexity and heterogeneity of the data.

[0069] If these multi-source heterogeneous data cannot be effectively processed, the subsequently constructed information model will fail to accurately reflect the actual operation of the substation, severely impacting the model's compatibility, completeness, and consistency. For example, inconsistent data structures across different devices may lead to information loss or errors during transmission and processing, affecting the model's accuracy and reliability. Furthermore, the lack of effective management and utilization of multi-source heterogeneous data can also result in a waste of data resources, failing to fully realize the data's value. Therefore, preprocessing the acquired multi-source heterogeneous data is a crucial step in constructing an accurate and reliable substation IoT information model.

[0070] In some specific implementations, preprocessing may include operations such as data cleaning, transformation, and integration. Data cleaning removes noise, duplicate values, and erroneous data to improve data quality. For example, in substation sensor data, there may be outliers caused by equipment malfunctions or communication interference. By setting reasonable threshold ranges, data exceeding these ranges are considered outliers and removed. Data transformation unifies data of different formats and standards into a consistent format for subsequent analysis and processing. For instance, equipment from different manufacturers may use different timestamp formats; converting them to a standard time format facilitates data integration and comparison. Data integration merges data from different data sources into a unified data warehouse, breaking down data barriers. Taking substations as an example, integrating the operating data, maintenance records, and monitoring data of various devices provides comprehensive data support for building information models. Through these preprocessing steps, the complexity of multi-source heterogeneous data can be effectively reduced, data availability and consistency can be improved, and a solid foundation can be laid for building accurate and reliable substation IoT information models.

[0071] It should be noted that this invention does not limit the specific types of multi-source heterogeneous data or the specific steps of preprocessing, as long as it meets the requirement of being able to process data from different sources and of different types, so that the data reaches the quality standard that can be used to build information models.

[0072] S102, A first device semantic recognition model is established based on the preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes, and communication capabilities, wherein:

[0073] It is important to note that after acquiring the preprocessed multi-source heterogeneous data, it is necessary to extract the semantic information from this data in order to accurately identify the type, functional attributes, and communication capabilities of the equipment. This is because different types of equipment perform different functions in substations and have different communication methods and capabilities. Only by accurately identifying this information can a reliable foundation be provided for subsequently building an information model.

[0074] In some specific implementations, machine learning algorithms, such as convolutional neural networks (CNN), recurrent neural networks (RNN), and their variants (such as LSTM and GRU) in deep learning, can be used to train preprocessed multi-source heterogeneous data to construct a first device semantic recognition model. These algorithms can automatically learn the features and patterns of data from large amounts of data, thereby achieving accurate recognition of device semantics.

[0075] For example, for sensor data, the type of sensor (such as temperature sensor, pressure sensor, etc.) and its functional attributes (such as measurement range, accuracy class, etc.) can be identified by analyzing its collected physical quantities, acquisition frequency, data accuracy, and other characteristics. For control command data, the type of control command (such as on / off control command, regulation control command, etc.) and its corresponding device communication capabilities (such as supported communication protocols, communication rates, etc.) can be identified by analyzing the command's format, content, transmission frequency, and other information.

[0076] In some specific implementations, data augmentation and model fusion techniques can be employed to improve the accuracy and generalization ability of the semantic recognition model of the first device. Data augmentation can increase the diversity of training data by transforming the original data (such as rotation, translation, scaling, etc.) and adding noise, thereby improving the robustness of the model. Model fusion combines multiple different models, comprehensively utilizing the advantages of each model to improve the final recognition performance.

[0077] In some specific implementations, continuous updates and optimizations are required to ensure the real-time performance and effectiveness of the first device semantic recognition model. With the upgrading of substation equipment, the emergence of new communication protocols, and changes in business requirements, the characteristics and patterns of multi-source heterogeneous data also change accordingly. Therefore, it is necessary to periodically collect new data to retrain and optimize the first device semantic recognition model, ensuring that the model can accurately identify the semantic information of the equipment.

[0078] In some specific implementations, expert knowledge and experience can be combined to verify and correct the output of the first device semantic recognition model. Expert knowledge can help identify complex device semantic information, improving the model's accuracy and reliability. For example, for some special types of equipment or complex communication protocols, experts can provide additional information and guidance to assist the model in accurate identification. In this way, the performance of the first device semantic recognition model can be further improved, providing strong support for building an accurate and reliable substation IoT information model.

