Construction method of intelligent power distribution network based on standard digital model

Abstract

The invention discloses an intelligent power distribution network construction method based on a standard digital model, and relates to the technical field of power distribution network construction, and the method comprises the following steps: selecting an international standard and a protocol of an IEC 61850 or IEEE standard as a design reference, verifying a design target of an intelligent power distribution network and key problems needing to be solved, analyzing demands, and formulating a standard; constructing an overall framework of the intelligent power distribution network; according to the invention, by taking international standards and protocols as design criteria, algorithm models of load prediction, optimization scheduling algorithm, asset management and state monitoring, and fault diagnosis and response are developed or integrated, and strategies of a PID controller, a deep learning prediction control model and a reinforcement learning model are set. Data protection and security strategies are formulated in the overall architecture of the intelligent power distribution network, the operation efficiency and reliability of the intelligent power distribution network are improved, the operation and maintenance cost is reduced, and the user satisfaction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network construction, and in particular to a method for constructing an intelligent power distribution network based on a standard digital model. Background Art

[0002] Smart distribution network uses advanced digital technology and intelligent algorithms to upgrade and optimize traditional distribution network to achieve more efficient, reliable, safe and sustainable power distribution and management. It uses real-time data collection, analysis and control technology to enable the power grid to better cope with challenges such as load fluctuations, fault isolation, energy management, etc., while also providing support for new applications such as new energy access, electric vehicle charging, smart home, etc.

[0003] The standard digital model is a model framework that describes the relationships and interactions between various components, devices and systems in the smart distribution network. It can include device data models, communication protocol models, control models, etc., and achieve interoperability and data exchange between different manufacturers and different types of devices through unified standardized representation.

[0004] The existing technology has the following deficiencies: In the existing technology, the construction of a smart distribution network based on a standard digital model, for power equipment and intelligent control systems from different manufacturers and with different technical standards, the differences between power equipment or software may lead to data interpretation errors, communication failures and incomplete functions, resulting in problems such as difficulty in integration and incompatibility in the construction of a smart distribution network. In addition, the constructed smart distribution network is difficult to prevent hacker attacks, resulting in data leakage, physical and network security vulnerabilities, and the system in which the smart distribution network is located usually relies on complex algorithms to achieve automated control. Once the algorithm is wrong or tampered with, the distribution network's management and control system may have very serious consequences, including equipment maintenance, economic losses and social harm. In addition, as the scale of the smart distribution network continues to expand and the number of new energy connections increases, the original digital models and control strategies are basically unable to flexibly respond to rapidly changing needs, which may limit its long-term sustainable development capabilities.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a method for constructing an intelligent distribution network based on a standard digital model. The present invention solves the problems in the above-mentioned background technology by means of the method steps for constructing an intelligent distribution network based on a standard digital model.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a method for constructing a smart distribution network based on a standard digital model, comprising the following steps:

[0008] S1. Analyze requirements and develop standards based on the design objectives and key issues of smart distribution networks;

[0009] S2. Build the overall architecture of the smart distribution network and formulate data protection and security strategies;

[0010] S3. Based on standard digital models, develop or integrate algorithm models for load forecasting, optimization scheduling algorithms, asset management and condition monitoring, fault diagnosis and response required for smart distribution networks;

[0011] S4. Test the smart distribution network and verify the effectiveness and security of the system and algorithm using the evaluation indicators of accuracy, recall rate and F1 value;

[0012] S5. Regularly update the power system and operation control algorithm of the smart distribution network;

[0013] S6. Continuously optimize the power system architecture and smart distribution network operation algorithms.

[0014] Optionally, in step S1, when reviewing the design objectives and key issues to be resolved for the smart distribution network, international standards and protocols such as IEC61850 or IEEE standards are selected as design benchmarks, and requirements are analyzed and standards are formulated;

[0015] In the process of building a smart distribution network, the design goals include stability, anti-interference, high efficiency, sustainability, intelligent automation management, timely user response, flexible adaptability and scalability;

[0016] Key issues that need to be addressed include data integration and management, system compatibility, security, accuracy and reliability of standard digital models, and compliance with policies and regulations;

[0017] Analyze needs including technical needs, user needs, operational needs, environmental and social needs;

[0018] The standards established include industry international standards, data exchange interface standards, security standards, and quality control standards.

[0019] Optionally, in step S2, data protection and security policies are formulated in the hardware layout and software architecture;

[0020] The security strategy for the hardware layout includes the need to physically protect power equipment in substations, control rooms, data storage facilities, security alarms, and surveillance cameras;

[0021] Redundant circuit design is used in the power equipment of servers and network exchanges to ensure that when some power equipment fails, the power system where the smart distribution network is located can still operate safely and stably.

[0022] Limit the physical access rights to important hardware components in the smart distribution network, such as distribution stations, substations, and control rooms, to allow only authorized personnel to access sensitive power equipment and communication interfaces;

[0023] And use a hardware security module HSM to perform hardware-level secure storage and processing of operating data and communication data in power equipment, where the HSM includes an AES encryption key, a programmable gate array GPGD and a secure digital signal processor SDSP.

[0024] The security strategy of the software architecture includes network security, data security and privacy protection, software development security and application layer security. For network security, it also includes the use of SSL / TLS and VPN technologies to encrypt data transmission in the network to ensure the security and integrity of data during transmission, and the use of firewalls and subnets to isolate the power control system components from the external network so that the power equipment is not interfered by signals from the external network;

[0025] Data security and privacy protection also includes the use of strong encryption algorithms to protect data stored on servers and backup media, as well as the use of desensitized data when analyzing and processing user information that does not require complete data;

[0026] Software development security also includes adopting secure programming standards and methods to prevent security vulnerabilities in the software, regularly checking and assessing security risks through code review and penetration testing, and updating the latest status of programs and power systems in real time and applying security patches in a timely manner;

[0027] Application layer security also includes implementing multi-factor authentication and role-based access control to ensure that only authorized personnel can access specific data and power system functions, and record audit logs of key operations in the smart distribution network for subsequent analysis and inspection.

[0028] Optionally, the logical steps of formulating data protection and security policies are as follows:

[0029] Define data classification and sensitivity: It is necessary to classify the relevant data stored, processed and transmitted in the smart distribution grid system, and determine the corresponding protection strategy based on the sensitivity of the data that affects the distribution and transmission status.

