An internet-of-things access, data storage and analysis platform for multi-protocol intelligent circuit breakers

Abstract

The present application relates to the technical field of power internet of things, and especially to a kind of internet of things access, data storage and analysis platform for multi-protocol intelligent circuit breaker, comprising multi-protocol adaptive security access module, data fusion and quality evaluation module, hybrid hierarchical storage module, intelligent analysis and diagnosis module and open service and visualization module. Through graphical configuration and dynamic protocol analysis, bidirectional authentication and secure access of heterogeneous circuit breakers are realized. Improved weighted evidence theory fusion and sliding window median filtering are used to improve data quality. Memory / chronological / hybrid relational hierarchical storage is used to optimize data management. Based on machine learning, a device behavior baseline model is constructed to realize fault prediction. The present application solves the problems of protocol fragmentation, weak security, low data quality and insufficient analysis depth, and realizes the unified access, data fusion and predictive maintenance of multi-protocol circuit breakers from multiple manufacturers.

Description

Technical Field

[0001] This invention relates to the field of power Internet of Things (IoT) technology, specifically to an IoT access, data storage, and analysis platform for multi-protocol smart circuit breakers. Background Technology

[0002] With the rapid development of smart grid and Internet of Things (IoT) technologies, smart circuit breakers, as core terminal equipment in power distribution systems, are evolving from traditional protective switches into comprehensive intelligent units with measurement, control, communication, and diagnostic functions. Currently, IoT management platforms for smart circuit breakers mainly adopt single-protocol interface mode, simple relational database storage mode, threshold alarm analysis mode, and independent management platform mode.

[0003] However, the aforementioned existing technologies have the following prominent technical defects in practical engineering applications: First, the communication protocols of smart circuit breakers on the market are complex, covering multiple protocols such as RS485 / ModbusRTU, IEC61850, DL / T698, MQTT, and CoAP. Existing platforms are mostly developed for single-protocol integration. When new brands or circuit breakers with new protocols need to be integrated, the core code of the platform must be modified and redeployed, resulting in poor scalability. Second, most existing systems lack strong authentication mechanisms on the device side, relying solely on simple device IDs or IP addresses for identification. This makes them vulnerable to protocol replay attacks, man-in-the-middle attacks, or forged device access, seriously threatening the safety of power grid operation. Third, in complex power grid environments, the collected raw messages contain a large amount of noisy data and outliers, and the same circuit breaker may communicate through multiple protocols. The parallel reporting of data creates redundancy. Existing platforms lack spatiotemporal alignment, quality assessment, and online real-time cleaning of data from different sources, resulting in low effective information density and a high risk of false alarms. Fourth, existing solutions use simple relational databases to store all data, which is insufficient to handle the writing pressure and query efficiency requirements of high-frequency time-series data, and lacks a hierarchical storage strategy, leading to high storage costs. Fifth, existing systems mostly rely on preset fixed thresholds for simple over-limit alarms, lacking in-depth analysis algorithms oriented towards equipment characteristics, failing to build a baseline model of normal equipment behavior, and failing to achieve the transformation from "post-fault maintenance" to "pre-fault early warning." Sixth, in scenarios such as tower base stations and industrial and mining enterprises, circuit breakers from different manufacturers correspond to independent management platforms, forming a fragmented pattern of "one manufacturer, one platform," resulting in severe data silos and difficulties in unified management.

[0004] Therefore, there is an urgent need for a comprehensive management platform that can achieve unified and secure access to multi-protocol smart circuit breakers, fusion and cleaning of heterogeneous data, hierarchical and standardized storage, and intelligent analysis. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide an IoT access, data storage and analysis platform for multi-protocol smart circuit breakers.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an IoT access, data storage, and analysis platform for multi-protocol smart circuit breakers, comprising:

[0007] The multi-protocol adaptive security access module is used to achieve unified secure access for intelligent circuit breakers with heterogeneous communication protocols. It includes a protocol access configuration unit, a multi-protocol monitoring and parsing unit, and a device bidirectional authentication and authorization unit. The protocol access configuration unit provides a graphical configuration interface for configuring access parameters for different communication protocols, and the configuration information supports hot loading. The multi-protocol monitoring and parsing unit starts a multi-threaded monitoring service based on configuration items, simultaneously receiving raw packets from different physical channels. It dynamically loads the matching protocol parsing component according to the device identifier, performs frame recognition, data extraction, and format conversion on the raw packets, and generates unified standardized data within the platform. The device bidirectional authentication and authorization unit uses an asymmetric encryption mechanism to achieve bidirectional authentication between the platform and the device, and binds, stores, and verifies the device's unique identification code and dynamically generated key to prevent forged device access.

[0008] The data fusion and quality assessment module is used to standardize, fuse, and improve the quality of circuit breaker data from multiple heterogeneous channels. It includes a unified information model mapping unit, a multi-source data fusion unit, and a data cleaning and filtering unit. The unified information model mapping unit establishes a mapping relationship between multi-source protocol data and a unified circuit breaker information model. The multi-source data fusion unit uses an improved weighted evidence theory fusion algorithm to fuse redundant data from multiple channels of the same circuit breaker, obtaining estimated values ​​and confidence levels of the fused data. The data cleaning and filtering unit uses a sliding window midpoint filtering algorithm for online real-time cleaning, determining and removing outliers based on the median and standard deviation of data points within the window.

