A multi-device compatible electric bicycle charging middleware management method and system

Abstract

This application discloses a method and system for managing middleware for multi-device compatible electric bicycle charging, relating to the field of battery charging management technology. This method employs a collaborative architecture of edge nodes and cloud management nodes to achieve intelligent and secure management of multi-protocol charging devices. On-site edge nodes establish communication links with compliant charging devices, collect raw data frames uploaded by the devices in real time, generate unique feature fingerprints adapted to the device's private protocol, match and filter corresponding semantic parsing models through a fingerprint database, parse the device's raw structured event data and effective model identifiers, and upload them to the cloud. The cloud, based on the model identifiers, calls standardized mapping rules to complete data processing, generates general standard event data, extracts corresponding business information, matches adapted standardized business templates, loads built-in event response rules and execution nodes to generate management instructions, and issues them for execution. This solves the problem of poor device compatibility in charging systems and ensures the safe operation of electric bicycle charging.

Description

Technical Field

[0001] This application relates to the field of battery charging control technology, and more specifically, to a method and system for controlling middleware charging of electric bicycles that is compatible with multiple devices. Background Technology

[0002] The current electric bicycle charging industry suffers from a diverse range of equipment brands and inconsistent communication protocols among manufacturers. Many devices utilize custom proprietary protocols, resulting in variations in message format, field definitions, transmission timing, and parameter units. This leads to significant challenges in adapting and integrating multi-brand devices, resulting in poor compatibility. Existing charging management technologies generally rely on manual static adaptation, which suffers from numerous industry pain points, including weak scalability, high development costs, high system complexity, and delayed iterations.

[0003] Existing single-protocol equipment solutions only support the manufacturer's proprietary protocol, with fixed data packet formats that are incompatible with other brands of equipment. Operators can only purchase equipment from the same brand, which not only results in high procurement costs and concentrated supply chain risks, but also requires the development of separate interface programs for new equipment when expanding projects or integrating systems, leading to long adaptation cycles and extremely poor universality.

[0004] To address protocol compatibility issues, the industry commonly employs a multi-protocol adapter solution, manually adapting various protocols through software and hardware. However, this solution requires custom-developed adapters for each new protocol and device, resulting in significant repetitive development work, long development cycles, and high costs. Furthermore, the deployment of multiple adapters significantly increases system architecture complexity, easily leading to parsing anomalies, data conflicts, and performance bottlenecks, resulting in poor overall system reliability.

[0005] Traditional centralized management platforms manage multiple devices by pre-setting parsing logic in the backend. However, the protocol conversion and data processing logic are centrally fixed, resulting in high code redundancy, system complexity, and high maintenance costs. The platform adopts a static configuration mode, lacks independent adaptation and iteration capabilities, and cannot quickly adapt to the access of new proprietary protocol devices, resulting in severely insufficient flexibility and scalability.

[0006] Conventional protocol conversion middleware is only compatible with a few general protocols such as Modbus, and can only meet standardized scenarios. It cannot adapt to a large number of non-standard proprietary protocols and extended protocols in the electric bicycle industry. Secondary development is required to connect to new devices. It lacks dynamic self-adaptation and automatic iteration mechanisms, making it difficult to adapt to the current multi-brand, rapidly iterating large-scale charging management and control scenarios.

[0007] In summary, existing charging management technologies cannot achieve unified management of multiple protocols without human intervention, are adaptive, fully compatible, and iterative. There is an urgent need for an intelligent, self-learning, and adaptive electric bicycle charging management solution. Summary of the Invention

[0008] To address the aforementioned technical challenges, this invention provides a multi-device compatible electric bicycle charging middleware management method and system. It employs an adaptive self-learning architecture that collaborates between the edge and cloud to achieve intelligent and unified management of devices from multiple brands and with multiple proprietary protocols. Edge nodes perform device security access verification and protocol feature fingerprint matching. For unknown proprietary protocols, it automatically learns semantics and generates a dedicated semantic parsing model, eliminating the need for manual adaptation programs. The cloud performs adaptive normalization processing on heterogeneous raw data and achieves adaptive adaptation to operating conditions through business template similarity matching, supporting autonomous iteration of new operating condition templates. Simultaneously, it combines device-specific charging baselines to identify hidden anomalies. Relying on edge offline autonomy and a full-process evidence storage mechanism, it effectively reduces maintenance costs and significantly improves system compatibility and management security.

[0009] On the one hand, this application provides a method for managing middleware for charging electric bicycles that is compatible with multiple devices, the method comprising: The on-site edge node receives network access requests initiated by electric bicycle charging devices and performs identity verification. For electric bicycle charging devices that pass the verification, a communication link is established between the on-site edge node and the device. The edge node continuously receives the real-time raw data frames of the electric bicycle charging device uploaded by the device through the communication link, generates a unique feature fingerprint of the device event that is uniquely bound to the device protocol, compares and matches the unique feature fingerprint of the device event with the event feature fingerprint database, determines the semantic parsing model bound to the corresponding fingerprint, and calls the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames of the charging device, outputting the raw structured event data of the device and the identifier of the currently effective semantic parsing model to the cloud management node; The cloud-based management node uses the identifier of the currently active semantic parsing model to call the standardized mapping rules of the semantic parsing model to normalize the original structured event data of the device, generating standard event data that can be universally recognized by the middleware system. Business information is extracted from the standard event data, and a standardized business template is matched based on the business information. The template's preset event response rules and execution nodes are loaded according to the standardized business template and sent to the field edge nodes to complete the management and control of electric bicycle charging. The business information includes event type, event level, associated device assets, and associated business node parameters.