[0030] Design secure communication protocols: To ensure that communication data in the power distribution network is securely transmitted between different power equipment components, it is necessary to design communication protocols that securely encrypt, authenticate, or define data formats and standardize headers;

[0031] Implement security vulnerability management: Conduct comprehensive security vulnerability scanning and risk assessment on the smart distribution network, and take measures to patch the vulnerabilities to avoid hacker attacks and external threats;

[0032] Monitor and manage security incidents: Monitor abnormal signals in real time, respond and report promptly when security incidents occur, and be used to quickly detect the operating status of power systems and power equipment, and be able to respond to security incidents.

[0033] Optionally, step S3 specifically includes developing or integrating algorithm models of load forecasting, optimization scheduling algorithm, asset management and condition monitoring, fault diagnosis and response required for the smart distribution network based on the standard digital model, and testing the fault tolerance and robustness of the required algorithms, wherein the logical steps of load forecasting are as follows:

[0034] Collect historical load data of the smart distribution network, clean the historical load data, obtain initial load data, calibrate it as Wld, and divide the initial load data into a training set and a test set;

[0035] Linear regression is used to calculate the influence of the initial load data Wld on the load prediction, and feature extraction is performed to obtain the load feature data and calibrate it as Lcd. The load prediction calculation formula of linear regression is Lcd=β0+β1×Lcd+γ×Lcd+b0, where β0 and β1 represent the characteristic parameter values ​​of linear regression calculation, γ represents the influence parameter of the initial load data Wld on the load prediction, and b0 represents the error term of the load prediction;

[0036] The load prediction model is trained using the load characteristic data Lcd and the training set Wld, and the prediction accuracy of the load prediction model is evaluated on the test set. The training load prediction model uses the least squares method to estimate the characteristic parameter value so that the sum of the squares of the difference between the actual value and the predicted value is minimized. The calculation formula for estimating the characteristic parameter value using the least squares method is: In the formula, β represents the estimated characteristic parameter value calculated by the least squares method, It is expressed as the partial derivative of the load characteristic data Lcd with respect to the characteristic parameter values ​​β0 and β1, and then the mean square error is used to evaluate the prediction accuracy of the load prediction model. The mean square error evaluation calculation formula is: Where MSE is the mean square error evaluation value, i and N are the values ​​corresponding to each actual load value in the test set and the total number of data points in the test set, respectively. (Lcd-Wld) i Represents the difference between the load value predicted by the load prediction model and the actual load value.

[0037] Optionally, in step S3, the logical steps of optimizing the scheduling algorithm are as follows:

[0038] Input the target data of load forecast result Lcd, power generation cost, and estimated output value of renewable energy, where power generation cost is calibrated as Pgc, which is the future power generation cost predicted based on historical cost data and trends; estimated output value of renewable energy is calibrated as Eor, which is the real-time data directly obtained by using the monitoring system of renewable energy equipment.

[0039] Construct a linear programming optimization model with the lowest cost as the goal, set the objective function and constraints, and calculate the optimal scheduling result of the objective function. The objective function calculation formula is: Where W min Expressed as the output value of the minimum component of the objective function, p t It is expressed as the actual power generation at time r, It is represented by the load forecast result Lcd, the power generation cost Pgc, the estimated value of renewable energy output Eor and the actual power generation p t The integral within the time interval t, the constraint expression is P t =Lcd, and 0≤P t ≤Eor.

[0040] Optionally, in step S3, the logical steps of asset management and status monitoring are as follows:

[0041] Use smart sensors to monitor the operating data of power equipment, calibrate it as OD, and standardize the operating data OD to obtain standardized data that can be input into the machine learning model, calibrated as SD. The calculation formula for standardized data is: and Where μ and σ represent the mean and standard deviation of n running data, respectively, n represents the total number of running data od, od1, od2, …, od n It is represented as the true value corresponding to n data in the running data od;

[0042] The machine learning model is used to detect anomalies of the power equipment according to the standardized data sd and analyze the operating status. The machine learning model uses the K-means clustering algorithm to train the model, perform anomaly detection steps, and analyze the operating status. The steps of the K-means clustering training model are as follows: select the K value of the number of clusters and initialize the cluster center 0 point;

[0043] Calculate the distance between each standardized data sd sample and each cluster center, and assign the sample to the cluster with the closest distance. The Euclidean distance calculation formula is d(sd, O K )=||sd-O K ||, where d(sd, O K) is expressed as the distance between the sample point sd at the Kth cluster center and the cluster center O of the initial cluster center;

[0044] Update the cluster center of each cluster and calculate the new cluster center position using the following formula: In the formula, O′ K It is represented as the updated position of the Kth cluster center, n(sd K ) represents all sample points in the Kth cluster, |n(sd K )| represents the number of sample points in the Kth cluster;

[0045] Repeat the training steps until the position of the cluster center no longer changes for the number of iterations;

[0046] The steps of K-means clustering anomaly detection are as follows: Calculate the distance d (sd, O) from each sample point to the center of the cluster to which it belongs. K ), when the distance exceeds the distance threshold, calibrated as Th d , that is, d(sd,O K )>Th d When , the sample is considered an outlier. The anomaly prediction evaluation of the K-means clustering algorithm includes the sum of squared errors SSE, the average distance within the cluster WCSS, and the distance between clusters BCSS. The calculation formula of the sum of squared errors SSE is: And i∈K; the calculation formula of the average distance WCSS within the cluster is The calculation formula of the inter-cluster distance BCSS is:

[0047] According to the number K of clusters to which the outlier belongs and the distance d (sd, O K ) is used to determine the degree of abnormality, that is, the outlier factor, denoted as Lof, is used to quantify the degree of abnormality. The calculation formula of the outlier factor is: And i∈K, where γ represents the normalization factor, D i It is represented as the diameter of the i-th cluster;

[0048] The K-means cluster analysis operation status is to further analyze the causes of the abnormalities for the data identified as abnormal points, and propose corresponding solutions by analyzing the characteristics of the abnormal data points to perform power equipment maintenance or smart distribution network parameter adjustments.

[0049] Optionally, in step S4, the test tools and frameworks used for testing the smart distribution network at the system level and module level include automated testing tools such as Selenium and Appium, performance testing tools such as JMeter and LoadRunner, security testing tools such as OWASP ZAP and Burp Suite, unit testing frameworks such as Junit and Pytest, integrated testing tools such as Postman and SoapUI, and code coverage tools such as JaCoCo and Coveralls;

[0050] System-level testing is testing of functionality, performance, integration, security, and reliability;

[0051] Module-level testing is for unit, integration, and regression testing;

[0052] In step S4, the effectiveness and security of the system and algorithm are verified using evaluation indicators such as accuracy, recall, and F1 value.

