A data processing method for intelligent gateway protocol

Abstract

The present invention proposes a data processing method for an intelligent gateway protocol, including: collecting data transmission data, building an adaptive deep learning model, and dynamically optimizing the protocol conversion strategy; building a load prediction model, optimizing the protocol conversion strategy and dynamically allocating network resources; building an edge computing model and training the edge computing model, and using the edge computing capability of the edge computing model to locally process data and make decisions; building a collaborative control model and a feedback mechanism, and using the collaborative control model to perform real-time collaborative control of intelligent gateway devices; using blockchain technology to encrypt and verify the data transmission process; designing the optimal configuration scheme of intelligent agent technology, and simultaneously monitoring the status of intelligent gateway devices in real time, predicting and pre-maintaining potential failures of intelligent gateway devices. The present invention improves data transmission efficiency, system performance and security by introducing adaptive deep learning, edge computing, reinforcement learning, blockchain and intelligent agent technologies.

Description

Technical Field

[0001] The invention belongs to the field of intelligent gateway protocols, and in particular relates to a data processing method for intelligent gateway protocols. Background Art

[0002] With the widespread application of the Internet of Things (IoT) and smart devices, smart gateways are becoming increasingly important as key nodes connecting various devices, sensors, and cloud platforms. Smart gateways not only need to process large amounts of data from different devices, but also need to support multiple communication protocols to ensure seamless data transmission and interaction. At present, smart gateway technology is relatively mature and can support multiple protocols (such as Modbus, MQTT, HTTP / HTTPS, DL / T698, IEC61850, etc.) to achieve data collection, protocol conversion, and transmission. However, the existing smart gateway technology still has some significant problems and shortcomings in practical applications.

[0003] First, when processing data of multiple protocols, existing intelligent gateways often need to pre-configure various protocol adapters and conversion rules. This static configuration method is difficult to cope with dynamically changing network environments and device requirements, resulting in low data transmission efficiency and low resource utilization. Secondly, with the continuous increase in the number of intelligent devices and the amount of data, the network load and data processing pressure have risen sharply. It is difficult for existing gateways to effectively perform load balancing and resource management, which is prone to data transmission bottlenecks and system performance degradation. In addition, the security of intelligent gateways is also an important challenge. Existing security measures (such as simple encrypted transmission) are insufficient in the face of complex network attacks and cannot fully guarantee the security and integrity of data. Finally, existing intelligent gateways are also difficult to configure and maintain. The wide variety of devices and complex protocols increase the complexity and cost of configuration and maintenance, making it difficult to achieve automated management and efficient operation and maintenance.

[0004] Therefore, there is an urgent need for a new data processing method that can dynamically adapt to changing network environments and equipment requirements, provide efficient protocol conversion and load balancing capabilities, ensure the security of data transmission, and simplify the configuration and maintenance process. In order to solve the above problems, the present invention proposes an intelligent gateway data processing method based on adaptive deep learning, edge computing, reinforcement learning and blockchain technology, aiming to overcome the shortcomings of the prior art and provide a smarter, more efficient and safer solution. Summary of the invention

[0005] The purpose of this invention is to propose a data processing method for an intelligent gateway protocol. By introducing adaptive deep learning, edge computing, reinforcement learning, blockchain and intelligent agent technology, an innovative intelligent gateway data processing method is provided, which can effectively solve various problems in the prior art, improve data transmission efficiency, system performance and security, simplify the configuration and maintenance process, and has broad application prospects and practical value.

[0006] In order to achieve the above object, the present invention provides a data processing method for an intelligent gateway protocol, the method comprising:

[0007] S1. Collect data transmission data of the intelligent gateway under different protocols, pre-process the collected data, build an adaptive deep learning model, analyze the data transmission characteristics of different protocols, and dynamically optimize the protocol conversion strategy according to the real-time changes of network conditions and data traffic; wherein the loss function of the adaptive deep learning model is It is expressed as follows:

[0008] ;

[0009] in, represents the model parameters, Indicates the actual performance index, Represents the model prediction value, N represents the total number of samples, and i represents the i-th sample;

[0010] The adaptive deep learning model uses an adaptive optimization algorithm to adjust the learning rate; the adaptive optimization algorithm is expressed as follows:

[0011] ;

[0012] in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the adaptive deep learning model, Represents a constant used to prevent division by zero;

[0013] S2. Collect the load data of the intelligent gateway in different time periods and extract the load data features, build a load prediction model, and use the load prediction model to optimize the protocol conversion strategy and dynamically allocate network resources according to the real-time monitored network traffic and data load;

[0014] S3. Build and train an edge computing model, and use the edge computing capabilities of the edge computing model to process data and make decisions locally;

[0015] S4. Construct a collaborative control model and a feedback mechanism, use the collaborative control model to perform real-time collaborative control on the intelligent gateway device, and use the feedback mechanism to dynamically adjust the control strategy; wherein the collaborative control model is expressed as follows:

[0016] ;

[0017] Where u(t) represents the control input, e(t) represents the error signal, , and represent the proportional, integral and differential control parameters at time t, respectively, represents the integral operation, dt represents the time differential, de(t) represents the differential of the error signal, and dt represents the time differential;

[0018] Through historical data and real-time feedback, the design target optimization algorithm adjusts the control parameters , and ; Wherein, the target optimization algorithm is expressed as follows:

[0019] ;

[0020] in, represents the target optimization function, represents the weight coefficient, and the control parameters are optimized by minimizing the objective function;

[0021] S5. Define smart contracts and use blockchain technology to encrypt and verify the data transmission process;

[0022] S6. Design intelligent agent technology, which automatically selects the optimal configuration solution for the intelligent gateway device according to the application scenario and type of the intelligent gateway device, monitors the status of the intelligent gateway device in real time, and predicts and maintains potential failures of the intelligent gateway device in advance.

