A mobile phone OTG load fingerprint identification adaptive boost control method and system

Abstract

The application discloses a mobile phone OTG self-adaptive boost control method and system based on load fingerprint identification and belongs to the technical field of power management. In view of the problem that when the mobile phone OTG reversely supplies power, the traditional boost control cannot adapt to the transient power consumption of heterogeneous peripherals and voltage drop is prone to occurring, the application proposes that after the mobile phone and the peripheral complete USB PD protocol negotiation and establish basic power supply, electrical fluctuation data in the standby state of the peripheral is collected, load electrical characteristic fingerprints are constructed in combination with E-Mark information and PD application parameters; the type of the peripheral is identified by using a lightweight neural network, and a typical power consumption atlas in the cloud is called; and the transient response parameters of the boost converter are dynamically configured according to this, including the current ramping rate and loop compensation, so that adaptive boost output is realized. The application can effectively inhibit the peripheral starting transient voltage drop and improve the OTG power supply stability and conversion efficiency.

Description

Technical Field

[0001] This invention belongs to the field of power management technology, specifically relating to a mobile phone OTG adaptive boost control method and system based on load fingerprint recognition. It is particularly suitable for reverse power supply scenarios where the transient response characteristics of the boost converter are predictively optimized when the mobile phone is connected to heterogeneous loads such as mechanical hard drives, solid-state drives, and cooling fans through the OTG interface. Background Technology

[0002] Currently, reverse power supply from mobile phones to peripherals via the OTG interface has become a mainstream function. The introduction of the USB PD protocol allows mobile phones and peripherals to negotiate output voltage and current levels (such as 5V / 9V), meeting the basic power supply needs of external hard drives, fans, and other peripherals.

[0003] However, in real-world scenarios where mobile OTG connects to heterogeneous loads such as external mechanical hard drives, solid-state drives, and cooling fans, existing solutions still have limitations. Firstly, the USB PD protocol addresses the negotiation and switching of steady-state voltage and current, failing to detect the dynamic power consumption surges during peripheral startup. The starting of a mechanical hard drive's read / write head and the acceleration of the fan motor generate millisecond-level large current transients. Since boost converters use fixed current ramp-up rates and loop response parameters, they struggle to respond promptly to such sudden load changes, leading to output voltage drops, peripheral startup failures, or repeated disconnections.

[0004] Secondly, existing solutions do not fully utilize the electrical characteristics of peripherals under basic power supply for load identification. After completing PD negotiation with the peripheral, the mobile phone can only obtain the requested voltage and current parameters, and cannot further identify the true type of the peripheral. Different loads (such as hard drives, fans, and mobile phones) have significantly different current fluctuation characteristics in standby mode, but existing technologies do not take advantage of this.

[0005] In addition, traditional control methods lack support from typical power consumption data of peripherals. The dynamic response parameters of boost converters are all fixed configurations, which cannot be customized for different models of peripherals, resulting in poor compatibility with heterogeneous loads, low conversion efficiency, and severe overheating on mobile devices.

[0006] Therefore, how to further perceive the load type and transient power consumption characteristics of peripherals based on PD protocol negotiation, and optimize the transient response parameters of the boost converter accordingly to suppress startup voltage drops, improve power supply stability and conversion efficiency, has become a technical problem that needs to be solved in the current OTG reverse power supply management of mobile terminals. Summary of the Invention

[0007] In a first aspect, the embodiments disclosed in this invention provide a mobile phone OTG load fingerprint recognition adaptive boost control method, comprising the following steps: After the mobile phone and OTG peripheral complete the USB PD protocol negotiation and establish basic power supply, obtain the E-mark electronic tag information and voltage and current application parameters of the connected peripheral. During the basic power supply maintenance period, the voltage and current of the OTG boost output terminal are continuously monitored at a sampling frequency of not less than 1kHz, and electrical fluctuation data of peripheral devices in standby state are collected to generate load electrical feature fingerprints. A lightweight neural network is used to fuse the E-mark information, voltage and current application parameters, and load electrical feature fingerprint to identify the type of peripheral device; Based on the identification results, retrieve the typical power consumption graph of the corresponding peripheral model from the cloud database; Based on the typical power consumption spectrum, the transient response parameters of the boost converter are dynamically configured to achieve adaptive boost output control.

[0008] Preferably, the step of acquiring the E-mark electronic tag information of the access peripheral, voltage and current application parameters, and generating the load electrical feature fingerprint specifically includes: The E-mark electronic tag information of the peripheral device is obtained through the CC logic detection unit, and the VID, PID, device type code and cable impedance are extracted. The voltage and current request parameters of the peripheral device are obtained through the PD protocol parsing module; After the basic power supply is established, the voltage and current of the OTG boost output terminal are continuously collected at a sampling frequency of not less than 1kHz for a duration of 200ms, generating a standby electrical fluctuation time series dataset containing timestamps.

[0009] Preferably, the step of fusing the E-mark information, voltage and current application parameters, and load electrical feature fingerprint using a lightweight neural network specifically includes: The collected raw data is cleaned, time-series aligned, and format-normalized. Extract load fingerprint feature indicators, which include electrical configuration features, electrical fluctuation pattern features, and protocol response features; input the load fingerprint feature indicators into a pre-set lightweight neural network model for forward inference, and output the peripheral type attribution probability value; If the classification confidence score is greater than the preset threshold, the classification result is deemed valid; if the classification confidence score is less than the preset threshold, it is determined to be an unknown heterogeneous peripheral and the default boost protection strategy is triggered.

[0010] Preferably, the electrical fluctuation morphological characteristics include standby current fluctuation variance, standby current peak value, standby current mean value, and fluctuation period; The protocol response characteristics include E-mark response latency, PD request response latency, and protocol handshake success rate.

[0011] Preferably, retrieving the typical power consumption graph of the corresponding peripheral model from the cloud database includes: The peripheral device's VID, PID, and classification results are encapsulated into an HTTP GET request and sent to the cloud; The cloud executes a dual matching algorithm to filter out typical power consumption profiles for the corresponding models and distributes them. The typical power consumption profiles include the startup transient power timing curve, steady-state operating power, peak current threshold, and power fluctuation period. If the local L2 cache is hit, the power consumption map in the cache is directly called; if the cloud interaction is abnormal, the local pre-stored general power consumption map of the same type is called.

[0012] Preferably, the dynamic configuration of the transient response parameters of the boost converter based on typical power consumption patterns specifically includes: For mechanical hard disk peripherals, based on their startup transient high current power consumption spectrum, high-speed current ramp-up and high-gain loop compensation parameters are configured; For solid-state drive peripherals, based on their steady-state power characteristics, medium-speed current ramping and medium-level loop compensation are configured. For cooling fan peripherals, low-speed current ramp-up and low-level ripple suppression are configured; For smartphone peripherals, match the output voltage according to the PD protocol and configure the output current limit value.

[0013] Preferably, after implementing adaptive boost output control, it also includes: Real-time acquisition of boost output operating status data to construct a power supply status time-series dataset; When abnormal operating parameters are detected, hierarchical causal inference diagnosis is performed based on load fingerprint characteristics, typical power consumption graphs in the cloud, and real-time operating data, and the root cause conclusion of the fault is output.