[0010] This application achieves rapid compatibility and real-time management and control of charging devices of different brands and protocols through a layered collaborative architecture of on-site edge nodes and cloud management nodes, combined with device fingerprint matching, dynamic invocation of semantic parsing models, and standardized business template loading mechanisms. This effectively reduces the development cost of multi-device adaptation and improves the standardization and response efficiency of charging event handling.

[0011] In one embodiment, the field edge node receives a network access request initiated by an electric bicycle charging device and performs identity verification. Through multi-factor authentication using the device hardware serial number, protocol signature, and access key, combined with a local whitelist for rapid matching and a cloud-based global blacklist for collaborative blocking, a dual security defense for charging device access is achieved, effectively preventing unauthorized device intrusion. Simultaneously, blockchain storage ensures the immutability of unauthorized access logs. Specifically, this includes: The edge node extracts the device hardware serial number, device protocol feature code and device access key uploaded by the charging device to form device identity verification information. The device identity verification information is then compared and verified with the locally stored trusted device whitelist. If identity verification fails, the edge node refuses to establish a communication link, records the illegal access log locally and performs blockchain notarization, and pushes an illegal device alarm event to the cloud management node. The cloud management node then adds the device's identity information to the global temporary blacklist and prohibits the device from initiating access requests again within the preset blocking time. If the identity verification is successful, the on-site edge node establishes a communication link and globally associates and binds the device's hardware serial number with the subsequently generated unique feature fingerprint of the device event and the corresponding semantic parsing model. The association and binding relationship is synchronously cached to the on-site edge node and synchronously sent to the cloud management node for storage.

[0012] In one embodiment, the step of comparing and matching the unique feature fingerprint of a device event with an event feature fingerprint database to determine the semantic parsing model bound to the corresponding fingerprint includes: The edge nodes on site calculate the feature matching degree between the unique feature fingerprint of the current device event and each archived fingerprint in the event feature fingerprint database, and compare the matching degree with the preset matching threshold. If there is an archived fingerprint with a matching degree exceeding the preset matching threshold, the semantic parsing model bound to the corresponding fingerprint will be retrieved. Otherwise, initiate semantic self-learning for electric bicycle charging, relying on a pre-stored general semantic field library to parse the field definitions, transmission interaction timing, and parameter value ranges within the original data frame of the charging device, and generate a custom semantic parsing model adapted to the device's private protocol; simultaneously generate a unique feature fingerprint of the newly added device event bound to the custom semantic parsing model, and associate the unique feature fingerprint of the newly added device event with the custom semantic parsing model and store it in the event feature fingerprint library.

[0013] This application achieves fast and accurate invocation of known protocols through a feature fingerprint matching degree threshold comparison mechanism. At the same time, it initiates a self-learning parsing process based on a general semantic field library for unmatched protocols, automatically generates a custom semantic parsing model and dynamically expands the fingerprint library, effectively improving the system's adaptive compatibility with unknown private protocols.

[0014] In one embodiment, the cloud management node calls the standardized mapping rules of the semantic parsing model according to the identifier of the currently effective semantic parsing model to perform normalization processing on the original structured event data of the device, including: field mapping adaptation processing, dynamic unit conversion processing and data correction processing; The field mapping adaptation process includes comparing each field in the original structured event data of the device obtained by parsing the field edge nodes with a preset charging standard field system as a benchmark; directly aligning the names of fields with preset mapping relationships; and generating temporary mapping relationships for fields not covered by the current mapping relationship by performing semantic comparison and reasoning through a general semantic dictionary. The dynamic unit conversion process includes converting voltage, current, battery temperature, and charging time parameters into standard units of measurement required by regulations according to the selected conversion factor. The data correction process includes calling the hierarchical safety threshold library to perform segmented judgment on the converted parameters and correcting the distorted parameters that deviate from the reasonable range; The standardized mapping rules are linked to a safety threshold library for electric bicycle charging grading and a dynamic unit conversion coefficient pool; the dynamic unit conversion coefficient pool can be switched and called according to the equipment model and the local regulatory reporting standards.

[0015] This embodiment employs a three-layer normalization processing mechanism—field mapping adaptation, dynamic unit conversion, and data correction—combined with semantic reasoning to complete mapping relationships and hierarchical safety thresholds to correct distortion parameters. It also supports flexible switching of conversion coefficients based on device model and local regulatory standards, achieving efficient conversion of multi-source heterogeneous charging data to a unified standard format. This effectively ensures data consistency, accuracy, and regulatory compliance.

[0016] In one embodiment, the step of extracting business information based on the standard event data and matching a standardized business template based on the business information includes: All extracted business information is used as matching and retrieval elements. The pre-configured standardized business template library is traversed, and the matching similarity between the current business information and the triggering conditions of each standardized business template in the library is calculated one by one. The standardized business template with the highest matching similarity value is selected and compared with the preset business matching threshold for judgment. When the maximum matching similarity is greater than the preset business matching threshold, the standardized business template with the highest similarity is directly determined as the matching result for this time; Otherwise, a temporary business handling template adapted to the current working conditions is generated based on the currently extracted business information. This temporary business handling template is then stored in the standardized business template library to complete the iterative update of the template library, and the temporary business handling template is determined as the standardized business template obtained in this matching.

[0017] This embodiment achieves rapid and accurate matching of standardized business templates through a traversal matching similarity calculation and threshold comparison mechanism; at the same time, it automatically generates temporary business handling templates for unmatched conditions and dynamically updates the template library, effectively improving the system's adaptive coverage of unknown business scenarios, reducing manual configuration intervention, and enhancing the intelligence and continuous evolution capabilities of charging business operations.