[0053] Optionally, in step S5, the method of regularly updating the power system and operation control algorithm of the smart distribution network includes regular evaluation, version control and continuous integration deployment methods;

[0054] The periodic evaluation method adopts a time series analysis model to predict the changing trend of system performance over time, so as to timely adjust and improve the power system of the smart distribution network. For the power system security performance, it is to establish a threat model and a risk assessment model, analyze various threats and attacks that the system may face, and analyze the confidentiality, integrity and availability of the system to evaluate the security of the system. The calculation formula for the performance index and security index of the periodic evaluation system includes the average response time. And i∈n, where AR represents the average response time, n represents the total number of requests, and Ti represents the response time of the i-th request;

[0055] Throughput In the formula, Bt represents the total data transmission volume, and T0 represents the total processing time;

[0056] Error rate In the formula, e0 represents the number of requests with errors, and R0 represents the total number of requests;

[0057] Vulnerability Density In the formula, Nv represents the number of vulnerabilities and Lc represents the number of lines of code;

[0058] Security breach rate In the formula, Nv represents the number of vulnerabilities, and Tt represents the total number of tests;

[0059] The version control method adopts Git and SVN control systems to evaluate the stability and security of the system by comparing historical versions and submission records of the code base;

[0060] The continuous integration deployment method uses Jenkins or Travis Cl algorithm to automate the construction, testing and deployment processes to achieve continuous integration, and measures the status of the power system based on indicators such as the success rate of the construction process, code coverage, and automated test coverage.

[0061] Optionally, in step S6, the method for continuously optimizing the power system architecture and the smart distribution network operation algorithm includes:

[0062] According to the feedback from the operation of the smart distribution network, a PID controller is set up in the power system of the smart distribution network, and the power system architecture and the smart distribution network operation algorithm are continuously optimized by combining the deep learning predictive control model and the reinforcement learning model;

[0063] The PID controller realizes closed-loop control of the system by adjusting the proportion, integration and differentiation. The calculation formula of PID feedback control is: In(PID) represents the output value of the control parameters of the power system adjusted by the PID controller, and K p It is expressed as proportional gain, e(t) is expressed as the error at the current moment, K i Expressed as integral gain, K d It is represented as the differential gain and t is the current time.

[0064] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0065] The present invention develops or integrates algorithm models for load forecasting, optimization scheduling algorithms, asset management and status monitoring, fault diagnosis and response based on international standards and protocols as design benchmarks, and realizes the formulation of data protection and security strategies in the overall architecture of the smart distribution network by setting strategies of PID controller, deep learning predictive control model and reinforcement learning model, thereby improving the operating efficiency and reliability of the smart distribution network while reducing operation and maintenance costs and improving user satisfaction, as well as the function of real-time monitoring of the power system of the smart distribution network and abnormal inspection and maintenance plans for power equipment, effectively protecting the smart distribution network from various security threats, ensuring the safety and stability of the smart distribution network, and ensuring that the smart distribution network continues to meet the latest safety and operation requirements, and improving the adaptability and performance of the expansion of the smart distribution network and the access of new energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a flow chart of the method for constructing a smart distribution network based on a standard digital model of the present invention. DETAILED DESCRIPTION

[0067] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0068] The present invention provides Figure 1 A method for constructing a smart distribution network based on a standard digital model is shown, comprising the following steps:

[0069] S1. Select international standards and protocols such as IEC61850 or IEEE as the design benchmark, review the design goals of the smart distribution network and the key issues that need to be solved, analyze the needs and formulate standards.

[0070] Specifically, in step S1, during the construction of the smart distribution network, the design goal is to determine the development direction of the smart distribution network and the functions that are ultimately expected to be achieved. The performance requirements of the IEC61850 standard are used to first define the macro goals of the smart distribution network, and then formulate specific KPIs. The design goals for the smart distribution network include stability, anti-interference, high efficiency, sustainability, intelligent automation management, user response timeliness, flexible adaptability and scalability, which are calibrated as St0, Im0, He0, Ed0, Ia0, Cr0, Fa0, and Sc0, respectively. Clear design goals help guide the planning and implementation of the entire project, ensure that all activities have clear directions and measurement standards, and ensure that the project proceeds as expected.

[0071] Among them, stability St0 is to ensure that the distribution network maintains normal continuous operation during the power supply process and avoid abnormal power outages caused by faults;

[0072] Anti-interference Im0 is the ability of the smart distribution network to cope with external interference such as weather changes and human damage, ensuring the safety of the power supply process;

[0073] High efficiency He0 means that the smart distribution network can maximize energy utilization and reduce energy consumption while ensuring stability and safety;

[0074] Sustainable Ed0 means that the smart distribution network supports the access of renewable energy and optimizes the use of distributed energy resources while meeting existing needs without harming the environment and natural resources;

[0075] Intelligent automation management Ia0 is a management method that realizes remote monitoring and automatic maintenance of intelligent distribution networks through advanced monitoring, prediction, and decision-making support grid management and control systems;

[0076] User response timeliness Cr0 is the rapid response capability and service quality of the smart distribution network to user demand response management and user-side management requirements;

[0077] Flexible adaptability Fa0 refers to the ability of the smart distribution network to quickly adjust to different environments and demands; scalability Sc0 refers to the ability of the smart distribution network to upgrade and expand the system based on existing power equipment and network technology to adapt to future technological development and market changes.

[0078] During the current situation analysis, problems in the existing system are discovered and classified. They are then prioritized according to severity and impact to identify key issues for demand analysis and risk management. The key issues that need to be addressed include: data integration and management, system compatibility, security, accuracy and reliability of standard digital models, and compliance with policies and regulations, which are denoted as Dim0, Syc0, Saf0, Arm0, and Cpr0 respectively.

[0079] Among them, data integration and management Dim0 specifies the communication protocol between intelligent terminal devices and master control devices, effectively integrates and manages real-time data and historical data from different sources, and realizes data management and information exchange;

[0080] System compatibility Syc0 sets standardized interfaces to ensure compatibility between different power equipment and different standards;

[0081] Security Saf0 is the data security, network security and physical security of the smart distribution network, preventing external attacks and internal loopholes and misoperation;

[0082] Accuracy and reliability of standard digital models Arm0 is a standard digital model for designing intelligent terminal devices. It can predict and optimize power systems and provide the effectiveness of control strategies, ensuring the accuracy of interoperability between power devices.

[0083] Comply with policies and regulations Cpr0 means that the construction of smart distribution networks must comply with local policies, industry standards, laws and regulations, and follow green environmental protection requirements for energy conservation, emission reduction, and waste management.