[0014] Secondly, embodiments of the present invention also provide a mobile phone OTG load fingerprint recognition adaptive boost control system, comprising: The OTG multi-source data acquisition module is used to acquire the E-mark electronic tag information and voltage and current application parameters of the connected peripheral after the mobile phone and the peripheral have completed the USB PD protocol negotiation and established basic power supply, and continuously collect the electrical fluctuation data of the peripheral in standby mode to generate load electrical feature fingerprint. The load fingerprint feature calculation module is used to clean and normalize the collected raw data, and extract electrical configuration features, electrical fluctuation pattern features and protocol response features. The lightweight neural network classification and cloud interaction module has a built-in lightweight neural network model, which is used to identify the type of peripheral device based on the load fingerprint characteristics, and retrieve the typical power consumption spectrum of the corresponding peripheral model in the cloud database according to the type. An adaptive boost control execution module is used to generate boost control commands based on the typical power consumption spectrum and configure the transient response parameters of the boost converter, including current ramp rate, loop compensation parameters and ripple suppression strength. The power supply monitoring and fault cause diagnosis module is used to monitor the power supply status in real time and locate the root cause of the fault based on map comparison and feature verification when an anomaly occurs.

[0015] Preferably, the system is integrated into a mobile phone power management chip, an OTG adapter control chip, or a mobile phone edge computing unit.

[0016] This invention has the following advantages: When the peripheral device is in standby mode, the electrical characteristic fingerprint of the load is continuously monitored to identify the type of peripheral device and its transient power consumption requirements. Based on this, the transient response parameters of the boost converter are configured, which can solve the problem of voltage drop caused by response lag when the mechanical hard drive starts up and improve the stability of OTG power supply. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the steps of a mobile phone OTG adaptive boost control method provided in an embodiment of the present invention; Figure 2 This is a structural diagram of a mobile phone OTG adaptive boost control system provided in an embodiment of the present invention. Detailed Implementation

[0019] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings to make the technical solution of the present invention clearer and more complete. It should be noted that the described embodiments are for illustrative purposes only and are not intended to limit the present invention. Other implementation methods that can be made by those skilled in the art based on the content of the present invention without creative effort should all fall within the protection scope of the present invention.

[0020] In this invention, unless otherwise expressly stated, the terms "comprising," "including," and similar expressions should be understood to indicate the presence of the listed items without excluding the presence or addition of other items. The words "an," "an," and similar terms should not be construed as limited to the singular in this invention and may include multiple items.

[0021] Furthermore, the accompanying drawings in this invention are merely illustrative and not necessarily drawn to scale. The same reference numerals denote components with the same or similar functions. To clearly illustrate this invention, specific details are provided in the following embodiments. Those skilled in the art should understand that these details are not essential for implementing this invention, and other methods can be used to implement it without affecting the basic idea of ​​the invention.

[0022] The overall flowchart of the method of the present invention is as follows: Figure 1 As shown in the attached diagram, the following is a detailed explanation: I. Steps and Procedures for Mobile OTG Adaptive Boost Control This embodiment uses a smartphone equipped with a Type-C OTG interface as the implementation scenario. The phone has a built-in power management chip (PMIC), OTG protocol processing unit, CC logic detection unit, high-frequency electrical sampling circuit, and edge computing unit. It can connect to four types of heterogeneous loads: a 2.5-inch mechanical hard drive, a solid-state drive, a portable cooling fan, and the smartphone itself. A cloud server deploys a database of typical power consumption patterns for peripheral devices, and the phone interacts with the cloud via Wi-Fi or cellular network.

[0023] S1: OTG Multi-Source Data Acquisition and Standby Electrical Fluctuation Monitoring (1) Peripheral device identification and electrical parameter acquisition Peripheral identity and electrical parameters are obtained through the CC logic detection unit and PD protocol parsing module of the Type-C OTG interface. Specifically, this includes E-mark electronic tag information and peripheral voltage and current request parameters. The acquisition process follows the timing rules of USB PD and OTG protocols, and data extraction is completed using standardized command interaction and protocol parsing algorithms.

[0024] ①Data source and metering points: The communication data from the CC pin of the Type-C OTG interface is used as the data source for peripheral identification information, which is used to obtain standardized data of peripheral E-mark electronic tags.

[0025] The electrical signal from the OTG boost output terminal is used as the data source for electrical parameters to collect voltage and current request data from peripherals. The metering point is the electrical signal acquisition node at the OTG interface output terminal.

[0026] ② Data collection and uploading: When the phone's OTG interface detects a connected peripheral device, it first completes basic negotiation according to the USB PD protocol to establish basic power supply (e.g., 5V / 0.9A). Based on this, the peripheral device's identity information and electrical request parameters are obtained through the CC logic detection unit and the PD protocol parsing module.

[0027] For E-mark electronic tag data: Based on the protocol parsing algorithm, the response data returned by the peripheral device is parsed in frames to extract the VID supplier ID, PID product ID, device type code, cable impedance, rated voltage, and rated current parameters, and generate a structured peripheral device identity dataset.

[0028] For voltage and current request parameters: using the PD protocol data packet capture algorithm, the Request packets sent by the peripheral device are parsed to extract the target output voltage, maximum operating current, and OTG power supply mode parameters, and a standardized electrical request dataset is generated.

[0029] A time-series alignment algorithm is used to add millisecond-level timestamps to the two types of data mentioned above, and the data is cached in the local RAM buffer to complete the collection and preprocessing of identity and electrical parameters.

[0030] (2) Peripheral device standby state electrical fluctuation data acquisition Standby electrical fluctuation data is acquired through the high-frequency electrical sampling unit at the OTG boost output terminal. A sampling algorithm with fixed duration and fixed frequency is used to collect voltage and current timing data of peripherals in standby mode, which is used to construct the electrical feature fingerprint of the load.

[0031] ①Data source and collection trigger: The sampling is triggered when the peripheral device reaches a stable state after completing basic power supply. After the mobile phone and the peripheral device complete USB PD protocol negotiation and establish basic power supply, wait for the peripheral device to enter standby mode (usually 10ms-50ms after power supply stabilizes), and immediately start the sampling process to ensure that the collected data covers the electrical fluctuations of the peripheral device in standby mode.

[0032] The voltage and current real-time electrical signals at the OTG boost output terminal are used as the data source for acquisition, and the two signals are acquired simultaneously to ensure timing consistency.

[0033] ② Data collection and uploading: A preset high-frequency sampling algorithm is adopted, with a sampling frequency of not less than 1kHz. A fixed sampling duration (e.g., 200ms) is used as a complete sampling cycle, and the instantaneous values ​​of voltage and current of peripheral devices in standby state are collected at equal time intervals.

[0034] Based on the electrical signal conversion algorithm, the sampled digital voltage is converted into the actual voltage value; based on the Ohm's law algorithm, the instantaneous current value is calculated from the sampled voltage signal.

[0035] Based on the instantaneous power calculation algorithm, the instantaneous power is calculated in real time according to the formula P=U×I, generating a standby electrical fluctuation time series dataset containing timestamps, instantaneous voltage, instantaneous current, and instantaneous power.

[0036] The aforementioned time-series dataset is transmitted to the data preprocessing unit for standardization processing.