[0079] In this embodiment of the invention, establishing a first device semantic recognition model includes:

[0080] A preset set of device semantic features, which includes several device semantic dimensions;

[0081] Feature vectors related to the semantic feature set of the device are extracted from preprocessed multi-source heterogeneous data;

[0082] Construct a classification model based on a deep neural network, with device semantic features as input and device semantic category labels as output.

[0083] Specifically, a set of device semantic features is defined, which includes multiple semantic dimensions used to distinguish different devices. Each dimension represents a quantifiable or categorizable attribute of the device.

[0084] For example, suppose the set of device semantic features is It consists of D independent device semantic dimensions:

[0085]

[0086] Among them, f d This represents the d-th semantic dimension (e.g., device type, measured physical quantity, communication protocol, data sampling rate, voltage level, etc.).

[0087] Specific dimension types can include the following:

[0088] For numerical dimensions (such as sampling rate f) sample Its range of values ​​is the set of real numbers.

[0089] For categorical dimensions (such as communication protocol f) protocol Its value range is a finite set of categories.

[0090] Furthermore, based on preprocessed multi-source heterogeneous data (Including sensor data, control commands, environmental monitoring, operation logs, etc.), extract and pre-set feature sets. Relevant feature vectors.

[0091] Define a feature extraction function φ, which extracts features from... The features of each dimension are analyzed and quantified:

[0092]

[0093] For numerical dimension f d x d Take the measured or calculated value directly (such as the average sampling rate).

[0094] The input to this feature extraction function is the preprocessed data. The output is a D-dimensional feature vector. Each component x d Corresponding to dimension f d The quantized value.

[0095] Furthermore, regarding the categorical dimension f d One-hot encoding is typically used. Assume f d There is Kd x has 1 possible category, then x d It will be expanded to K d A binary component. For example, if f protocol There are three categories: {Modbus, IEC61850, MQTT}. If the device uses IEC61850, the corresponding encoding is [0, 1, 0]. After this processing, the original D-dimensional feature vector may be expanded into a higher-dimensional vector.

[0096] Furthermore, we can construct a deep neural network classification model. Extracted feature vectors As input, the semantic category label of the output device is used, and the output can be represented as a C-dimensional probability vector p∈[0,1]. C Where C is the total number of predefined semantic categories (e.g., transformer, circuit breaker, temperature sensor, humidity sensor, etc.), and each component p is a component of p. c Let represent the probability that the input data belongs to the c-th semantic category, and satisfy .

[0097] Furthermore, deep neural networks can be represented as a complex nonlinear function g:

[0098]

[0099] Where θ represents all trainable parameters of the model (weights and biases).

[0100] Furthermore, the function g is often composed of multiple layers of neural networks, for example:

[0101] h (1) =σ (1) (W (1) x+b (1) )

[0102] h (2) =σ (2) (W (2) h (1) +b (2) )

[0103]

[0104] p = softmax(W (L) h (L-1) +b (L) )

[0105] Among them, h (l) W is the hidden layer output vector of layer l. (l) and b (l)σ represents the weight matrix and bias vector of the l-th layer. (l) It is the activation function of the l-th layer (such as ReLU, Sigmoid). The softmax function is used in the last layer to ensure that the output p is an efficient probability distribution.

[0106] Furthermore, using a labeled training dataset (where y) i (This is the one-hot encoded vector of the true class), and the model parameters θ are optimized by minimizing a loss function (such as cross-entropy loss):

[0107]

[0108] in, The optimization process typically uses gradient descent (such as Adam).

[0109] Furthermore, given a new input feature vector x new The semantic category labels predicted by the model The category with the highest probability:

[0110]

[0111] It should be noted that, through a pre-set feature set Key attributes for device identification were defined. Feature vectors x were extracted from the raw data using a function φ. Finally, a deep neural network model was developed. It is trained to learn the complex mapping relationship from feature vector x to semantic category label c, realizing automated and high-precision semantic recognition of IoT devices in substations.