[0018] In one embodiment, the standardized business template loads preset event response rules and execution nodes and distributes them to the field edge nodes to complete the electric bicycle charging management and control, including: The cloud-based control node loads the preset event response rules and corresponding execution nodes within the template, and encapsulates the device identity identifier and event tracing credentials associated with this business into a control instruction message and sends it to the field edge node. The edge nodes on site execute corresponding control actions according to the division of execution nodes: if the execution node points to the currently connected electric bicycle charging equipment, it sends on / off control, power adjustment, and start / stop charging commands to the charging equipment according to the event response rules; if the execution node points to the background alarm push node, it pushes the corresponding level of alarm information in sequence according to the preset push level. After the edge node completes all control actions, it collects the results of the action execution, generates an execution receipt message, and sends it back to the cloud control node.

[0019] In one embodiment, the multi-device compatible electric bicycle charging middleware management method described in this application further includes: For the same electric bicycle charging device, the cloud-based management node continuously collects standard event data of each charging process and fits and generates a baseline model of charging characteristics adapted to the device. After acquiring new standard event data for each charge, the dynamic deviation of the charging characteristic baseline model is calculated. Based on the dynamic deviation of the parameters, the normal fluctuation of individual devices or hidden abnormalities of devices are judged, and corresponding alarms are triggered.

[0020] On the other hand, this application also proposes a multi-device compatible electric bicycle charging middleware management system, the system comprising: field edge nodes and cloud management nodes; The field edge node is used to receive network access requests from electric bicycle charging devices and perform identity verification, establish a communication link for the verified charging devices, continuously receive raw data frames from the charging devices, generate a unique feature fingerprint of the device event and compare it with the event feature fingerprint database, determine and call the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames, and output the raw structured event data of the device and the identifier of the effective semantic parsing model; when cloud communication is interrupted, it independently completes local data parsing, business judgment and charging control execution, and caches offline data and control logs. The cloud-based management and control node is used to adaptively normalize the original structured event data of the device by calling the corresponding standardized mapping rules according to the identifier of the semantic parsing model, and generate standard event data; extract business information from the standard event data and match the corresponding standardized business template through a similarity matching mechanism; automatically iterate and generate temporary business handling templates for working conditions not covered in the database; load the event response rules and execution nodes of the business template and generate management and control instructions to be sent to the field edge nodes.

[0021] In one embodiment, the control system further includes: the field edge node has a built-in local resource library and a local offline execution unit; The local resource library is used to locally store a trusted device whitelist, a global temporary blacklist, an event feature fingerprint library, a general semantic field library for electric bicycle charging, and the association and binding relationships of device hardware serial numbers, device event unique feature fingerprints, and semantic parsing models. The local offline execution unit is used to independently complete the parsing of the original data frames of the charging device, the judgment of device anomalies and the execution of hierarchical control actions by relying on the locally cached frame parsing rules, device binding relationships and historical control strategies when communication with the cloud control node is interrupted. It also caches illegal access logs, charging data and execution receipt messages offline. After communication is restored, it batch synchronizes all offline data to the cloud control node to complete unified archiving and iteration.

[0022] In one embodiment, the control system further includes: the cloud control node is configured with a cloud rule iteration unit, a baseline diagnosis unit, and a blockchain evidence storage unit; The cloud-based rule iteration unit has a built-in hierarchical security threshold library, a dynamic unit conversion coefficient pool, and a standardized business template library. The intelligent baseline diagnostic unit is used to collect standard event data of each charging of a single electric bicycle, fit and generate a baseline model of device charging characteristics, calculate the dynamic deviation of the current charging parameters in real time, distinguish between normal fluctuations in individual devices and hidden anomalies of devices, and trigger corresponding level alarms. The blockchain evidence storage unit is used to store evidence of illegal access logs, semantic model iteration records, data normalization processing records, business template update records, control execution receipts, and equipment hidden anomaly diagnosis records.

[0023] The beneficial effects of the embodiments in this application compared with the prior art are: This invention employs an edge-cloud collaborative adaptive self-learning architecture, overcoming the technical bottlenecks of traditional solutions that rely on manual adaptation and static protocol configuration. Addressing the challenges of complex and difficult adaptation of proprietary protocols across multiple brands of devices, this invention can autonomously complete protocol fingerprint matching, semantic self-learning, and iterative generation of parsing models, eliminating the need for manual development of adaptation programs and significantly reducing device access costs and timelines. Simultaneously, through adaptive data normalization and intelligent business template matching and iteration mechanisms, it overcomes the drawbacks of poor scalability and high platform maintenance costs associated with traditional middleware. Relying on device-specific charging characteristic baselines, it can accurately identify hidden device anomalies. Combined with edge offline autonomy and a full-process blockchain evidence storage mechanism, it effectively improves the stability, intelligence, and regulatory traceability of charging management, adapting to large-scale charging scenarios with mixed multi-device access. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of a multi-device compatible electric bicycle charging middleware management method provided in one embodiment of the present invention.

[0026] Figure 2 This is a schematic diagram of a multi-device compatible electric bicycle charging middleware management system provided in one embodiment of the present invention. Detailed Implementation

[0027] It should be understood that, when used in this application specification and the appended claims, the term includes indicating the presence of the described feature, integral, step, operation, element and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0028] It should also be understood that the terms used in this application specification and the appended claims refer to any combination of one or more of the associated listed items and all possible combinations, and include such combinations.