[0037] Furthermore, to eliminate interference from internal mobile phone radio frequency signals, CPU high-frequency switching, or power supply ripple on the sampling data and ensure the accuracy of load fingerprint feature extraction, this embodiment introduces an anti-interference processing mechanism in the acquisition link. Specifically, at the hardware level, an RC low-pass filter circuit is configured at the ADC sampling front end to filter out high-frequency noise; at the software level, the acquired raw time-series data is subjected to noise reduction processing using a moving average filtering algorithm or a Kalman filtering algorithm to extract stable electrical feature components.

[0038] (3) Unified data entry and standardization All collected raw data are integrated into a unified data access unit. A three-step standardization algorithm, consisting of data cleaning, time-series alignment, and format normalization, is used to eliminate heterogeneous data differences and provide a standardized data source for subsequent load fingerprint feature calculation.

[0039] The data access unit receives E-mark identity data, electrical application parameters, and power fluctuation timing data in a unified manner through a built-in cache queue, ensuring the orderly access of data.

[0040] The data standardization unit performs the following standardization algorithm: Outlier removal algorithm: using The principle is to identify and remove outliers in the sampled data to ensure data validity; Missing value completion algorithm: For missing data points in the sampling process, a linear interpolation algorithm is used to complete the completion, ensuring the continuity of time series data; Time sequence alignment algorithm: calibrate the timestamps of all data to the mobile phone system UTC time to achieve time sequence alignment of multi-source data; Format normalization algorithm: Converts heterogeneous data into a unified JSON structured format, thus achieving data format normalization.

[0041] The standardized data is written to the local time-series buffer according to a time-series structure, and the data organization rules are as follows: Peripheral identity data is stored in the device_identity node, which includes the fields vid, pid, device_type, and cable_impedance; Electrical application data is stored in the electric_apply node and includes the fields target_voltage, max_current, and otg_mode. Power fluctuation data is stored in the power_fluctuation node, which contains the timestamp, voltage, current, and power fields.

[0042] S2: Load fingerprint feature index calculation This step uses S1 to collect and standardize multi-source data. Through fixed time-series processing rules, morphological feature extraction rules, and response statistics rules, it performs full-quantitative analysis of the peripheral's electrical parameter sequence and startup power fluctuation sequence, extracting three types of load fingerprint feature indicators that uniquely identify the peripheral type. This step is executed automatically immediately after the OTG protocol handshake is completed and the startup transient sampling ends. The analysis window is strictly limited to 200 milliseconds throughout the peripheral startup process to ensure the consistency and reproducibility of the fingerprint features.

[0043] (1) Data Preprocessing This step follows a fixed data processing flow to standardize the data output by S1, eliminating temporal misalignment, missing data, and abnormal interference, thus providing stable, continuous, and reliable input data for subsequent feature calculations.

[0044] First, retrieve the complete acquisition data corresponding to the current peripheral from the local timing buffer. The data includes E-mark identity resolution data, peripheral voltage and current request parameters, and power fluctuation timing data within a 200-millisecond startup cycle.

[0045] ① Data alignment and integration According to the timestamp matching rules, the voltage time series data, current time series data, and power time series data are bound point by point to ensure that the voltage, current, and power data at the same moment are associated one by one.

[0046] All time-series data are uniformly rearranged into a continuous sequence with one sampling point per millisecond. A total of 200 consecutive time-series points are formed in the entire 200-millisecond period, with the time-series points numbered from 0 to 199, forming a complete and time-uniform load characteristic sequence.

[0047] ② Data cleaning Missing data repair: For missing sampling points in the time series, the linear trend of adjacent valid data points is used to fill them. The median value of the two valid data points before and after the missing point is directly used as the repair value to ensure that the sequence has no breaks or interruptions.

[0048] Outlier data removal: First, calculate the overall numerical distribution range of the entire 200-millisecond power sequence. Isolation points that significantly exceed the normal distribution range are marked as outliers and directly removed. The patched values ​​of adjacent data are then used to replace them, ensuring that feature calculation is not affected by outlier sampling.

[0049] (2) Calculation of load fingerprint characteristic indicators Based on the pre-processed continuous load feature sequence, electrical configuration feature indicators, power fluctuation pattern feature indicators, and protocol response feature indicators are extracted according to the preset fixed calculation rules. These three types of indicators together constitute the complete load fingerprint of the peripheral device.

[0050] ① First category of indicators: Electrical configuration characteristic indicators This category of indicators reflects the inherent electrical attributes and power supply requirements of peripheral devices. The calculation rules all adopt standardized numerical processing methods, and the specific extraction steps are as follows: Rated electrical matching degree: Calculate the ratio of peripheral requested voltage to the default OTG output voltage of the mobile phone and the ratio of peripheral requested current to the maximum allowable output current of the mobile phone to obtain two sets of standardized electrical matching characteristics.

[0051] Identification features: The supplier ID, product ID, and device type code obtained from E-mark parsing are converted into numerical features of a unified dimension according to a fixed character-to-numerical rule, which serve as the basic identification features of peripheral devices.

[0052] Cable impedance characteristics: The cable impedance values ​​carried in the E-mark information are directly extracted and normalized according to the preset standardized range, which serves as the basic characteristics of power supply link loss.

[0053] ② Second category of indicators: Electrical fluctuation pattern characteristic indicators Electrical fluctuation morphology characteristics are the core distinguishing features of load fingerprints. By analyzing the current fluctuation data of peripherals in standby mode, the following features are extracted: Standby current fluctuation variance: The variance of standby current data at 200 timing points is calculated to reflect the stability of the load's standby power consumption; Peak standby current: Traverse the standby current timing sequence and record the maximum current value to reflect the instantaneous current demand of the load in standby mode; Average standby current: Calculates the average value of the standby current timing sequence, reflecting the basic standby power consumption of the load; Fluctuation period: Perform spectral analysis on the standby current time sequence to extract the main frequency period, which reflects the periodic behavior of the load's internal power management strategy.

[0054] The above characteristics can effectively distinguish different peripheral types such as mechanical hard drives (with large fluctuations in standby current and periodic seek behavior), solid-state drives (with stable standby current), fans (with near-zero standby current), and mobile phones (with periodic wake-up pulses).

[0055] ③ Third category of indicators: Protocol response characteristic indicators This category of metrics is used to assist in verifying the identity of peripheral devices and improve the accuracy of load identification. All metrics are calculated using time-series statistics. The specific extraction steps are as follows: E-mark response latency: The time interval from when the mobile phone sends the Get_Identity identity query command to when it successfully receives the response data returned by the peripheral device is used to obtain the E-mark response latency characteristics.

[0056] PD request response delay: The time interval from when the mobile phone triggers the boost output enable signal to when it successfully receives the Request power request data packet sent by the peripheral device is used to obtain the PD request response delay characteristics.

[0057] Protocol handshake success rate: The success rate of the protocol handshake is obtained by statistically analyzing the ratio of the number of successful command exchanges to the total number of commands sent during the OTG handshake process.

[0058] (3) Result Storage The three types of indicators are combined in a fixed order and dimension to form a standardized load fingerprint feature sequence that uniquely corresponds to the current peripheral device.

[0059] This feature sequence is stored in the device_fingerprint_features table of the local database, using a unique index of the combination of supplier ID and product ID.

[0060] The data table contains a unique index field, a complete feature sequence field, and a feature generation timestamp field. The data format is uniform and the structure is fixed, and it can be directly called by the lightweight neural network classification process of the S3 step and the cloud power consumption map retrieval process.