[0112] S103, Based on the output of the semantic recognition model of the first device, a second information structure mapping model is constructed. The second information structure mapping model is used to map the data structures of different devices into a unified information body format, wherein:

[0113] It's important to note that after obtaining the output of the semantic recognition model from the first device, the data structures of different devices need to be unified. This is because devices from different manufacturers may use different data structures and formats, which poses significant challenges to data integration and analysis. For example, some devices may store data in binary format, others in text format, and still others may use custom data structures. This diversity makes it difficult for data to be directly interacted and shared between different devices, affecting the construction and application of information models.

[0114] In some specific implementations, constructing a second information structure mapping model can effectively solve this problem. This model, based on information such as device type, functional attributes, and communication capabilities determined by the first device semantic recognition model, finds the correspondence between different device data structures and maps them into a unified information body format. For example, for temperature data collected by different sensors, although their representation in the original data structure may differ, the second information structure mapping model can uniformly convert them into a standard temperature value representation, facilitating subsequent processing and analysis.

[0115] In some specific implementations, data mapping rules and algorithms can be used to achieve this conversion. First, the data structures of different devices are analyzed in detail to determine their key data fields and their meanings. Then, according to the unified information body format requirements, corresponding mapping rules are formulated to map the data fields of different devices to the unified information body fields. For example, the "temperature value" field in the data structure of device A and the "current temperature" field in the data structure of device B can both be mapped to the "temperature" field in the unified information body format.

[0116] In an embodiment of the present invention, constructing the second information structure mapping model includes:

[0117] A pre-defined standard information body template library is provided, which contains standard data structure definitions applicable to different device types.

[0118] Design an information mapping strategy based on a rule engine to match the semantic category labels of the device output by the first device semantic recognition model with templates in the standard information body template library.

[0119] Specifically, a standard information body template library is first defined, which predefines a unified data structure for different types of devices.

[0120] Furthermore, let the standard information body template library be T, which contains C standard information body templates, corresponding to the C semantic categories of the first device semantic recognition model:

[0121] T = {T1,T2,...,T} C}

[0122] Among them, T c This represents the standard information body template corresponding to the c-th semantic category (e.g., "temperature sensor").

[0123] Furthermore, each template T c A structured data format is defined, typically a collection of field-value pairs. Let template T be... c Includes K c One predefined field:

[0124]

[0125] in, This represents the name of the kth standard field of the c-th device (e.g., "temperature", "device_id", "timestamp"). Indicates the data type of the field (such as float, string, int, datetime) and possible constraints (such as units of ℃, %).

[0126] Furthermore, design an information mapping strategy based on a rule engine. The engine searches for the corresponding template T based on the semantic category label c output by the semantic recognition model of the first device. c And define a set of rules to process the raw device data d raw Mapping to template T c On the field.

[0127] Furthermore, the rules engine The input is raw device data d raw Raw device data is a data packet (such as JSON, XML, or binary stream) from a specific device, with its own proprietary or protocol-specific format. It can be represented as a collection of field value pairs:

[0128] d raw ={(src_field1,v1),(src_field2,v2),...,(src_field N ,v N )}

[0129] Furthermore, the semantic category label c is determined by the first device semantic recognition model. Output.

[0130] For example, a rules engine The core function is to execute a mapping function ψ, which receives d raw and c, and generate a template T that matches. c Standardized information body of structure std :

[0131]

[0132] Among them, s std It is a structured object, whose structure is similar to T. c Consistent.

[0133] Furthermore, for each semantic category c, a rule set is defined. This rule set contains a series of mapping rules r c,k Used to fill template T c Each field in c,k :

[0134]

[0135] Each rule r c,k Defines how to get from d raw Get or calculate field c,k The value of .

[0136] For example, rules can include the following: ① direct field mapping, ② field renaming mapping, ③ expression calculation mapping, ④ constant filling, ⑤ default value filling, and ⑥ unit conversion. Direct field mapping directly maps a field in the original data to the corresponding field in the template without additional processing. For example, the "voltage value" field in the original data can be directly mapped to the "voltage" field in the template. Field renaming mapping modifies field names to conform to template requirements while keeping the data content unchanged. For example, the "current intensity" field in the original data might be renamed to "current" in the template. Expression calculation mapping involves calculating the values ​​of fields in the template using specific mathematical expressions based on multiple fields in the original data. For example, the value of the "current" field in the template can be calculated using the formula "current = power / voltage" based on the "power" and "voltage" fields in the original data. Constant filling involves filling a fixed constant value into a field in the template. This constant value does not depend on the original data and may be determined based on business requirements or industry standards. For example, "substation equipment" is always filled into the "equipment type" field in the template. Default value filling assigns a pre-defined default value to the corresponding field in the template when a field is missing in the original data. This ensures the integrity of the template and avoids incomplete information due to missing data. For example, if the original data lacks a "temperature" field, the template can fill in the default value "25℃" for the "temperature" field. Unit conversion converts the units of the values ​​in the original data to match the units required by the template. For example, if the original data uses "centimeters" for length, but the template requires "meters," the length values ​​in the original data need to be divided by 100 for unit conversion.