[0023] Furthermore, the preprocessing includes processing missing values ​​using mean filling or interpolation methods, removing outliers using the IQR method, and standardization.

[0024] Furthermore, the deep learning model includes a long short-term memory network or a convolutional neural network; in S1, the real-time data of the intelligent gateway is input into the trained deep learning model to obtain the protocol conversion optimization strategy, which is expressed as follows:

[0025] ;

[0026] in, represents the real-time input feature vector, represents the optimization strategy of the model output, Represents a deep learning model;

[0027] According to the optimization strategy output by the trained deep learning model, the protocol conversion parameters of the intelligent gateway are adjusted in real time, as shown below:

[0028] ;

[0029] in, represents the optimized protocol parameters, g represents the parameter adjustment function;

[0030] Establish a feedback mechanism for the deep learning model, monitor the protocol conversion effect in real time, and collect new data to update and train the model, as shown below:

[0031] ;

[0032] in, Represents a new data set for continuous training and optimization of the model. Indicates the transmission characteristics under different protocols. Represents the corresponding performance indicator, N represents the total number of samples, and M represents the number of newly collected samples.

[0033] Furthermore, in S2, a load data set L is constructed and the loss function of the load prediction model is designed. as follows:

[0034] ;

[0035] in, represents the model parameters of the load prediction model, represents a negative prediction value, represents the negative true load value, and T represents the total number of samples in the prediction time period;

[0036] The learning rate is adjusted using an adaptive optimization algorithm, as shown below:

[0037] ;

[0038] in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the load prediction model, Represents a constant used to prevent division by zero;

[0039] Deploy a real-time load monitoring system on the smart gateway to collect CPU usage in real time , memory usage , bandwidth usage and I / O usage , define the real-time load vector ;

[0040] Input the real-time load vector into the trained load prediction model , the load situation in the future is predicted as follows:

[0041] ;

[0042] in, represents the load prediction model, represents the negative predicted load value, Indicates the load prediction result;

[0043] According to the load prediction results , dynamically adjust the resource allocation strategy of the intelligent gateway , which is expressed as follows:

[0044] ;

[0045] in, represents the resource allocation strategy, g represents the resource allocation function;

[0046] Design a load balancing strategy and adjust the resource allocation of each task, as shown below:

[0047] ;

[0048] in, Indicates the set resource usage threshold. Indicates the current CPU usage. Indicates the current bandwidth usage. Indicates the current memory usage. Indicates the current I / O usage. Indicates reallocation of CPU resources. Indicates reallocation of memory resources. Indicates reallocation of broadband resources. Indicates the reallocation of I / O usage resources.

[0049] Furthermore, the loss function of the edge computing model is It is expressed as follows:

[0050] ;

[0051] in, represents the model parameters, represents the predicted value, represents the real data value, and M represents the number of newly collected samples;

[0052] Deploy the trained edge computing model to the edge computing node, as shown below:

[0053] ;

[0054] in, represents the edge computing model, Represents input data, Indicates the processing result.

[0055] Furthermore, the edge device data collected in real time is input into the deployed edge computing model for local processing and decision-making, as shown below:

[0056] ;

[0057] in, Indicates the timestamp, Indicates real-time input data, Represents a real-time dataset;

[0058] The edge computing model is used to process real-time input data and output the processing results, which are expressed as follows:

[0059] ;

[0060] in, Indicates the processing result.

[0061] Furthermore, in S4, the real-time data fusion result is used to perform association analysis to build the relationship between the intelligent gateway devices, which is expressed as follows:

[0062] ;

[0063] in, represents the device association matrix, represents the strength of association between device i and device j, which is obtained by calculating the correlation coefficient; wherein the correlation coefficient is calculated as follows:

[0064] ;

[0065] in, and Represent the data means of device i and device j respectively, represents the data from the kth data source at the i-th time point, Represents the data from the kth data source at the jth time point.

[0066] Furthermore, real-time monitoring of the equipment's operating status and control effects, and collection of control effect data , which is expressed as follows:

[0067] ;

[0068] in, represents the control effect index of device j at the i-th time point, represents the control effect index of device n at the i-th time point, Indicates a timestamp;

[0069] Using adaptive control algorithm, the control parameters are optimized through real-time data feedback, which is expressed as follows:

[0070] ;

[0071] in, Represents the learning rate, which dynamically adjusts the control parameters through the gradient descent method; , and They represent the proportional, integral and differential control parameters at time t+1 respectively, represents the partial derivative of the objective function J, Represents the partial derivative operator.

[0072] Furthermore, the S5 specifically includes:

[0073] The real-time data generated by the intelligent gateway is packaged to form a data block B to be transmitted, and the data block is digitally signed using the private key of the gateway;

[0074] Pack multiple data blocks B to generate block C in the blockchain, and distribute the generated block C to each node in the blockchain network for verification and storage;

[0075] The Byzantine fault-tolerant consensus mechanism is used to verify the received block C by each verification node;

[0076] Define rules for the smart contract to automatically execute data transmission on the blockchain; the rules are that when the data transmission meets the triggering conditions of the smart contract, the transmission operation is automatically executed;

[0077] The verified data blocks B and C are stored in the blockchain, and a blockchain-based data query interface is provided.