[0084] User needs are collected through questionnaires, interviews, and on-site measurements. Visual modeling and analysis of needs are performed in combination with data collected by sensors in the existing smart distribution network system. Software engineers use the UML language to write a requirement specification, recording the description, priority, and acceptance criteria of the requirements in detail to meet user needs and provide a scientific basis for the formulation of design goals and standards. The analyzed needs include: technical needs, user needs, operational needs, and environmental and social needs, which are calibrated as Tr0, Ur0, Or0, and Es0 respectively.

[0085] Among them, technical requirements Tr0 is to determine the technical solutions for communication protocols, sensor networks, and data analysis and processing required for smart distribution networks based on the evaluation of the maturity and feasibility of existing technologies, as well as the technical route for determining system performance indicators, supported equipment types, and software platform requirements;

[0086] User demand Ur0 is to understand the user's needs and expectations for power safety, power supply reliability, energy efficiency, and customized services of the smart distribution network;

[0087] Operational requirements Or0 is to clarify the operation mode and management process of the smart distribution network, including the convenience, cost control, and personnel training of the operation and maintenance process of data collection, transmission, processing, analysis, and application;

[0088] Environmental and social requirements Es0 is the combination of the impact and acceptability of the smart distribution network on the surrounding environment and society.

[0089] Refer to the selected IEC61850 international standard to formulate technical standards and operating specifications and customize standard manuals and technical guidelines to ensure the consistency of the smart distribution network, reduce implementation risks, and improve the efficiency of subsequent integration and maintenance work. The formulated standards include industry international standards, data exchange interface standards, safety standards, and quality control standards, which are marked as Is0, Ds0, Ss0, and Qs0 respectively.

[0090] Among them, the industry international standard Is0 is an international general standard and specification for the field of smart distribution networks, covering the basic concepts, design principles, test methods, and equipment performance requirements of smart distribution networks;

[0091] The data exchange interface standard Ds0 defines the data format and protocol interface for external interaction between the power system and different components of the smart distribution network;

[0092] Security standard Ss0 defines the requirements and guidelines for data security, network security, and physical security of smart distribution networks. It is used to protect smart distribution networks from external and internal threats and ensure the reliability and stability of the system.

[0093] The quality control standard Qs0 defines the power system design quality, implementation quality and maintenance quality of the smart distribution network.

[0094] S2. Build the overall architecture of the smart distribution network, and formulate data protection and security strategies in the hardware layout and software architecture to effectively protect the smart distribution network from various security threats, thereby supporting reliable and efficient operation.

[0095] Specifically, in step S2, the security strategy for the hardware layout includes the need to physically protect the power equipment in the substation, control room, data storage facility, security alarm, and surveillance camera;

[0096] Redundant circuit design is used in the power equipment of servers and network exchanges to ensure that when some power equipment fails, the power system where the smart distribution network is located can still operate safely and stably.

[0097] Limit the physical access rights to important hardware components in the smart distribution network, such as distribution stations, substations, and control rooms, to allow only authorized personnel to access sensitive power equipment and communication interfaces;

[0098] And use a hardware security module HSM to perform hardware-level secure storage and processing of operating data and communication data in power equipment, where the HSM includes an AES encryption key, a programmable gate array GPGD and a secure digital signal processor SDSP.

[0099] The security strategy of the software architecture includes network security, data security and privacy protection, software development security and application layer security. For network security, it includes the use of SSL / TLS and VPN technologies to encrypt data transmission in the network, ensure the security and integrity of data during transmission, and use firewalls and subnets to isolate the power control system components from the external network, so that the power equipment is not interfered by the signals of the external network;

[0100] Data security and privacy protection also includes the use of strong encryption algorithms to protect data stored on servers and backup media, as well as the use of desensitized data when analyzing and processing user information that does not require complete data;

[0101] Software development security also includes adopting secure programming standards and methods to prevent security vulnerabilities in the software, regularly checking and assessing security risks through code review and penetration testing, and updating the latest status of programs and power systems in real time and applying security patches in a timely manner;

[0102] Application layer security also includes implementing multi-factor authentication and role-based access control to ensure that only authorized personnel can access specific data and power system functions, and record audit logs of key operations in the smart distribution network for subsequent analysis and inspection.

[0103] Specifically, the logical steps to developing a data protection and security strategy are as follows:

[0104] Define data classification and sensitivity: It is necessary to classify the relevant data stored, processed and transmitted in the smart distribution network system, and determine the corresponding protection strategy based on the sensitivity of the data that affects the distribution and transmission status. The relevant data includes user personal information, power equipment operating status, power distribution data and power system management and control strategies. The protection strategy includes data encryption, user identity authentication, access control, logging and auditing;

[0105] Design secure communication protocols: To ensure that communication data in the power distribution network is securely transmitted between different power equipment components, it is necessary to design communication protocols that securely encrypt, authenticate, or define data formats and standardize headers;

[0106] Implement security vulnerability management: Conduct comprehensive security vulnerability scanning and risk assessment on the smart distribution network, and take measures to patch the vulnerabilities to avoid hacker attacks and external threats;

[0107] Monitor and manage security incidents: Monitor abnormal signals in real time, respond and report promptly when security incidents occur, and be used to quickly detect the operating status of power systems and power equipment, and be able to respond to security incidents.

[0108] S3. Based on the standard digital model, develop or integrate the algorithm models for load forecasting, optimization scheduling algorithms, asset management and condition monitoring, fault diagnosis and response required for the smart distribution network, and test the fault tolerance and robustness of the required algorithms to ensure that the algorithm models can maintain stable and reliable operation when faced with inaccurate data, equipment failures or network attacks, thereby improving the operating efficiency and reliability of the smart distribution network, while reducing operation and maintenance costs and improving user satisfaction. In other words, by improving and applying the standard digital model, effectively solve the problems of integration, compatibility and security of the existing smart distribution network.

[0109] Specifically, in step S3, the logical steps of load prediction are as follows:

[0110] Load forecasting predicts the short-term, medium-term or long-term load of the power grid in the future by analyzing the influencing factors of historical load data. Based on the load forecast results, the configuration of power generation resources and the dispatch of electricity are optimized to improve energy efficiency and reduce operating costs.

[0111] Collect historical load data of the smart distribution network, clean the historical load data, obtain initial load data, calibrate it as Wld, and divide the initial load data into a training set and a test set;

[0112] Linear regression is used to calculate the influence of the initial load data Wld on the load prediction, and feature extraction is performed to obtain load feature data and calibrate it as Lcd. The load prediction calculation formula of linear regression is:

[0113] Lcd=β0+β1×Lcd+γ×Lcd+b0 (1)

[0114] Where:

[0115] β0 and β1 represent the characteristic parameter values ​​calculated by linear regression,

[0116] γ represents the influence parameter of the initial load data Wld on the load prediction.

