Super device capability calling method and system based on distributed technology, terminal and storage medium

Abstract

The invention discloses a super device capability calling method and system based on a distributed technology, a terminal and a storage medium, and the method comprises the steps: obtaining device operation state data and calling data, and carrying out the preprocessing, and obtaining target device operation state data and target calling data; performing calling trend prediction processing according to the target calling data to obtain a calling trend result, and performing equipment state prediction processing according to the target equipment operation state data to obtain an abnormal equipment state; performing feature fusion processing and calling failure reason calculation on the calling trend result and the abnormal equipment state to obtain a target calling failure reason; and performing optimization strategy matching and optimization strategy dynamic adjustment according to the target calling failure reason to obtain a target optimization suggestion, and adjusting the target fault equipment according to the target optimization suggestion. According to the method, the optimization strategy is adopted for adjustment according to the calling failure reason, and the efficiency and stability of super device capacity calling can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, system, terminal, and computer-readable storage medium for invoking the capabilities of super devices based on distributed technology. Background Technology

[0002] With the rapid development of IoT technology, super devices have emerged as a new type of virtual device model. Super devices seamlessly integrate multiple physical devices (such as screens, speakers, lights, and air conditioners) into a single logical device through the soft bus technology of the open-source HarmonyOS operating system, providing abstract capabilities such as lighting, screen projection, and audio playback. Its hardware and software decoupling design concept significantly reduces the complexity of development in traditional IoT systems.

[0003] In multi-device collaboration scenarios, the process of calling a super device involves multiple devices and communication links. However, existing monitoring systems often lack the ability to process call data in real time, making it difficult to detect and respond to problems in a timely manner. This results in low super device calling capabilities, which cannot meet the efficient operation and maintenance needs of complex IoT scenarios.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a method, system, terminal, and computer-readable storage medium for calling super devices based on distributed technology. This invention aims to solve the problem that existing monitoring systems often lack the ability to process call data in real time, making it difficult to detect and respond to problems in a timely manner. As a result, the super devices have low calling capabilities and cannot meet the needs of efficient operation and maintenance in complex IoT scenarios.

[0006] To achieve the above objectives, the present invention provides a method for invoking the capabilities of super devices based on distributed technology, the method comprising the following steps: Acquire device operating status data and call data, and preprocess the device operating status data and call data to obtain target device operating status data and target call data; Based on the target call data, call trend prediction processing is performed to obtain call trend results, and based on the target device operating status data, device status prediction processing is performed to obtain abnormal device status. The call trend results and the abnormal device status are subjected to feature fusion processing and call failure reason calculation to obtain the target call failure reason; Based on the reasons for the target call failure, optimization strategies are matched and dynamically adjusted to obtain target optimization suggestions, and the target faulty equipment is adjusted according to the target optimization suggestions.

[0007] Optionally, the super device capability invocation method based on distributed technology, wherein obtaining device operating status data and invocation data, and preprocessing the device operating status data and the invocation data to obtain target device operating status data and target invocation data, specifically includes: Determine the target acquisition source. If the target acquisition source reaches the preset acquisition frequency, acquire the device operation status data and call data of the super device. The device operation status data includes CPU utilization, memory utilization, and target network parameters. The call data includes the number of calls, call success rate, call failure reason, and call duration. The device operating status data and the call data are preprocessed to obtain target device operating status data and target call data. The preprocessing includes data cleaning, format conversion and data encryption.

[0008] Optionally, the super device capability invocation method based on distributed technology, wherein the step of performing invocation trend prediction processing based on the target invocation data to obtain the invocation trend result specifically includes: The target call data is subjected to time-series alignment processing to obtain a call timing sequence; The missing value completion and outlier deletion processes are performed on the call sequence to obtain the target call sequence. Determine the initial long short-term memory network model and obtain historical call time series data; The initial long short-term memory network model is trained based on the historical call time series data, and the trained initial long short-term memory network model is optimized using the cross-entropy loss function to obtain the call trend model. The target call time sequence is input into the call trend model to obtain the call trend result.

[0009] Optionally, the super device capability invocation method based on distributed technology, wherein the step of performing device status prediction processing based on the target device operating status data to obtain abnormal device status specifically includes: A state feature matrix is ​​constructed based on the target device's operating status data, and the state feature matrix is ​​normalized to obtain a normalized state feature matrix. Construct a target convolutional neural network and input the normalized state feature matrix into the target convolutional neural network to output a device operating state feature vector; A preset normal equipment state feature library is determined, and the cosine similarity between the equipment operating state feature vector and the preset normal equipment state feature library is calculated to obtain the cosine similarity result. If the cosine similarity result is less than a preset threshold, the target device's operating status data is determined to be abnormal, and the abnormal device status is obtained.

[0010] Optionally, the super device capability invocation method based on distributed technology, wherein the step of performing feature fusion processing and invocation failure reason calculation on the invocation trend result and the abnormal device state to obtain the target invocation failure reason specifically includes: The call trend results and the abnormal device status are subjected to feature fusion processing to obtain fused features; Construct a fault feature library, wherein the fault feature library includes the correspondence between call failure reasons and fault features; Calculate the Euclidean distance between the fused feature and multiple fault features in the fault feature library, extract the target fault feature with the smallest Euclidean distance from the fault feature library to the fused feature, and use the target fault feature as the target call failure reason.

[0011] Optionally, the super device capability invocation method based on distributed technology, wherein the step of matching and dynamically adjusting optimization strategies based on the target invocation failure reason to obtain target optimization suggestions, and adjusting the target faulty device according to the target optimization suggestions, specifically includes: A preset rule engine is determined, and optimization strategies are matched to the reasons for the failure of the target call using the preset rule engine to obtain preliminary optimization suggestions. The initial optimization suggestions are dynamically adjusted using a reinforcement learning algorithm to obtain the target optimization suggestions. Based on the target optimization suggestions, a target optimization instruction is generated and sent to the corresponding target faulty device.