[0137] These rules play a crucial role in the construction of the IoT information model for substations. They can transform raw, potentially inconsistent data into standardized information bodies that conform to specific template structures, thereby facilitating subsequent data analysis, processing, and application.

[0138] It should be noted that the specific execution flow of the rule engine is as follows: Based on the input semantic category label c, it retrieves data from the template library. Retrieve the corresponding template T from the middle c Load the rule set associated with category c. For template T c Each field in c,k In the rule set Find the corresponding rule r in c,k Application rule r c,k To the original data d raw Above, calculate or extract the field. c,k The value v c,k Ensure v c,k The data types and units conform to type c,k Requirements (which may require implicit or explicit type conversion in the rules). Combine all the mapped field-value pairs to form the final standardized information body s. std :

[0139]

[0140] It should be noted that by establishing a standard template library T, a unified data "language" is defined for different devices. The rule engine acts as a "translator," utilizing a predefined set of rules. Based on the semantic category c of the device, the original, heterogeneous device data d is... raw Accurately "translate" into a standard template T c Unified Information Body std This process standardizes and unifies data structures, laying a solid foundation for subsequent consistency checks and inter-system communication.

[0141] S104, based on the second information structure mapping model and the target substation business logic rules, construct a third consistency verification and optimization model, wherein:

[0142] It should be noted that after the second information structure mapping model is completed, although the data structures of different devices have been mapped to a unified information body format, these information bodies still need to conform to the business logic rules of the target substation in practical applications. The business logic rules of the target substation cover many aspects such as equipment operation specifications, safety standards, and operating procedures. For example, in the process of power transmission, there are strict voltage range requirements for equipment of different voltage levels; during equipment maintenance, there are prescribed maintenance cycles and operating procedures.

[0143] The purpose of building the third consistency verification and optimization model is to ensure that data based on the unified information body format conforms to these business logic rules. This model performs consistency checks on the data processed by the second information structure mapping model to see if the data meets the requirements of the business logic rules. If data is found to be inconsistent with the rules, such as the operating parameters of a device exceeding the specified safety range, or the operating procedure not conforming to the standard, the model will issue an alert in a timely manner.

[0144] In some specific implementations, the third consistency verification and optimization model can employ rule engine technology to implement consistency verification. The rule engine can quickly match and judge the input data based on preset business logic rules. Furthermore, to improve the model's intelligence and adaptability, machine learning algorithms can be combined to automatically discover potential patterns and anomalies in the data by learning from historical data and business logic rules, further optimizing the verification process.

[0145] In this embodiment of the invention, the third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization.

[0146] In this embodiment of the invention, constructing the third consistency verification and optimization model includes:

[0147] Establish a verification mechanism based on constraint satisfaction problems, and set several constraint conditions;

[0148] The output data of the second information structure mapping model is input into the solver to detect whether there are any violations of constraints.

[0149] Optimization suggestions are generated for detected inconsistencies, including field renaming, unit conversion, data completion, and structure reorganization.

[0150] In embodiments of the present invention, dynamic optimization includes:

[0151] An improved genetic algorithm is introduced to perform multi-objective optimization of the information model structure;

[0152] The multiple objectives include minimizing communication overhead and maximizing data readability;

[0153] Output the final information model version after consistency verification and structural optimization.

[0154] Specifically, a verification mechanism based on the constraint satisfaction problem (CSP) is established, which models the consistency verification problem of the information model as a constraint satisfaction problem.