[0029] Example 1 like Figure 1 As shown, this application provides a method for managing middleware for charging electric bicycles that is compatible with multiple devices, the method comprising: This application provides a multi-device compatible electric bicycle charging middleware management method, applicable to scenarios involving unified access and intelligent security management of multi-brand, multi-proprietary protocol electric bicycle charging devices. For example... Figure 1 As shown, the management method for the multi-device compatible electric bicycle charging middleware specifically includes the following steps: The on-site edge node receives network access requests initiated by electric bicycle charging devices and performs identity verification. For electric bicycle charging devices that pass the verification, a communication link is established between the on-site edge node and the device. The edge node continuously receives the real-time raw data frames of the electric bicycle charging device uploaded by the device through the communication link, generates a unique feature fingerprint of the device event that is uniquely bound to the device protocol, compares and matches the unique feature fingerprint of the device event with the event feature fingerprint database, determines the semantic parsing model bound to the corresponding fingerprint, and calls the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames of the charging device, outputting the raw structured event data of the device and the identifier of the currently effective semantic parsing model to the cloud management node; The cloud-based management node uses the identifier of the currently active semantic parsing model to call the standardized mapping rules of the semantic parsing model to normalize the original structured event data of the device, generating standard event data that can be universally recognized by the middleware system. Based on the standard event data, business information is extracted, and a standardized business template is matched based on the business information. The template's preset event response rules and execution nodes are loaded according to the standardized business template and distributed to the field edge nodes to complete the management and control of electric bicycle charging. The business information includes event type, event level, associated device assets, and associated business node parameters.

[0030] In one specific embodiment, the device access identity verification process is as follows: the on-site edge node extracts the device hardware serial number, device protocol feature code, and device access key uploaded by the charging device to form device identity verification information. The device identity verification information is compared and verified with a locally pre-stored trusted device whitelist. If the identity verification fails, the on-site edge node refuses to establish a communication link, records an illegal access log locally and performs blockchain notarization, and pushes an illegal device alarm event to the cloud management node. The cloud management node simultaneously adds the device identity information to a global temporary blacklist and prohibits the device from initiating access requests again within a preset blocking period. If the identity verification passes, the on-site edge node establishes a communication link and globally associates and binds the device hardware serial number with the subsequently generated device event unique feature fingerprint and corresponding semantic parsing model. The association and binding relationship is synchronously cached at the on-site edge node and synchronously sent to the cloud management node for storage.

[0031] In one specific embodiment, the semantic parsing model matching and adaptive learning process is as follows: the edge node calculates the feature matching degree between the current device event unique feature fingerprint and each archived fingerprint in the event feature fingerprint database, and compares the matching degree with a preset matching threshold; if there is an archived fingerprint with a matching degree exceeding the preset matching threshold, the semantic parsing model bound to the corresponding fingerprint is retrieved; otherwise, the electric bicycle charging semantic self-learning is started, relying on the pre-stored general semantic field database, parsing the field definitions, transmission interaction timing, and parameter value range within the original data frame of the charging device, and generating a custom semantic parsing model adapted to the device's private protocol; simultaneously, a new device event unique feature fingerprint bound to the custom semantic parsing model is generated, and the new device event unique feature fingerprint is associated with the custom semantic parsing model and stored in the event feature fingerprint database.

[0032] The process includes: Field definition parsing: Edge nodes scan and parse each field of the raw data frames uploaded by the charging device, using field feature patterns (such as field length, data type, and numerical range) from a general semantic field library to identify and label the semantic meaning of each data field. Transmission interaction timing analysis: Edge nodes record and analyze the data interaction timing between the charging device and the edge node, including request-response intervals, heartbeat packet cycles, and event reporting trigger conditions, forming unique timing behavior characteristics for the device. Parameter numerical range learning: Edge nodes collect and statistically analyze the numerical distribution of each data field over a period of time, generating normal operating numerical ranges and abnormal fluctuation thresholds for each field, serving as a benchmark for subsequent data verification. Based on the comprehensive parsing results, the edge node generates a dynamic custom semantic parsing model adapted to the device's private protocol. This model is described in a structured format, such as JSON Schema or XML, recording the device's data frame format definition, field mapping relationships, timing rules, and parameter range constraints. After the custom semantic parsing model is generated, the edge node synchronously generates a unique feature fingerprint for newly added device events corresponding to this model. The fingerprint is generated based on information such as the device hardware serial number, initial access timestamp, and learned protocol signature, and is calculated using a hash algorithm, such as SHA-256 or SimHash, to ensure the fingerprint's uniqueness and stability. The edge node establishes an association between the newly added device event unique feature fingerprint and a custom semantic parsing model, and stores this association record in the event feature fingerprint database, completing the dynamic expansion of the fingerprint database. Simultaneously, this association is synchronously cached locally on the edge node according to the aforementioned global binding process and uploaded to the cloud management node for persistent storage, enabling real-time identification and reuse of new device protocols.

[0033] In one specific embodiment, the cloud-based data normalization process is as follows: the cloud-based management node calls the standardized mapping rules of the semantic parsing model based on the identifier of the currently effective semantic parsing model to perform field mapping adaptation, dynamic unit conversion, and data correction on the original structured event data of the device. Specifically, the field mapping adaptation is based on a preset charging standard field system, comparing each field in the original structured event data of the device obtained from the parsing of the field edge nodes. Fields with preset mapping relationships are directly aligned by name. For fields not covered by the current mapping relationship, semantic comparison and reasoning are performed through a general semantic lexicon to generate a temporary mapping relationship corresponding to that field. The dynamic unit conversion process converts voltage, current, battery temperature, and charging time parameters into standard units of measurement required by regulations according to selected conversion coefficients. The data correction process calls the graded safety threshold library to segment the converted parameters and corrects distorted parameters that deviate from the reasonable range. The standardized mapping rules are bound to the electric bicycle charging graded safety threshold library and the dynamic unit conversion coefficient pool, which can be switched and called according to the device model and local regulatory reporting standards.