[0078] Furthermore, the S6 specifically includes:

[0079] Deploy intelligent agents on the intelligent gateway and associated devices, initialize the intelligent agents, and deploy the intelligent agents to the intelligent gateway and associated devices at the same time; wherein each intelligent agent includes initial configuration parameters and management rules;

[0080] The intelligent agent monitors the intelligent gateway and associated devices in real time, collects the operating status and performance data of the intelligent gateway and associated devices, and transmits the collected monitoring data to the intelligent agent for processing and storage to form a historical data set H;

[0081] Based on the historical data set H, the intelligent agent constructs an adaptive configuration model; wherein the objective function F of the adaptive configuration model is expressed as follows:

[0082] ;

[0083] in, Indicates the associated device i at time The calculated optimal configuration parameters, Indicates that device i is at time Actual monitoring data;

[0084] The optimization algorithm is used to dynamically adjust the configuration parameters of the associated devices, as shown below:

[0085] ;

[0086] in, represents the velocity vector, represents the inertia weight, and They represent learning factors, and Represents a random number, represents the optimal position of an individual, represents the global optimal position, represents the velocity vector at the next moment, Indicates the individual position at the next moment;

[0087] The intelligent agent uses monitoring data and adaptive configuration models to detect anomalies in the operating status of associated devices. When the intelligent agent detects an anomaly, it immediately triggers the fault handling mechanism and performs recovery operations.

[0088] The intelligent agent adjusts the configuration parameters of the associated devices in real time according to the monitoring data and anomaly detection results. At the same time, a feedback mechanism is established between the intelligent agent and the associated devices to optimize the configuration parameters of the associated devices through real-time monitoring and data analysis.

[0089] The beneficial technical effects of the present invention are at least as follows:

[0090] (1) The present invention adopts an adaptive deep learning algorithm, which can dynamically analyze and optimize the transmission performance of different protocols (such as Modbus, MQTT, HTTP / HTTPS, DL / T698, IEC61850, etc.), and autonomously adjust the protocol conversion strategy according to changes in network conditions and data traffic to ensure the efficiency and stability of data transmission. At the same time, through intelligent load balancing technology, real-time monitoring and dynamic allocation of network resources can be carried out to avoid overload and bottlenecks, thereby improving the overall performance of the system.

[0091] (2) The present invention utilizes the edge computing capabilities of the intelligent gateway to achieve local data processing and rapid decision-making, reduce dependence on the central server, and improve the response speed of the system. Edge computing can not only achieve local energy optimization and control strategy decomposition, but also conduct real-time collaborative control with IoT devices, dynamically adjust control strategies according to actual needs, and improve the flexibility and autonomy of the system.

[0092] (3) The present invention introduces a reinforcement learning algorithm. The intelligent gateway can autonomously adjust and optimize data processing strategies based on historical data and real-time feedback, and continuously learn and adapt to new data patterns and network environments. The self-optimization mechanism of reinforcement learning ensures that the intelligent gateway always maintains the best operating state in a complex and dynamic environment, improving the efficiency and accuracy of data processing.

[0093] (4) The present invention uses blockchain technology to encrypt and verify data to ensure the integrity and tamper-proofness of data during transmission. Each data packet is encrypted and recorded in the blockchain before transmission. The recipient decrypts and verifies the data through the blockchain to ensure the authenticity and security of the data. In addition, the decentralized management of smart gateways is achieved through blockchain, reducing the risk of single point failures and enhancing the robustness and fault tolerance of the system.

[0094] (5) The present invention uses intelligent agent technology to achieve automatic configuration and intelligent maintenance of intelligent gateways. The intelligent agent automatically selects and applies the optimal configuration solution according to different application scenarios and device types, simplifies the configuration process, and improves configuration efficiency. At the same time, the intelligent agent monitors the system status in real time, predicts potential failures, and performs maintenance and repairs in advance, reducing system downtime and maintenance costs, and improving system stability and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0096] Figure 1 The present invention is a flow chart of a data processing method for an intelligent gateway protocol. DETAILED DESCRIPTION

[0097] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0098] like Figure 1 As shown, an embodiment of the present invention provides a data processing method for an intelligent gateway protocol, the method comprising the following steps S1-S6:

[0099] S1. Collect data transmission data of intelligent gateways under different protocols, pre-process the collected data, build an adaptive deep learning model, analyze the data transmission characteristics of different protocols, and dynamically optimize the protocol conversion strategy according to the real-time changes of network conditions and data traffic.

[0100] Specifically, S1 includes the following steps:

[0101] Collect data transmission records of smart gateways under different protocols (such as Modbus, MQTT, HTTP / HTTPS, DL / T698, IEC61850, etc.), including key performance indicators such as transmission delay (T), packet loss rate (P), bandwidth utilization (B), data packet size (S), and network congestion (C).

[0102] Defining Datasets ,in represents the i-th feature under different protocols, Represents the corresponding performance index, N represents the total number of samples, and i represents the i-th sample.

[0103] Among them, the missing values ​​are processed using mean filling or interpolation methods:

[0104] (1);

[0105] Where n represents the number of non-missing samples in the data set. represents the processed eigenvalue, represents the characteristic value of the j-th non-missing sample, and j represents the j-th sample;

[0106] Use IQR (interquartile range) method to remove outliers:

[0107] (2);

[0108] Among them, Lower Bound represents the lower boundary, Q3 represents the third quartile, Q1 represents the first quartile, and Upper Bound represents the upper boundary;

[0109] Standardization:

[0110] (3);

[0111] Among them, among them, is the sample mean, is the sample standard deviation, Represents the standardized eigenvalue.

[0112] Furthermore, the key features that affect transmission performance are extracted from the preprocessed data, including transmission delay (T), packet loss rate (P), bandwidth utilization (B), data packet size (S), network congestion (C), etc. Define the feature matrix ,in, represents the i-th feature.

[0113] Principal component analysis (PCA) is used to select the most representative features, reduce data dimensions, and reduce computational complexity.

[0114] (4);

[0115] Among them, Z is the feature matrix after dimensionality reduction, and W is the principal component matrix.

[0116] Furthermore, a deep learning model suitable for processing time series data and multi-dimensional features is selected, such as long short-term memory network (LSTM) or convolutional neural network (CNN). The model structure is: input layer (feature vector) → hidden layer (LSTM / CNN) → output layer (optimization strategy).