[0061] S3: Load classification and cloud-based typical power consumption map retrieval based on lightweight neural networks This step, based on the standardized load fingerprint feature sequence generated by S2, uses the invention's unique load fingerprint-adaptive lightweight neural network inference algorithm to accurately classify peripheral types. Simultaneously, it employs a cloud-based interactive algorithm combining VID, PID, classification result joint retrieval, and local secondary caching to obtain a typical power consumption map specific to each peripheral. This is the core innovative step in achieving "accurate load identification—power consumption map matching—adaptive boost control." This step is automatically triggered immediately after the load fingerprint feature calculation is completed, relying entirely on the mobile terminal's local computing power without additional hardware dependencies. The overall execution latency is strictly controlled within 10ms, fully compatible with OTG power supply standard timings, ensuring no interruption or delay to the peripheral power supply startup process.

[0062] (1) Lightweight Neural Network Load Classification Processing This section represents the core algorithmic innovation of this invention. It abandons the general-purpose neural network model and adopts a lightweight, fully connected neural network customized for load fingerprint features. It can complete peripheral classification with low computing power, high accuracy, and high real-time performance using only the terminal edge computing unit, thus solving the technical pain points of poor feature adaptability, low classification accuracy, and insufficient terminal computing power in traditional load identification methods.

[0063] ① Model solidification deployment and input feature preprocessing The customized lightweight neural network model is pre-stored in the local non-volatile storage unit of the mobile terminal. The model structure is a dedicated three-layer structure: input layer (matching the load fingerprint feature dimension of S2 output), lightweight hidden layer (using neuron pruning algorithm to remove redundant computing units), and output layer (corresponding to the classification results of four types of target peripherals). The model weight file has been quantized and compressed, so there is no need for real-time loading or cloud computing power assistance, and the inference calculation is completed directly on the local terminal.

[0064] The load fingerprint feature sequence output by S2 (including electrical configuration features, response waveform morphology features, and protocol response features) is used as the sole input data for the model, and a three-step feature preprocessing is performed: The first step is to perform extreme value normalization on all values ​​in the feature sequence, mapping all feature values ​​to a unified interval of 0-1; The second step is to remove invalid zero-value dimensions from the feature sequence and retain the effective distinguishing features. The third step is to perform dimension completion on the normalized feature sequence to ensure that the input dimension is completely matched with the model input layer, thereby completely eliminating the differences in dimensions and dimensionality of the three types of features: electrical configuration, power fluctuation pattern, and protocol response, and ensuring the stability and accuracy of model inference.

[0065] The model outputs the probability values ​​of four types of peripherals: 2.5-inch mechanical hard drive, solid-state drive, portable cooling fan, and smartphone.

[0066] ② Neural Network Forward Inference and Classification Result Generation The model receives the preprocessed load fingerprint feature sequence and performs hierarchical forward inference calculation: First, the input layer receives the feature data and transmits it to the lightweight hidden layer for feature association and weight calculation, extracting the inherent correlation patterns of core distinguishing features such as standby current fluctuation variance, standby current peak value, and fluctuation period in the load fingerprint; then, the hidden layer transmits the calculation results to the output layer, outputting the probability values ​​of four types of peripherals. The four types of peripherals are 2.5-inch mechanical hard drives, solid-state drives, portable cooling fans, and smartphones. Each type of peripheral corresponds to a probability value between 0 and 1, with a higher probability value indicating a higher matching degree.

[0067] The final classification result is generated using the maximum probability selection rule: the four probability values ​​of the output layer are traversed, and the peripheral type corresponding to the one with the largest value is selected as the final classification result of the currently connected peripheral; at the same time, the maximum probability value is retained as the classification confidence level for subsequent power supply reliability verification and fault determination.

[0068] ③ Validation of classification results and handling of anomalies A preset classification confidence threshold (fixed at 0.85) is used to perform double validity checks on the classification results: if the classification confidence is ≥0.85, the classification result is deemed valid and the process proceeds directly to the subsequent cloud interaction process; if the classification confidence is <0.85, the classification result is deemed invalid, the peripheral is marked as "unknown heterogeneous peripheral", the default boost protection strategy is immediately triggered, and the local general power consumption map is called to perform basic power supply to avoid power supply abnormalities caused by classification errors.

[0069] The classification results are finally encapsulated into a standardized data structure containing three core components: peripheral type code (fixed code corresponding to mechanical hard drive / SSD / fan / smartphone), classification confidence score, and peripheral unique identifier (VID+PID combined code), providing accurate and unique retrieval basis for subsequent cloud power consumption map retrieval.

[0070] (2) Cloud-based typical power consumption graph interaction and acquisition This step employs an innovative cloud-based interaction mechanism that combines local classification results with unique peripheral identifiers for dual verification and retrieval, replacing the traditional single VID+PID matching method. It also incorporates a local secondary caching strategy to achieve accurate matching and rapid retrieval of typical power consumption graphs, solving the problems of incorrect matching, high response latency, and redundant duplicate requests in traditional cloud-based interaction, and ensuring accurate adaptation between boost control and peripheral power consumption.

[0071] ① Encapsulation of cloud communication connection and retrieval request Mobile terminals establish standardized RESTful communication connections with cloud-based peripheral power consumption databases via Wi-Fi or cellular networks. The communication process uses the universal HTTP protocol, requiring no customized hardware or dedicated drivers, and is compatible with all mobile terminals equipped with Type-C OTG interfaces.

[0072] After the communication connection is established, the data synchronization service encapsulates three pieces of data—the peripheral unique identifier (VID+PID), the peripheral classification result, and the feature version number—into a standardized HTTP GET query request. The request parameters are strictly concatenated according to a fixed format to ensure that the cloud can directly parse and match them without ambiguity. At the same time, a terminal identifier and a timestamp are added to the request header for the cloud to verify the legitimacy of the request and prevent unauthorized access and data tampering.

[0073] ② Cloud-based multi-dimensional precise matching and power consumption map distribution After receiving a query request from a mobile terminal, the cloud database executes a two-layer precise matching algorithm: The first layer uses the VID+PID combination code as the core search condition to filter peripheral data of the corresponding model in the database; the second layer uses the peripheral classification result as the verification condition to perform a second verification on the filtered results, eliminating the error of matching different types of peripherals with the same VID+PID, and ensuring the uniqueness and accuracy of the search results.

[0074] After successful matching, the cloud database retrieves the exclusive typical power consumption spectrum of this peripheral model. The spectrum includes five core power consumption characteristics: startup transient power timing curve, steady-state operating power standard value, peak current safety threshold, power fluctuation cycle parameter, and transient response delay value, which fully covers the power consumption characteristics of the peripheral throughout the startup, steady-state, and transient fluctuation cycles, and is fully adapted to the adaptive boost control requirements.

[0075] The cloud encapsulates the aforementioned power consumption graph data into a fixed-structure JSON format and returns it to the mobile terminal via the original communication link. The data field names are fixed and the format is uniform, so the terminal can directly call it without additional parsing or conversion after receiving it.

[0076] ③ Optimization of local second-level caching and repeated access A dual-caching mechanism is established, consisting of a local first-level cache and a memory second-level cache: the first-level cache stores the successfully matched power consumption map in a local non-volatile storage unit, using VID+PID as a unique index; the second-level cache loads the power consumption map of the currently connected peripheral into the running memory, enabling millisecond-level access.