[0012] Optionally, the super device capability invocation method based on distributed technology, wherein the step of matching and dynamically adjusting optimization strategies based on the target invocation failure reason to obtain target optimization suggestions, and adjusting the target faulty device according to the target optimization suggestions, further includes: Obtain the permission information of the super device, wherein the permission information includes the device ID, the operation and maintenance personnel account, and the scope of permission; The permission information is hashed to obtain a permission digest; The permission digest is verified through consensus and written into the blockchain to obtain a distributed permission library; When a call request from a target person is received and the target person has carried preset information, the identity of the target person is verified, the scope of permissions is matched, and the anti-replay verification is performed based on the preset information and the distributed permission library to obtain the verification result. The preset information includes a digital certificate, the request operation content, and a randomly generated 32-bit temporary random number. If the verification result is successful, the authorization approval result will be sent to the target personnel. If the verification result is unsuccessful, the authorization failure result will be sent to the target personnel.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a super device capability invocation system based on distributed technology, wherein the super device capability invocation system based on distributed technology includes: The data preprocessing module is used to acquire device operating status data and call data, and preprocess the device operating status data and the call data to obtain target device operating status data and target call data; The device status prediction module is used to perform call trend prediction processing based on the target call data to obtain call trend results, and to perform device status prediction processing based on the target device operating status data to obtain abnormal device status. The failure reason calculation module is invoked to perform feature fusion processing and failure reason calculation on the call trend result and the abnormal device status to obtain the target call failure reason; The optimization suggestion generation module is used to match optimization strategies and dynamically adjust optimization strategies based on the reasons for the target call failure, obtain target optimization suggestions, and adjust the target faulty equipment according to the target optimization suggestions.

[0014] In addition, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a super device capability invocation program based on distributed technology stored on the memory and executable on the processor, wherein when the super device capability invocation program based on distributed technology is executed by the processor, it implements the steps of the super device capability invocation method based on distributed technology as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a super device capability invocation program based on distributed technology, and when the super device capability invocation program based on distributed technology is executed by a processor, it implements the steps of the super device capability invocation method based on distributed technology as described above.

[0016] In this invention, device operating status data and call data are acquired, and preprocessed to obtain target device operating status data and target call data. Call trend prediction processing is performed based on the target call data to obtain call trend results, and device status prediction processing is performed based on the target device operating status data to obtain abnormal device status. Feature fusion processing and call failure reason calculation are performed on the call trend results and abnormal device status to obtain target call failure reasons. Optimization strategy matching and dynamic adjustment are performed based on the target call failure reasons to obtain target optimization suggestions, and the target faulty device is adjusted according to the target optimization suggestions. This invention, by predicting call trends and device status based on device operating status data and call data, thereby determining the cause of call failure and employing optimization strategies for adjustment, can effectively improve the efficiency and stability of super device capability invocation. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 2 This is a schematic diagram of the system architecture of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 3 This is a schematic diagram of the data acquisition process of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 4 This is a schematic diagram of the data analysis process of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 5 This is a schematic diagram of the optimization suggestion generation process of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 6 This is a schematic diagram of the security monitoring module structure of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 7 This is a schematic diagram of the first visualization interface of the visualization display module of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 8 This is a schematic diagram of the second visualization interface of the visualization display module of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 9 This is a schematic diagram of the third visualization interface of the visualization display module of a preferred embodiment of the super device capability invocation method based on distributed technology of the present invention; Figure 10This is a structural diagram of a preferred embodiment of the super device capability invocation system based on distributed technology of the present invention; Figure 11 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] With the rapid development of IoT technology, super devices have emerged as a new type of virtual device model. Super devices seamlessly integrate multiple physical devices (such as screens, speakers, lights, and air conditioners) into a single logical device using the open-source HarmonyOS operating system's soft bus technology, providing abstract capabilities such as lighting, screen projection, and audio playback. Their hardware-software decoupling design significantly reduces the development complexity of traditional IoT systems. However, in practical use, the efficiency and stability of super device capability invocation still face the following challenges: 1. Complexity: In multi-device collaboration scenarios, the calling process involves multiple devices and communication links, and the causes of failure are often complex and diverse, such as network congestion, excessive device load, or resource contention.

[0020] 2. Real-time performance: Existing monitoring systems often lack the ability to process called data in real time, making it difficult to detect and respond to problems in a timely manner.

[0021] 3. Insufficient optimization capabilities: Most traditional monitoring systems can only provide basic call data statistics and lack in-depth analysis of the reasons for call failures and targeted optimization capabilities.

[0022] Therefore, there is an urgent need for an intelligent monitoring and optimization system based on distributed technology to improve the efficiency and stability of super device capability calls and meet the high-efficiency operation and maintenance needs of complex IoT scenarios.

[0023] To address the aforementioned issues, this invention proposes a system for real-time monitoring, intelligent analysis, and adaptive optimization of super device capability calls based on distributed technology, involving distributed systems, the Internet of Things, artificial intelligence, big data processing, and the open-source Harmony operating system.

[0024] The preferred embodiment of the super device capability invocation method based on distributed technology described in this invention, such as... Figure 1 and Figure 2 As shown, the super device capability invocation method based on distributed technology includes the following steps: Step S10: Obtain device operating status data and call data, and preprocess the device operating status data and call data to obtain target device operating status data and target call data.

[0025] like Figure 2 As shown, this invention includes a data acquisition module, a data analysis module, an optimization suggestion module, a security monitoring module, and a visualization module. This invention is applicable to complex IoT environments such as smart homes, smart offices, smart conference rooms, and smart cities, and is particularly suitable for application scenarios involving high-frequency calls and multi-device collaboration.

[0026] The implementation process for the data acquisition module is as follows: Distributed data acquisition agents are deployed on various terminal devices of the super device. These agents acquire call data in real time through the API interface provided by the open-source HarmonyOS system at a preset acquisition frequency (dynamically adjusted according to the importance and real-time requirements of different capability calls).