[0117] b0 represents the error term of load prediction;

[0118] The load prediction model is trained using the load characteristic data Lcd and the training set Wld, and the prediction accuracy of the load prediction model is evaluated on the test set. The training load prediction model uses the least squares method to estimate the characteristic parameter value so that the sum of the squares of the difference between the actual value and the predicted value is minimized. The calculation formula for estimating the characteristic parameter value using the least squares method is:

[0119]

[0120] Where:

[0121] β represents the estimated characteristic parameter value calculated by the least squares method,

[0122] It is expressed as the partial derivative of the load characteristic data Lcd with respect to the characteristic parameter values ​​β0 and β1,

[0123] The mean square error is then used to evaluate the prediction accuracy of the load prediction model, where the mean square error evaluation calculation formula is:

[0124]

[0125] Where:

[0126] MSE is the mean square error.

[0127] i and N represent the value corresponding to each actual load value in the test set and the total number of data points in the test set, respectively.

[0128] (Lcd-Wld) i Represents the difference between the load value predicted by the load prediction model and the actual load value.

[0129] Specifically, in step S3, the logical steps of the optimization scheduling algorithm are as follows:

[0130] The optimization scheduling algorithm ensures the balance of power supply and demand and reduces power loss by dynamically adjusting power generation and distribution strategies. It optimizes operation strategies to reduce power costs according to power market price fluctuations, and can efficiently integrate renewable energy such as wind and solar energy to improve the utilization rate of renewable energy.

[0131] Input the target data of load forecast result Lcd, power generation cost, and estimated output value of renewable energy, where power generation cost is calibrated as Pgc, which is the future power generation cost predicted based on historical cost data and trends; estimated output value of renewable energy is calibrated as Eor, which is the estimated output value of renewable energy by collecting weather data through equipment such as meteorological stations, anemometers, and solar radiometers, so as to estimate the output of wind and solar power generation, or directly use the monitoring system of renewable energy equipment to obtain real-time data;

[0132] Construct a linear programming optimization model with the lowest cost as the goal, set the objective function and constraints, and calculate the optimization scheduling result of the objective function. The objective function calculation formula is:

[0133]

[0134] Where:

[0135] W min Represented as the output value of the minimum component of the objective function,

[0136] p t It is expressed as the actual power generation at time t,

[0137] It is represented by the load forecast result Lcd, the power generation cost Pgc, the estimated value of renewable energy output Eor and the actual power generation p t The integral over the time interval t is

[0138] The constraint expression is:

[0139] P t =Lcd, and 0≤P t ≤Eor (5)

[0140] Specifically, in step S3, the logical steps of asset management and status monitoring are as follows:

[0141] Asset management and condition monitoring monitor the operating status of power grid equipment through sensors and intelligent devices, detect abnormal conditions of equipment in a timely manner, and predict equipment failures based on statistical and machine learning models.

[0142] The operation data of the power equipment is monitored by intelligent sensors and calibrated as OD. The operation data OD is standardized to obtain standardized data that can be input into the machine learning model and calibrated as SD. The calculation formula of the standardized data is:

[0143]

[0144] Where:

[0145] μ and σ represent the mean and standard deviation of n running data, respectively.

[0146] n represents the total number of running data od,

[0147] od1, od2, …, od n It is represented as the true value corresponding to n data in the running data od;

[0148] The machine learning model is used to detect anomalies of the power equipment according to the standardized data sd and analyze the operating status. The machine learning model uses the K-means clustering algorithm to train the model, perform anomaly detection steps, and analyze the operating status. The steps of the K-means clustering training model are as follows: select the K value of the number of clusters and initialize the cluster center O point of the cluster center;

[0149] Calculate the distance between each standardized data sd sample and each cluster center, and assign the sample to the cluster with the closest distance. The Euclidean distance calculation formula is:

[0150] d(sd,O K )=||sd-O K || (9)

[0151] Where: d(sd, O K ) is expressed as the distance between the sample point sd at the Kth cluster center and the cluster center O of the initial cluster center;

[0152] Update the cluster center of each cluster and calculate the new cluster center position using the following formula:

[0153]

[0154] Where:

[0155] O′ K Represents the updated position of the Kth cluster center,

[0156] n(sd K ) represents all sample points in the Kth cluster,

[0157] |n(sd K )| represents the number of sample points in the Kth cluster;

[0158] Repeat the training steps until the position of the cluster center no longer changes for the number of iterations;

[0159] The steps of K-means clustering anomaly detection are as follows: Calculate the distance d (sd, O) from each sample point to the center of the cluster to which it belongs. K ), when the distance exceeds the distance threshold, calibrated as Th d , that is, d(sd,O K )>Th d When , the sample is considered an outlier. The anomaly prediction evaluation of the K-means clustering algorithm includes the sum of squared errors SSE, the average distance within the cluster WCSS, and the distance between clusters BCSS. The calculation formula of the sum of squared errors SSE is:

[0160]

[0161] The calculation formula of the average distance WCSS within a cluster is:

[0162]

[0163] The calculation formula of cluster distance BCSS is:

[0164]

[0165] According to the number K and distance d (sd, Q K ) is used to determine the degree of abnormality, that is, the outlier factor, calibrated as Lof, is used to quantify the degree of abnormality. The calculation formula of the outlier factor is:

[0166]

[0167] Where:

[0168] γ is the normalization factor, D i It is expressed as the diameter of the i-th cluster, that is, the larger the outlier factor is, the more abnormal the sample point is;

[0169] The K-means cluster analysis operation status is to further analyze the causes of the abnormalities for the data identified as abnormal points, and propose corresponding solutions by analyzing the characteristics of the abnormal data points to perform power equipment maintenance or smart distribution network parameter adjustments.

[0170] Specifically, in step S3, the logical steps of fault diagnosis and response are as follows:

[0171] Fault diagnosis and response is to use standard digital models and algorithms to quickly locate the fault point and diagnose the cause of the fault when a fault occurs in the power grid. After diagnosing the fault, the fault isolation operation is automatically performed to restore power supply to non-faulty areas and reduce the impact of the fault.