[0009] A hybrid hierarchical storage module is used to hierarchically store the cleaned and fused data. It includes a real-time status storage unit, a time-series historical data storage unit, a device information and relationship storage unit, and a data query and playback interface. The real-time status storage unit uses an in-memory database to store the latest telemetry data and online status. The time-series historical data storage unit uses a time-series database to store high-frequency sampled data and supports downsampling. The device information and relationship storage unit uses a relational database to persistently store device files and configuration parameters. The data query and playback interface provides a unified data query API and historical data playback functionality.

[0010] The intelligent analysis and diagnosis module is used for in-depth mining of stored data, including an equipment status assessment unit, a fault prediction and diagnosis unit, an energy efficiency analysis unit, and a report generation and push unit. The equipment status assessment unit uses a rule engine and statistical model to evaluate the health of circuit breakers based on fused data. The fault prediction and diagnosis unit uses machine learning algorithms to construct a baseline model of normal equipment operation characteristics, inputs the feature vectors of real-time data into the model for judgment, and generates an abnormal status trend warning when the deviation score exceeds a preset threshold. The energy efficiency analysis unit statistically analyzes electricity consumption trends and peak-valley distribution to identify high-energy-consuming equipment and inefficient operating periods. The report generation and push unit periodically generates and pushes equipment health reports.

[0011] It also includes open services and visualization modules, including an open API service unit and a visualization monitoring unit; the open API service unit provides a RESTful API based on OAuth2.0 authentication for third-party applications to call; the visualization monitoring unit provides web and mobile visualization interfaces, supporting power distribution topology diagram display, real-time data monitoring, historical trend curves, alarm event list and equipment health dashboard.

[0012] In some embodiments, the multi-protocol monitoring and parsing unit has built-in libraries for parsing various communication protocols, supporting Modbus RTU / TCP, IEC61850, DL / T698, DL / T645, MQTT, CoAP, HTTP standard protocols, as well as vendor-specific serial port protocols. It also establishes a circuit breaker feature mapping table that includes hardware model, firmware version, protocol type, and register definitions. After device authentication is successful, it automatically retrieves the hardware type based on the device identifier and dynamically loads the matching protocol parsing component.

[0013] In some embodiments, in the device's two-way identity authentication and authorization unit, the asymmetric encryption mechanism adopts the SM2 national cryptographic algorithm or the RSA-2048 algorithm, and the digital certificate is issued by the platform's private CA and supports certificate revocation list updates; for MQTT and CoAP protocols, the device carries a unique identification code and a dynamic key allocated by the platform when connecting. During the first connection, the platform binds and stores the device's unique identification code with the dynamically generated ClientID / Token. If the ClientID does not match the binding relationship or the certificate verification fails during subsequent connections, the connection is rejected.

[0014] In some embodiments, the protocol parsing component in the multi-protocol adaptive secure access module adopts a plug-in architecture design. When adding new protocol support, it is only necessary to develop a parsing plugin that conforms to the interface specification and upload it to the platform. The platform completes the plugin registration through a dynamic class loading mechanism, without modifying the platform's core code or restarting the service.

[0015] In some embodiments, the improved weighted evidence theory fusion algorithm employed by the multi-source data fusion unit includes:

[0016] Assume it exists There are several data sources, each corresponding to random evidence of a circuit breaker parameter. Its basic probability assignment function and focal element are respectively and ;

[0017] Calculation evidence and Distance between :

[0018]

[0019] in This is a focal element similarity-conflict matrix, whose elements are... Measuring the focal length and Similarity, when and When they do not intersect ;

[0020] Construct a similarity matrix based on the distance matrix Its elements ;

[0021] Calculate each piece of evidence support That is, the similarity matrix of the th Sum of row elements:

[0022]

[0023] Calculation evidence Composite weight ,in Prior weights are set based on protocol reliability. , It is a regulating factor and 0 < <1; After applying a weighted average to the evidence, the DS synthesis rule is used for fusion to obtain the final fusion value and uncertainty range.

[0024] In some embodiments, the data cleaning and filtering unit maintains a length of [length missing] for each device. The time window for newly collected data points Sort the point and its historical data points within the window and calculate the median. Simultaneously calculate the standard deviation of all data points within the window. ;like Then determine Outlier points are identified and removed; otherwise, the data point is retained and the sliding window is updated. The length N of the sliding window is adaptively adjusted according to the device's data sampling frequency. For high-frequency sampling devices... Take 60, for low-frequency sampling equipment Take 30.

[0025] In some embodiments, the time-series historical data storage unit in the hybrid hierarchical storage module supports an automatic downsampling strategy: original high-frequency data is retained for 7 days, downsampled to 1-minute granularity and retained for 30 days, further downsampled to 1-hour granularity and retained for 1 year, and then downsampled again to 1-day granularity for long-term archiving; the real-time status storage unit uses a Redis in-memory database, the time-series historical data storage unit uses an InfluxDB or TDengine time-series database, and the device information and relational storage unit uses a MySQL or PostgreSQL relational database.

[0026] In some embodiments, the fault prediction and diagnosis unit utilizes machine learning algorithms including LSTM long short-term memory networks, random forests, or isolated forests to learn and model historical electrical features, including current harmonic content, temperature change rate, opening and closing time deviation, and three-phase imbalance. The machine learning model adopts an online learning mechanism, periodically performing incremental training and parameter updates on the model based on newly added historical data to adapt to the feature distribution drift caused by equipment aging.

[0027] In some embodiments, the prior weights in the multi-source data fusion unit The settings are as follows: Fiber Ethernet channel weight 0.9, RS485 wired channel weight 0.7, Wi-Fi channel weight 0.5, NB-IoT channel weight 0.4, Zigbee channel weight 0.3, and adjustment factor. The default value is 0.6, which can be dynamically adjusted through the configuration interface according to the on-site communication quality.