[0034] Preferably, for instantaneous distortion parameters, the cloud-based control node uses linear interpolation or a weighted average of adjacent data points to correct them, in order to smooth data fluctuations; Preferably, for persistent deviation parameters, the cloud-based management node clamps the distortion value to the corresponding safety boundary value based on the upper and lower safety limits of the device model and battery type defined in the hierarchical safety threshold library, and generates a data anomaly alarm event, which is then pushed to the operation and maintenance monitoring platform for manual review.

[0035] The specific conversions include, but are not limited to: Voltage: Convert the raw voltage values ​​reported by the equipment (such as millivolts mV, volts V) into the standard units required by the regulator (such as volts V). Current: Convert the original current value (such as milliampere mA, ampere A) into a standard unit (such as ampere A); Battery temperature: Convert the original temperature values ​​(such as Fahrenheit °F, Celsius °C) to the units required by regulations (such as Celsius °C). Charging time: Convert the original time units (such as seconds s, minutes min) into standard units (such as minutes min or hours h).

[0036] The conversion factor is automatically matched based on the device model field or the region identifier of the access node carried when the device reports its information. For example, when the same device is deployed in different regions, the cloud can dynamically switch the unit conversion configuration according to the local regulatory standards to meet the differentiated requirements of regulatory reporting in multiple regions.

[0037] In one specific embodiment, the specific implementation process of adaptive matching and iteration of business templates is as follows: The cloud management node uses all extracted business information as matching retrieval elements, traverses the pre-configured standardized business template library, calculates the matching similarity between the current business information and the triggering conditions of each standardized business template in the library, filters out the standardized business template with the largest matching similarity value, and compares the maximum matching similarity with the preset business matching threshold; when the maximum matching similarity is greater than the preset business matching threshold, the standardized business template with the highest similarity is directly determined as the matching result for this time; otherwise, a temporary business handling template adapted to the current working condition is generated based on the currently extracted business information, the temporary business handling template is stored in the standardized business template library to complete the iterative update of the template library, and the temporary business handling template is determined as the standardized business template obtained in this matching.

[0038] After completing data normalization processing, the cloud-based management node extracts all business information for business template matching from the standardized event data and assembles it into a set of matching retrieval elements. These retrieval elements cover, but are not limited to, the following dimensions: Equipment dimensions: equipment model, equipment manufacturer, battery type (lead-acid / lithium battery), battery rated capacity, and equipment access region; Charging conditions: current charging stage (constant current / constant voltage / trickle current), real-time voltage and current values, battery temperature, charging time, and remaining charge (SOC). Environmental dimensions: ambient temperature, humidity, and type of charging location (indoor / outdoor / underground parking garage). Time dimension: charging initiation time (peak / off-peak / low-peak), seasonal information.

[0039] The above elements are organized in the form of structured key-value pairs, which serve as input parameters for subsequent template matching.

[0040] In one specific embodiment, based on the currently extracted set of search elements, and combined with the system's built-in business rule primitive library and security baseline policy, a temporary business processing template adapted to the current operating conditions is dynamically generated. Specific steps include: Rule primitive assembly: Based on key dimensions such as device type, battery type, and charging stage in the search elements, the cloud management node selects applicable basic rule fragments (such as the current upper limit rule for lithium battery constant current stage, the over-temperature current reduction rule, and the charging timeout protection rule) from the business rule primitive library. Adaptive parameter calculation: Based on the specific parameter values ​​of the current operating conditions (such as battery rated capacity and real-time temperature), the threshold parameters in the rule primitives are adaptively adjusted. For example, a reasonable upper limit for charging current is dynamically calculated based on the battery capacity, and the over-temperature protection threshold is corrected based on the ambient temperature. Template structured encapsulation: The assembled rule set and corresponding trigger conditions are structured and encapsulated to generate a complete temporary business processing template, and a temporary template identifier and version number are assigned.

[0041] After the temporary business processing template is generated, the cloud management node performs the following operations: Iterative Input: This temporary business processing template is added as a new record to the standardized business template library, completing the dynamic iterative update of the template library. The triggering conditions for the new template are directly taken from the current matching and retrieval element set, and the execution strategy is the generated rule set. At the same time, it is marked as "pending review" for subsequent manual confirmation and optimization by business experts. Result confirmed: This temporary service processing template will be selected as the standardized service template obtained from this matching and will be directly used for the service processing of the current charging session.

[0042] Through the aforementioned adaptive matching and iteration mechanism, the standardized business template library can be continuously enriched and improved during system operation.

[0043] In one specific embodiment, the specific implementation process of issuing control commands and executing them in a closed loop is as follows: the cloud control node loads the preset event response rules and corresponding execution nodes inside the template, and encapsulates the device identity identifier and event tracing credentials associated with this business into a control command message and sends it to the field edge node; The on-site edge nodes execute corresponding control actions in layers according to the execution node: if the execution node points to the currently connected electric bicycle charging equipment, it issues on / off control, power adjustment, and start / stop charging commands to the charging equipment according to the event response rules; if the execution node points to the background alarm push node, it pushes alarm information of the corresponding level in sequence according to the preset push hierarchy; after the on-site edge nodes have executed all control actions, they collect the execution results of this action, generate an execution receipt message, and send it back to the cloud control node. Specifically: First level: Direct control on the equipment side; If the execution node points to the currently connected electric bicycle charging device, the edge node sends corresponding control commands to the charging device through the device communication link according to the control action type defined in the event response rules, including but not limited to: On / off control: When a charging abnormality is detected (such as over-temperature, over-voltage, or short circuit), a relay disconnect command is issued to immediately cut off the charging circuit and ensure charging safety; after the abnormality is recovered and a safety verification is performed, a closing command is issued to restore power supply. Power regulation: Based on the grid load status or battery charging stage demand, power regulation commands are issued to adjust the output power of the charging pile, thereby achieving dynamic load management; Start or stop charging: Issue charging start or stop commands according to business rules (such as insufficient user balance or the scheduled charging time).

