A method and system for identifying electronic information of Internet of Things

Abstract

The present invention discloses an Internet of Things electronic information identification method and system, which relates to the field of Internet of Things information processing technology, including deploying multimodal sensors to collect electronic information data through Internet of Things devices and storing them in a database after preprocessing; extracting the time characteristics and frequency domain characteristics of the electronic information data collected by each sensor, and performing feature fusion to form fusion features; constructing an information recognition model, deploying the trained information classification and recognition model to the database, and classifying and recognizing the data based on the fusion features; and visually displaying the data classification and recognition results. The present invention collects Internet of Things electronic information data for preprocessing, extracts data features for feature fusion, constructs an information classification and recognition model to classify and recognize Internet of Things electronic information data based on fusion features, realizes comprehensive processing and recognition of electronic information data, and improves classification and recognition accuracy and flexibility by updating the classification and recognition model in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things information processing, and in particular to an Internet of Things electronic information identification method and system. Background Art

[0002] The rapid development of Internet of Things (IoT) technology has promoted the interconnection of various devices and systems, and the real-time collection, transmission and processing of data can be realized through IoT devices. With the advancement of sensor technology, multimodal sensors can simultaneously collect multiple environmental parameters such as temperature, humidity, pressure, light intensity, etc., thereby providing more comprehensive and detailed data support for various applications. The wide application of IoT covers smart homes, smart transportation, smart cities, Industry 4.0 and other fields. These applications collect and analyze a large amount of data in real time through IoT devices, thereby realizing intelligent management and decision-making. However, the existing IoT electronic information recognition technology still has many shortcomings. When processing high-dimensional and multimodal data, the existing technology faces the problems of data dimensionality disaster and difficulty in feature extraction, and often does not update the classification and recognition model in real time, resulting in errors in electronic information recognition, making it difficult to achieve efficient and accurate classification and recognition. Summary of the invention

[0003] In view of the problems existing in the above-mentioned existing Internet of Things electronic information identification methods and systems, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that the existing technology faces the problems of data dimensionality disaster and difficulty in feature extraction when processing high-dimensional, multimodal data, and often fails to update the classification and recognition model in real time, resulting in errors in electronic information recognition, making it difficult to achieve efficient and accurate classification and recognition.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a method for identifying electronic information in the Internet of Things, comprising: deploying multimodal sensors to collect electronic information data through Internet of Things devices and storing them in a database after pre-processing; extracting the time characteristics and frequency domain characteristics of the electronic information data collected by each sensor, and performing feature fusion to form fusion features; constructing an information recognition model, deploying the trained information classification and recognition model to the database, and classifying and identifying the data based on the fusion features; and visually displaying the data classification and recognition results.

[0006] As a preferred solution of the Internet of Things electronic information identification method described in the present invention, the deployment of multimodal sensors to collect electronic information data through Internet of Things devices and store it in a database after pre-processing refers to collecting data generated by the devices by deploying multimodal sensors, setting independent classifications for the data collected by each sensor, and directly cleaning and filtering the collected data based on edge computing technology after collecting the data and converting it into a unified format. After pre-processing, the data is transmitted to the database through an encrypted network and stored according to sensor classification, and security access control is set, and the database performs cloud backup of the stored data.

[0007] As a preferred solution of the method for identifying electronic information of the Internet of Things of the present invention, wherein: the extraction of time characteristics and frequency domain characteristics of electronic information data collected by each sensor includes:

[0008] Calculate the mean μ and variance σ of the collected data, and perform autocorrelation analysis on the collected data:

[0009]

[0010] where x t is the tth data point, μ is the mean of the data, N is the total number of data points, k is the number of lags, and R is the autocorrelation coefficient;

[0011] The data mean, variance and autocorrelation coefficient are combined to form the time feature T = {R, μ, σ};

[0012] Use Fast Fourier Transform to extract the frequency domain representation of the data and calculate the spectral density:

[0013]

[0014] S=(P) 2 ,

[0015] Where i is the imaginary unit, ω is the frequency, P is the frequency domain representation, and S is the spectral density;

[0016] The frequency domain representation and the spectral density are combined to form the frequency domain feature F = {P, S}.