[0028] In some embodiments, the equipment condition assessment unit generates a health score by weighting the contact wear amount and insulation aging index; the fault prediction and diagnosis unit identifies and initially locates fault types based on a fault feature pattern library; and the energy efficiency analysis unit identifies high-energy-consuming equipment and inefficient operating periods and generates an energy efficiency optimization suggestion report.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. Secure and seamless unified access for multi-protocol devices is achieved. Through a multi-protocol adaptive secure access module, the platform shields the differences in underlying communication protocols. New devices can be connected simply by configuring protocol parameters and mapping rules through a graphical interface, without requiring modification to the platform's core code. Combined with a two-way authentication mechanism, the physical identity of connected devices and communication security are ensured, effectively resisting protocol replay attacks and spoofed device access.

[0031] 2. Significantly improved the quality and reliability of multi-source heterogeneous data. Through an improved weighted evidence theory fusion algorithm, the problem of fusing redundant data from multiple channels of the same device is effectively handled. It comprehensively considers prior knowledge of protocol reliability and posterior information on the mutual support between data points, providing more reliable data estimates and confidence intervals. Simultaneously, combined with a sliding window midpoint filtering algorithm, grid noise and sensor anomalies are eliminated online, avoiding false alarms caused by dirty data and laying a solid data quality foundation for upper-level analysis.

[0032] 3. Unified and centralized management of equipment and data has been achieved. Breaking down the management silos of "one manufacturer, one platform", intelligent circuit breakers from different manufacturers, with different protocols and communication methods are uniformly connected to the same platform, realizing unified monitoring, unified alarm, unified configuration, and unified operation and maintenance, which significantly reduces the operation and maintenance complexity and management costs in power distribution scenarios.

[0033] 4. It has enabled a shift from passive maintenance to proactive prediction. By introducing machine learning algorithms to build a baseline model of normal equipment behavior, it can identify trend anomalies that traditional threshold alarms cannot capture, achieving early warning of faults and prediction of remaining lifespan. Combined with energy efficiency analysis capabilities, the platform has been upgraded from a simple data aggregation tool to an intelligent decision support system, empowering predictive maintenance and energy-saving optimization.

[0034] 5. Optimized data storage costs and query efficiency. A hybrid tiered storage architecture is adopted, selecting the optimal storage engine for different data characteristics, and combined with an automatic downsampling strategy, significantly reducing storage costs and hardware investment while ensuring query efficiency.

[0035] 6. Excellent openness and scalability. It adopts a modular, microservice architecture, with the protocol parsing component using a plug-in architecture. Adding new protocol support only requires developing the parsing plugin and uploading it to the platform; no service restart is needed. Through open APIs and a standardized information model, it achieves seamless integration with third-party systems, adapting to the needs of power distribution IoT applications of different scales and scenarios.

[0036] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. The embodiments of this application will provide a detailed description and understanding of the application. Attached Figure Description

[0037] Figure 1 This is a diagram of the overall system architecture of the present invention;

[0038] Figure 2 This is a timing diagram of the device bidirectional identity authentication interaction of the multi-protocol adaptive secure access module of the present invention;

[0039] Figure 3 This is a flowchart of the improved weighted evidence theory fusion algorithm in the data fusion and quality assessment module of this invention;

[0040] Figure 4 This is a flowchart of the sliding window mid-value filtering algorithm in the data cleaning and filtering unit of the present invention;

[0041] Figure 5 This is a sequence diagram of the data reporting, storage, and analysis interaction in this invention;

[0042] Figure 6 This is a flowchart of the fault prediction and early warning process based on machine learning in the intelligent analysis and diagnosis module of this invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1: Platform Overall Architecture and Single Device Access Workflow

[0045] like Figure 1 As shown, the platform of this invention adopts a layered modular architecture, mainly composed of a multi-protocol adaptive secure access module, a data fusion and quality assessment module, a hybrid hierarchical storage module, an intelligent analysis and diagnosis module, and an open service and visualization module. Its core design concept is as follows: a two-way secure authentication mechanism ensures that only legitimate hardware devices can access the platform; dynamic protocol parsing and a plug-in architecture solve the problem of heterogeneous device access; multi-source data fusion and sliding window filtering improve data quality; hybrid hierarchical storage optimizes data management efficiency; and finally, machine learning algorithms are combined to perform in-depth analysis and predictive early warning of device status.

[0046] This embodiment describes in detail the entire process of a specific smart circuit breaker (which has both RS485 / Modbus RTU local interface and NB-IoT remote reporting capability) being connected to the platform for real-time data acquisition, processing and analysis.

[0047] Step 1: Device Security Authentication and Protocol Configuration

[0048] When the intelligent circuit breaker is powered on for the first time and attempts to connect to the platform, the multi-protocol adaptive security access module initiates... Figure 2 The two-way authentication process is shown below:

[0049] (1) The circuit breaker sends a connection request to the platform through the NB-IoT channel, carrying the pre-set SM2 digital certificate and the device unique identification code (IMEI).

[0050] (2) The platform device two-way identity authentication and authorization unit verifies the legality of the certificate (verifies the CA signature, certificate validity period, and certificate revocation status), and at the same time verifies whether the IMEI is in the platform whitelist.

[0051] (3) After the certificate verification is successful, the platform generates a dynamic token and binds it to the IMEI and stores it in the MySQL database, and then sends the token to the device.