[0155] Furthermore, we define the CSP triple: CSP = (V, D, C), where the variable set V represents the normalized information body s to be verified.std Verifiable elements in the domain. The domain set D is defined for each variable v. k Define its allowed range of values, where C represents the constraint set.

[0156] Furthermore, the constraint set includes value range constraints to ensure field values ​​are within their domain, type constraints to ensure field values ​​are of the correct data type, required constraints to ensure critical fields are not empty, logical / business constraints to define the logical relationships between fields, and unit consistency constraints to ensure fields involved in calculations or comparisons have consistent units. Structural integrity constraints ensure the information body contains all required fields.

[0157] Furthermore, the output data is input into the solver for verification. The CSP solver is used to detect whether the normalized information volume violates any constraints.

[0158] Optimization suggestions are generated for detected inconsistencies. Based on the violation report, specific optimization suggestions are generated to guide how to modify the information model or mapping rules.

[0159] Furthermore, an improved genetic algorithm is introduced for multi-objective structural optimization of the information model's structure (mainly referring to the standard information body template T). C and mapping rule R C The abstract structure is dynamically optimized to balance multiple objectives.

[0160] For example, multiple objectives may include minimizing communication overhead (O comm The communication overhead is mainly determined by the size of the information body. The goal is to generate a more compact information body.

[0161] Define the communication overhead function:

[0162]

[0163] Where `size(type)` is the size in bytes of the data type (e.g., `int32 = 4`, `float64 = 8`, string length is relevant). The optimization goal is minO. comm (T c );

[0164] Furthermore, multiple objectives can also include maximizing data readability / availability (O) read Data readability refers to the ease with which information can be understood and used by technical personnel or upper-level applications. More complete and semantically clear information is more readable.

[0165] Define a readability function:

[0166]

[0167] Among them, w kThis represents the importance weight of the k-th field (determined by business rules or expert knowledge), and relevance is a function that measures the relevance of a field (which can be simply represented by 1). The optimization objective is maxO. read (T c )

[0168] Furthermore, it is necessary to solve the multi-objective function obtained above, using an improved genetic algorithm to transform the information model structure (such as template T) into a more efficient one. c The set of fields is encoded as a chromosome (e.g., a binary string, where 1 indicates the field is included and 0 indicates it is not included), specifically:

[0169] Population initialization: Randomly generate an initial population P(0) containing multiple candidate information model structures.

[0170] Furthermore, fitness evaluation: for each individual (candidate structure) in the population. ):

[0171] Furthermore, calculate its communication overhead.

[0172] Furthermore, calculate its data readability.

[0173] Furthermore, it is substituted into the CSP model for verification to calculate a constraint violation degree. (For example, the number of constraints violated or the weighted sum).

[0174] Furthermore, we define the comprehensive fitness function F, typically using a weighted sum or Pareto optimal method. A simple weighted sum form (minimizing and maximizing must be unified):

[0175]

[0176] Where α, β, γ > 0 are weighting coefficients. The objective is to minimize F.

[0177] Furthermore, after multiple rounds of iterative optimization, the genetic algorithm converges to a set of non-dominated solutions (Pareto front) or a solution with optimal overall performance.

[0178] Select the optimal candidate structure or the one that best fits the current context (such as operational requirements) from the final population as the final information model version after consistency verification and structural optimization.

[0179] The optimized final core settings are based on It can be represented as:

[0180]

[0181] Among them, P(T) max ) is the final generation of the population.

[0182] Furthermore, The optimization results (such as new template definitions and optimized mapping rules) are fed back and updated into the second information structure mapping model for subsequent data processing.

[0183] It should be noted that by establishing a CSP model, inconsistencies in the information body are systematically detected, and specific remediation suggestions are generated. Based on this, an improved genetic algorithm is used to formalize the information model structure optimization problem into a multi-objective optimization problem. By balancing the two key indicators of communication overhead (efficiency) and data readability (quality), and combining consistency constraints, a better information model structure is automatically searched for. The final output... It is a more stable, efficient, and reliable version of the standard information model in actual communication.