[0044] The command interaction between edge nodes and charging devices follows the device's private communication protocol. Before issuing commands, edge nodes convert standardized commands into a raw command format that the device can recognize, based on the bound semantic parsing model.

[0045] Second level: Background alarm push; If the execution node points to the background alarm push node, the edge node will push alarm information sequentially according to the preset push hierarchy in the event response rules. The preset push hierarchy is based on the alarm severity level: Level 1 Alarm (Severe): Pushes notifications to maintenance staff via SMS / phone, community property manager, and cloud-based maintenance dashboard, triggering audible and visual alarms; Level 2 Alarm (Warning): Pushed to the operation and maintenance monitoring platform and community property management system, generating a work order to be processed; Level 3 Alarm (Notification): Pushed to system logs and mobile notification bar for maintenance personnel as reference information.

[0046] Edge nodes execute alarm pushes step by step in the above hierarchical order. After each level push is successful, the execution status is recorded. If a level push fails, the edge node starts a retry mechanism. If the maximum number of retries is reached and the failure still fails, the failure status is recorded and reported to the cloud.

[0047] In one specific embodiment, the method further includes a device-specific baseline fitting and latent anomaly identification step: the cloud management node continuously collects standard event data of each charging process of the same electric bicycle charging device, and fits and generates a charging characteristic baseline model adapted to the device; after acquiring new standard event data for each charging, the dynamic deviation of the parameters of the charging characteristic baseline model is calculated, and the device's normal individual variation fluctuation or latent anomaly is judged based on the dynamic deviation of the parameters, and the corresponding level of alarm is triggered.

[0048] In one specific embodiment, the construction of the baseline model includes: extracting local fluctuation features of the charging current curve using a CNN, capturing long-term temporal dependencies using a BiLSTM network, and focusing on key abnormal periods through a multi-head self-attention mechanism to form a multi-dimensional feature vector; using the MiniBatchKMeans algorithm to cluster historical charging feature vectors to form multiple pattern clusters representing the normal charging behavior of the device. The cluster center represents a typical charging pattern, and the cluster radius, such as based on the 95th percentile distance, defines the normal fluctuation boundary of the pattern. The set of cluster centers formed by clustering, the radius of each cluster, and the corresponding charging stage label are persistently stored as the charging feature baseline model of the device and periodically incrementally updated with each collection of new charging data.

[0049] In one specific embodiment, calculating the dynamic deviation of the parameters of the charging feature baseline model includes: calculating the distance between the current charging feature vector and the centers of each cluster in the baseline model, and determining its nearest cluster. Based on the radius of the cluster and the dynamic weight allocation strategy, a comprehensive anomaly score is calculated. Dynamic weights assign higher weights to feature dimensions with smaller fluctuations, making minor deviations in these dimensions easier to detect. If the current feature vector is within the radius of a cluster, but some key dimensions (such as the current decrease rate during the constant voltage phase) continuously deviate from the historical average within the cluster, and the weighted deviation exceeds a preset intra-cluster anomaly threshold, it is determined to be a normal fluctuation in the device's individual characteristics, such as a slow change in the charging curve caused by natural battery aging. Such fluctuations are recorded for incremental updates to the baseline model but do not trigger alarms. If the current feature vector is far from all normal cluster centers, exceeding the cluster radius, and the anomaly score exceeds the global threshold, it is determined to be a latent device anomaly. This anomaly may indicate safety hazards such as internal battery short circuits, precursors to thermal runaway, or charger malfunctions.

[0050] When a device is identified as having a latent anomaly, the cloud-based control node triggers an alarm event of the corresponding level based on the severity of the deviation: Minor deviation: Pushed to the operation and maintenance monitoring platform as a reminder, recorded in the device health record for subsequent trend analysis.

[0051] Moderate deviation: Push to the mobile terminal of maintenance personnel, generate a work order to be checked, and suggest conducting on-site equipment inspection.

[0052] Severe deviation: If the abnormal score continues to rise or exceeds the emergency threshold, an emergency alarm is triggered and pushed to the operation and maintenance personnel and community property managers. At the same time, the control command issuance process is linked to issue power adjustment or on / off control commands to edge nodes to implement proactive safety protection.

[0053] Through the aforementioned personalized baseline fitting and dynamic deviation analysis mechanism, the system can effectively avoid false alarms and missed alarms based on a unified tangent for the group, accurately identify the unique aging trajectory and abnormal signs of each device, and achieve a safety management upgrade from passive response to proactive early warning.

[0054] Example 2: like Figure 2 As shown, this application also proposes a multi-device compatible electric bicycle charging middleware management system, the system comprising: field edge nodes and cloud management nodes; The field edge node is used to receive network access requests from electric bicycle charging devices and perform identity verification, establish a communication link for the verified charging devices, continuously receive raw data frames from the charging devices, generate a unique feature fingerprint of the device event and compare it with the event feature fingerprint database, determine and call the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames, and output the raw structured event data of the device and the identifier of the effective semantic parsing model; when cloud communication is interrupted, it independently completes local data parsing, business judgment and charging control execution, and caches offline data and control logs. The cloud-based management and control node is used to adaptively normalize the original structured event data of the device by calling the corresponding standardized mapping rules according to the identifier of the semantic parsing model, and generate standard event data; extract business information from the standard event data and match the corresponding standardized business template through a similarity matching mechanism; automatically iterate and generate temporary business handling templates for working conditions not covered in the database; load the event response rules and execution nodes of the business template and generate management and control instructions to be sent to the field edge nodes.