[0052] (4) The device uses a token to perform reverse authentication on the platform server to confirm the legitimacy of the platform's identity.

[0053] (5) After two-way authentication is successful, an encrypted communication channel is established, and the platform marks the device as "online" in Redis.

[0054] This step ensures the physical identity of the connected device is legitimate, and the device also verifies the legitimacy of the platform it is connected to, effectively resisting the risks of protocol replay attacks and forged device access.

[0055] Before authentication, maintenance personnel had already completed the protocol access configuration for the device through the graphical configuration interface: the local Modbus RTU channel was configured as serial port COM1, baud rate 9600, and register mapping table (40001=A-phase current, 40002=B-phase current, 40003=C-phase current, 40004=A-phase voltage, etc.); the NB-IoT channel was configured as MQTTBroker address, subscription Topic format, and JSON data field mapping (e.g., {"U_a":220.1} corresponds to the U_a attribute in the CBIM model). The configuration information is persistently stored in a relational database, supporting hot reloading without restarting the platform service.

[0056] Step 2: Dynamic Protocol Resolution and Standardization

[0057] After device authentication is successful, the multi-protocol monitoring and parsing unit retrieves the specific model of the device (e.g., "XX brand - XX model - dual-mode circuit breaker") from the circuit breaker feature mapping table based on the device identifier. The engine automatically and dynamically loads the ModbusRTU parsing plugin and the NB-IoTJSON parsing plugin from the built-in protocol parsing library.

[0058] The engine receives raw messages from the circuit breaker and performs frame recognition, data extraction, and format conversion. For example, the Modbus parser extracts physical quantities such as phase A current, phase B current, phase C current, and phase A voltage corresponding to registers 40001-40004 from the function code 03 response frame according to the protocol specification; the NB-IoT parser extracts corresponding fields from the JSON object of the MQTT message payload. Finally, these scattered physical quantities are converted into the platform's internally defined CBIM unified JSON format, such as:

[0059] {"device_id":"CB_2024_001","timestamp":"2024-06-10T15:47:00Z","U_a":220.1, "U_b":219.8,"U_c":220.5,"I_a":12.5,"I_b":12.3,"I_c":12.4,"status":"closed"}

[0060] Step 3: Multi-source data fusion and cleaning / filtering

[0061] The data fusion and quality assessment module fuses the two A-phase current values ​​within the same time window:

[0062] (1) Invoking the improved weighted evidence theory algorithm: Assuming Modbus data evidence NB-IoT data evidence .

[0063] (2) Prior weight setting: (RS485 wired channels are more reliable). (NB-IoT wireless channels have relatively low reliability).

[0064] (3) Calculate the distance between the two pieces of evidence. Similarity matrix and support , .

[0065] (4) Calculate the posterior weights , And synthesize composite weights , .

[0066] (5) After weighted averaging of the two current values, the uncertainty is handled using the DS synthesis rule to obtain the final fused value. The confidence level is 0.95.

[0067] The fused data enters the data cleaning and filtering unit. Taking phase A voltage U_a as an example, a sliding window of length N=60 is maintained (corresponding to 1 minute, sampling frequency 1 time / second). Newly acquired data points... Median of historical data within the window after sorting Standard deviation Calculated Therefore, this point was determined to be an outlier caused by instantaneous fluctuations in the power grid and was removed from the downstream storage and analysis process. This step ensures that the stored historical data accurately reflects the steady-state operating status of the circuit breaker.

[0068] Step 4: Hybrid Tiered Storage

[0069] The hybrid hierarchical storage module performs hierarchical storage on the cleaned and merged data:

[0070] (1) Real-time data storage: The latest collected device telemetry data (including fused I_a_fused, U_a, etc.) is written to the Redis cache database and the expiration time is set to 5 minutes to meet the low-latency query requirements of the real-time monitoring screen.

[0071] (2) Data storage of archives: The static information of the equipment (model, manufacturer, installation location, protocol type, configuration parameters, etc.) is persistently stored in a MySQL relational database, which supports transactional updates.

[0072] (3) Historical data storage: The cleaned time series data (electrical quantities and state quantities with timestamps) are written in batches into the InfluxDB time series database. The original high-frequency data is retained for 7 days, and the data is automatically downsampled to 1 minute granularity and retained for 30 days. The 1 hour granularity is retained for 1 year, which is convenient for long-term trend analysis.

[0073] The time-series historical data storage unit supports an automatic downsampling strategy: original high-frequency data is retained for 7 days, downsampled to 1-minute granularity and retained for 30 days, further downsampled to 1-hour granularity and retained for 1 year, and then downsampled to 1-day granularity for long-term archiving; the real-time status storage unit uses a Redis in-memory database, the time-series historical data storage unit uses an InfluxDB or TDengine time-series database, and the device information and relational storage unit uses a MySQL or PostgreSQL relational database.

[0074] Step 5: Intelligent Analysis and Diagnosis

[0075] The intelligent analysis and diagnostic module periodically (hourly) extracts the device's historical data from InfluxDB for the past 7 days and performs actions such as... Figure 6 The analysis process shown is as follows:

[0076] (1) Feature extraction: Calculate a series of statistical features, including mean, variance, maximum value, minimum value, kurtosis, skewness, current harmonic content, three-phase unbalance, opening and closing time, etc., to form the feature vector of the equipment.