[0027] like Figure 3 As shown, the collected data includes the number of calls, call success rate, reasons for call failure, call duration, information on the initiating and target devices, and network status parameters during the call process. The data acquisition agent encapsulates the acquired data in JSON format, performs preprocessing (such as data cleaning and format conversion), and then stores it in a distributed caching system. For example, in a conference room super device, for emergency lighting control calls, the data acquisition agent collects data once per second, while for audio playback calls, it collects data once per minute, ensuring high real-time performance of critical capability call data.

[0028] Specifically, a target acquisition source is determined. If the target acquisition source reaches a preset acquisition frequency, the device operation status data and call data of the super device are acquired. The device operation status data includes CPU utilization, memory utilization, and target network parameters. The call data includes the number of calls, call success rate, call failure reasons, and call duration. The device operation status data and the call data are preprocessed to obtain target device operation status data and target call data. The preprocessing includes data cleaning, format conversion, and data encryption.

[0029] It is understood that the data acquisition module is a fundamental function of this invention, used to acquire the real-time operating status of the super device (i.e., the device operating status data in this invention) and to retrieve data. Its features include: 1. Data Scope: The collected data includes, but is not limited to, the number of calls, success rate, reasons for failure, call duration, and device status data (such as CPU utilization, memory utilization, and network parameters (latency, packet loss rate, bandwidth utilization)). This data provides a comprehensive foundation for subsequent analysis and optimization.

[0030] 2. Dynamic Acquisition Frequency: This invention dynamically adjusts the acquisition frequency based on the importance of different calls. For example, emergency lighting calls are acquired once per second to ensure timely response; routine capability calls, such as audio playback, are acquired once per minute.

[0031] 3. Distributed data acquisition architecture: This invention adopts a distributed architecture, which distributes the acquisition tasks to various sub-devices, reduces the load on the central node, and encapsulates the data in a unified JSON format to ensure the consistency and real-time performance of data from multiple devices.

[0032] 4. Security: The collected data is encrypted using AES encryption technology to prevent data leakage or tampering during transmission.

[0033] The "dynamic acquisition frequency" in the data acquisition process is intended to balance equipment resource consumption, data validity, and overall system load. Its core necessity is as follows: 1. Avoid overloading device resources. Super devices (such as smart home devices and smart office terminals) are mostly lightweight hardware (with limited CPU, memory and battery resources). If the sampling frequency is not limited (such as uninterrupted sampling), it will continuously occupy the device's computing or power resources, causing the device's own ability to call (such as audio playback, lighting control) to lag, respond late, or even shorten the device's battery life.

[0034] 2. Reduce data redundancy and storage or reduce analysis pressure. If the collection frequency is too high (e.g., 10 times per second), the call data and device status data of adjacent collection cycles will be highly repetitive. Such redundant data will not only fail to improve the analysis accuracy, but will also occupy the space of the distributed storage system, while increasing the computing load of subsequent data analysis modules and reducing the overall processing efficiency.

[0035] 3. To adapt to the real-time requirements of different scenarios, this invention presets differentiated collection frequencies for different capability call scenarios (such as "emergency lighting call" requires high-frequency collection, while "low-load audio playback" can be collected at low frequency). Determining "whether the collection frequency has been reached" is the core logic for realizing "on-demand collection", which ensures the real-time data requirements of high-priority scenarios while avoiding resource waste in low-priority scenarios.

[0036] 4. Stable system data flow: By controlling the acquisition frequency, acquisition requests from the device side can be evenly distributed over time, avoiding a large influx of data into the cloud in a short period of time. This prevents network bandwidth from being saturated and data accumulation from occurring in the distributed caching system, thereby ensuring the stability of the entire system's data flow.

[0037] Step S20: Perform call trend prediction processing based on the target call data to obtain call trend results, and perform device status prediction processing based on the target device operating status data to obtain abnormal device status.

[0038] like Figure 4 As shown, the implementation process of the data analysis module is as follows: This invention builds a deep learning algorithm training platform, adopting a hybrid architecture of Long Short-Term Memory (LSTM) network combined with Convolutional Neural Network (CNN). First, the collected call data is preprocessed, including data normalization and missing value imputation. Then, the time series features (such as the number of calls and duration) and device status features (such as network parameters and device load) of the call data are extracted as input to the algorithm. Using super device call data simulated in various typical IoT scenarios under the open-source HarmonyOS system, the model is trained and validated to ensure that the model can accurately identify complex patterns and underlying causes of call failures. The trained model is deployed in the data analysis module, and then the call data in the distributed caching system is analyzed in real time to identify the specific reasons for call failures, abnormal status indicators of related devices, and the weights of possible influencing factors, and the analysis results are stored in a distributed database.

[0039] Specifically, the target call data is time-aligned to obtain a call time sequence; the call time sequence is then filled with missing values ​​and outliers are removed to obtain a target call time sequence; an initial long short-term memory network model is determined, and historical call time sequence data is obtained; the initial long short-term memory network model is trained based on the historical call time sequence data, and the trained initial long short-term memory network model is optimized using the cross-entropy loss function to obtain a call trend model; the target call time sequence is input into the call trend model to obtain the call trend result.

[0040] Understandably, the data analytics module uses AI and big data technologies to deeply process the collected data, helping to identify potential problems and provide predictive capabilities.

[0041] The data analysis module employs the following algorithm architecture: using a Long Short-Term Memory (LSTM) network to analyze time series data and identify trends in call failures; and using a Convolutional Neural Network (CNN) to extract device operating status features and detect abnormal patterns.