[0055] In one embodiment, the control system further includes: the field edge node has a built-in local resource library and a local offline execution unit; The local resource library is used to locally store a trusted device whitelist, a global temporary blacklist, an event feature fingerprint library, a general semantic field library for electric bicycle charging, and the association and binding relationships of device hardware serial numbers, device event unique feature fingerprints, and semantic parsing models. The local offline execution unit is used to independently complete the parsing of the original data frames of the charging device, the judgment of device anomalies and the execution of hierarchical control actions by relying on the locally cached frame parsing rules, device binding relationships and historical control strategies when communication with the cloud control node is interrupted. It also caches illegal access logs, charging data and execution receipt messages offline. After communication is restored, it batch synchronizes all offline data to the cloud control node to complete unified archiving and iteration.

[0056] In one embodiment, the control system further includes: the cloud control node is configured with a cloud rule iteration unit, a baseline diagnosis unit, and a blockchain evidence storage unit; The cloud-based rule iteration unit has a built-in hierarchical security threshold library, a dynamic unit conversion coefficient pool, and a standardized business template library. The intelligent baseline diagnostic unit is used to collect standard event data of each charging of a single electric bicycle, fit and generate a baseline model of device charging characteristics, calculate the dynamic deviation of the current charging parameters in real time, distinguish between normal fluctuations in individual devices and hidden anomalies of devices, and trigger corresponding level alarms. The blockchain evidence storage unit is used to store evidence of illegal access logs, semantic model iteration records, data normalization processing records, business template update records, control execution receipts, and equipment hidden anomaly diagnosis records.

[0057] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for managing and controlling a charging middleware for multi-device compatible electric bicycles, characterized in that, The method includes: The on-site edge node receives network access requests initiated by electric bicycle charging devices and performs identity verification. For electric bicycle charging devices that pass the verification, a communication link is established between the on-site edge node and the device. The edge node continuously receives the real-time raw data frames of the electric bicycle charging device uploaded by the device through the communication link, generates a unique feature fingerprint of the device event that is uniquely bound to the device protocol, compares and matches the unique feature fingerprint of the device event with the event feature fingerprint database, determines the semantic parsing model bound to the corresponding fingerprint, and calls the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames of the charging device, outputting the raw structured event data of the device and the identifier of the currently effective semantic parsing model to the cloud management node; The cloud-based management node uses the identifier of the currently active semantic parsing model to call the standardized mapping rules of the semantic parsing model to normalize the original structured event data of the device, generating standard event data that can be universally recognized by the middleware system. Business information is extracted from the standard event data, and a standardized business template is matched based on the business information. The template's preset event response rules and execution nodes are loaded according to the standardized business template and sent to the field edge nodes to complete the management and control of electric bicycle charging. The business information includes event type, event level, associated device assets, and associated business node parameters.

2. The method for managing and controlling a multi-device compatible electric bicycle charging middleware according to claim 1, characterized in that, The on-site edge node receives network access requests initiated by electric bicycle charging devices and performs identity verification, including: The edge node extracts the device hardware serial number, device protocol feature code and device access key uploaded by the charging device to form device identity verification information. The device identity verification information is then compared and verified with the locally stored trusted device whitelist. If identity verification fails, the edge node refuses to establish a communication link, records the illegal access log locally and performs blockchain notarization, and pushes an illegal device alarm event to the cloud management node. The cloud management node then adds the device's identity information to the global temporary blacklist and prohibits the device from initiating access requests again within the preset blocking time. If the identity verification is successful, the on-site edge node establishes a communication link and globally associates and binds the device's hardware serial number with the subsequently generated unique feature fingerprint of the device event and the corresponding semantic parsing model. The association and binding relationship is synchronously cached to the on-site edge node and synchronously sent to the cloud management node for storage.

3. The method for managing and controlling a multi-device compatible electric bicycle charging middleware according to claim 2, characterized in that, The step of comparing and matching the unique feature fingerprint of a device event with the event feature fingerprint database to determine the semantic parsing model bound to the corresponding fingerprint includes: The edge nodes on site calculate the feature matching degree between the unique feature fingerprint of the current device event and each archived fingerprint in the event feature fingerprint database, and compare the matching degree with the preset matching threshold. If there is an archived fingerprint with a matching degree exceeding the preset matching threshold, the semantic parsing model bound to the corresponding fingerprint will be retrieved. Otherwise, initiate semantic self-learning for electric bicycle charging, relying on a pre-stored general semantic field library to parse the field definitions, transmission interaction timing, and parameter value ranges within the original data frame of the charging device, and generate a custom semantic parsing model adapted to the device's private protocol; simultaneously generate a unique feature fingerprint of the newly added device event bound to the custom semantic parsing model, and associate the unique feature fingerprint of the newly added device event with the custom semantic parsing model and store it in the event feature fingerprint library.