[0077] If the current peripheral device repeatedly connects to the device within a short period of time, the terminal automatically determines that the cache has been hit and directly calls the power consumption map in the second-level cache, skipping the cloud request process and minimizing the interaction latency; if the cache times out or the data becomes invalid, the cloud retrieval request is re-initiated to ensure the real-time performance and validity of the map data.

[0078] ④ Handling Cloud Interaction Anomalies If cloud communication is interrupted, the request times out, or there is no matching power consumption map data, the terminal immediately triggers the exception fallback mechanism, calls the locally pre-stored general power consumption map of the same type to ensure that the OTG power supply process is not interrupted, and records the interaction exception log. After the network is restored, the request will be automatically re-initiated.

[0079] (3) Standardized storage of classification and atlas data This step integrates the classification results data with the cloud-based power consumption map data, aligns the timing, and solidifies the format to form a unique, standardized, and directly callable boost control input data source, ensuring the feasibility, stability, and continuity of the S4 adaptive boost output control process.

[0080] ① Data integration and time-series alignment The peripheral classification data (peripheral type code, classification confidence, VID+PID) and cloud power consumption map data (start-up transient power curve, steady-state power, peak current, fluctuation period) are integrated in a unified manner. Based on the load fingerprint generation timestamp, the timing alignment of the two types of data is completed, and the timing deviation between the data is eliminated.

[0081] ② Standardized data format encapsulation The integrated data is encapsulated according to a fixed data structure and divided into two standardized nodes: Peripheral category node: contains three fixed fields: device_type (peripheral type code), confidence (category confidence), and vid_pid (peripheral unique identifier); Typical power consumption graph node: contains four fixed fields: startup_power_curve (startup transient power curve), steady_power (steady-state operating power), peak_current (peak current threshold), and fluctuation_period (power fluctuation period).

[0082] The encapsulated data has a unified format, fixed fields, and no redundant information, fully adapting to the input requirements of the adaptive boost control unit.

[0083] ③ Data storage and push access The standardized data is first stored in a dedicated cache area in the local memory of the mobile terminal to ensure real-time data retrieval; at the same time, the classification results and the core parameters of the power consumption graph are backed up to the local database for subsequent power supply log recording and fault tracing.

[0084] Once the data is stored, it is immediately pushed to the adaptive boost control execution unit as the sole and standardized input data source for step S4, ensuring that the subsequent boost output parameter configuration and dynamic adjustment are perfectly matched with the power consumption characteristics of the peripheral load.

[0085] S4: Adaptive boost output control parameter configuration and execution Based on the peripheral classification results, classification confidence, and cloud-specific typical power consumption graphs output by S3, this step dynamically configures the transient response parameters of the boost converter when the peripheral is starting up at full power. It also uses a graph-driven dynamic closed-loop adjustment algorithm to achieve precise matching between the boost output characteristics and the transient power consumption of the peripheral.

[0086] This step is triggered immediately after the power consumption map is acquired and standardized in the cloud. All control logic and peripheral power-on timing are strictly synchronized, with no additional timing insertion and no power supply delay introduced, fully meeting the timing specifications of USB OTG and PD protocols.

[0087] (1) Boost control parameter matching configuration This step relies on the triple constraints of peripheral type, typical power consumption graph, and terminal rated capacity to execute a differentiated parameter automatic matching algorithm. It abandons the traditional fixed parameter configuration mode and generates exclusive boost control parameters for each type of peripheral, solving the problems of power supply mismatch, transient voltage drop, and low efficiency from the source.

[0088] ① Parameter configuration, data source, and constraint rules The parameter configuration in this step strictly follows two types of constraints. All constraints are fixed in the form of algorithm rules and have no hardware dependency. The first category is the rated constraints of the terminal boost output (safety boundaries at the software algorithm level): the adjustable range of output voltage, the upper limit of continuous output current, the upper limit of instantaneous peak current, the minimum threshold of boost conversion efficiency, and the maximum allowable value of output voltage ripple, which are safety red lines that cannot be broken as parameter configurations; The second category is typical cloud power consumption graph inputs: transient power timing curves during peripheral startup, standard values ​​of steady-state operating power, peak current safety thresholds, and power fluctuation cycles, which serve as the core basis for parameter configuration.

[0089] During the configuration process, the algorithm automatically matches and verifies the power consumption characteristics of the spectrum with the terminal's rated constraints to ensure that all configuration parameters are within a safe range and prevent out-of-specification output.

[0090] ②Classification of boost parameter configuration rules This section represents the core innovation. It implements customized parameter configuration rules based on the power consumption characteristics of different peripherals. All rules are practical algorithmic logic, as detailed below: Mechanical hard drive peripherals: These peripherals are characterized by high transient current surges during startup. The algorithm is configured to set the output voltage to 5V. The current ramp rate is calculated based on the peak current of the cloud spectrum and configured to be high-speed ramp. The pre-boost duty cycle increase ratio is automatically calculated based on the transient power value during startup from the spectrum to increase the output capability in advance. The ripple suppression level is configured to be high to suppress transient ripple interference during startup. Solid-state drive peripherals: These peripherals have stable power consumption but require voltage adaptation. The algorithm automatically switches between 5V and 9V output voltage according to the parameters requested by the peripheral's PD; the current ramp rate is configured with medium speed based on the steady-state power spectrum; the ripple suppression level is configured to medium level to balance efficiency and stability. Portable cooling fan peripherals: These peripherals have low power consumption and low fluctuation, with a fixed output voltage of 5V configured by the algorithm; low current ramp rate configured; low ripple suppression level configured to prioritize improving boost conversion efficiency. Smartphone peripherals: These peripherals are bidirectional power supply loads. The algorithm automatically matches the 5V / 9V output voltage according to the PD protocol; the output current limit value is configured based on terminal safety specifications; the dynamic response bandwidth is configured to a medium level, compatible with both charging and power supply scenarios.

[0091] ③ Parameter distribution and timing fixation The algorithm encapsulates the matched boost control parameters into a standardized boost control command frame according to a fixed protocol format. The command frame contains all the core parameters, including output voltage, current ramp rate, pre-boost duty cycle, ripple suppression level, and current limiting threshold. The instruction frame is sent to the boost control execution unit via the terminal's internal bus. The sending process follows strict timing rules and is synchronized with the OTG power supply enable signal. After receiving the instruction, the boost control execution unit loads the parameters into the running configuration queue, completes the parameter solidification, and waits for the boost start trigger signal to ensure that the parameters are seamlessly connected with the power-on sequence.

[0092] ④ Pre-configure execution timing After obtaining the peripheral classification results and typical power consumption spectrum, the mobile phone configures the transient response parameters of the boost converter according to the identification results. When the peripheral sends a high-power request through the PD protocol, the boost converter starts to operate with the optimized transient response parameters.

[0093] (2) Boost Output Execution and Dynamic Adjustment This stage employs a three-stage boost output control algorithm consisting of pre-execution, transient following, and steady-state locking. It dynamically adjusts the output in real time based on typical power consumption graphs in the cloud, making it a core component for achieving stable power supply and suppressing voltage drops.