[0042] The specific implementation process of the algorithm architecture is as follows: The Long Short-Term Memory (LSTM) network analyzes time-series data and identifies call failure trends. For time-series call data such as "number of calls, call success rate, and call duration," LSTM is used to identify trends. The steps are as follows: 1. Time series data preprocessing: The collected call data is aligned according to "1 second / time step" to construct a call time series sequence for a single device (such as "call success rate, call duration" data for 60 consecutive time steps); missing values ​​are filled with linear interpolation, and outliers (such as call duration > 3 times the mean) are removed according to the 3σ principle; 2. LSTM Model Construction and Training: A bidirectional LSTM network with two hidden layers (128 neurons per layer) is constructed, including: Input layer: receiving a 2D temporal feature vector composed of "call success rate and call duration"; Hidden layer: capturing temporal dependencies through input gate, forget gate, and output gate (such as the correlation between "call duration increasing for 5 consecutive steps and call success rate decreasing"); Output layer: outputting "call failure probability" (probability in the range of 0-1); Training process: training the model with historical call time series data of 30 days (labeled "failure / normal"), optimizing parameters with cross-entropy loss function to achieve a test set accuracy of ≥93%; 3. Call failure trend identification: Input the real-time call time series data (i.e. the target call time series sequence in this invention) into the trained LSTM model (i.e. the call trend model in this invention). If the "call failure probability" is ≥0.7 for 10 consecutive time steps, it is determined to be a "call failure trend", and the trend association features are marked simultaneously (such as "call duration increase rate ≥20ms / step").

[0043] A state feature matrix is ​​constructed based on the target device's operating status data, and the state feature matrix is ​​normalized to obtain a normalized state feature matrix. A target convolutional neural network is constructed, and the normalized state feature matrix is ​​input into the target convolutional neural network to output a device operating status feature vector. A preset normal device status feature library is determined, and cosine similarity is calculated between the device operating status feature vector and the preset normal device status feature library to obtain a cosine similarity result. If the cosine similarity result is less than a preset threshold, the target device's operating status data is determined to be abnormal, and an abnormal device status is obtained.

[0044] The process of using a Convolutional Neural Network (CNN) to extract device operating status features and detect abnormal patterns involves extracting features and detecting anomalies from device status data such as CPU utilization, memory usage, and network bandwidth. The steps are as follows: 1. State feature engineering: Sample the device state data at “5 seconds / time step” and construct a state feature matrix of “time step (30 steps) × state dimension (CPU, memory, network)”; normalize the feature matrix (scale to the [0, 1] interval) to obtain the normalized state feature matrix. 2. CNN Feature Extraction: Construct a CNN network with 2 convolutional layers and 1 pooling layer: Convolutional Layer 1: Uses 3×3 convolutional kernels (32 kernels) to extract "local change features in a single state dimension" (such as "fluctuation amplitude of CPU utilization over 3 consecutive steps"); Convolutional Layer 2: Uses 3×3 convolutional kernels (64 kernels) to extract "correlated features in multiple states" (such as the collaborative feature of "increased CPU utilization + increased memory usage"); Pooling Layer: Compresses the feature dimension through max pooling (2×2 window) to output a 128-dimensional device state feature vector; 3. Abnormal Pattern Detection: The cosine similarity of the state feature vector output by the CNN with the "normal device state feature library" (based on the feature statistics of 1000 normally operating devices) is calculated; if the similarity is <0.6 (preset threshold), it is determined as "device state abnormal"; at the same time, combined with the key dimensions in the feature vector (such as "CPU utilization feature value >0.9"), the abnormal associated dimensions (such as "CPU resource abnormal") are marked.

[0045] Furthermore, feature extraction can also be performed based on the cleaned call data and device status data, divided into two data dimensions. The specific steps are as follows: 1. Data Feature Extraction: For call data such as "number of calls, call success rate, and call duration", three core features are extracted, including: a) Basic statistical features: Calculate the "average number of calls, minimum call success rate, and maximum call duration within a 5-minute time window" to obtain three statistical features; b) Time-series correlation features: Calculate the "difference in call success rate and rate of change in call duration between two consecutive time windows" to obtain two time-series correlation features; c) Anomaly marking features: If the call success rate in a certain time window is <80% (preset threshold), it is marked as "1", otherwise it is marked as "0", resulting in one binary feature. 2. Device Status Data Feature Extraction: For device status data such as "CPU utilization, memory usage, and network bandwidth", three types of core features are extracted: a) Dimensional features: Calculate "peak CPU utilization, average memory usage, and minimum network bandwidth within a 5-minute time window" to obtain three dimensional features; b) Correlation features: Calculate "Pearson correlation coefficient between CPU utilization and memory usage, and covariance between network bandwidth and call duration" to obtain two correlation features; c) Threshold features: If CPU utilization in a certain time window is >90%, it is marked as "1"; otherwise, it is marked as "0", resulting in one binary feature. 3. Feature Standardization and Integration: The 12 features extracted above (3+2+1)+(3+2+1) are scaled to the [0,1] interval through Min-Max normalization and integrated into a 12-dimensional feature vector, which is used as the input data for the deep learning model.

[0046] Furthermore, this invention employs an "LSTM+CNN hybrid model" to process feature vectors, with the specific process as follows: 1. Model input preparation: The integrated 12-dimensional feature vector is used to construct a time series feature matrix in the format of "time step (30 consecutive 5-minute windows) × feature dimension (12 dimensions)" to serve as the input data for the model.

[0047] 2. The LSTM sub-model processes call time-series features, including: a) Model structure: Constructing a unidirectional LSTM network with one hidden layer (64 neurons); b) Processing logic: After inputting the time-series feature matrix, the LSTM captures the "temporal change trend of call data features" (such as "call success rate decreases for 5 consecutive time steps") through gating units; c) Output result: Outputs a 64-dimensional call time-series feature vector, corresponding to the "feature representation of call trend".