[0117] Use the collected dataset to train the model and adjust the model parameters to minimize the prediction error:

[0118] (5);

[0119] in, are model parameters, is the actual performance indicator, is the model prediction value, Represents the model loss.

[0120] During model training, adaptive optimization algorithms (such as Adam or RMSprop) are used to adjust the learning rate to improve the convergence speed and accuracy of the model:

[0121] (6);

[0122] (7);

[0123] (8);

[0124] (9);

[0125] in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the adaptive deep learning model, Represents a constant used to prevent division by zero;

[0126] Preferably, the real-time data of the intelligent gateway is input into the trained deep learning model to obtain the protocol conversion optimization strategy:

[0127] (10);

[0128] in, represents the real-time input feature vector, represents the optimization strategy of the model output, Represents a deep learning model.

[0129] Then, according to the optimization strategy output by the model, the protocol conversion parameters of the intelligent gateway are adjusted in real time to improve the efficiency and stability of data transmission:

[0130] (11);

[0131] in, is the optimized protocol parameter, and g is the parameter adjustment function.

[0132] Establish a feedback mechanism to monitor the protocol conversion effect in real time, collect new data to update and train the model, and achieve continuous optimization:

[0133] (12);

[0134] in, It is a new data set for continuous training and optimization of the model. N represents the total number of samples, and M represents the number of newly collected samples.

[0135] S2. Collect the load data of the intelligent gateway in different time periods and extract the load data features to build a load prediction model. Use the load prediction model to optimize the protocol conversion strategy and dynamically allocate network resources based on the real-time monitored network traffic and data load.

[0136] Specifically, S2 includes the following steps:

[0137] Collect load data of smart gateways in different time periods, including CPU usage , memory usage , Network bandwidth usage , I / O operation rate etc. Define the load data set ,in Indicates the timestamp, Indicates the CPU usage at the i-th time point, represents the memory usage at the i-th time point, represents the bandwidth usage at the i-th time point, represents the I / O operation rate at the i-th time point, and T represents the total number of time points in the time period.

[0138] Extract key features from the load data set and represent it in the form of a time series. Feature vector matrix ,in Respectively represent the time series data of CPU, memory, bandwidth, and I / O.

[0139] Furthermore, a model suitable for processing time series forecasts is selected, such as a recurrent neural network (RNN) or a long short-term memory network (LSTM). The model structure is input layer (feature vector) → multiple hidden layers (LSTM / RNN) → output layer (load forecast value). Use the load data set L for model training and optimize the model parameters to minimize the forecast error:

[0140] (13);

[0141] in, are model parameters, is the predicted value, is the actual load value

[0142] During the model training process, adaptive optimization algorithms (such as Adam) are used to adjust the learning rate to ensure that the model converges quickly and has good predictive ability.

[0143] (14);

[0144] (15);

[0145] (16);

[0146] (17);

[0147] in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the adaptive deep learning model, Represents a constant used to prevent division by zero.

[0148] Deploy a real-time load monitoring system on the smart gateway to collect CPU, memory, bandwidth, and I / O usage in real time. Define real-time load vectors .

[0149] The real-time load vector is input into the trained load prediction model to predict the load conditions in the future.

[0150] (18);

[0151] in, is the load forecasting model, is the predicted load value.

[0152] According to the load forecast results, the resource allocation strategy of the intelligent gateway is dynamically adjusted to ensure the optimal utilization of various resources.

[0153] (19);

[0154] in, is the resource allocation strategy, and g is the resource allocation function.

[0155] Implement load balancing strategies and adjust resource allocation for each task to avoid overload and bottlenecks.

[0156] (20);

[0157] (twenty one);

[0158] (twenty two);

[0159] (twenty three);

[0160] in, Indicates the set resource usage threshold. Indicates the current CPU usage. Indicates the current bandwidth usage. Indicates the current memory usage. Indicates the current I / O usage. Indicates reallocation of CPU resources. Indicates reallocation of memory resources. Indicates reallocation of broadband resources. Indicates the reallocation of I / O usage resources.

[0161] Monitor resource allocation and load balancing effects in real time, and collect new load data to update and train load prediction models:

[0162] (twenty four);

[0163] in, Represents a new load dataset, which is used to update and train the load prediction model. represents the CPU at the i-th time point, represents the memory usage at the i-th time point, represents the bandwidth usage at the i-th time point, Indicates the I / O operation rate at the i-th time point;

[0164] Utilize newly collected load data to continuously optimize the load prediction model to ensure that the model maintains high-precision prediction capabilities in an ever-changing network environment.

[0165] (25);

[0166] in, represents the updated parameters, Indicates the parameters before updating.

[0167] S3. Build and train an edge computing model, and use the edge computing capabilities of the edge computing model to locally process data and make decisions.

[0168] In the application scenario of smart gateways, the real-time and diversity of data are required to be high, including sensor data (such as temperature, humidity, pressure, etc.), device status data (such as running time, fault information, etc.), and network traffic data (such as bandwidth usage, latency, etc.). Define the data set ;in Indicates the timestamp, Indicates at time Collected edge device data. These data reflect the real-time status of devices and networks, and can help the system detect and respond to abnormal situations in a timely manner.

[0169] Due to the large amount and diversity of data, the collected data is partitioned by time and device type and stored in the distributed storage system of the edge computing node to improve data access efficiency. , where key represents the data partition key and value represents the data value after partitioning. Partition storage can speed up data access, improve processing efficiency, and facilitate subsequent real-time processing.

[0170] Specifically, the S3 specifically includes:

[0171] The edge device dataset E is used for model training and the model parameters are optimized to minimize the processing error.

[0172] (26);

[0173] in, are model parameters, is the predicted value, is the real data value. By defining the loss function, we can quantify the error of model prediction, guide the optimization of model parameters, and ensure that the model has high accuracy when processing actual data. N represents the total number of samples.

[0174] Deploy the trained model to the edge computing node to realize local data processing and decision-making.