[0094] ①Pre-execution of boost start-up After receiving the start trigger signal, the boost control execution unit immediately starts the boost conversion process according to the fixed configuration parameters; According to the configured current ramp rate, the output current is increased smoothly and linearly without jumps or overshoots; For peripherals such as mechanical hard drives that have transient high current, the algorithm calls the pre-boost configuration in advance and raises the output drive capability in advance according to the calculated duty cycle. The output capability is preset before the current surge of the peripheral is reached, thus completely avoiding the problem of voltage drop at the moment of startup from the algorithm level. The entire startup process strictly follows the startup power timing of the cloud graph to ensure that the output curve closely matches the graph curve.

[0095] ② Transient dynamic follow-up adjustment The boost output employs a closed-loop tracking algorithm that combines high-frequency real-time sampling, point-by-point spectrum comparison, and dynamic duty cycle fine-tuning throughout the entire process. The boost output voltage and output current are sampled in real time at a fixed high frequency, and the current output power is calculated in real time. The real-time power, voltage, and current data obtained from sampling are compared point by point with the transient timing values ​​of typical power consumption spectra in the cloud, and the real-time deviation value is calculated. If the real-time power is close to the peak threshold of the graph, the algorithm automatically fine-tunes the PWM duty cycle to improve the output drive capability; if the real-time data is lower than the graph value, the duty cycle is appropriately reduced. By dynamically fine-tuning, the output voltage fluctuation range is strictly controlled within the allowable range, enabling the boost output to follow the load power consumption in real time without lag or overshoot.

[0096] ③Steady-state output maintenance Once the peripheral device completes the startup process and the real-time power enters the steady-state range marked on the spectrum, the algorithm automatically switches to steady-state lockout mode. Based on the steady-state operating power value of the cloud graph, lock the core parameters of the boost output and stop large-scale duty cycle adjustment; Automatically optimizes the boost switching frequency, reduces frequency fluctuations, and improves boost conversion efficiency to the optimal range; According to the configured ripple suppression level, the steady-state ripple suppression strategy is implemented to ensure that the output voltage is stable and free from interference. In steady-state mode, the algorithm makes only minor parameter adjustments to maintain output stability while reducing terminal computing power consumption.

[0097] (3) Real-time operation status acquisition and closed-loop feedback This step establishes a closed-loop management mechanism that integrates full-dimensional status acquisition, hierarchical anomaly protection, and standardized data feedback. It monitors the boost operation status in real time, ensures power supply safety, and provides data support for fault diagnosis.

[0098] ① Operational status acquisition The boost control unit has a built-in status acquisition algorithm that continuously acquires core operating parameters at fixed short intervals. All acquired data carries a millisecond-level timestamp to ensure time sequence traceability. The collected parameters include: real-time output voltage, real-time output current, real-time output power, and boost unit operating temperature, covering all dimensions of electrical performance and hardware status indicators; The collected data is transmitted to the local data processing unit in real time, without the need for caching or waiting, ensuring real-time feedback.

[0099] ② Layered anomaly protection judgment The algorithm presets multi-level protection thresholds, including overcurrent protection threshold, undervoltage protection threshold, and overtemperature protection threshold. All thresholds are matched with the terminal's rated constraints. The collected data is compared with the protection threshold in real time, and the hierarchical protection logic is executed: Mild anomaly: The parameter is close to the threshold but has not exceeded it. The algorithm automatically adjusts the output parameter slightly to suppress the anomaly from spreading without interrupting the power supply. Severe anomaly: When the parameter exceeds the threshold, the graded protection is immediately triggered. The current limiting protection is executed first. If it is ineffective, the boost output is quickly shut down, and a corresponding abnormal status flag is generated. Once an anomaly is triggered, the algorithm automatically records the time of the anomaly and the anomaly parameter values ​​for easy tracing later.

[0100] ③ Feedback data standardization and storage The data processing unit will process real-time running data, adjust parameters, and anomaly flags, and encapsulate them in a standardized manner according to a fixed structure. The encapsulation format is uniform, the fields are fixed, and there is no redundant data. The data is stored in three standardized nodes: Boost operation node: includes fields for real-time output voltage, real-time output current, real-time output power, and operating temperature; Adjustment recording node: includes fields for real-time PWM duty cycle, current ramp rate, and ripple suppression level; Abnormal status node: contains overcurrent flag, undervoltage flag, and overtemperature flag fields; The encapsulated data is written to the local operation log cache and simultaneously synchronized to the fault diagnosis unit, providing raw data support for the fault cause diagnosis of S5 and realizing full monitoring and traceability of the power supply process.

[0101] S5: Real-time power supply status monitoring, fault cause diagnosis, and operation log output This step continues to run online after the boost output enters a steady state. Through a full-cycle electrical state sampling algorithm, a multi-dimensional fault causal reasoning algorithm, and a standardized log archiving algorithm, it achieves real-time monitoring, intelligent fault location, and full-process data traceability of the OTG power supply process. Unlike the simple protection logic of traditional methods that only perform threshold alarms, it relies on load fingerprints and typical power consumption maps to achieve accurate attribution of fault root causes. It is a key closed-loop link to ensure OTG power supply safety and improve user experience.

[0102] (1) Real-time power supply status closed-loop monitoring This step employs a monitoring algorithm that combines high-frequency timed sampling with parallel calculation of multiple indicators to continuously collect operational data throughout the entire power supply process. This constructs a complete power supply status time-series dataset, providing a real-time, continuous, and traceable data foundation for fault diagnosis. The entire process does not rely on dedicated hardware monitoring units; instead, it utilizes built-in electrical sampling data from the terminal to complete algorithmic monitoring.

[0103] ① Definition of data sources and monitoring dimensions The monitoring data sources are uniformly taken from the real-time electrical sampling data, OTG protocol interaction status data, and boost control operation parameter data of the terminal boost output end. They do not involve specific hardware registers and physical pin definitions, and are all standardized data that can be called by the algorithm.

[0104] The monitoring dimensions cover three core indicators: electrical operating parameters, control and regulation parameters, and peripheral connection status. Specifically, these include: real-time output voltage, real-time output current, instantaneous output power, boost PWM duty cycle, voltage fluctuation amplitude, boost conversion efficiency, module operating temperature, and peripheral online connection status, comprehensively covering all dimensions of OTG power supply operation.

[0105] ② Data Acquisition and Time-Series Refresh Rules The monitoring algorithm is configured with a fixed 2ms sampling period and triggers sampling actions at equal time intervals to ensure the continuity and uniformity of monitoring data. After each sampling is triggered, real-time data of all monitoring dimensions are read in parallel and synchronously bound with millisecond-level timestamps in ISO8601 format to prevent data time sequence misalignment.

[0106] Real-time calculation of derived operating indicators: Instantaneous output power is calculated based on real-time voltage and current, boost conversion efficiency is calculated based on output and input parameters, and voltage fluctuation rate is calculated based on the average output voltage. All derived indicators are obtained in real time through algorithms without the need for hardware-assisted calculation.

[0107] All collected and calculated monitoring data are written to the local circular data buffer in time stamp order. The buffer has a fixed storage capacity and adopts an overwrite writing strategy to always retain the latest power supply operation sequence data, ensuring that complete historical data can be quickly retrieved when a fault is triggered.