[0172] Abnormal signals are detected in the power system where the smart distribution network is located, and the abnormal prediction results of the power equipment are combined with the machine learning model as the initial diagnostic data;

[0173] A machine learning model is established, and the initial diagnostic data is brought in to use the machine learning algorithm to determine the fault type and location, and corresponding automatic response measures are taken according to the results of the fault location. Among them, the machine learning algorithm fault location is to take each fault point identified as an abnormality by K-means clustering as a sample point, and then perform anomaly detection again through the K-means clustering algorithm to determine the fault type and location based on the distribution of the abnormal points and the outlier factor.

[0174] S4. Use testing tools and frameworks to conduct comprehensive and effective testing of the smart distribution network at the system and module levels, and use the accuracy, recall, and F1 value evaluation indicators to verify the effectiveness and security of the system and algorithm to complete the development and implementation of the smart distribution network system;

[0175] Specifically, in step S4, the testing tools and frameworks include automated testing tools such as Selenium and Appium, performance testing tools such as JMeter and LoadRunner, security testing tools such as OWASP ZAP and Burp Suite, unit testing frameworks such as Junit and Pytest, integration testing tools such as Postman and SoapUI, and code coverage tools such as JaCoCo and Coveralls;

[0176] System-level testing is a test of function, performance, integration, security and reliability. Functional testing verifies whether the system operates normally according to the design specifications; performance testing verifies the performance of the system under various loads and conditions; integration testing verifies whether the interfaces and interactions between different modules are normal; security testing verifies the system's ability to prevent potential threats and attacks; reliability testing verifies the stability and reliability of the system under long-term operation.

[0177] Module-level testing is unit, integration and regression testing. Unit testing is to independently test each module in the system to verify whether its function is correct; integration testing is to test whether the communication and data transmission between different modules are normal; regression testing is to test the impact of new functions or bug fixes on other functions of the system.

[0178] S5. Regularly update the power system and operation control algorithms of the smart distribution network by using regular evaluation, version control and continuous integration deployment methods to monitor the power system of the smart distribution network and abnormal maintenance and maintenance plans of power equipment in real time to respond to new challenges and threats;

[0179] Specifically, in step S5, the periodic evaluation method uses a time series analysis model to predict the changing trend of system performance over time, so as to timely adjust and improve the power system of the smart distribution network. For the safety performance of the power system, it is to establish a threat model and a risk assessment model, analyze the various threats and attacks that the system may face, and analyze the confidentiality, integrity and availability of the system to evaluate the security of the system. The calculation formulas for the periodic evaluation of system performance indicators and security indicators include average response time, throughput, error rate, vulnerability density, and security vulnerability rate.

[0180] The average response time is expressed as follows:

[0181]

[0182] Where:

[0183] AR is the average response time.

[0184] n represents the total number of requests,

[0185] T i Represented as the response time of the i-th request;

[0186] The throughput is expressed as follows:

[0187]

[0188] Where:

[0189] Bt represents the total data transmission volume,

[0190] T0 represents the total processing time;

[0191] The error rate is expressed as follows:

[0192]

[0193] Where:

[0194] e0 represents the number of requests with errors.

[0195] R0 represents the total number of requests;

[0196] The vulnerability density is expressed as follows:

[0197]

[0198] Where:

[0199] Nv represents the number of vulnerabilities,

[0200] Lc is expressed as the number of lines of code;

[0201] The security vulnerability rate is expressed as follows:

[0202]

[0203] Where:

[0204] Nv represents the number of vulnerabilities, and Tt represents the total number of tests.

[0205] The version control method uses Git and SVN control systems to evaluate the stability and security of the system by comparing historical versions and submission records of the code base;

[0206] The continuous integration and deployment method uses Jenkins or Travis Cl algorithm to automate the build, test and deployment processes to achieve continuous integration, and measures the status of the power system based on indicators such as the success rate of the build process, code coverage, and automated test coverage.

[0207] S6. According to the feedback from the operation of the smart distribution network, a PID controller is set in the power system of the smart distribution network. Combined with the deep learning predictive control model and the reinforcement learning model, the power system architecture and the smart distribution network operation algorithm are continuously optimized to ensure that the power system continues to meet the latest safety and operational requirements.

[0208] Specifically, in step S6, the PID controller in the power system of the smart distribution network can control voltage regulation, frequency control, reactive power control and load balancing, wherein voltage regulation is to adjust the tap position of the transformer to stabilize the voltage of the distribution network, frequency control is to adjust the output frequency of the generator to maintain the stability of the system frequency, reactive power control is to control reactive compensation equipment, such as capacitor banks or synchronous phase regulators, to maintain the voltage level, load balancing is to adjust the load distribution to ensure the load balance of each feeder and substation, and the PID controller realizes closed-loop control of the system by adjusting the three parts of proportion, integration and differentiation, and the calculation formula of PID feedback control is:

[0209]

[0210] Where:

[0211] In(PID) represents the output value of the control parameters of the power system adjusted by the PID controller.

[0212] K p Expressed as proportional gain, it is used to adjust the influence of the proportional part.

[0213] e(t) is the error at the current moment, that is, the difference between the expected value and the actual value.

[0214] K iExpressed as integral gain, it is used to adjust the influence of the integral part.

[0215] K d It is expressed as the differential gain, which is used to adjust the influence of the differential part, and t represents the current moment.

[0216] It is further explained that the deep learning predictive control model uses historical data and real-time data for prediction and control, and generates a control strategy based on the prediction results of the deep learning predictive control model, and the reinforcement learning model learns the optimal control strategy through interaction with the environment, which are all provided to the PID controller for execution. Among them, the input of the deep learning predictive control model is the historical data of the smart distribution network and the voltage, frequency, and load data collected by the real-time sensors. The output of the deep learning predictive control model is the prediction result of the future state of the smart distribution network system. The input of the PID controller is the output of the deep learning predictive control model, and the output of the PID controller is the control signal of the smart distribution network system. The input of the reinforcement learning model is the current state and historical actions of the smart distribution network, and the output of the reinforcement learning model is the execution instruction of the optimal control action. The PID controller in the smart distribution network system is combined with the deep learning predictive control model and the reinforcement learning model to autonomously combine and optimize through system machine learning to improve the stability, response speed and optimization capability of the smart distribution network system.

[0217] The specific method and process of the method for constructing a smart distribution network based on a standard digital model provided in an embodiment of the present invention are detailed in the embodiment of the method for constructing a smart distribution network based on a standard digital model, and will not be repeated here.

[0218] A system for constructing a smart distribution network based on a standard digital model, running a method for constructing a smart distribution network based on a standard digital model as described in Example 1, includes the following modules:

[0219] Model building module, load forecasting and optimization module, grid status monitoring module, fault diagnosis and response module, testing module, risk assessment module and system operation optimization module.