4. The method for managing a multi-device compatible electric bicycle charging middleware according to claim 3, characterized in that, The cloud-based management node calls the standardized mapping rules of the semantic parsing model based on the identifier of the currently effective semantic parsing model to normalize the original structured event data of the device, including: field mapping adaptation processing, dynamic unit conversion processing, and data correction processing. The field mapping adaptation process includes comparing each field in the original structured event data of the device obtained by parsing the field edge nodes with a preset charging standard field system as a benchmark; directly aligning the names of fields with preset mapping relationships; and generating temporary mapping relationships for fields not covered by the current mapping relationship by performing semantic comparison and reasoning through a general semantic dictionary. The dynamic unit conversion process includes converting voltage, current, battery temperature, and charging time parameters into standard units of measurement required by regulations according to the selected conversion factor. The data correction process includes calling the hierarchical safety threshold library to perform segmented judgment on the converted parameters and correcting the distorted parameters that deviate from the reasonable range; The standardized mapping rules are linked to a safety threshold library for electric bicycle charging grading and a dynamic unit conversion coefficient pool; the dynamic unit conversion coefficient pool can be switched and called according to the equipment model and the local regulatory reporting standards.

5. A method for managing and controlling a multi-device compatible electric bicycle charging middleware according to claim 4, characterized in that, The step of extracting business information based on the standard event data and matching the business information to obtain a standardized business template includes: All extracted business information is used as matching and retrieval elements. The pre-configured standardized business template library is traversed, and the matching similarity between the current business information and the triggering conditions of each standardized business template in the library is calculated one by one. The standardized business template with the highest matching similarity value is selected and compared with the preset business matching threshold for judgment. When the maximum matching similarity is greater than the preset business matching threshold, the standardized business template with the highest similarity is directly determined as the matching result for this time; Otherwise, a temporary business handling template adapted to the current working conditions is generated based on the currently extracted business information. This temporary business handling template is then stored in the standardized business template library to complete the iterative update of the template library, and the temporary business handling template is determined as the standardized business template obtained in this matching.

6. A method for managing and controlling a multi-device compatible electric bicycle charging middleware according to claim 5, characterized in that, Based on the standardized business template, the template's preset event response rules and execution nodes are loaded and distributed to the field edge nodes to complete this electric bicycle charging management and control, including: The cloud-based control node loads the preset event response rules and corresponding execution nodes within the template, and encapsulates the device identity identifier and event tracing credentials associated with this business into a control instruction message and sends it to the field edge node. The edge nodes on site execute corresponding control actions according to the division of execution nodes: if the execution node points to the currently connected electric bicycle charging equipment, it sends on / off control, power adjustment, and start / stop charging commands to the charging equipment according to the event response rules; if the execution node points to the background alarm push node, it pushes the corresponding level of alarm information in sequence according to the preset push level. After the edge node completes all control actions, it collects the results of the action execution, generates an execution receipt message, and sends it back to the cloud control node.

7. A method for managing a multi-device compatible electric bicycle charging middleware according to any one of claims 1-6, characterized in that, Also includes: For the same electric bicycle charging device, the cloud-based management node continuously collects standard event data of each charging process and fits and generates a baseline model of charging characteristics adapted to the device. After acquiring new standard event data for each charge, the dynamic deviation of the charging characteristic baseline model is calculated. Based on the dynamic deviation of the parameters, the normal fluctuation of individual devices or hidden abnormalities of devices are judged, and corresponding alarms are triggered.

8. A control system for a multi-device compatible electric bicycle charging middleware control method according to any one of claims 1-7, characterized in that, The system includes: on-site edge nodes and cloud-based control nodes; The field edge node is used to receive network access requests from electric bicycle charging devices and perform identity verification, establish a communication link for the verified charging devices, continuously receive raw data frames from the charging devices, generate a unique feature fingerprint of the device event and compare it with the event feature fingerprint database, determine and call the frame parsing rules of the semantic parsing model to complete the parsing of the raw data frames, and output the raw structured event data of the device and the identifier of the effective semantic parsing model; when cloud communication is interrupted, it independently completes local data parsing, business judgment and charging control execution, and caches offline data and control logs. The cloud-based management and control node is used to adaptively normalize the original structured event data of the device by calling the corresponding standardized mapping rules according to the identifier of the semantic parsing model, and generate standard event data; extract business information from the standard event data and match the corresponding standardized business template through a similarity matching mechanism; automatically iterate and generate temporary business handling templates for working conditions not covered in the database; load the event response rules and execution nodes of the business template and generate management and control instructions to be sent to the field edge nodes.

9. The control system according to claim 8, characterized in that, Also includes: The field edge node has a built-in local resource library and a local offline execution unit; The local resource library is used to locally store a trusted device whitelist, a global temporary blacklist, an event feature fingerprint library, a general semantic field library for electric bicycle charging, and the association and binding relationships of device hardware serial numbers, device event unique feature fingerprints, and semantic parsing models. The local offline execution unit is used to independently complete the parsing of the original data frames of the charging device, the judgment of device anomalies and the execution of hierarchical control actions by relying on the locally cached frame parsing rules, device binding relationships and historical control strategies when communication with the cloud control node is interrupted. It also caches illegal access logs, charging data and execution receipt messages offline. After communication is restored, it batch synchronizes all offline data to the cloud control node to complete unified archiving and iteration.

10. The control system according to claim 8, characterized in that, Also includes: The cloud-based management node is equipped with a cloud-based rule iteration unit, a baseline diagnostic unit, and a blockchain evidence storage unit. The cloud-based rule iteration unit has a built-in hierarchical security threshold library, a dynamic unit conversion coefficient pool, and a standardized business template library. The intelligent baseline diagnostic unit is used to collect standard event data of each charging of a single electric bicycle, fit and generate a baseline model of device charging characteristics, calculate the dynamic deviation of the current charging parameters in real time, distinguish between normal fluctuations in individual devices and hidden anomalies of devices, and trigger corresponding level alarms. The blockchain evidence storage unit is used to store evidence of illegal access logs, semantic model iteration records, data normalization processing records, business template update records, control execution receipts, and equipment hidden anomaly diagnosis records.

Images