[0108] (2) Cause-and-effect diagnosis of power supply faults based on load characteristics This section is the core innovation of the invention. It adopts a fault causal diagnosis algorithm of graph comparison + feature verification + hierarchical reasoning to replace the simple protection alarm triggered by the traditional single threshold. It combines load fingerprint features, typical power consumption graphs in the cloud, and real-time operation data to perform multi-level attribution reasoning, so as to achieve accurate location of the root cause of the fault, rather than just outputting abnormal alarms.

[0109] ① Fault Trigger Determination Algorithm Preset quantitative abnormal trigger thresholds, all thresholds are fixed by algorithms and there is no manual adjustment: output voltage drops more than ±10% relative to the rated value, real-time output current exceeds 1.2 times the peak current of the cloud spectrum, operating temperature exceeds the safety threshold, peripheral protocol response timeout exceeds 500ms, and peripheral connection level is continuously abnormal.

[0110] The monitoring algorithm compares the running data with the above thresholds in real time. When any trigger condition is met, the fault cause diagnosis process is immediately started. At the same time, the complete running data and load fingerprint data before and after the current moment are automatically frozen, the automatic rollback of boost control parameters is prohibited, and the original fault state is preserved for accurate diagnosis.

[0111] ② Hierarchical causal inference diagnostic process The diagnostic process follows a four-layer inference logic: load side → link side → protocol side → power supply side. Each layer has clearly defined algorithm rules for judgment, eliminating possibilities and pinpointing the exact location step by step. The specific steps are as follows: The first layer: load power consumption matching determination. The real-time collected power consumption time series data is compared with the typical power consumption spectrum in the cloud point by point. If the real-time power consumption continuously exceeds the fluctuation range of the spectrum and there are no compliant transient impact characteristics, it is determined that the power consumption of the load itself is abnormal. The second layer: power supply link loss determination. The cable impedance characteristic value extracted by S2 is retrieved and combined with the output voltage attenuation amplitude for algorithm calculation. If the line loss conversion value exceeds the compliance threshold and the voltage drop is positively correlated with the line loss, it is determined that the cable impedance is too large or the contact is poor. The third layer: Determine the integrity of the protocol interaction, trace back the entire OTG handshake log, and verify the integrity and timeliness of the E-mark response data and PD request data packets. If there is no response to the command or abnormal data frame, it is determined that the protocol handshake has failed. The fourth layer: power supply module operation judgment, monitors the boost control parameters and operating status. If the parameters are not abnormal but the operating temperature and overcurrent flag trigger protection, it is determined that the power supply module itself has taken protection action.

[0112] The entire reasoning process is executed automatically by the algorithm without human intervention. Each step generates a confidence level, and the result with the highest confidence level is selected as the root cause conclusion of the fault.

[0113] ③ Standardized fault location and output After diagnosis, the algorithm automatically generates standardized fault results, including a unique fault code, fault type, fault cause, confidence level, and handling suggestions. These results are categorized into four types of precise location results with no ambiguity: Load body abnormality: The real-time power consumption of the peripheral device deviates from the threshold of the typical power consumption spectrum, which is determined to be a peripheral hardware abnormality or a non-standard load. Power supply link transmission failure: Excessive cable impedance characteristics lead to excessive line loss, which is determined to be due to poor cable quality, poor contact, or loose interface. Protocol interaction anomaly: E-mark has no valid response, PD request data packet is abnormal, which is determined to be due to peripheral protocol incompatibility or handshake failure; Power supply module protection fault: The boost unit triggers over-temperature and over-current hardware protection, which is determined to be an overload or abnormal operation of the power supply module.

[0114] (3) Operational data reporting and standardized log storage This step utilizes structured log encapsulation algorithms, hierarchical storage strategies, and upper-layer system integration rules to uniformly archive and store monitoring data, adjustment records, and fault diagnosis results, forming a complete OTG power supply archive that is traceable, queryable, and reportable, thus meeting the needs of equipment operation management and fault tracing.

[0115] ① Log structured encapsulation rules All operational data is standardized and encapsulated in a single-frame log format. Each log frame has a fixed field structure with no redundant information. The core fields include: standard timestamp, peripheral VID / PID combination identifier, peripheral classification type, boost configuration parameters, real-time operating electrical parameters, fault code, fault diagnosis results, and adjustment action record.

[0116] Log frames are stored in a unified structured format to ensure consistency in subsequent reading, parsing, and reporting. All numeric fields use the same unit, and character fields use the same encoding format, eliminating format ambiguity.

[0117] ② Local Tiered Storage and Circular Management Log data employs a dual storage strategy: local Flash partition solidified storage plus temporary cache in running memory. The temporary cache is used for real-time reading, while solidified storage is used for long-term traceability.

[0118] The maximum number of frames stored in a single log file is set to 10,000. A first-in-first-out (FIFO) cyclic overwrite strategy is adopted to automatically clean up the oldest historical logs and always retain the complete OTG power supply log data of the most recent 30 days to avoid storage overflow.

[0119] Log files are named and partitioned by date, and can be quickly retrieved by peripheral type, fault status, and time range. The retrieval logic is automatically implemented by the algorithm.

[0120] ③Standardized reporting of upper-level systems Logs and status data are reported to the upper framework layer through the terminal's standard system interface. The reported data uses a lightweight format, transmitting only core status and fault results without consuming system resources.

[0121] The reported data can be directly accessed by the system power management application and UI display module without additional parsing and adaptation, achieving system-level synchronization of power supply status.

[0122] (4) Visualization of power supply status and user prompts Based on the operating status and fault diagnosis results, this step provides users with clear and accurate status prompts through a prompt message generation algorithm and multi-channel synchronous output rules. This is different from traditional vague abnormal prompts and provides a straightforward explanation of the cause of the fault.

[0123] ① Normal operating status prompt generation When there is no abnormality in the power supply status, the algorithm generates a standardized status prompt based on the peripheral type and real-time output power, which includes the peripheral type, current power supply status, and real-time power consumption information. The prompt content has a fixed format and no arbitrary expressions.

[0124] The prompts automatically match the peripheral device category results, such as: "Hard disk drive connected, OTG power supply is stable, current power consumption is 2.5W" and "Solid state drive connected, OTG power supply is normal".

[0125] ② Accurate fault status prompts generated When a fault is diagnosed, the algorithm matches a standardized fault code with a corresponding user-understandable prompt, directly pointing to the root cause and solution of the fault, without ambiguous alarms. Load anomaly message: "External device power consumption is abnormal and power supply is not normal. Please check the peripheral device." Cable error message: "OTG cable impedance is too high, voltage attenuation is abnormal. It is recommended to replace with original cable." Protocol error message: "Peripheral protocol handshake failed, incompatible with current OTG mode"; Power supply protection message: "Power supply module temperature too high / current overload, output has been suspended."

[0126] ③ Multi-channel prompt output execution The prompts are output simultaneously through three channels: status bar text prompts, system pop-up prompts, and vibration feedback. The output logic is uniformly controlled by an algorithm: in normal state, only the status bar provides a silent prompt; in fault state, pop-ups and vibrations are triggered simultaneously to ensure that users can quickly perceive abnormalities and improve user safety.