[0175] (27);

[0176] in, It is an edge computing model. is the input data, Deploying trained models on edge nodes can achieve local data processing, reduce the delay caused by data transmission, and improve system response speed.

[0177] Input the edge device data collected in real time into the deployed edge computing model for local processing and decision-making.

[0178] (28);

[0179] in, Indicates the timestamp, Represents real-time input data. This data is input into the edge computing model for processing to ensure that the system can respond quickly.

[0180] Use edge computing models to process real-time input data and output processing results for local decision-making and control.

[0181] (29);

[0182] in, The real-time processing of edge device data can ensure that the system can respond quickly and adjust and control related equipment in a timely manner when an abnormality or emergency occurs.

[0183] Specifically, related devices refer to various devices controlled and managed by edge computing nodes in the intelligent gateway network. These devices can be any devices that need to be monitored and controlled in real time, such as:

[0184] Sensors: used to monitor environmental parameters (such as temperature, humidity, pressure, etc.).

[0185] Actuator: Used to execute control commands (such as opening or closing a valve, adjusting motor speed, etc.).

[0186] Communication equipment: used to transmit data and control instructions (such as routers, switches, etc.).

[0187] Power equipment: such as transformers and circuit breakers, which are used to manage power distribution and protection in smart grids.

[0188] Other industrial equipment: such as machinery and equipment in factories, monitoring their operating status and performance.

[0189] Based on the processing results, local decision-making algorithms (such as rule engines or simple logical judgments) are used to make decisions and generate control instructions.

[0190] (30);

[0191] in, is the local control instruction, and g is the decision function. Local decision-making can significantly reduce the system's dependence on the central server and improve the timeliness and reliability of decision-making.

[0192] Send local control instructions to related equipment to perform corresponding control operations, such as adjusting equipment operating parameters or issuing alarm signals.

[0193] (31);

[0194] in, Indicates that the local control command Sent to relevant devices and perform corresponding control operations.

[0195] Executing control instructions is the ultimate goal of the entire real-time data processing process. By controlling the equipment, it can ensure that the system runs in the best state.

[0196] The corresponding control operations refer to the specific operations generated and executed based on the processing results. These operations are designed to ensure the efficient, safe and stable operation of the system, including but not limited to:

[0197] Adjust equipment operating parameters: such as adjusting the speed of the motor, modifying the sampling frequency of the sensor, changing the bandwidth allocation of the communication equipment, etc.

[0198] Send out an alarm signal: When an abnormal situation is detected, an alarm is triggered to notify maintenance personnel to check and handle it.

[0199] Enabling or shutting down equipment: such as starting a backup power supply in an emergency or shutting down a faulty device to prevent damage.

[0200] Execute predetermined control strategies: such as automatically adjusting system parameters according to predetermined rules to achieve optimized control.

[0201] Monitor edge computing processing results and control effects in real time, and collect new data to update and train edge computing models.

[0202] (32);

[0203] in, Represents a new dataset, represents the timestamp of the kth time point, represents the newly collected real-time input data at the kth time point;

[0204] Through real-time monitoring, errors in the processing process can be discovered and corrected in a timely manner to ensure the stability and reliability of the system.

[0205] Utilize newly collected edge device data to continuously optimize edge computing models to ensure that the models maintain efficient processing capabilities in a changing environment:

[0206] (33);

[0207] Continuously optimizing the model can ensure that it always maintains efficient and accurate processing capabilities in a dynamically changing environment and improve the adaptability of the system.

[0208] S4. Construct a collaborative control model and a feedback mechanism, use the collaborative control model to perform real-time collaborative control on the intelligent gateway devices, and use the feedback mechanism to dynamically adjust the control strategy.

[0209] Specifically, the S4 specifically includes:

[0210] Collect real-time data from multiple data sources (such as sensors, device status, network traffic, etc.) and fuse them into a unified real-time data set.

[0211] (34);

[0212] in, Indicates the timestamp, represents the data from the jth data source at the i-th time point, Represents the data from the nth data source at the i-th time point. Use the real-time data fusion results to perform association analysis to find out the associations and influencing factors between devices. These relationships are described by constructing a device association matrix:

[0213] (35);

[0214] in, represents the device association matrix, Indicates the strength of association between device i and device j, which is obtained by calculating the correlation coefficient:

[0215] (36);

[0216] in, and Represent the data means of device i and device j respectively.

[0217] Based on the results of data association analysis, a collaborative control model is constructed. Using feedback control theory, a controller model is designed to achieve collaborative work between devices:

[0218] (37);

[0219] Where u(t) is the control input, e(t) is the error signal, , and are the proportional, integral and differential control parameters respectively.

[0220] Using historical data and real-time feedback, control parameters can be adjusted using optimization algorithms (such as genetic algorithms or particle swarm optimization) , and , to ensure optimal control effect:

[0221] (38);

[0222] in, is the target optimization function, is the weight coefficient, and the control parameters are optimized by minimizing the objective function;

[0223] Utilize the optimized control parameters to calculate the control instructions for each device. Combined with real-time data, generate specific control instructions:

[0224] (39);

[0225] in, is the control instruction of device i, f represents the control instruction generation function, Represents the fused real-time dataset.

[0226] Preferably, the generated control instructions are published to each device through an intelligent gateway to ensure that the device can receive the control instructions in real time and perform corresponding operations.

[0227] Specifically, corresponding operations refer to specific operations generated and executed based on the processing results. These operations are intended to ensure the efficient, safe and stable operation of the system, including but not limited to:

[0228] Adjust equipment operating parameters: such as adjusting the speed of the motor, modifying the sampling frequency of the sensor, changing the bandwidth allocation of the communication equipment, etc.

[0229] Send out an alarm signal: When an abnormal situation is detected, an alarm is triggered to notify maintenance personnel to check and handle it.