[0017] As a preferred solution of the method for identifying electronic information of the Internet of Things of the present invention, the feature fusion to form a fusion feature refers to connecting the time features and frequency domain features extracted from the electronic information data collected by each sensor to form an initial fusion feature C:

[0018] C = {R, μ, σ, P, S};

[0019] Calculate the covariance matrix G of the initial fusion feature C:

[0020]

[0021] Where M is the total number of fusion features, C j is the jth initial fusion feature, is the mean of the initial fusion features, and T is the transposition operation;

[0022] Perform eigenvalue decomposition on the covariance matrix G:

[0023] G=V∧V T ,

[0024] Where V is the eigenvector matrix, ∧ is the eigenvalue matrix, V T is the transpose of V;

[0025] Select the first k largest eigenvalues ​​and their corresponding eigenvectors to construct the dimension reduction matrix W = {v1, v2, ..., v k}, the initial fusion feature C is reduced in dimension based on the dimensionality reduction matrix:

[0026] C''=CW,

[0027] Where C” is the fusion feature after dimensionality reduction.

[0028] As a preferred solution of the method for identifying electronic information of the Internet of Things of the present invention, wherein: the construction of the information classification and identification model refers to the use of a neural network to construct a classification and identification model, which includes an input layer, a hidden layer and an output layer;

[0029] Input layer: The number of neurons is set according to the total number of extracted time features and frequency domain features;

[0030] Hidden layer: It has two layers, each layer contains 100 neurons. The hidden layer uses the ReLU function as the activation function, which is expressed as:

[0031]

[0032] where Φ i is the weight of the i-th feature, x i is the i-th feature of the input, b is the bias term, m is the total number of input features, and h is the output of the neuron;

[0033] Output layer: The number of neurons in the output layer is consistent with the number of categories in the classification task, and the softmax function is used as the output layer activation function:

[0034]

[0035] Where zj is the input of the jth output neuron, zl is the input of the lth output neuron, L is the number of output neurons, that is, the total number of categories, σ j is the predicted probability of the jth class;

[0036] The model training loss function is defined as:

[0037]

[0038] Where Q is the model loss, y o，l is the true label of data o belonging to category l, σ o，l The model predicts the probability that data o belongs to category l;

[0039] The Adam optimizer is used to optimize the model and the loss gradient is calculated to update the model parameters:

[0040]

[0041] where θ t is the model parameter updated for the tth time, η is the learning rate, is the loss gradient;

[0042] Use the training data set to iteratively train the model, annotate the data in the training data set with real labels, calculate the model loss in each iteration, and stop model training after reaching the set loss or the preset number of iterations to obtain a classification recognition model;

[0043] After each electronic information recognition, the input data is saved as a training set for model training and updating.

[0044] As a preferred solution of the method for identifying electronic information in the Internet of Things described in the present invention, wherein: the information identification model that has been trained is deployed to the database, and the data is classified and identified based on the fusion features, which means that the classification and identification model is directly deployed to the database after training, and the database inputs the fusion feature C" into the classification and identification model to classify and identify the electronic information data, and the classification results and the identification data are synchronously stored in the database.

[0045] As a preferred solution of the method for identifying electronic information in the Internet of Things described in the present invention, the visual display of the data classification and identification results refers to the database summarizing the classification results of all electronic information data and calculating the proportional diagram of the classification results of all electronic information data, and synchronously displaying the electronic information data, classification results and proportional diagram through a visualization port, and making customized adjustments based on user needs.

[0046] Another object of the present invention is to provide an Internet of Things electronic information identification system, which comprises:

[0047] Data acquisition module, used to collect and pre-process the electronic information data of the Internet of Things;

[0048] A feature fusion module is used to extract and fuse features of pre-processed electronic information data;

[0049] Information recognition module, used to build an information classification and recognition model and classify and recognize electronic information data based on fusion features;

[0050] Visual display module, used to calculate the classification ratio diagram of the classification results and display the data;

[0051] The data storage module is used to store the collected and analyzed data and set security access measures.

[0052] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned method for identifying electronic information of the Internet of Things when executing the computer program.

[0053] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for identifying electronic information of the Internet of Things.

[0054] The beneficial effects of the present invention are as follows: the present invention collects electronic information data of the Internet of Things for preprocessing, extracts data features for feature fusion, constructs an information classification and recognition model to classify and recognize the electronic information data of the Internet of Things based on the fusion features, thereby realizing comprehensive processing and recognition of electronic information data, and improving classification and recognition accuracy and flexibility by updating the classification and recognition model in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0056] Figure 1 The figure is a flowchart of the method for identifying electronic information of the Internet of Things.