[0077] (2) Model training / update: Based on the feature vectors during normal operation of the device, an LSTM network is used to train a behavior baseline model unique to the device. The model adopts an online learning mechanism, and incremental training and parameter updates are performed weekly based on new data to adapt to the feature distribution drift caused by device aging.

[0078] (3) Real-time judgment: For the data collected and cleaned and fused in real time, the corresponding feature vectors are extracted and input into the trained LSTM model for judgment. The model outputs the deviation score.

[0079] (4) Warning generation: If the deviation score exceeds the preset threshold (e.g., 0.85), it indicates that the current operating status has deviated from the historical normal mode. For example, if the circuit breaker's opening and closing time feature vector continues to deviate from its historical normal distribution, even if the current current value has not yet reached the traditional overcurrent threshold alarm condition, the platform will determine that there is mechanical wear or aging trend, generate "abnormal status trend - mechanical wear warning", and push it to the operation and maintenance APP.

[0080] (5) Energy efficiency analysis: Statistically analyze the power consumption trend and peak-valley distribution of the circuit where the circuit breaker is located over the past 30 days, identify periods of abnormally high standby power consumption at night, and generate a "Nighttime Standby Power Consumption Optimization Suggestion" report.

[0081] The equipment condition assessment unit generates a health score by weighting the contact wear amount and insulation aging index; the fault prediction and diagnosis unit identifies and initially locates fault types based on the fault feature pattern library; and the energy efficiency analysis unit identifies high-energy-consuming equipment and inefficient operating periods and generates an energy efficiency optimization suggestion report.

[0082] Example 2: Multi-protocol adaptive secure access and plug-in extension

[0083] This embodiment focuses on explaining the detailed working mechanism of the multi-protocol adaptive secure access module, especially the unified secure access, two-way identity authentication, and plug-in protocol extension capabilities for intelligent circuit breakers with heterogeneous communication protocols.

[0084] like Figure 2As shown, the multi-protocol monitoring and parsing unit has built-in libraries for various communication protocols, supporting Modbus RTU / TCP, IEC61850, DL / T698, DL / T645, MQTT, CoAP, HTTP standard protocols, as well as vendor-specific serial port protocols. It also establishes a circuit breaker feature mapping table containing hardware model, firmware version, protocol type, and register definitions. After device authentication, the hardware type is automatically retrieved based on the device identifier, and the matching protocol parsing component is dynamically loaded.

[0085] The protocol access configuration unit provides a graphical configuration interface, allowing maintenance personnel to configure access parameters for different communication protocols, including but not limited to: communication port (RS485 serial port, TCP / IP port, MQTTBroker address, CoAPURI, etc.), baud rate, data bits, stop bits, parity check, protocol specifications (such as Modbus register address mapping table, IEC61850 logical node model, DL / T698 data identifier, MQTTTopic subscription rules, CoAP resource path, etc.), and custom message parsing scripts. Configuration information is persistently stored in a relational database, supporting hot reloading without requiring a platform service restart.

[0086] The device's two-way identity authentication and authorization unit employs an asymmetric encryption mechanism to achieve two-way identity authentication between the platform and the device. This asymmetric encryption mechanism uses either the SM2 national cryptographic algorithm or the RSA-2048 algorithm. The digital certificate is issued by the platform's private CA and supports Certificate Revocation List (CRL) updates, ensuring the security and compliance of the authentication system.

[0087] For MQTT and CoAP protocols, devices carry a unique identifier (such as IMEI, MAC address, or device serial number) and a dynamic key assigned by the platform when connecting. Upon initial connection, the platform binds and stores the device's unique identifier with a dynamically generated ClientID / Token. Subsequent connections will be rejected if the ClientID does not match the binding or if certificate verification fails, effectively preventing device impersonation, protocol replay attacks, and unauthorized device access. Furthermore, the binding, storage, and verification based on the device's unique identifier and the dynamically generated key further prevent unauthorized device access.

[0088] The protocol parsing component in the multi-protocol adaptive secure access module adopts a plug-in architecture. When new protocol support is required, developers only need to develop a parsing plugin that conforms to the interface specification and upload it to the platform. The platform completes plugin registration through a dynamic class loading mechanism, without requiring modification of the platform's core code or a service restart. For example, when a new type of circuit breaker uses a vendor's proprietary serial port protocol, developers only need to implement the protocol parsing interface, define frame header recognition rules, data field length, verification algorithm, and field mapping relationships, package it into a plugin JAR file, and upload it to the platform's plugin directory. When the platform detects a new device access, it automatically loads the plugin to complete the protocol parsing.

[0089] Example 3: Implementation of Multi-Source Data Fusion and Quality Assessment Algorithm

[0090] This embodiment focuses on the detailed algorithm implementation of the data fusion and quality assessment module, including the specific execution process of the unified information model mapping, the improved weighted evidence theory fusion algorithm, and the sliding window value filtering algorithm.

[0091] The unified information model mapping unit establishes a mapping relationship between multi-source protocol data and the unified circuit breaker information model (CBIM). This model defines a common set of attributes for circuit breakers, including: identification information (device ID, manufacturer, model, installation location), electrical parameters (A / B / C phase voltages U_a / U_b / U_c, A / B / C phase currents I_a / I_b / I_c, active power P, reactive power Q), protection parameters (overcurrent threshold I_over, leakage current threshold I_leak, short-circuit instantaneous multiple), status information (switch position, closing / opening status, fault flag, alarm level), and statistics (positive active power E_p, positive reactive power E_q, cumulative number of operations N_op), etc. Through model mapping, heterogeneous data from different protocol sources are converted into semantically consistent and format-unified standardized data.