[0220] Specifically, the model building module is used to build the overall architecture of the smart distribution network, including hardware layout and software architecture;

[0221] The load prediction and optimization module uses the load prediction model to obtain load prediction results, and optimizes the power generation resource allocation and power dispatch through the optimization scheduling algorithm;

[0222] The power grid status monitoring module monitors the operating status of power grid equipment through sensors and intelligent devices, detects abnormal conditions of equipment in a timely manner, and predicts equipment failures based on statistical and machine learning models;

[0223] Fault diagnosis and response module, which uses standard digital models and algorithms to quickly locate the fault point, diagnose the cause of the fault and automatically perform fault isolation operations when a fault occurs in the power grid;

[0224] Testing modules, including automated testing tools, performance testing tools, security testing tools, integration testing tools, code coverage tools, and unit testing frameworks, are used to conduct comprehensive testing of smart distribution networks at the system and module levels;

[0225] The risk assessment module uses a time series analysis model to predict the changing trend of system performance over time, and analyzes and evaluates the security of the system by establishing a threat model and a risk assessment model;

[0226] The system operation optimization module sets up a PID controller and combines deep learning predictive control models and reinforcement learning models to continuously optimize the power system architecture and smart distribution network operation algorithms.

[0227] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for constructing a smart distribution network based on a standard digital model, characterized in that: The steps include: S1. Analyze requirements and develop standards based on the design objectives and key issues of smart distribution networks; S2. Build the overall architecture of the smart distribution network and formulate data protection and security strategies; S3. Based on standard digital models, develop or integrate algorithm models for load forecasting, optimization scheduling algorithms, asset management and condition monitoring, fault diagnosis and response required for smart distribution networks; S4. Test the smart distribution network and verify the effectiveness and security of the system and algorithm using the evaluation indicators of accuracy, recall rate and F1 value; S5. Regularly update the power system and operation control algorithm of the smart distribution network; S6. Continuously optimize the power system architecture and smart distribution network operation algorithms.

2. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S1, when reviewing the design objectives and key issues to be resolved for the smart distribution network, international standards and protocols such as IEC61850 or IEEE standards are selected as design benchmarks, and requirements are analyzed and standards are formulated; In the process of building a smart distribution network, the design goals include stability, anti-interference, high efficiency, sustainability, intelligent automation management, timely user response, flexible adaptability and scalability; Key issues that need to be addressed include data integration and management, system compatibility, security, accuracy and reliability of standard digital models, and compliance with policies and regulations; Analyze needs including technical needs, user needs, operational needs, environmental and social needs; The standards established include industry international standards, data exchange interface standards, security standards, and quality control standards.

3. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S2, data protection and security strategies are formulated in the hardware layout and software architecture; The security strategy for the hardware layout includes the need to physically protect the power equipment in the substation, control room, data storage facilities, security alarms and surveillance cameras; Redundant circuit design is used in the power equipment of servers and network exchanges to ensure that when some power equipment fails, the power system where the smart distribution network is located can still operate safely and stably. Limit the physical access rights to important hardware components in the smart distribution network, such as distribution stations, substations, and control rooms, to allow only authorized personnel to access sensitive power equipment and communication interfaces; and using a hardware security module HSM to securely store and process operating data and communication data in power equipment at the hardware level, wherein the HSM includes an AES encryption key, a programmable gate array GPGD, and a secure digital signal processor SDSP; The security strategy of the software architecture includes network security, data security and privacy protection, software development security and application layer security. For network security, it also includes the use of SSL / TLS and VPN technologies to encrypt data transmission in the network to ensure the security and integrity of data during transmission, and the use of firewalls and subnets to isolate the power control system components from the external network so that the power equipment is not interfered by signals from the external network; Data security and privacy protection also includes the use of strong encryption algorithms to protect data stored on servers and backup media, as well as the use of desensitized data when analyzing and processing user information that does not require complete data; Software development security also includes adopting secure programming standards and methods to prevent security vulnerabilities in the software, regularly checking and assessing security risks through code review and penetration testing, and updating the latest status of programs and power systems in real time and applying security patches in a timely manner; Application layer security also includes implementing multi-factor authentication and role-based access control to ensure that only authorized personnel can access specific data and power system functions, and record audit logs of key operations in the smart distribution network for subsequent analysis and inspection.

4. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S2, the logical steps for formulating data protection and security policies are as follows: Define data classification and sensitivity: It is necessary to classify the relevant data stored, processed and transmitted in the smart distribution grid system, and determine the corresponding protection strategy based on the sensitivity of the data that affects the distribution and transmission status; Design secure communication protocols: To ensure that communication data in the power distribution network is securely transmitted between different power equipment components, it is necessary to design communication protocols that securely encrypt, authenticate, or define data formats and standardize headers; Implement security vulnerability management: Conduct comprehensive security vulnerability scanning and risk assessment on the smart distribution network, and take measures to patch the vulnerabilities to avoid hacker attacks and external threats; Monitor and manage security incidents: monitor abnormal signals in real time, respond and report in a timely manner when security incidents occur, detect the operating status of power systems and power equipment, and respond to security incidents.

5. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: The step S3 specifically includes developing or integrating the algorithm models of load prediction, optimization scheduling algorithm, asset management and condition monitoring, fault diagnosis and response required for the smart distribution network based on the standard digital model, and testing the fault tolerance and robustness of the required algorithms. The logical steps of load prediction are as follows: Collect historical load data of the smart distribution network, clean the historical load data, obtain initial load data, calibrate it as Wld, and divide the initial load data into a training set and a test set; Linear regression is used to calculate the influence of the initial load data Wld on the load prediction, and feature extraction is performed to obtain the load feature data and calibrate it as Lcd. The load prediction calculation formula of linear regression is Lcd=β0+β1×Lcd+γ×Lcd+b0, where β0 and β1 represent the characteristic parameter values ​​of linear regression calculation, γ represents the influence parameter of the initial load data Wld on the load prediction, and b0 represents the error term of the load prediction; The load prediction model is trained using the load characteristic data Lcd and the training set Wld, and the prediction accuracy of the load prediction model is evaluated on the test set. The training load prediction model uses the least squares method to estimate the characteristic parameter value so that the sum of the squares of the difference between the actual value and the predicted value is minimized. The calculation formula for estimating the characteristic parameter value using the least squares method is: In the formula, β represents the estimated characteristic parameter value calculated by the least squares method, It is expressed as the partial derivative of the load characteristic data Lcd with respect to the characteristic parameter values ​​β0 and β1, and then the mean square error is used to evaluate the prediction accuracy of the load prediction model. The mean square error evaluation calculation formula is: Where MSE is the mean square error evaluation value, i and N are the values ​​corresponding to each actual load value in the test set and the total number of data points in the test set, respectively. (Lcd-Wld) i Represents the difference between the load value predicted by the load prediction model and the actual load value.

6. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S3, the logical steps of the optimization scheduling algorithm are as follows: Input the target data of load forecast result Lcd, power generation cost, and estimated output value of renewable energy, where power generation cost is calibrated as Pgc, which is the future power generation cost predicted based on historical cost data and trends; estimated output value of renewable energy is calibrated as Eor, which is the real-time data directly obtained by using the monitoring system of renewable energy equipment. Construct a linear programming optimization model with the lowest cost as the goal, set the objective function and constraints, and calculate the optimal scheduling result of the objective function. The objective function calculation formula is: Where W min Expressed as the output value of the minimum component of the objective function, p t It is expressed as the actual power generation at time r, It is represented by the load forecast result Lcd, the power generation cost Pgc, the estimated value of renewable energy output Eor and the actual power generation p t The integral within the time interval t, the constraint expression is P t =Lcd, and 0≤P t ≤Eor.

7. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S3, the logical steps of asset management and status monitoring are as follows: Use smart sensors to monitor the operating data of power equipment, calibrate it as OD, and standardize the operating data OD to obtain standardized data that can be input into the machine learning model, calibrated as SD. The calculation formula for standardized data is: and Where μ and σ represent the mean and standard deviation of n running data respectively, n represents the total number of running data od, od1, od2, ..., od n It is represented as the true value corresponding to n data in the running data od; The machine learning model is used to detect anomalies of the power equipment according to the standardized data sd and analyze the operating status. The machine learning model uses the K-means clustering algorithm to train the model, perform anomaly detection steps, and analyze the operating status. The steps of the K-means clustering training model are as follows: select the K value of the number of clusters and initialize the cluster center O point of the cluster center; Calculate the distance between each standardized data sd sample and each cluster center, and assign the sample to the cluster with the closest distance. The Euclidean distance calculation formula is d(sd, O K )=||sd-O K ||, where d(sd, O K ) is expressed as the distance between the sample point sd at the Kth cluster center and the cluster center O of the initial cluster center; Update the cluster center of each cluster and calculate the new cluster center position using the following formula: In the formula, O′ K It is represented as the updated position of the Kth cluster center, n(sd K ) represents all sample points in the Kth cluster, |n(sd K )| represents the number of sample points in the Kth cluster; Repeat the training steps until the position of the cluster center no longer changes for the number of iterations; The steps of K-means clustering anomaly detection are as follows: Calculate the distance d (sd, O) from each sample point to the center of the cluster to which it belongs. K ), when the distance exceeds the distance threshold, calibrated as Th d , that is, d(sd,O K )>Th d When , the sample is considered an outlier. The anomaly prediction evaluation of the K-means clustering algorithm includes the sum of squared errors SSE, the average distance within the cluster WCSS, and the distance between clusters BCSS. The calculation formula of the sum of squared errors SSE is: And i∈K; the calculation formula of the average distance WCSS within the cluster is The calculation formula of the inter-cluster distance BCSS is: According to the number K of clusters to which the outlier belongs and the distance d (sd, O K ) is used to determine the degree of abnormality, that is, the outlier factor, denoted as Lof, is used to quantify the degree of abnormality. The calculation formula of the outlier factor is: And i∈K, where γ represents the normalization factor, D i It is represented as the diameter of the i-th cluster; The K-means clustering analysis operation status is to further analyze the causes of the abnormalities for the data identified as abnormal points, and propose corresponding solutions by analyzing the characteristics of the abnormal data points to perform power equipment maintenance or smart distribution network parameter adjustments.

8. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S4, the test tools and frameworks used for testing the smart distribution network at the system level and module level include automated testing tools such as Selenium and Appium, performance testing tools such as JMeter and LoadRunner, security testing tools such as OWASPZAP and Burp Suite, unit testing frameworks such as Junit and Pytest, integrated testing tools such as Postman and SoapUI, and code coverage tools such as JaCoCo and Coveralls; System-level testing is testing of functionality, performance, integration, security, and reliability; Module-level testing is for unit, integration, and regression testing; In step S4, the effectiveness and security of the system and algorithm are verified using evaluation indicators such as accuracy, recall, and F1 value.

9. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S5, the method of regularly updating the power system and operation control algorithm of the smart distribution network includes regular evaluation, version control and continuous integration deployment methods; The periodic evaluation method adopts a time series analysis model to predict the changing trend of system performance over time, so as to timely adjust and improve the power system of the smart distribution network. For the power system security performance, it is to establish a threat model and a risk assessment model, analyze various threats and attacks that the system may face, and analyze the confidentiality, integrity and availability of the system to evaluate the security of the system. The calculation formula for the performance index and security index of the periodic evaluation system includes the average response time. And i∈n, where AR represents the average response time, n represents the total number of requests, T i Represented as the response time of the i-th request; Throughput In the formula, Bt represents the total data transmission volume, and T0 represents the total processing time; Error rate In the formula, e0 represents the number of requests with errors, and R0 represents the total number of requests; Vulnerability Density In the formula, Nv represents the number of vulnerabilities and Lc represents the number of lines of code; Security breach rate In the formula, Nv represents the number of vulnerabilities, and Tt represents the total number of tests; The version control method adopts Git and SVN control systems to evaluate the stability and security of the system by comparing historical versions and submission records of the code base; The continuous integration deployment method uses Jenkins or Travis CI algorithm to automate the construction, testing and deployment processes to achieve continuous integration, and measures the status of the power system based on indicators such as the success rate of the construction process, code coverage, and automated test coverage.

10. The method for constructing a smart distribution network based on a standard digital model according to claim 1, characterized in that: In step S6, the method for continuously optimizing the power system architecture and the smart distribution network operation algorithm includes: According to the feedback from the operation of the smart distribution network, a PID controller is set up in the power system of the smart distribution network, and the power system architecture and the smart distribution network operation algorithm are continuously optimized by combining the deep learning predictive control model and the reinforcement learning model; The PID controller realizes closed-loop control of the system by adjusting the proportion, integration and differentiation. The calculation formula of PID feedback control is: In(PID) represents the output value of the control parameters of the power system adjusted by the PID controller, and K p It is expressed as proportional gain, e(t) is expressed as the error at the current moment, K i Expressed as integral gain, K d It is represented as the differential gain and t is the current time.

Images