[0048] 3. The CNN sub-model processes device status features, including: a) Model structure: Constructing a CNN network with 1 convolutional layer (3×3 convolutional kernels, 32 kernels) + 1 pooling layer (2×2 window); b) Processing logic: After inputting the temporal feature matrix, the convolutional layer extracts the "local correlation patterns of device status features" (such as "CPU utilization increases while memory usage increases"), and the pooling layer compresses the feature dimension; c) Output result: Outputs a 32-dimensional device status feature vector, corresponding to the "abnormal correlation feature representation of device status".

[0049] 4. Feature Fusion and Result Determination: The 64-dimensional vector output by the LSTM (i.e., the call trend result in this invention) and the 32-dimensional vector output by the CNN (i.e., the abnormal device status in this invention) are concatenated to obtain a 96-dimensional fused feature vector (i.e., the fused feature in this invention). This vector is input into a fully connected layer (2 layers, with 64 and 2 neurons respectively), and the softmax activation function outputs two types of results: Result 1: Classification result (probability value) of "normal or abnormal call"; Result 2: Classification result (probability value) of "normal or abnormal device status". If the probability of a certain result is ≥0.8, it is determined to be the corresponding status (e.g., "abnormal call").

[0050] 5. The results output will include the determined "call status, device status" and corresponding feature association information (such as "call anomaly association feature: call success rate decrease rate ≥10% / step"), as the model processing results output for subsequent optimization suggestion modules to use.

[0051] Step S30: Perform feature fusion processing and call failure reason calculation on the call trend result and the abnormal device status to obtain the target call failure reason.

[0052] This invention analyzes the reasons for call failures. The specific process is as follows: by analyzing the data and device status of the failed call, complex reasons such as network latency and insufficient resources are identified, and failure cases are classified to summarize the conditions for the occurrence of high-frequency failures.

[0053] Specifically, the call trend result and the abnormal device status are subjected to feature fusion processing to obtain fused features; a fault feature library is constructed, wherein the fault feature library includes the correspondence between call failure reasons and fault features; the Euclidean distance between the fused features and multiple fault features in the fault feature library is calculated, and the target fault feature with the smallest Euclidean distance to the fused features in the fault feature library is extracted, and the target fault feature is used as the target call failure reason.

[0054] The specific process for analyzing call failure reasons (including identifying complex reasons such as network latency and insufficient resources) is as follows: This invention identifies the reasons for failure based on the call trend results of LSTM and the device status features of CNN through "multi-feature fusion matching". The specific process is as follows: 1. Feature Fusion: The “call failure trend feature” (such as “call duration increase feature”) output by LSTM is concatenated with the “device status feature vector” output by CNN to form a 130-dimensional fused feature; 2. Fault Feature Database Matching: Construct a "Failure Reason-Feature Mapping Database" (e.g., "Network latency → Increased call duration + Network bandwidth feature value < 0.3; Insufficient resources → CPU / memory feature value > 0.9 + Decreased call success rate"). 3. Cause identification: Calculate the Euclidean distance between the fused feature and each cause feature in the fault feature library. The entry with the smallest distance and less than the preset threshold (e.g., 0.2) is the specific cause of the current call failure (e.g., "network latency").

[0055] The specific process of classifying failure cases and summarizing high-frequency failure conditions is as follows: 1. Failure Case Classification Dimensions Definition: 3 core classification dimensions are defined: Dimension 1: Reason for failure (e.g., "network latency, insufficient resources, interface abnormality"); Dimension 2: Device type (e.g., "smart speaker, smart camera"); Dimension 3: Application scenario (e.g., "peak hour call, low load call"). 2. Automatic case classification: Match the "fusion features, device type, and scene tags" of each failed case with the classification dimensions and automatically classify it into the corresponding category (such as "network latency - smart camera - peak period call" category). 3. Summary of high-frequency failure conditions: Statistically analyze the occurrence frequency of each type of case, and extract the common characteristics of the top 5 most frequent cases. For example, the common characteristic of the "network latency - smart camera - peak period call" category is "call concurrency > 50 times / second + network bandwidth < 10Mbps", which is the condition for the occurrence of this type of high-frequency failure.

[0056] Furthermore, this invention also incorporates trend prediction. Based on historical data, the system predicts potential resource bottlenecks or call anomalies and issues early warnings. The specific process of trend prediction (predicting resource bottlenecks or call anomalies) is based on the time-series prediction capabilities of LSTM, combined with device status data to achieve trend warnings. The steps are as follows: 1. Predictive data preparation: Combine "call time series data + device status data" to construct a "call-status" fused time series sequence (such as "call success rate, CPU utilization" data for 60 consecutive time steps). 2. LSTM prediction model training: The output layer is extended on the basis of the original LSTM network to add prediction outputs of "call success rate and CPU utilization in the next 10 minutes"; the model is trained with the historical "call-state" time series data of 30 days and the parameters are optimized with mean squared error (MSE); 3. Resource bottleneck or call anomaly prediction: Input the real-time "call-status" sequence into the model to obtain the prediction results for the next 10 minutes: if the prediction is "call success rate < 80%", it is judged as "call anomaly trend"; if the prediction is "CPU utilization > 90% / memory utilization > 85%", it is judged as "resource bottleneck trend". 4. Warning Trigger: When the prediction results meet the above conditions, a warning message will be pushed to the optimization suggestion module in advance.

[0057] Step S40: Based on the reason for the target call failure, perform optimization strategy matching and dynamic adjustment of the optimization strategy to obtain target optimization suggestions, and adjust the target faulty device according to the target optimization suggestions.

[0058] like Figure 5 As shown, the specific implementation process of the optimization suggestion module is as follows: Based on the analysis results stored in the distributed database, the rule-based inference engine in the optimization suggestion module quickly matches and infers the analysis results according to the preset optimization rule base (such as suggesting network configuration optimization when network latency exceeds a set threshold; suggesting load balancing when device load is consistently higher than 80%), and generates preliminary optimization suggestions.