[0092] When the same circuit breaker is used through different protocol channels (such as simultaneously supporting local RS485 and remote NB-IoT reporting) or when redundant sampling exists, an improved weighted evidence theory fusion algorithm is employed to fuse multi-source data to obtain more reliable data estimates and confidence levels. For example... Figure 3 As shown, the improved weighted evidence theory fusion algorithm includes the following steps:

[0093] Assume it exists There are several data sources, each corresponding to random evidence of a circuit breaker parameter. Its basic probability assignment function and focal element are respectively and .

[0094] Calculation evidence and Distance between :

[0095]

[0096] in This is a focal element similarity-conflict matrix, whose elements are... Measuring the focal length and Similarity, when and When they do not intersect .

[0097] Construct a similarity matrix based on the distance matrix Its elements :

[0098]

[0099] Calculate each piece of evidence support That is, the similarity matrix of the th Sum of row elements:

[0100]

[0101] Calculation evidence Composite weight ,in Prior weights are set based on protocol reliability. For posterior weights, It is a regulating factor and 0 < <1:

[0102]

[0103] Where the posterior weight Calculated by the following formula:

[0104]

[0105] After applying a weighted average to the evidence, the DS synthesis rule is used to fuse the evidence, resulting in the final fusion value and uncertainty range.

[0106] Prior weights in the multi-source data fusion unit The settings are as follows: Fiber Ethernet channel weight 0.9, RS485 wired channel weight 0.7, Wi-Fi channel weight 0.5, NB-IoT channel weight 0.4, Zigbee channel weight 0.3, and adjustment factor. The default value is 0.6, which can be dynamically adjusted through the configuration interface according to the on-site communication quality.

[0107] The data cleaning and filtering unit employs a sliding window midpoint filtering algorithm for online real-time cleaning. For example... Figure 4 As shown, maintain a length of [length missing] for each device. The time window for newly collected data points Sort the point and its historical data points within the window and calculate the median. Simultaneously calculate the standard deviation of all data points within the window. ;like Then determine If a data point is an outlier, remove it; otherwise, keep the data point and update the sliding window.

[0108] The length of the sliding window The frequency of data sampling is adaptively adjusted based on the device's data sampling frequency, for high-frequency sampling devices (e.g., 1 time / second). Take 60, for low-frequency sampling devices (e.g., 1 time / minute). We set it to 30 to balance filtering effect and response sensitivity.

[0109] Example 4: Hybrid Hierarchical Storage and Intelligent Analysis and Diagnosis

[0110] This embodiment focuses on illustrating the detailed data management strategy of the hybrid hierarchical storage module and the implementation of the machine learning algorithm of the intelligent analysis and diagnosis module.

[0111] The hybrid hierarchical storage module employs a hybrid database architecture to hierarchically store the cleaned and fused data, adapting to different data characteristics and optimizing storage costs and query efficiency. The real-time status storage unit uses an in-memory database (such as Redis) to store the latest telemetry data collected by the circuit breaker, real-time alarm status, and device online status, supporting high-concurrency, low-latency read and write operations to meet the rapid query needs of scenarios such as real-time monitoring dashboards and maintenance apps. The time-series historical data storage unit uses a time-series database (such as InfluxDB or TDengine) to store high-frequency sampled electrical quantities, power consumption, temperature, and other time-series data from the circuit breaker, supporting efficient aggregation queries by time range, downsampling, and continuous queries, providing data support for long-term trend analysis and machine learning model training. The device information and relational storage unit uses a relational database (such as MySQL or PostgreSQL) to persistently store structured data such as device files (model, manufacturer, installation location, protocol configuration, mapping rules), user permissions, maintenance work orders, and configuration parameters, ensuring data consistency and transaction integrity. The data query and playback interface provides a unified data query API, supporting multi-dimensional searches based on device, time range, data type, and other criteria, and also supports historical data playback, facilitating fault tracing and maintenance auditing.

[0112] The intelligent analysis and diagnosis module utilizes preset threshold models, statistical models, and machine learning algorithms to deeply mine stored historical and real-time data, enabling equipment health status assessment, fault early warning, and energy efficiency analysis. For example... Figure 6 As shown, the fault prediction and diagnosis unit utilizes machine learning algorithms, including LSTM (Long Short-Term Memory) networks, random forests, or isolated forests, to learn and model historical electrical characteristics. These historical electrical characteristics include current harmonic content, temperature change rate, switching time deviation, and three-phase imbalance. The machine learning model employs an online learning mechanism, periodically (e.g., weekly) incrementally training and updating parameters based on newly added historical data to adapt to feature distribution drift caused by equipment aging.

[0113] The equipment status assessment unit comprehensively evaluates the health of circuit breakers based on fused real-time and historical data, using a rule engine and statistical model. For example, it generates a health score of 0-100 by weighting contact wear (estimated based on cumulative operation count and breaking current) and insulation aging index (based on the trend of dielectric loss angle change). The fault prediction and diagnosis unit identifies and initially locates fault types based on a fault feature pattern library. The energy efficiency analysis unit statistically analyzes the power consumption trends, peak-valley distribution, and power factor changes of each region, circuit, and time period, identifies high-energy-consuming equipment and inefficient operating periods, and generates energy efficiency optimization suggestion reports to provide data support for energy conservation and consumption reduction. The report generation and push unit periodically generates periodic equipment health reports containing equipment health scores, fault type predictions, remaining life estimates, operating statistics, event records, and energy-saving suggestions, and notifies maintenance personnel through visual interfaces, emails, SMS, and APP push notifications.