[0057] Figure 2 This is a structural diagram of the Internet of Things electronic information identification system. DETAILED DESCRIPTION

[0058] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0059] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0060] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0061] Example 1

[0062] Reference Figure 1 , which is the first embodiment of the present invention, and this embodiment provides an Internet of Things electronic information identification method, the Internet of Things electronic information identification method includes,

[0063] S1. Deploy multimodal sensors to collect electronic information data through IoT devices and store them in the database after preprocessing;

[0064] Specifically, deploying multimodal sensors to collect electronic information data through IoT devices and storing it in a database after pre-processing means collecting the data generated by the equipment by deploying multimodal sensors, setting independent classifications for the data collected by each sensor, and directly cleaning and filtering the collected data based on edge computing technology after collecting the data and converting it into a unified format. After pre-processing, the data is transmitted to the database through an encrypted network and stored by sensor classification, and security access control is set. The database performs cloud backup of the stored data.

[0065] By deploying multimodal sensors, multiple types of data can be collected at the same time to provide more comprehensive and detailed environmental information. This multimodal data collection method can improve data coverage and accuracy, avoid the limitations of a single sensor, and independently classify the data collected by each sensor, which is helpful for data organization and management, and improves the efficiency of data retrieval and analysis. The classified and stored data structure is clearer, which is helpful for subsequent data processing and analysis. Data cleaning, filtering and format conversion through edge computing technology can greatly reduce the amount of data transmission, reduce network bandwidth pressure, and improve the real-time performance of data processing. At the same time, edge computing can process data at the source of data generation, reducing the delay and cost of data transmission. Cloud backup can ensure the persistence and reliability of data and prevent data loss. At the same time, cloud storage is highly scalable and flexible, and can dynamically adjust storage resources according to demand, reducing the cost of data storage and management.

[0066] S2, extracting the time characteristics and frequency domain characteristics of the electronic information data collected by each sensor, and performing feature fusion to form fusion features;

[0067] Specifically, the time characteristics and frequency domain characteristics of the electronic information data collected by each sensor are extracted, including:

[0068] Calculate the mean μ and variance σ of the collected data, and perform autocorrelation analysis on the collected data:

[0069]

[0070] where x t is the tth data point, μ is the mean of the data, N is the total number of data points, k is the number of lags, and R is the autocorrelation coefficient;

[0071] The data mean, variance and autocorrelation coefficient are combined to form the time feature T = {R, μ, σ};

[0072] Use Fast Fourier Transform to extract the frequency domain representation of the data and calculate the spectral density:

[0073]

[0074] S=(P) 2 ,

[0075] Where i is the imaginary unit, ω is the frequency, P is the frequency domain representation, and S is the spectral density;

[0076] The frequency domain representation and the spectral density are combined to form the frequency domain feature F = {P, S}.

[0077] By calculating the mean, variance and autocorrelation coefficient of the collected data, the statistical characteristics and time dependence of the data can be fully described. The mean reflects the central trend of the data, the variance indicates the degree of dispersion of the data, and the autocorrelation coefficient reveals the inherent laws and periodicity of the data. Combining these characteristics can improve the accuracy and robustness of data analysis and model building. The mean and variance are basic statistical characteristics that can be used to initially understand the distribution and fluctuation of the data. Calculating the mean and variance provides a basis for subsequent autocorrelation analysis and feature fusion. Autocorrelation analysis helps to identify patterns and trends in the data by revealing the inherent correlation and periodicity of time series data. The autocorrelation coefficient provides an in-depth understanding of the data structure and helps to improve the performance of the prediction model. Combining the mean, variance and autocorrelation coefficient of the data to form a comprehensive time feature can more comprehensively characterize the statistical and time characteristics of the data. The comprehensive time feature helps to improve the accuracy of data classification and prediction. The spectral density describes the energy distribution of the signal at each frequency and provides a quantitative analysis of the frequency characteristics of the signal. By calculating the spectral density, the main frequency components and energy distribution in the signal can be identified, which is helpful for signal classification and feature extraction.