[0059] Meanwhile, the reinforcement learning algorithm continuously learns and optimizes the implementation effect of suggestions through interaction with the super device, dynamically adjusting the optimization strategy. For example, when the analysis results show that the screen projection capability of a certain conference room super device has a high failure rate due to network congestion, the optimization suggestion module first generates preliminary suggestions for optimizing network configuration based on the rule engine, such as adjusting network bandwidth allocation and optimizing network routing. Then, the reinforcement learning algorithm dynamically adjusts the network parameter configuration based on the actual optimized call effect, further reducing the call failure rate. The generated optimization suggestions are fed back to the super device in the form of a standardized instruction set through the device management interface of the open-source HarmonyOS system, guiding it to perform corresponding optimization operations, and the detailed content and expected effect of the optimization suggestions are synchronized to the visualization display module.

[0060] Specifically, a preset rule engine is determined, and optimization strategies are matched to the reasons for the target call failure using the preset rule engine to obtain preliminary optimization suggestions; a reinforcement learning algorithm is used to dynamically adjust the optimization strategies of the preliminary optimization suggestions to obtain target optimization suggestions; a target optimization instruction is generated based on the target optimization suggestions, and the target optimization instruction is sent to the corresponding target faulty device.

[0061] The optimization suggestion module in this invention combines data analysis results to generate actionable optimization strategies to improve equipment operating efficiency, including: 1. Rule Engine: Quickly generate optimization suggestions through preset rules (such as optimizing routing configuration when network latency exceeds 100ms), and the rule base is continuously expanded based on actual application scenarios.

[0062] 2. Reinforcement learning algorithm: This invention learns optimization strategies through interaction between the device and the environment, and gradually improves the optimization scheme. Example optimization measures include: dynamically adjusting the call priority and reallocating device resources.

[0063] 3. Execution Feedback: Optimization suggestions are fed back to the device in the form of instructions via API. The system monitors the implementation effect in real time and records the improvement results.

[0064] like Figure 6As shown, the specific implementation process of the security monitoring module in this invention is as follows: When a capability call request from a super device is initiated, the security monitoring module performs blockchain-based permission verification on each request. The call request must carry the caller's identity information and call permission credentials encrypted with blockchain. The security monitoring module verifies the authenticity and validity of the credentials through the blockchain network, ensuring that only legitimate callers can initiate capability calls. During data transmission, the AES encryption algorithm is used to encrypt the call data, ensuring the security of data transmission in the distributed network. Simultaneously, the operating status of the super device and its terminal devices is monitored in real time. Through security monitoring agents deployed on the devices, information such as system logs and network traffic is collected. The collected information is analyzed using an intrusion detection system (IDS) and a security information and event management system (SIEM) to detect any abnormal behavior or security vulnerabilities. Once a security threat is detected, such as illegal device intrusion or abnormal data leakage, security protection measures are immediately initiated, such as cutting off network connections and locking the device. The security event is recorded and stored in the blockchain to ensure the traceability of the security event.

[0065] Further, the system obtains the permission information of the super device, including the device ID, the operation and maintenance personnel account, and the permission scope; it hashes the permission information to obtain a permission digest; it performs consensus verification on the permission digest and writes it into the blockchain to obtain a distributed permission library; when a call request from a target personnel is received and the target personnel have carried preset information, the system performs identity verification, permission scope matching, and anti-replay verification on the target personnel based on the preset information and the distributed permission library to obtain a verification result, wherein the preset information includes a digital certificate, the requested operation content, and a randomly generated 32-bit temporary random number; if the verification result is successful, a call authorization success result is sent to the target personnel; if the verification result is unsuccessful, a call authorization failure result is sent to the target personnel.

[0066] The security monitoring module ensures the security of the call process and provides multi-layered protection. It also includes identity authentication, a distributed permission verification mechanism based on blockchain technology to prevent unauthorized access.

[0067] The specific authentication process is as follows: This invention adopts a consortium blockchain architecture (nodes include super devices, operation and maintenance management platforms, and distributed authentication centers) to implement permission verification. The specific process is as follows: 1. Initialization of permission information on the blockchain: The unique identifier (device ID), operation and maintenance personnel account, and permission scope (e.g., "device A is only allowed to call audio capabilities by operation and maintenance account B") of all super devices are pre-processed into permission digests using SHA-256 hashing. After consensus verification by each node of the consortium blockchain, the digests are written into the blockchain ledger to form an immutable distributed permission library. At the same time, a digital certificate based on the ECDSA algorithm (binding the entity identifier and public key) is assigned to each subject (device or operation and maintenance personnel).

[0068] 2. Authentication Request Initiation: When an entity (such as maintenance personnel or equipment) initiates a capability call request, it must carry: a. its own digital certificate (to prove the legitimacy of its identity); b. the content of the requested operation (such as "call the lighting capability of device C"); c. a randomly generated 32-bit temporary random number (to prevent replay attacks).

[0069] 3. On-chain permission verification: After receiving the request, the authentication center node in the consortium blockchain performs the following verification steps: a. Identity legitimacy verification: Verify the signature validity of the requester's digital certificate through the public key library in the blockchain ledger to confirm the authenticity of the subject's identity; b. Permission scope matching: Hash the requester's identifier (device ID / operation account) and query its corresponding permission scope in the blockchain permission library to determine whether the "request operation" is within the permission scope (e.g., "Does operation account B have the right to call the audio capabilities of device A"); c. Anti-replay verification: Query the historical authentication records in the blockchain to confirm that the temporary random number for this request has not been used.

[0070] 4. Verification Result Feedback and Recording: If all the above verifications pass, the authentication center node returns a "authorization passed" signature result (including the operation validity period, such as 5 minutes) to the requester, allowing the initiation of capability calls; if the verification fails (such as insufficient permissions or identity forgery), a "authorization denied" result is returned, and the abnormal request is marked; regardless of the verification result, the complete information of this authentication (requester, operation content, verification result, and timestamp) will be written into the blockchain ledger for subsequent auditing and traceability.