[0230] Enabling or shutting down equipment: such as starting a backup power supply in an emergency or shutting down a faulty device to prevent damage.

[0231] Execute predetermined control strategies: such as automatically adjusting system parameters according to predetermined rules to achieve optimized control.

[0232] Monitor the operating status and control effect of the equipment in real time and collect control effect data. Define control effect indicators (such as equipment stability, response time, error, etc.):

[0233] (40);

[0234] in, represents the control effect index of device j at the i-th time point, Represents the control effect index of device n at the i-th time point.

[0235] According to the control effect evaluation results, the control parameters are adjusted dynamically. The adaptive control algorithm is used to continuously optimize the control parameters through real-time data feedback to ensure that the control system is always in the best state:

[0236] (41);

[0237] (42);

[0238] (43);

[0239] in, Represents the learning rate, dynamically adjusts the control parameters through the gradient descent method, represents the partial derivative of the objective function J, Represents the partial derivative operator.

[0240] S5. Define smart contracts and use blockchain technology to encrypt and verify the data transmission process.

[0241] Specifically, step S5 specifically includes:

[0242] The real-time data generated by the intelligent gateway is packaged to form a data block to be transmitted. The structure of data block B is defined as follows:

[0243] (44);

[0244] The header contains metadata of the data block (such as timestamp, data source identifier, etc.), the payload contains the actual transmitted data, and the signature contains the signature information of the data block.

[0245] The gateway's private key is used to digitally sign the data block to ensure the integrity and authenticity of the data. The signing process is as follows:

[0246] (45);

[0247] in, Indicates the use of a private key for signing operations, and hash indicates the hash calculation of the data. It represents the data concatenation operation, header represents the metadata of the data block, and signature represents the digital signature to ensure the integrity and authenticity of the data.

[0248] Pack multiple data blocks to generate blocks in the blockchain. The structure of block C is defined as follows:

[0249] (46);

[0250] Among them, the block header contains the metadata of the block (such as the hash of the previous block, timestamp, etc.), the blockbody contains multiple data blocks, and the block signature contains the signature information of the block.

[0251] The generated blocks are distributed to various nodes in the blockchain network for verification and storage. The distribution process includes sending the blocks to the designated network nodes and waiting for the node's confirmation.

[0252] Furthermore, the Byzantine Fault Tolerance (PBFT) consensus mechanism is adopted to ensure the consistency and security of blocks in the network. The steps of the consensus mechanism are as follows:

[0253] Pre-preparation phase: The block generation node sends the block proposal to all verification nodes.

[0254] Preparation phase: The verification node verifies the block and broadcasts the preparation message.

[0255] Commit phase: After the verification node collects enough prepare messages, it broadcasts the commit message.

[0256] Confirmation phase: After all verification nodes collect enough submission messages, they confirm the block and store it.

[0257] Each verification node verifies the received block, including the integrity of the block, the signature of the data block, and the signature of the block. The verification process is as follows:

[0258] (47);

[0259] Among them, Verify represents the verification operation, and public key is the public key of the block generation node.

[0260] Preferably, a smart contract is defined on the blockchain to automatically execute logic and rules related to data transmission. The structure of the smart contract is as follows:

[0261] (48);

[0262] Among them, conditions represent the triggering conditions of the contract, actions represent the execution operations of the contract, and events represent the event notifications of the contract.

[0263] When data transmission meets the triggering conditions of the smart contract, the corresponding operations are automatically executed, such as data verification, record storage, and exception handling. The execution process is as follows:

[0264] (49);

[0265] Furthermore, the verified data blocks and blocks are stored on the blockchain to ensure the data is tamper-proof and traceable. The on-chain process is as follows:

[0266] (50);

[0267] Among them, blockchain.append means adding a block to the blockchain.

[0268] Preferably, a data query interface based on blockchain is provided, allowing users and systems to query data stored on blockchain as needed. The query process is as follows:

[0269] (51);

[0270] Among them, blockchain.query represents the blockchain data query operation, and conditions represents the query conditions.

[0271] In the application scenarios of smart gateways, the security and reliability of data transmission are crucial. For example, in smart grids, the status data of power equipment needs to be securely transmitted to the control center, and the integrity and non-tamperability of the data during transmission must be ensured.

[0272] S6. Design intelligent agent technology, which automatically selects the optimal configuration solution for the intelligent gateway device according to the application scenario and type of the intelligent gateway device, monitors the status of the intelligent gateway device in real time, and predicts and maintains potential failures of the intelligent gateway device in advance.

[0273] Specifically, step S6 includes:

[0274] Deploy smart agents on smart gateways and associated devices. Each smart agent contains initial configuration parameters and management rules. Define the initial configuration matrix of the smart agent , indicating the configuration status of the agent when the system is initialized:

[0275] (52);

[0276] in, represents the initial value of the i-th configuration parameter, Indicates the initial value of the nth configuration parameter. For intelligent gateways, these configuration parameters may include network bandwidth, storage capacity, processor load, etc. The initialization and deployment of intelligent agents ensure that the device has basic functions and performance during initial operation.

[0277] Deploy the intelligent agent to the intelligent gateway and associated devices to ensure that the agent can communicate and interact with the devices in real time. After the agent is deployed, the communication matrix C between the agent and the device is:

[0278] (53);

[0279] in, Indicates the communication connection status between intelligent agent i and device j. These connection states reflect the real-time interaction capabilities between the intelligent gateway and each device, ensuring that the device can obtain and transmit data in a timely manner during operation.

[0280] Furthermore, the intelligent agent monitors the intelligent gateway and associated devices in real time, collecting device operation status and performance data. The monitoring data matrix M contains the status data of each device:

[0281] (54);

[0282] in, Indicates that device i is at time For example, monitoring data may include CPU usage, memory usage, network traffic, etc. Real-time monitoring enables the intelligent agent to promptly detect problems in device operation and ensure that the device is running in the best condition.