[0078] Furthermore, feature fusion to form fusion features refers to connecting the time features and frequency domain features extracted from the electronic information data collected by each sensor to form an initial fusion feature C:

[0079] C = {R, μ, σ, P, S};

[0080] Calculate the covariance matrix G of the initial fusion feature C:

[0081]

[0082] Where M is the total number of fusion features, C j is the jth initial fusion feature, is the mean of the initial fusion features, and T is the transposition operation;

[0083] Perform eigenvalue decomposition on the covariance matrix G:

[0084] G=V∧V T ,

[0085] Where V is the eigenvector matrix, and its column vector is the corresponding eigenvector V i , ∧ is the eigenvalue matrix, and its diagonal is the eigenvalue λ i , V T is the transpose of V;

[0086] Select the first k largest eigenvalues ​​and their corresponding eigenvectors to construct the dimension reduction matrix W = {v1, v2, ..., v k}, the initial fusion feature C is reduced in dimension based on the dimensionality reduction matrix:

[0087] C''=CW,

[0088] Where C” is the fusion feature after dimensionality reduction.

[0089] Connecting the time features and frequency domain features collected by each sensor to form the initial fusion feature C can comprehensively reflect the multi-dimensional information of the data and improve the expressiveness of the data. This combination of multimodal features helps to capture the complex patterns and associations in the data and improve the accuracy of subsequent analysis and modeling. By calculating the covariance matrix G of the initial fusion feature, the correlation between the features can be quantified, providing a basis for subsequent eigenvalue decomposition and dimensionality reduction processing. The covariance matrix helps to identify the main change direction of the data and the relationship between the features. Performing eigenvalue decomposition on the covariance matrix to obtain the eigenvector matrix V and the eigenvalue matrix, Λ, can reveal the main change direction of the data and the importance of the features. Eigenvalue decomposition helps to identify the main components in the data and simplify the data structure. Selecting the first k largest eigenvalues ​​and the corresponding eigenvectors to construct the dimensionality reduction matrix W can retain the main information of the data and reduce the dimension and complexity of the data. Dimensionality reduction processing helps to improve the training speed and generalization ability of the model and reduce the computational cost.

[0090] S3. Build an information recognition model, deploy the trained information classification and recognition model to the database, and perform data classification and recognition based on fusion features;

[0091] Specifically, constructing an information classification and recognition model refers to constructing a classification and recognition model using a neural network, which includes an input layer, a hidden layer, and an output layer;

[0092] Input layer: The number of neurons is set according to the total number of extracted time features and frequency domain features;

[0093] Hidden layer: It has two layers, each layer contains 100 neurons. The hidden layer uses the ReLU function as the activation function, which is expressed as:

[0094]

[0095] where Φ i is the weight of the i-th feature, x i is the i-th feature of the input, b is the bias term, m is the total number of input features, and h is the output of the neuron;

[0096] Output layer: The number of neurons in the output layer is consistent with the number of categories in the classification task, and the softmax function is used as the output layer activation function:

[0097]

[0098] Where zj is the input of the jth output neuron, zl is the input of the lth output neuron, L is the number of output neurons, that is, the total number of categories, σ j is the predicted probability of the jth class;

[0099] The model training loss function is defined as:

[0100]

[0101] Where Q is the model loss, y o，l is the true label of data o belonging to category l, σ o，l The model predicts the probability that data o belongs to category l;

[0102] The Adam optimizer is used to optimize the model and the loss gradient is calculated to update the model parameters:

[0103]

[0104] where θ t is the model parameter updated for the tth time, η is the learning rate, is the loss gradient;

[0105] Use the training data set to iteratively train the model, annotate the data in the training data set with real labels, calculate the model loss in each iteration, and stop model training after reaching the set loss or the preset number of iterations to obtain a classification recognition model;

[0106] After each electronic information recognition, the input data is saved as a training set for model training and updating.

[0107] A neural network is used to build a classification and recognition model. Through hierarchical structure and weight adjustment, complex pattern recognition tasks can be achieved. Neural networks have strong nonlinear mapping capabilities and can capture complex patterns in data and improve classification accuracy. The number of neurons in the input layer is set according to the total number of extracted time features and frequency domain features to ensure that all feature information can be fully input into the model. Reasonable input layer settings can improve the model's expressiveness and training efficiency. The hidden layer has a two-layer structure, each containing 100 neurons. The ReLU function is used as the activation function, which can effectively alleviate the gradient disappearance problem and accelerate model training. The multi-layer hidden layer design can improve the model's nonlinear representation ability and capture complex patterns in data. The number of neurons in the output layer is consistent with the number of categories in the classification task. The Softmax function is used to convert the output value into a probability distribution to ensure that the sum of all output category probabilities is 1, which is convenient for Probabilistic interpretation and decision-making of multi-classification tasks, using the cross-entropy loss function to measure the difference between the model prediction results and the true label, can effectively guide the optimization of model parameters. The cross-entropy loss function is suitable for classification tasks and has a good theoretical basis and practical effect. The Adam optimizer calculates the first-order and second-order moment estimates and adjusts the learning rate to improve the efficiency and stability of model training. It combines the advantages of the momentum method and RMSprop and is suitable for the optimization of most deep learning models. It uses the training data set to iteratively train the model, annotates the training data with real labels, and updates the parameters by calculating the model loss. Through multiple iterative training, the model gradually converges and stops training after reaching the set loss or the preset number of iterations to obtain a classification recognition model. After each electronic information recognition, the input data is saved as a training set for model training and update to ensure continuous optimization and performance improvement of the model.