[0114] Example 5: Unified Access and Interlocking Application of Circuit Breakers from Multiple Manufacturers in Shared Base Stations

[0115] This embodiment uses a communication base station power distribution cabinet as an application scenario to illustrate the unified access, data fusion, intelligent linkage analysis, and energy efficiency optimization process of circuit breakers from multiple manufacturers in a shared base station.

[0116] The power distribution cabinet of this communication base station is equipped with smart circuit breakers from three different manufacturers: Manufacturer A's circuit breaker uses the Zigbee protocol and is responsible for powering the air conditioning circuit; Manufacturer B's circuit breaker uses the RS485 / ModbusRTU protocol and is responsible for powering the main equipment circuit; Manufacturer C's circuit breaker uses the NB-IoT protocol and is responsible for powering the backup power circuit.

[0117] Access process

[0118] The maintenance personnel configured the corresponding protocol parameters and mapping rules for the three circuit breakers through the platform's graphical configuration interface: Manufacturer A configured the Zigbee coordinator address, endpoints, cluster ID, and data field mapping; Manufacturer B configured the RS485 serial port parameters and Modbus register table; and Manufacturer C configured the MQTTBroker connection parameters and Topic mapping. After configuration, the platform started the corresponding multi-protocol listening service, and the three circuit breakers connected to the platform through their respective protocol channels, all undergoing a two-way authentication process.

[0119] Data fusion process

[0120] Manufacturer B's main equipment circuit breaker simultaneously supports local RS485 and remote NB-IoT dual-channel reporting. The platform's data fusion module fuses the current data from both channels to obtain highly reliable electrical measurement values.

[0121] Intelligent linkage analysis

[0122] The intelligent analysis module detected a continuous increase in the current harmonic content of the air conditioner circuit breaker from Manufacturer A (from the normal 3% to 12%). Based on historical trends, it predicted that the load equipment (air conditioner compressor) might experience abnormalities. It generated a "Potential Overcurrent Risk - Air Conditioner Circuit" warning work order in advance and automatically linked it with the historical load data of the main equipment circuit breaker from Manufacturer B. It assessed the impact on the main equipment circuit if the air conditioner failure caused a sudden change in current and generated a linkage response suggestion, which was pushed to the operation and maintenance APP.

[0123] Energy efficiency optimization

[0124] The energy efficiency analysis unit statistically analyzed the base station's annual power consumption data and found that the standby power consumption of the air conditioning circuit at night (00:00-06:00) accounted for 35% of the total power consumption of the daytime, which was significantly higher than the normal level (15%). A report entitled "Optimization Suggestions for Air Conditioning Standby Power Consumption at Night" was generated, which suggested adjusting the air conditioning timing strategy or replacing it with a high-efficiency device.

[0125] Open API Applications

[0126] The base station's upper-level energy management system uses the OAuth2.0-based RESTful API provided by the platform to obtain the real-time status, historical energy consumption data, and health scores of all circuit breakers in batches. The system then displays the power distribution operation status of the base station and all its subordinate sites on the EMS dashboard, achieving cross-platform data sharing and centralized management.

[0127] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An IoT access, data storage, and analysis platform for multi-protocol intelligent circuit breakers, characterized in that, include: The multi-protocol adaptive security access module is used to realize unified security access for intelligent circuit breakers with heterogeneous communication protocols. It includes a protocol access configuration unit, a multi-protocol monitoring and parsing unit, and a device two-way identity authentication and authorization unit. The protocol access configuration unit provides a graphical configuration interface for configuring access parameters of different communication protocols, and the configuration information supports hot loading. The multi-protocol monitoring and parsing unit starts a multi-threaded monitoring service based on configuration items, and simultaneously receives raw packets from different physical channels. It dynamically loads the matching protocol parsing component according to the device identifier, performs frame recognition, data extraction and format conversion on the raw packets, and generates unified standardized data within the platform. The device two-way identity authentication and authorization unit adopts an asymmetric encryption mechanism to realize two-way identity authentication between the platform and the device, and binds, stores and verifies the device's unique identification code and dynamically generated key to prevent counterfeit device access. The data fusion and quality assessment module is used to standardize, fuse, and improve the quality of circuit breaker data from multiple heterogeneous channels. It includes a unified information model mapping unit, a multi-source data fusion unit, and a data cleaning and filtering unit. The unified information model mapping unit establishes a mapping relationship between multi-source protocol data and the unified circuit breaker information model; The multi-source data fusion unit uses an improved weighted evidence theory fusion algorithm to fuse multi-channel redundant data from the same circuit breaker, obtaining fused data estimates and confidence levels. The data cleaning and filtering unit uses a sliding window midpoint filtering algorithm for online real-time cleaning, determining and removing outliers based on the median and standard deviation of data points within the window. The hybrid hierarchical storage module is used to store the cleaned and fused data in a hierarchical manner, including a real-time status storage unit, a time-series historical data storage unit, a device information and relationship storage unit, and a data query and playback interface. The real-time status storage unit uses an in-memory database to store the latest telemetry data and online status. The time-series historical data storage unit uses a time-series database to store high-frequency sampled data and supports downsampling; the device information and relational storage unit uses a relational database to persistently store device files and configuration parameters; the data query and playback interface provides a unified data query API and historical data playback function. The intelligent analysis and diagnosis module is used for in-depth mining of stored data, including equipment status assessment unit, fault prediction and diagnosis unit, energy efficiency analysis unit, and report generation and push unit; The equipment status assessment unit evaluates the health of circuit breakers based on fused data using a rule engine and statistical model; the fault prediction and diagnosis unit uses machine learning algorithms to construct a baseline model of normal equipment operation characteristics, inputs the feature vectors of real-time data into the model for judgment, and generates an abnormal status trend warning when the deviation score exceeds a preset threshold; the energy efficiency analysis unit statistically analyzes electricity consumption trends and peak-valley distribution to identify high-energy-consuming equipment and inefficient operating periods; the report generation and push unit periodically generates and pushes equipment health reports. And open services and visualization modules, including open API service unit and visualization monitoring unit; The open API service unit provides a RESTful API based on OAuth2.0 authentication for third-party applications to call; The visualization monitoring unit provides web and mobile visualization interfaces, supporting the display of power distribution topology diagrams, real-time data monitoring, historical trend curves, alarm event lists, and equipment health dashboards.