[0184] S105, through the preset configuration strategy engine, calls the output results of the third consistency verification and optimization model to generate a standardized target substation IoT information model template, and triggers deployment instructions to realize the deployment and updating of the model on the edge gateway or platform side, wherein:

[0185] It should be noted that once the third consistency verification and optimization model is built and the output results are obtained, the preset configuration strategy engine begins to function. The configuration strategy engine processes the final information model version output by the third consistency verification and optimization model according to pre-defined rules and algorithms, transforming it into a standardized target substation IoT information model template. This template has a unified format and specifications, clearly and accurately describing the relationships between various devices and systems within the substation, data flow, and business logic.

[0186] During the generation of standardized templates, the configuration strategy engine fully considers the actual needs and characteristics of the target substation, and further adjusts and optimizes the model. For example, based on factors such as the substation's scale, voltage level, and equipment type, the parameters and configurations in the template are customized to ensure that the template can better adapt to specific application scenarios.

[0187] Once the standardized IoT information model template for the target substation is generated, the configuration policy engine triggers a deployment command. This command sends the template to the edge gateway or platform, enabling model deployment and updates. On the edge gateway side, the deployment process may involve loading the template into the local device and performing corresponding configuration and initialization operations. This allows the edge gateway to collect, process, and transmit data from the local device based on the new model, improving data processing efficiency and accuracy.

[0188] On the platform side, deployment may involve uploading the template to a cloud server and integrating it with the existing platform system. The platform system will analyze and process the collected data based on the new model, providing more accurate decision support and monitoring services. Simultaneously, the platform system will regularly update and maintain the model to ensure it reflects the latest operating status and business needs of the substation in a timely manner.

[0189] This approach ensures that the IoT information model of the target substation can be deployed and updated in a timely and accurate manner on both the edge gateway and the platform side, thus providing strong support for the intelligent management and operation of the substation. As the substation continues to develop and change, the configuration strategy engine can also dynamically adjust and optimize the model based on new needs and data, ensuring that the model always maintains good performance and adaptability.

[0190] In this embodiment of the invention, the output of the third consistency verification and optimization model invoked through a preset configuration strategy engine includes:

[0191] The configuration strategy engine selects the matching information model to generate a strategy based on the context information, which includes the target substation's region, voltage level, and operation and maintenance requirements.

[0192] Receive the final information model version from the third consistency verification and optimization model;

[0193] Automatically generate model deployment configuration files suitable for edge gateways, main station systems, or cloud platforms;

[0194] Trigger the automated deployment process to push the information model to relevant devices and systems, and record version change logs.

[0195] In this embodiment of the invention, an information model operation monitoring mechanism is established to continuously monitor the performance data of the final information model version in actual communication. The performance data includes at least parsing success rate, transmission delay and error rate.

[0196] The monitoring data is fed back to the semantic recognition model of the first device and the mapping model of the second information structure for iterative updates of the model;

[0197] When a new device is detected or a change in business requirements is detected, the information model reconstruction process is automatically initiated.

[0198] In summary, this invention discloses a method for constructing an IoT information model for substations. This method effectively solves the problems of poor compatibility, insufficient completeness, and difficulty in dynamic optimization in the construction of IoT information models for substations in existing technologies. By constructing a third consistency verification and optimization model, the compatibility, completeness, and consistency of the information model in actual communication can be accurately verified, and dynamic optimization can be achieved, greatly improving the stability and reliability of the information model in complex communication environments. This invention further promotes the development and application of IoT technology for substations, providing strong support for the intelligent and automated management of power systems.

[0199] Example 2, in a preferred embodiment, the detailed steps for calling the output of the third consistency verification and optimization model through the preset configuration strategy engine can be as follows:

[0200] The configuration strategy engine selects the matching information model and generates a strategy based on the context information. The configuration strategy engine first analyzes the current context information to determine the most suitable model deployment and configuration strategy.

[0201] Furthermore, a predefined strategy library Includes multiple information model generation and adaptation strategies:

[0202]

[0203] Each strategy P m Defines how to apply the base model based on the context ctx. Customization is required. For example, each strategy... Defines how to apply the underlying model M based on the context ctx. final Customization is available. For example: P edge A streamlined strategy for edge gateways may involve removing non-critical fields to reduce model size and communication overhead. cloud A complete policy for cloud platforms, retaining all fields and potentially adding metadata. high_sec Encryption strategies suitable for high security requirements embed encrypted fields or requirements into the model. regional Localization strategies applicable to specific regions adjust field naming or units to conform to regional standards.