[0071] In addition, this invention also includes: a) Data transmission encryption: using the AES encryption algorithm to protect the security of data transmission. b) Abnormal behavior detection: monitoring the device's operating status, promptly identifying and preventing abnormal operations, such as unauthorized intrusion or data leakage.

[0072] like Figure 7 , Figure 8 ,as well as Figure 9As shown, this invention also includes a visualization module. This module features an interactive visualization interface that visually displays the super device's capability call data and optimization suggestions through various formats such as charts, dashboards, and topology diagrams. Specifically, the visualization module includes: 1. Real-time call data statistics charts displaying key indicators such as call frequency, success rate, and failure rate for each capability; 2. A device status heatmap visually presenting information such as the load and network status of each terminal device in the super device; 3. A call failure reason analysis report detailing the reasons for call failures identified by the analysis module and related abnormal device indicators; 4. A detailed list of optimization suggestions displaying the content, basis, and expected effects of the optimization suggestions generated by the optimization suggestion module; 5. An implementation effect evaluation area evaluating and displaying the call effect after the optimization suggestions are implemented. Maintenance personnel can perform data filtering, querying, and drill-down operations through the interface, such as filtering call data by time period, device type, and capability type, clicking on charts to view detailed data records, and supporting data export functionality for convenient further analysis and archiving by maintenance personnel. Meanwhile, the visualization module and the optimization suggestion module work together in real time. When optimization suggestions are generated or their implementation status is updated, the interface automatically refreshes to display the latest information, helping maintenance personnel to understand the system's operating status and optimization progress in a timely manner, and achieving comprehensive visual management of the super device's capability calls.

[0073] The functions of the super device capability call monitoring and optimization system based on distributed technology in this invention include: 1. Real-time monitoring function: The system uses distributed acquisition technology to reflect the operating status of super equipment in real time, which facilitates rapid detection and response to problems.

[0074] 2. In-depth analysis of call failures: Utilizes AI technology to deeply mine data on call failures, helping to pinpoint the root cause of complex failures.

[0075] 3. Adaptive optimization function: Combining rule engine and reinforcement learning, it automatically generates and implements optimization schemes to reduce the failure rate.

[0076] 4. Security protection function: Provides full-process security protection from identity authentication to anomaly detection to ensure the reliability of the equipment operating environment.

[0077] 5. Visual management function: Displays data and analysis results intuitively in various forms, improving operation and maintenance efficiency and supporting decision-making.

[0078] Technical effects: 1. Improve call efficiency: Optimization strategies significantly reduce latency and resource conflicts, increase call success rate to over 99%, and reduce call latency by 20%.

[0079] 2. Enhance system stability: Precise analysis and optimization improve operational reliability and increase equipment resource utilization by 15%.

[0080] 3. Ensure security: Multiple protection measures ensure data security and reliable operation, preventing unauthorized access and data leakage.

[0081] 4. Simplify operation and maintenance management: Comprehensive visualization and intelligent tools reduce management complexity and improve operation and maintenance efficiency.

[0082] In summary, the super device capability call monitoring and optimization system based on distributed technology in this invention can effectively monitor the capability call status of super devices in real time, accurately analyze the reasons for call failures, generate and implement optimization suggestions, and at the same time ensure the security of the call process, significantly improve the efficiency and stability of super device capability calls, and provide strong support for the development and operation of IoT applications based on the open-source HarmonyOS system.

[0083] Furthermore, such as Figure 10 As shown, based on the above-described method for invoking super device capabilities using distributed technology, this invention also provides a system for invoking super device capabilities using distributed technology, wherein the system includes: The data preprocessing module 51 is used to acquire device operating status data and call data, and preprocess the device operating status data and the call data to obtain target device operating status data and target call data; The device status prediction module 52 is used to perform call trend prediction processing based on the target call data to obtain call trend results, and to perform device status prediction processing based on the target device operating status data to obtain abnormal device status. The failure reason calculation module 53 is used to perform feature fusion processing and failure reason calculation on the call trend result and the abnormal device status to obtain the target call failure reason; The optimization suggestion generation module 54 is used to match optimization strategies and dynamically adjust optimization strategies based on the reasons for the target call failure, obtain target optimization suggestions, and adjust the target faulty equipment according to the target optimization suggestions.

[0084] Furthermore, such as Figure 11 As shown, based on the above-mentioned method and system for calling the capabilities of super devices based on distributed technology, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 11 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0085] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a distributed super device capability invocation program 40, which can be executed by the processor 10 to implement the distributed super device capability invocation method of this application.

[0086] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the super device capability invocation method based on distributed technology.

[0087] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface.

[0088] In one embodiment, when the processor 10 executes the distributed technology-based super device capability invocation program 40 in the memory 20, it implements the steps of the distributed technology-based super device capability invocation method as described above.

[0089] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a super device capability invocation program based on distributed technology, and when the super device capability invocation program based on distributed technology is executed by a processor, it implements the steps of the super device capability invocation method based on distributed technology as described above.

[0090] In summary, this invention provides a method, system, terminal, and storage medium for super device capability invocation based on distributed technology. The method includes: acquiring device operating status data and invocation data; preprocessing the device operating status data and the invocation data to obtain target device operating status data and target invocation data; performing invocation trend prediction processing based on the target invocation data to obtain invocation trend results; and performing device status prediction processing based on the target device operating status data to obtain abnormal device status. The invention performs feature fusion processing and calculates call failure reasons based on the call trend results and the abnormal device status to obtain the target call failure reason. Based on the target call failure reason, optimization strategies are matched and dynamically adjusted to obtain target optimization suggestions. The target faulty device is then adjusted according to these suggestions. This invention, by predicting call trends and device status based on device operating status data and call data, and then determining the call failure reason and applying optimization strategies, can effectively improve the efficiency and stability of super device capability invocation.