[0108] Furthermore, the trained information recognition model is deployed to the database, and data classification and recognition is performed based on the fusion features. The classification and recognition model is directly deployed to the database after training. The database inputs the fusion features C" into the classification and recognition model to perform classification and recognition of electronic information data, and stores the classification results and recognition data synchronously in the database.

[0109] Deploying the trained classification and recognition model to the database can realize real-time call and online service of the model. Database deployment can make full use of the storage and computing resources of the database to ensure efficient operation of the model and rapid processing of data. By inputting fusion features into the classification and recognition model, it can make full use of the comprehensive information of multimodal data and improve the accuracy and robustness of classification and recognition. Fusion features combine the advantages of time features and frequency domain features to capture complex patterns and associations in the data. Storing the classification and recognition results and the original data synchronously in the database can facilitate subsequent data query and analysis. Synchronous storage ensures the consistency of data and classification results, which helps with system maintenance and management.

[0110] S4, visualize the data classification and recognition results;

[0111] Specifically, visually displaying the data classification and recognition results means that the database summarizes the classification results of all electronic information data and calculates a proportional graph of the classification results of all electronic information data, and synchronously displays the electronic information data, classification results and proportional graph through a visualization port, and customizes and adjusts based on user needs.

[0112] Summarizing the classification results of all electronic information data into a database helps to centrally manage and analyze data. By summarizing the classification results, we can fully understand the distribution of data and provide a basis for subsequent visualization. By calculating the proportion of all electronic information data classification results, we can intuitively display the proportion of each category of data in the whole. This visualization method helps to quickly identify the main categories and abnormal situations in the data and improve the efficiency and effectiveness of data analysis. The electronic information data, classification results and proportion charts are displayed simultaneously through the visualization port, which can provide users with an intuitive and interactive data analysis platform. Users can easily view and understand data through the visualization port, improve the convenience and accuracy of data analysis, and allow users to customize the visualization content according to their own needs, which can improve the flexibility of visualization and user experience. Users can choose the appropriate chart type, adjust the chart style and color, and even add or delete data according to specific analysis needs to obtain a more personalized display effect.

[0113] Example 2

[0114] Reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides an Internet of Things electronic information identification system, which includes:

[0115] Data acquisition module, used to collect and pre-process the electronic information data of the Internet of Things;

[0116] A feature fusion module is used to extract and fuse features of pre-processed electronic information data;

[0117] Information recognition module, used to build an information classification and recognition model and classify and recognize electronic information data based on fusion features;

[0118] Visual display module, used to calculate the classification ratio diagram of the classification results and display the data;

[0119] The data storage module is used to store the collected and analyzed data and set security access measures.