[0204] Furthermore, the configuration policy engine executes a policy selection function λ, which takes the context information ctx as input and outputs the best-matching policy P. * :

[0205]

[0206] This function can be a rule engine, a simple lookup table, or a classification model.

[0207] Furthermore, the configuration strategy engine receives the final information model version output from the previous stage, which has undergone consistency verification and optimization.

[0208] Furthermore, based on the selected strategy P * and received Automatically generate deployment configuration files for specific target platforms.

[0209] Furthermore, define a model adaptation function ζ, which applies the selected policy P* to the model. The transformation is performed to generate a platform-specific intermediate model representation M. adapted :

[0210] M adapted =ζ(M final ,P * )

[0211] For example, if P* = P_edge, then ζ might remove fields marked "non_essential", reduce the precision of floating-point fields from float64 to float32, or generate a simplified mapping rule.

[0212] Furthermore, the configuration file generation function (ξ) defines a configuration file generation function ξ that converts the adapted model M_adapted into a configuration file format that can be recognized and loaded by the specific platform.

[0213] cfg platform =ξ(M adapted (target platform)

[0214] Output cfg platform It is a specific file or data structure.

[0215] Furthermore, for an edge gateway, it might be a .json or .xml file defining the data point table, communication protocol configuration, and model version number. For the main system, it might be a database script (.sql) used to update the data table structure, or a .conf file. For a cloud platform, it might be a JSON payload for an API call, or a microservice configuration file (such as a .yaml file).

[0216] Furthermore, the configuration strategy engine triggers an automated deployment process, pushing the generated configuration file to the target device and system.

[0217] Furthermore, after the entire process is completed, a detailed version change log must be recorded to ensure traceability.

[0218] It's worth noting that the information model deployment is intelligent and automated through a configuration strategy engine. First, it selects the optimal strategy based on the context. Then, it receives the final model and uses adaptation and generation functions (ξ) to create a platform-specific configuration file. Next, an automated process is triggered by a deployment function to push the model to the target system. Finally, a logging function records the complete version change history. This series of operations ensures that the information model can be flexibly, reliably, and traceably updated and managed in the complex and ever-changing substation IoT environment.

[0219] Example 3, referring to Figure 2 This embodiment also provides a substation Internet of Things (IoT) information model construction system, including:

[0220] The data acquisition and processing module is used to acquire multi-source heterogeneous data from IoT devices in the target substation and to preprocess the multi-source heterogeneous data.

[0221] Multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data;

[0222] The first model building module is used to build a first device semantic recognition model based on preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities.

[0223] The second model building module is used to construct a second information structure mapping model based on the output results of the first device semantic recognition model. The second information structure mapping model is used to map the data structures of different devices into a unified information body format.

[0224] The third model building module is used to construct a third consistency verification and optimization model based on the second information structure mapping model and the business logic rules of the target substation.

[0225] The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization.

[0226] The template generation module is used to call the output of the third consistency verification and optimization model through the preset configuration strategy engine to generate a standardized target substation IoT information model template, and trigger deployment instructions to realize the deployment and update of the model on the edge gateway or platform side.

[0227] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0228] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows: Figure 2 As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for constructing an Internet of Things (IoT) information model for a substation. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0229] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0230] Acquire multi-source heterogeneous data from IoT devices in the target substation and preprocess the multi-source heterogeneous data;

[0231] Multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data;

[0232] A first device semantic recognition model is established based on the preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities.

[0233] Based on the output of the semantic recognition model of the first device, a second information structure mapping model is constructed. The second information structure mapping model is used to map the data structures of different devices into a unified information body format.

[0234] Based on the second information structure mapping model and the business logic rules of the target substation, a third consistency verification and optimization model is constructed.

[0235] The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization.

[0236] By calling the output of the third consistency verification and optimization model through the preset configuration strategy engine, a standardized target substation IoT information model template is generated, and deployment instructions are triggered to realize the deployment and updating of the model on the edge gateway or platform side.