[0091] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0092] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0093] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for invoking the capabilities of a super device based on distributed technology, characterized in that, The method for invoking the capabilities of super devices based on distributed technology includes: Acquire device operating status data and call data, and preprocess the device operating status data and call data to obtain target device operating status data and target call data; Based on the target call data, call trend prediction processing is performed to obtain call trend results, and based on the target device operating status data, device status prediction processing is performed to obtain abnormal device status. The call trend results and the abnormal device status are subjected to feature fusion processing and call failure reason calculation to obtain the target call failure reason; Based on the reasons for the target call failure, optimization strategies are matched and dynamically adjusted to obtain target optimization suggestions, and the target faulty equipment is adjusted according to the target optimization suggestions.

2. The method for invoking super device capabilities based on distributed technology according to claim 1, characterized in that, The process of acquiring device operating status data and call data, and preprocessing the device operating status data and call data to obtain target device operating status data and target call data, specifically includes: Determine the target acquisition source. If the target acquisition source reaches the preset acquisition frequency, acquire the device operation status data and call data of the super device. The device operation status data includes CPU utilization, memory utilization, and target network parameters. The call data includes the number of calls, call success rate, call failure reason, and call duration. The device operating status data and the call data are preprocessed to obtain target device operating status data and target call data. The preprocessing includes data cleaning, format conversion and data encryption.

3. The method for invoking super device capabilities based on distributed technology according to claim 1, characterized in that, The step of performing call trend prediction processing based on the target call data to obtain the call trend result specifically includes: The target call data is subjected to time-series alignment processing to obtain a call timing sequence; The missing value completion and outlier deletion processes are performed on the call sequence to obtain the target call sequence. Determine the initial long short-term memory network model and obtain historical call time series data; The initial long short-term memory network model is trained based on the historical call time series data, and the trained initial long short-term memory network model is optimized using the cross-entropy loss function to obtain the call trend model. The target call time sequence is input into the call trend model to obtain the call trend result.

4. The method for invoking the capabilities of a super device based on distributed technology according to claim 1, characterized in that, The step of performing equipment status prediction processing based on the target equipment operating status data to obtain abnormal equipment status specifically includes: A state feature matrix is ​​constructed based on the target device's operating status data, and the state feature matrix is ​​normalized to obtain a normalized state feature matrix. Construct a target convolutional neural network and input the normalized state feature matrix into the target convolutional neural network to output a device operating state feature vector; A preset normal equipment state feature library is determined, and the cosine similarity between the equipment operating state feature vector and the preset normal equipment state feature library is calculated to obtain the cosine similarity result. If the cosine similarity result is less than a preset threshold, the target device's operating status data is determined to be abnormal, and the abnormal device status is obtained.

5. The method for invoking super device capabilities based on distributed technology according to claim 1, characterized in that, The step of performing feature fusion processing and call failure reason calculation on the call trend result and the abnormal device status to obtain the target call failure reason specifically includes: The call trend results and the abnormal device status are subjected to feature fusion processing to obtain fused features; Construct a fault feature library, wherein the fault feature library includes the correspondence between call failure reasons and fault features; Calculate the Euclidean distance between the fused feature and multiple fault features in the fault feature library, extract the target fault feature with the smallest Euclidean distance from the fault feature library to the fused feature, and use the target fault feature as the target call failure reason.

6. The method for invoking the capabilities of a super device based on distributed technology according to claim 1, characterized in that, The process of matching and dynamically adjusting optimization strategies based on the reasons for the target call failure to obtain target optimization suggestions, and adjusting the target faulty device according to the target optimization suggestions, specifically includes: A preset rule engine is determined, and optimization strategies are matched to the reasons for the failure of the target call using the preset rule engine to obtain preliminary optimization suggestions. The initial optimization suggestions are dynamically adjusted using a reinforcement learning algorithm to obtain the target optimization suggestions. Based on the target optimization suggestions, a target optimization instruction is generated and sent to the corresponding target faulty device.

7. The method for invoking super device capabilities based on distributed technology according to claim 1, characterized in that, The process of matching and dynamically adjusting optimization strategies based on the cause of the target call failure to obtain target optimization suggestions, and adjusting the target faulty device according to the target optimization suggestions, further includes: Obtain the permission information of the super device, wherein the permission information includes the device ID, the operation and maintenance personnel account, and the scope of permission; The permission information is hashed to obtain a permission digest; The permission digest is verified through consensus and written into the blockchain to obtain a distributed permission library; When a call request from a target person is received and the target person has carried preset information, the identity of the target person is verified, the scope of permissions is matched, and the anti-replay verification is performed based on the preset information and the distributed permission library to obtain the verification result. The preset information includes a digital certificate, the request operation content, and a randomly generated 32-bit temporary random number. If the verification result is successful, the authorization approval result will be sent to the target personnel. If the verification result is unsuccessful, the authorization failure result will be sent to the target personnel.

8. A super device capability invocation system based on distributed technology, characterized in that, The super device capability invocation system based on distributed technology includes: The data preprocessing module is used to acquire device operating status data and call data, and preprocess the device operating status data and the call data to obtain target device operating status data and target call data; The device status prediction module is used to perform call trend prediction processing based on the target call data to obtain call trend results, and to perform device status prediction processing based on the target device operating status data to obtain abnormal device status. The failure reason calculation module is invoked to perform feature fusion processing and failure reason calculation on the call trend result and the abnormal device status to obtain the target call failure reason; The optimization suggestion generation module is used to match optimization strategies and dynamically adjust optimization strategies based on the reasons for the target call failure, obtain target optimization suggestions, and adjust the target faulty equipment according to the target optimization suggestions.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and a super device capability invocation program based on distributed technology stored in the memory and executable on the processor. When the super device capability invocation program based on distributed technology is executed by the processor, it implements the steps of the super device capability invocation method based on distributed technology as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a super device capability invocation program based on distributed technology, which, when executed by a processor, implements the steps of the super device capability invocation method based on distributed technology as described in any one of claims 1-7.

Images