[0127] II. Mobile OTG Adaptive Boost Control System Reference Figure 2 This disclosure also provides a mobile phone OTG adaptive boost control system based on load fingerprint recognition, including: The OTG multi-source data acquisition module is used to acquire the E-mark electronic tag information and voltage and current application parameters of the connected peripheral after the mobile phone and the peripheral have completed the USB PD protocol negotiation and established basic power supply, and continuously collect the electrical fluctuation data of the peripheral in standby mode to generate load electrical feature fingerprint. The load fingerprint feature calculation module is used to clean and normalize the collected raw data, and extract electrical configuration features, power fluctuation pattern features and protocol response features. The lightweight neural network classification and cloud interaction module has a built-in lightweight neural network model, which is used to identify the type of peripheral device based on the load fingerprint characteristics, and retrieve the typical power consumption spectrum of the corresponding peripheral model in the cloud database according to the type. An adaptive boost control execution module is used to generate boost control commands based on the typical power consumption spectrum, and adjust the output voltage, current ramp rate and ripple suppression strength of the boost converter. The power supply monitoring and fault cause diagnosis module is used to monitor the power supply status in real time and locate the root cause of the fault based on map comparison and feature verification when an anomaly occurs.

[0128] The modules of this system communicate with each other via I2C, SPI, and system bus. The module logic corresponds completely to the aforementioned method steps. It can be independently deployed on mobile phone power management chips, OTG adapter control chips, or mobile phone edge computing units. It is applicable to all smartphones, tablets, and other mobile terminals equipped with Type-C OTG interfaces. Without creative effort, those skilled in the art can perform hardware adaptation and software porting based on the structure of this system, all of which fall within the protection scope of this invention.

[0129] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0130] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0131] In this disclosure, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The block diagrams of devices, apparatuses, devices, and systems involved in this disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as "comprising," "including," "having," etc., are open-ended terms meaning "including but not limited to," and are used interchangeably with them. The terms "or" and "and" as used herein refer to the terms "and / or," and are used interchangeably with them unless the context clearly indicates otherwise. The term "such as" as used herein refers to the phrase "such as but not limited to," and is used interchangeably with it.

[0132] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0133] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0134] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0135] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0136] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A mobile phone OTG load fingerprint recognition adaptive boost control method, characterized in that, Includes the following steps: After the mobile phone and OTG peripheral complete the USB PD protocol negotiation and establish basic power supply, obtain the E-mark electronic tag information and voltage and current application parameters of the connected peripheral. During the basic power supply maintenance period, the voltage and current of the OTG boost output terminal are continuously monitored at a sampling frequency of not less than 1kHz, and electrical fluctuation data of peripheral devices in standby state are collected to generate load electrical feature fingerprints. A lightweight neural network is used to fuse the E-mark information, voltage and current application parameters, and load electrical feature fingerprint to identify the type of peripheral device; Based on the identification results, retrieve the typical power consumption graph of the corresponding peripheral model from the cloud database; Based on the typical power consumption spectrum, the transient response parameters of the boost converter are dynamically configured to achieve adaptive boost output control.

2. The method according to claim 1, characterized in that, The acquisition of E-mark electronic tag information, voltage and current request parameters, and generation of load electrical feature fingerprints for the access peripherals specifically includes: The E-mark electronic tag information of the peripheral device is obtained through the CC logic detection unit, and the VID, PID, device type code and cable impedance are extracted. The voltage and current request parameters of the peripheral device are obtained through the PD protocol parsing module; After the basic power supply is established, the voltage and current of the OTG boost output terminal are continuously collected at a sampling frequency of not less than 1kHz for a duration of 200ms, generating a standby electrical fluctuation time series dataset containing timestamps.

3. The method according to claim 1, characterized in that, The process of fusing the E-mark information, voltage and current request parameters, and load electrical feature fingerprint using a lightweight neural network specifically includes: The collected raw data is cleaned, time-series aligned, and format-normalized. Extract load fingerprint feature indicators, which include electrical configuration features, electrical fluctuation pattern features, and protocol response features; input the load fingerprint feature indicators into a pre-set lightweight neural network model for forward inference, and output the peripheral type attribution probability value; If the classification confidence score is greater than the preset threshold, the classification result is deemed valid; if the classification confidence score is less than the preset threshold, it is determined to be an unknown heterogeneous peripheral and the default boost protection strategy is triggered.

4. The method according to claim 3, characterized in that, The electrical fluctuation morphological characteristics include standby current fluctuation variance, standby current peak value, standby current mean value, and fluctuation period; The protocol response characteristics include E-mark response latency, PD request response latency, and protocol handshake success rate.

5. The method according to claim 1, characterized in that, The process of retrieving typical power consumption graphs for the corresponding peripheral models from the cloud database includes: The peripheral device's VID, PID, and classification results are encapsulated into an HTTP GET request and sent to the cloud; The cloud executes a dual matching algorithm to filter out typical power consumption profiles for the corresponding models and distributes them. The typical power consumption profiles include the startup transient power timing curve, steady-state operating power, peak current threshold, and power fluctuation period. If the local L2 cache is hit, the power consumption map in the cache is directly called; if the cloud interaction is abnormal, the local pre-stored general power consumption map of the same type is called.

6. The method according to claim 1, characterized in that, The transient response parameters of the boost converter, which are dynamically configured based on typical power consumption patterns, specifically include: For mechanical hard disk peripherals, based on their startup transient high current power consumption spectrum, high-speed current ramp-up and high-gain loop compensation parameters are configured; For solid-state drive peripherals, based on their steady-state power characteristics, medium-speed current ramping and medium-level loop compensation are configured. For cooling fan peripherals, low-speed current ramp-up and low-level ripple suppression are configured; For smartphone peripherals, match the output voltage according to the PD protocol and configure the output current limit value.

7. The method according to claim 1, characterized in that, After implementing adaptive boost output control, it also includes: Real-time acquisition of boost output operating status data to construct a power supply status time-series dataset; When abnormal operating parameters are detected, hierarchical causal inference diagnosis is performed based on load fingerprint characteristics, typical power consumption graphs in the cloud, and real-time operating data, and the root cause conclusion of the fault is output.

8. A mobile phone OTG load fingerprint recognition adaptive boost control system, characterized in that, include: The OTG multi-source data acquisition module is used to acquire the E-mark electronic tag information and voltage and current application parameters of the connected peripheral after the mobile phone and the peripheral have completed the USB PD protocol negotiation and established basic power supply, and continuously collect the electrical fluctuation data of the peripheral in standby mode to generate load electrical feature fingerprint. The load fingerprint feature calculation module is used to clean and normalize the collected raw data, and extract electrical configuration features, electrical fluctuation pattern features and protocol response features. The lightweight neural network classification and cloud interaction module has a built-in lightweight neural network model, which is used to identify the type of peripheral device based on the load fingerprint characteristics, and retrieve the typical power consumption spectrum of the corresponding peripheral model in the cloud database according to the type. An adaptive boost control execution module is used to generate boost control commands based on the typical power consumption spectrum and configure the transient response parameters of the boost converter, including current ramp rate, loop compensation parameters and ripple suppression strength. The power supply monitoring and fault cause diagnosis module is used to monitor the power supply status in real time and locate the root cause of the fault based on map comparison and feature verification when an anomaly occurs.

9. The system according to claim 8, characterized in that, The system is integrated into the mobile phone power management chip, OTG adapter control chip, or mobile phone edge computing unit.

Images