[0237] It should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0238] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0239] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for constructing an Internet of Things (IoT) information model for a substation, characterized in that, include: Acquire multi-source heterogeneous data from IoT devices in the target substation, and preprocess the multi-source heterogeneous data; The multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data; A first device semantic recognition model is established based on the preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities. Based on the output of the semantic recognition model of the first device, a second information structure mapping model is constructed. The second information structure mapping model is used to map the data structures of different devices into a unified information body format. Based on the second information structure mapping model and the business logic rules of the target substation, a third consistency verification and optimization model is constructed. The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization. By calling the output of the third consistency verification and optimization model through the preset configuration strategy engine, a standardized target substation IoT information model template is generated, and a deployment command is triggered to realize the deployment and update of the model on the edge gateway or platform side.

2. The method for constructing a substation Internet of Things (IoT) information model as described in claim 1, characterized in that, The establishment of the first device semantic recognition model includes: A preset set of device semantic features, which includes several device semantic dimensions; Feature vectors related to the semantic feature set of the device are extracted from preprocessed multi-source heterogeneous data; Construct a classification model based on a deep neural network, with device semantic features as input and device semantic category labels as output.

3. The method for constructing a substation Internet of Things (IoT) information model as described in claim 2, characterized in that, The construction of the second information structure mapping model includes: A pre-defined standard information body template library is provided, which contains standard data structure definitions applicable to different device types. Design an information mapping strategy based on a rule engine to match the semantic category labels of the device output by the first device semantic recognition model with templates in the standard information body template library.

4. The method for constructing a substation Internet of Things (IoT) information model as described in claim 3, characterized in that, The construction of the third consistency verification and optimization model includes: Establish a verification mechanism based on constraint satisfaction problems, and set several constraint conditions; The output data of the second information structure mapping model is input into the solver to detect whether there are any violations of constraints. Optimization suggestions are generated for the detected inconsistencies, including field renaming, unit conversion, data completion, and structure reorganization.

5. The method for constructing a substation Internet of Things (IoT) information model as described in claim 4, characterized in that, The aforementioned dynamic optimization includes: An improved genetic algorithm is introduced to perform multi-objective optimization of the information model structure; The multiple objectives include minimizing communication overhead and maximizing data readability; Output the final information model version after consistency verification and structural optimization.

6. The method for constructing a substation Internet of Things (IoT) information model as described in claim 5, characterized in that, The output results of calling the third consistency verification and optimization model through the preset configuration strategy engine include: The configuration strategy engine selects a matching information model to generate a strategy based on context information, including the target substation's region, voltage level, and operation and maintenance management requirements. Receive the final information model version from the third consistency verification and optimization model; Automatically generate model deployment configuration files suitable for edge gateways, main station systems, or cloud platforms; Trigger the automated deployment process to push the information model to relevant devices and systems, and record version change logs.

7. The method for constructing a substation Internet of Things (IoT) information model as described in claim 6, characterized in that, Also includes: Establish an information model operation monitoring mechanism to continuously monitor the performance data of the final information model version in actual communication. The performance data includes at least parsing success rate, transmission delay and error rate. The monitoring data is fed back to the semantic recognition model of the first device and the mapping model of the second information structure for iterative updates of the model; When a new device is detected or a change in business requirements is detected, the information model reconstruction process is automatically initiated.

8. A substation Internet of Things (IoT) information model construction system, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition and processing module is used to acquire multi-source heterogeneous data from IoT devices in the target substation and to preprocess the multi-source heterogeneous data. The multi-source heterogeneous data includes sensor data, control command data, environmental monitoring data, and equipment operation log data; The first model building module is used to build a first device semantic recognition model based on preprocessed multi-source heterogeneous data. The first device semantic recognition model is used to perform semantic parsing and classification of device type, functional attributes and communication capabilities. The second model building module is used to construct a second information structure mapping model based on the output results of the first device semantic recognition model. The second information structure mapping model is used to map the data structures of different devices into a unified information body format. The third model building module is used to construct a third consistency verification and optimization model based on the second information structure mapping model and the business logic rules of the target substation. The third consistency verification and optimization model is used to verify the compatibility, integrity and consistency of the information model in actual communication, and to perform dynamic optimization. The template generation module is used to call the output results of the third consistency verification and optimization model through the preset configuration strategy engine, generate a standardized target substation IoT information model template, and trigger deployment instructions to realize the deployment and update of the model on the edge gateway or platform side.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the substation Internet of Things information model construction method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the substation Internet of Things information model construction method according to any one of claims 1 to 7.

Images