[0283] At the same time, the collected monitoring data is transmitted to the intelligent agent for processing and storage to form a historical data set H:

[0284] (55);

[0285] in, Indicates the timestamp, Indicates that device i is at time The historical data can help analyze the load patterns of power equipment in smart grids, thereby optimizing energy distribution.

[0286] Preferably, based on the historical data set H, the intelligent agent constructs an adaptive configuration model. The model dynamically adjusts the configuration parameters of the device by analyzing the historical data. The objective function F of the adaptive configuration model is:

[0287] (56);

[0288] in, Represents the intelligent agent as device i at time Calculated optimal configuration parameters. For example, in a smart grid, the model can adjust the operating parameters of power equipment based on historical load data to optimize energy distribution and ensure that power equipment can operate efficiently under different load conditions.

[0289] Furthermore, an optimization algorithm (such as particle swarm optimization algorithm) is used to optimize the configuration parameters of the equipment to achieve optimal system performance. The iterative formula for parameter optimization is:

[0290] (57);

[0291] (58);

[0292] in, is the velocity vector, is the inertia weight, and is the learning factor, and is a random number, is the individual optimal position, It is the global optimal position. For smart gateways, parameters such as bandwidth allocation, cache size, etc. can be optimized to ensure high performance under different network load conditions.

[0293] Furthermore, the intelligent agent uses monitoring data and adaptive configuration models to detect anomalies in the device operation status. The anomaly detection formula is as follows:

[0294] (59);

[0295] in, Indicates that device i is at time Abnormal state, For example, in a smart grid, if the load of a certain power device exceeds a preset threshold, it is marked as abnormal and triggers an early warning mechanism.

[0296] When the intelligent agent detects an anomaly, it immediately triggers the fault handling mechanism and performs the corresponding recovery operation. The fault handling operation R is defined as:

[0297] (60);

[0298] in, Indicates the kth recovery operation. For example, you can restart the device, adjust operating parameters, or switch to the backup system to ensure that the device quickly resumes normal operation.

[0299] Preferably, the intelligent agent adjusts the configuration parameters of the device in real time based on the monitoring data and anomaly detection results to ensure that the device is always in the best operating state. Configuration Adjustment Matrix for:

[0300] (61);

[0301] in, represents the configuration parameters of device i at time t. For example, bandwidth allocation is adjusted according to real-time network traffic to avoid network congestion and improve data transmission efficiency.

[0302] Preferably, a feedback mechanism is established between the intelligent agent and the device to continuously optimize the device configuration parameters through real-time monitoring and data analysis. The feedback control equation is:

[0303] (62);

[0304] in, For example, in smart grids, intelligent agents can adjust the operating parameters of power equipment based on real-time load data to optimize energy distribution and ensure the stability and efficiency of the power system.

[0305] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0306] In addition, for technical details that are not described in detail in this embodiment, reference can be made to the parameter operation method provided in any embodiment of the present invention, and will not be repeated here.

[0307] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0308] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0309] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0310] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A data processing method for an intelligent gateway protocol, characterized in that: The method comprises: S1. Collect data transmission data of the intelligent gateway under different protocols, pre-process the collected data, build an adaptive deep learning model, analyze the data transmission characteristics of different protocols, and dynamically optimize the protocol conversion strategy according to the real-time changes of network conditions and data traffic; wherein the loss function of the adaptive deep learning model is It is expressed as follows: ; in, represents the model parameters, Indicates the actual performance index, Represents the model prediction value, N represents the total number of samples, and i represents the i-th sample; The adaptive deep learning model uses an adaptive optimization algorithm to adjust the learning rate; the adaptive optimization algorithm is expressed as follows: ; in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the adaptive deep learning model, Represents a constant used to prevent division by zero; S2. Collect the load data of the intelligent gateway in different time periods and extract the load data features, build a load prediction model, and use the load prediction model to optimize the protocol conversion strategy and dynamically allocate network resources according to the real-time monitored network traffic and data load; S3. Build and train an edge computing model, and use the edge computing capabilities of the edge computing model to process data and make decisions locally; S4. Construct a collaborative control model and a feedback mechanism, use the collaborative control model to perform real-time collaborative control on the intelligent gateway device, and use the feedback mechanism to dynamically adjust the control strategy; wherein the collaborative control model is expressed as follows: ; Where u(t) represents the control input, e(t) represents the error signal, , and represent the proportional, integral and differential control parameters at time t, respectively, represents the integral operation, dt represents the time differential, de(t) represents the differential of the error signal, and dt represents the time differential; Through historical data and real-time feedback, the design target optimization algorithm adjusts the control parameters , and ; Wherein, the target optimization algorithm is expressed as follows: ; in, represents the target optimization function, represents the weight coefficient, and the control parameters are optimized by minimizing the objective function; S5. Define smart contracts and use blockchain technology to encrypt and verify the data transmission process; S6. Design intelligent agent technology, which automatically selects the optimal configuration solution for the intelligent gateway device according to the application scenario and type of the intelligent gateway device, monitors the status of the intelligent gateway device in real time, and predicts and maintains potential failures of the intelligent gateway device in advance.

2. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: The preprocessing includes using mean filling or interpolation methods to process missing values, using the IQR method to remove outliers and standardization.

3. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: The deep learning model includes a long short-term memory network or a convolutional neural network. In S1, the real-time data of the intelligent gateway is input into the trained deep learning model to obtain the protocol conversion optimization strategy, which is expressed as follows: ; in, represents the real-time input feature vector, represents the optimization strategy of the model output, Represents a deep learning model; According to the optimization strategy output by the trained deep learning model, the protocol conversion parameters of the intelligent gateway are adjusted in real time, as shown below: ; in, represents the optimized protocol parameters, g represents the parameter adjustment function; Establish a feedback mechanism for the deep learning model, monitor the protocol conversion effect in real time, and collect new data to update and train the model, as shown below: ; in, Represents a new data set for continuous training and optimization of the model. Indicates the transmission characteristics under different protocols. Represents the corresponding performance indicator, N represents the total number of samples, and M represents the number of newly collected samples.

4. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: In S2, construct the load data set L and design the loss function of the load prediction model as follows: ; in, represents the model parameters of the load prediction model, represents a negative prediction value, represents the negative true load value, and T represents the total number of samples in the prediction time period; The learning rate is adjusted using an adaptive optimization algorithm, as shown below: ; in, and denote the momentum term at time t, represents the gradient, represents the learning rate, and They represent the momentum term at time t-1, and represents the exponential decay rate, and Represents the exponential decay rate to the power of t, which is used for deviation correction; represents the bias-corrected momentum term, represents the bias-corrected second-order moment estimate, represents the updated model parameters, represents the model parameters of the load prediction model, Represents a constant used to prevent division by zero; Deploy a real-time load monitoring system on the smart gateway to collect CPU usage in real time , memory usage , bandwidth usage and I / O usage , define the real-time load vector ; Input the real-time load vector into the trained load prediction model , the load situation in the future is predicted as follows: ; in, represents the load prediction model, represents the negative predicted load value, Indicates the load prediction result; According to the load prediction results , dynamically adjust the resource allocation strategy of the intelligent gateway , which is expressed as follows: ; in, represents the resource allocation strategy, g represents the resource allocation function; Design a load balancing strategy and adjust the resource allocation of each task, as shown below: ; in, Indicates the set resource usage threshold. Indicates the current CPU usage. Indicates the current bandwidth usage. Indicates the current memory usage. Indicates the current I / O usage. Indicates reallocation of CPU resources. Indicates reallocation of memory resources. Indicates reallocation of broadband resources. Indicates the reallocation of I / O usage resources.

5. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: The loss function of the edge computing model It is expressed as follows: ; in, represents the model parameters, represents the predicted value, represents the real data value, and M represents the number of newly collected samples; Deploy the trained edge computing model to the edge computing node, as shown below: ; in, represents the edge computing model, Represents input data, Indicates the processing result.

6. A data processing method for intelligent gateway protocol according to claim 5, characterized in that: The edge device data collected in real time is input into the deployed edge computing model for local processing and decision-making, as shown below: ; in, Indicates the timestamp, Indicates real-time input data, Represents a real-time dataset; The edge computing model is used to process real-time input data and output the processing results, which are expressed as follows: ; in, Indicates the processing result.

7. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: In S4, the real-time data fusion result is used to perform association analysis and build the relationship between the intelligent gateway devices, which is expressed as follows: ; in, represents the device association matrix, represents the strength of association between device i and device j, which is obtained by calculating the correlation coefficient; wherein the correlation coefficient is calculated as follows: ; in, and Represent the data means of device i and device j respectively, represents the data from the kth data source at the i-th time point, Represents the data from the kth data source at the jth time point.

8. A data processing method for intelligent gateway protocol according to claim 7, characterized in that: Monitor the operating status and control effect of the equipment in real time and collect control effect data , which is expressed as follows: ; in, represents the control effect index of device j at the i-th time point, represents the control effect index of device n at the i-th time point, Indicates a timestamp; Using adaptive control algorithm, control parameters are optimized through real-time data feedback, as shown below: ; in, Represents the learning rate, which dynamically adjusts the control parameters through the gradient descent method; , and They represent the proportional, integral and differential control parameters at time t+1 respectively, represents the partial derivative of the objective function J, Represents the partial derivative operator.

9. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: The S5 specifically includes: The real-time data generated by the intelligent gateway is packaged to form a data block B to be transmitted, and the data block is digitally signed using the private key of the gateway; Pack multiple data blocks B to generate block C in the blockchain, and distribute the generated block C to each node in the blockchain network for verification and storage; The Byzantine fault-tolerant consensus mechanism is used to verify the received block C by each verification node; Define rules for the smart contract to automatically execute data transmission on the blockchain; the rules are that when the data transmission meets the triggering conditions of the smart contract, the transmission operation is automatically executed; The verified data blocks B and C are stored in the blockchain, and a blockchain-based data query interface is provided.

10. A data processing method for intelligent gateway protocol according to claim 1, characterized in that: The S6 specifically includes: Deploy intelligent agents on the intelligent gateway and associated devices, initialize the intelligent agents, and deploy the intelligent agents to the intelligent gateway and associated devices at the same time; wherein each intelligent agent includes initial configuration parameters and management rules; The intelligent agent monitors the intelligent gateway and associated devices in real time, collects the operating status and performance data of the intelligent gateway and associated devices, and transmits the collected monitoring data to the intelligent agent for processing and storage to form a historical data set H; Based on the historical data set H, the intelligent agent constructs an adaptive configuration model; wherein the objective function F of the adaptive configuration model is expressed as follows: ; in, Indicates the associated device i at time The calculated optimal configuration parameters, Indicates that device i is at time Actual monitoring data; The optimization algorithm is used to dynamically adjust the configuration parameters of the associated devices, as shown below: ; in, represents the velocity vector, represents the inertia weight, and They represent learning factors, and Represents a random number, represents the optimal position of an individual, represents the global optimal position, represents the velocity vector at the next moment, Indicates the individual position at the next moment; The intelligent agent uses monitoring data and adaptive configuration models to detect anomalies in the operating status of associated devices. When the intelligent agent detects an anomaly, it immediately triggers the fault handling mechanism and performs recovery operations. The intelligent agent adjusts the configuration parameters of the associated devices in real time according to the monitoring data and anomaly detection results. At the same time, a feedback mechanism is established between the intelligent agent and the associated devices to optimize the configuration parameters of the associated devices through real-time monitoring and data analysis.

Images