[0120] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0121] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0122] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0123] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for identifying electronic information of the Internet of Things, characterized in that: include, Deploy multimodal sensors to collect electronic information data through IoT devices and store it in the database after preprocessing; Extract the time characteristics and frequency domain characteristics of the electronic information data collected by each sensor, and perform feature fusion to form fusion features; Build an information recognition model, deploy the trained information classification and recognition model to the database, and perform data classification and recognition based on fusion features; Visualize the data classification and recognition results; The deployment of multimodal sensors to collect electronic information data through IoT devices and store it in a database after preprocessing means that the data generated by the device is collected by deploying multimodal sensors, and the data collected by each sensor is set to an independent category. After the data is collected, the collected data is directly cleaned and filtered based on edge computing technology and converted into a unified format. After the preprocessing is completed, the data is transmitted to the database through an encrypted network and stored by sensor category, and security access control is set. The database performs cloud backup of the stored data; The extracting of time characteristics and frequency domain characteristics of electronic information data collected by each sensor includes: Calculate the mean μ and variance σ of the collected data, and perform autocorrelation analysis on the collected data: where x t is the tth data point, μ is the mean of the data, N is the total number of data points, k is the number of lags, and R is the autocorrelation coefficient; The data mean, variance and autocorrelation coefficient are combined to form the time feature T = {R, μ, σ}; Use Fast Fourier Transform to extract the frequency domain representation of the data and calculate the spectral density: S=(P) 2 , Where i is the imaginary unit, ω is the frequency, P is the frequency domain representation, and S is the spectral density; The frequency domain representation and the spectral density are combined to form the frequency domain feature F = {P, S}; The feature fusion to form fusion features refers to connecting the time features and frequency domain features extracted from the electronic information data collected by each sensor to form an initial fusion feature C: C = {R, μ, σ, P, S}; Calculate the covariance matrix G of the initial fusion feature C: Where M is the total number of fusion features, C j is the jth initial fusion feature, is the mean of the initial fusion features, and T is the transposition operation; Perform eigenvalue decomposition on the covariance matrix G: G=V∧V T , Where V is the eigenvector matrix, ∧ is the eigenvalue matrix, V T is the transpose of V; Select the first k largest eigenvalues ​​and their corresponding eigenvectors to construct the dimension reduction matrix W = {v1, v2, ..., v k }, the initial fusion feature C is reduced in dimension based on the dimensionality reduction matrix: C″=CW, Where C” is the fusion feature after dimensionality reduction; The construction of the information classification and recognition model refers to the use of a neural network to construct a classification and recognition model, which includes an input layer, a hidden layer, and an output layer; Input layer: The number of neurons is set according to the total number of extracted time features and frequency domain features; Hidden layer: It has two layers, each layer contains 100 neurons. The hidden layer uses the ReLU function as the activation function, which is expressed as: where Φ i is the weight of the i-th feature, x i is the i-th feature of the input, b is the bias term, m is the total number of input features, and h is the output of the neuron; Output layer: The number of neurons in the output layer is consistent with the number of categories in the classification task, and the softmax function is used as the output layer activation function: Where zj is the input of the jth output neuron, zl is the input of the lth output neuron, L is the number of output neurons, that is, the total number of categories, σ j is the predicted probability of the jth class; The model training loss function is defined as: Where Q is the model loss, y o，l is the true label of data o belonging to category l, σ o，l The model predicts the probability that data o belongs to category l; The Adam optimizer is used to optimize the model and the loss gradient is calculated to update the model parameters: where θ t is the model parameter updated for the tth time, η is the learning rate, is the loss gradient; Use the training data set to iteratively train the model, annotate the data in the training data set with real labels, calculate the model loss in each iteration, and stop model training after reaching the set loss or the preset number of iterations to obtain a classification recognition model; After each electronic information recognition, the input data is saved as a training set for model training and updating.

2. The method for identifying electronic information of the Internet of Things according to claim 1, characterized in that: The method of deploying the trained information recognition model to the database and classifying and identifying the data based on the fusion features means that the classification recognition model is directly deployed to the database after training, and the database inputs the fusion feature C" into the classification recognition model to classify and identify the electronic information data, and stores the classification result and the recognition data synchronously in the database.

3. The method for identifying electronic information of the Internet of Things as claimed in claim 2, characterized in that: The visual display of data classification and identification results refers to the database summarizing the classification results of all electronic information data and calculating the proportion diagram of all electronic information data classification results, and synchronously displaying the electronic information data, classification results and proportion diagram through the visualization port, and customizing and adjusting based on user needs.

4. An Internet of Things electronic information identification system according to any one of claims 1 to 3, characterized in that: include, Data acquisition module, used to collect and pre-process the electronic information data of the Internet of Things; A feature fusion module is used to extract and fuse features of pre-processed electronic information data; Information recognition module, used to build an information classification and recognition model and classify and recognize electronic information data based on fusion features; Visual display module, used to calculate the classification ratio diagram of the classification results and display the data; The data storage module is used to store the collected and analyzed data and set security access measures.

5. A computer device comprising: A memory and a processor; the memory stores a computer program, wherein the processor implements the steps of the method for identifying electronic information of the Internet of Things according to any one of claims 1 to 3 when executing the computer program.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for identifying electronic information of the Internet of Things described in any one of claims 1 to 3 are implemented.

Images