2. The platform according to claim 1, characterized in that, The multi-protocol monitoring and parsing unit has built-in libraries for various communication protocols, supporting Modbus RTU / TCP, IEC61850, DL / T698, DL / T645, MQTT, CoAP, HTTP standard protocols, as well as vendor-specific serial port protocols. It also establishes a circuit breaker feature mapping table that includes hardware model, firmware version, protocol type, and register definitions. After device authentication, it automatically retrieves the hardware type based on the device identifier and dynamically loads the matching protocol parsing component.

3. The platform according to claim 1, characterized in that, In the device's two-way identity authentication and authorization unit, the asymmetric encryption mechanism adopts the SM2 national cryptographic algorithm or the RSA-2048 algorithm. The digital certificate is issued by the platform's private CA and supports certificate revocation list updates. For MQTT and CoAP protocols, the device carries a unique identification code and a dynamic key allocated by the platform when connecting. During the first connection, the platform binds and stores the device's unique identification code with the dynamically generated ClientID / Token. If the ClientID does not match the binding relationship or the certificate verification fails during subsequent connections, the connection will be rejected.

4. The platform according to claim 1, characterized in that, The protocol parsing component in the multi-protocol adaptive secure access module adopts a plug-in architecture design. When adding new protocol support, it is only necessary to develop a parsing plugin that conforms to the interface specification and upload it to the platform. The platform completes the plugin registration through a dynamic class loading mechanism, without modifying the platform's core code or restarting the service.

5. The platform according to claim 1, characterized in that, The improved weighted evidence theory fusion algorithm used by the multi-source data fusion unit includes: Assume it exists There are several data sources, each corresponding to random evidence of a circuit breaker parameter. Its basic probability assignment function and focal element are respectively and ; Calculation evidence and Distance between : in This is a focal element similarity-conflict matrix, whose elements are... Measuring the focal length and Similarity, when and When they do not intersect ; Construct a similarity matrix based on the distance matrix Its elements ; Calculate each piece of evidence support That is, the similarity matrix of the th Sum of row elements: Calculation evidence Composite weight ,in Prior weights are set based on protocol reliability. , It is a regulating factor and 0 < <1; After applying a weighted average to the evidence, the DS synthesis rule is used for fusion to obtain the final fusion value and uncertainty range.

6. The platform according to claim 1, characterized in that, The data cleaning and filtering unit maintains a length of [length missing] for each device. The time window for newly collected data points Sort the point and its historical data points within the window and calculate the median. Simultaneously calculate the standard deviation of all data points within the window. ;like Then determine Outliers were identified and removed. Otherwise, the data point is retained and the sliding window is updated; the length N of the sliding window is adaptively adjusted according to the device's data sampling frequency, for high-frequency sampling devices. Take 60, for low-frequency sampling equipment Take 30.

7. The platform according to claim 1, characterized in that, The time-series historical data storage unit in the hybrid hierarchical storage module supports an automatic downsampling strategy: original high-frequency data is retained for 7 days, downsampled to 1-minute granularity and retained for 30 days, further downsampled to 1-hour granularity and retained for 1 year, and then downsampled again to 1-day granularity for long-term archiving; the real-time status storage unit uses a Redis in-memory database, the time-series historical data storage unit uses an InfluxDB or TDengine time-series database, and the device information and relational storage unit uses a MySQL or PostgreSQL relational database.

8. The platform according to claim 1, characterized in that, The fault prediction and diagnosis unit utilizes machine learning algorithms including LSTM (Long Short-Term Memory) networks, random forests, or isolated forests to learn and model historical electrical features, such as current harmonic content, temperature change rate, opening and closing time deviation, and three-phase imbalance. The machine learning model employs an online learning mechanism, periodically performing incremental training and parameter updates based on newly added historical data to adapt to feature distribution drift caused by equipment aging.

9. The platform according to claim 5, characterized in that, Prior weights in the multi-source data fusion unit The settings are as follows: Fiber Ethernet channel weight 0.9, RS485 wired channel weight 0.7, Wi-Fi channel weight 0.5, NB-IoT channel weight 0.4, Zigbee channel weight 0.3, and adjustment factor. The default value is 0.6, which can be dynamically adjusted through the configuration interface according to the on-site communication quality.

10. The platform according to claim 1, characterized in that, The equipment condition assessment unit generates a health score by weighting the contact wear amount and insulation aging index; the fault prediction and diagnosis unit identifies and initially locates fault types based on the fault feature pattern library; and the energy efficiency analysis unit identifies high-energy-consuming equipment and inefficient operating periods and generates an energy efficiency optimization suggestion report.

Images