Data management platform of industrial internet

Abstract

The invention discloses a data management platform of an industrial internet, and relates to the technical field of industrial internet data management. Comprising a data acquisition module which is responsible for collecting original data from various industrial devices and sensors; the module adopts a self-adaptive sampling algorithm, and dynamically adjusts the sampling frequency and precision according to the equipment state and the network condition; the data preprocessing module is used for cleaning, formatting and preliminarily analyzing the collected data; the module uses a mathematical formula to carry out abnormal value detection and correction, such as an abnormal value detection method based on a Gaussian mixture model. According to the invention, the platform can efficiently collect and preprocess data from various industrial devices and sensors through a self-adaptive sampling algorithm, a Gaussian mixture model and other methods, so that the quality and availability of the data are ensured; by adopting a distributed database and an enhanced encryption technology, the security and expandability of the data are guaranteed, and meanwhile, the continuous availability of the data is ensured by a backup and recovery mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial Internet data management, and in particular to a data management platform for industrial Internet. Background Art

[0002] With the continuous advancement of industrial technology, more and more industrial equipment and systems are connected through the Internet, generating a large amount of data. The effective management and analysis of this data is of great significance for improving production efficiency, reducing operating costs, optimizing resource allocation, etc. However, the data management platform in the existing technology often cannot meet the high requirements for real-time, accuracy and security of data processing in the industrial Internet environment.

[0003] After searching, the Chinese patent application number CN202311806458.1 discloses an industrial Internet cloud platform based on data intelligence, including an integration and expansion layer, an edge computing layer, and a supervision and compliance layer. The edge computing layer is connected to a security enhancement layer, the security enhancement layer is connected to a data management layer, the data management layer is connected to a platform service layer, the platform service layer is connected to an application and analysis layer, the integration and expansion layer is connected to an application and analysis layer, the supervision and compliance layer is connected to an edge computing layer, a security enhancement layer, a data management layer, a platform service layer, an application and analysis layer, and an integration and expansion layer, and the security enhancement layer is connected to a security decision algorithm. The industrial Internet cloud platform in the above patent is still insufficient in terms of in-depth data analysis and processing capabilities. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a data management platform for the industrial Internet.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] An industrial Internet data management platform, comprising:

[0007] Data acquisition module: responsible for collecting raw data from various industrial equipment and sensors; this module uses an adaptive sampling algorithm to dynamically adjust the sampling frequency and accuracy according to the device status and network conditions;

[0008] Data preprocessing module: cleans, formats and preliminarily analyzes the collected data; this module uses mathematical formulas to detect and correct outliers, such as outlier detection methods based on Gaussian mixture models;

[0009] Data storage module: stores processed data in a distributed database to ensure high availability and scalability of data; this module uses encrypted storage and access control mechanisms to ensure data security;

[0010] Data analysis module: performs in-depth analysis of stored data, including machine learning model training, trend prediction, and pattern recognition; for example, using time series analysis models to predict the probability of equipment failure;

[0011] Intelligent decision support module: uses machine learning and artificial intelligence algorithms to further interpret the analysis results and provide decision support; this module can automatically adjust production parameters, optimize resource allocation, and make maintenance recommendations; for example, it uses deep learning algorithms for image recognition to automatically detect product quality issues on the production line;

[0012] Edge computing module: implements data computing on devices near the data collection source to reduce data transmission volume and latency. This module uses edge computing frameworks such as Apache Edgent or AWS Greengrass to achieve local data processing and rapid response.

[0013] Security control module: responsible for the implementation of security policies for the entire platform, including data encryption, access control, and intrusion detection; this module uses encryption algorithms and authentication mechanisms to ensure the security of data transmission and storage.

[0014] Preferably: the data preprocessing module comprises:

[0015] Data cleaning unit: responsible for removing noise, correcting errors and processing missing values;

[0016] Data formatting unit: converts the collected data into a unified format for subsequent processing;

[0017] Preliminary analysis unit: Use statistical methods and pattern recognition techniques to conduct preliminary data analysis and extract key features;

[0018] Based on principal component analysis, key features are extracted by finding the main change direction of the data, specifically:

[0019]

[0020] Where S is the covariance matrix of the data, and the eigenvalues ​​and eigenvectors of S are solved.

[0021] Preferably, the data storage module comprises:

[0022] Distributed database unit: Use NoSQL database to store large-scale data to ensure scalability and high performance;

[0023] Data encryption unit: encrypt sensitive data to protect data privacy;

[0024] Backup and recovery unit: implement data backup strategy and support rapid data recovery;

[0025] The data analysis module comprises:

[0026] Machine Learning Unit: Apply supervised and unsupervised learning algorithms for pattern recognition and predictive analysis;

[0027] Time Series Analysis Unit: Use ARIMA and State Space Model models for trend prediction and anomaly detection;

[0028] Optimization algorithm unit: Use genetic algorithms for parameter optimization and resource scheduling.

[0029] Preferably, the optimization algorithm unit uses the genetic algorithm to perform parameter optimization and resource scheduling in the following specific manner:

[0030] S1: Representation: Each individual is encoded as a chromosome, represented by a binary string or a real-valued vector;

[0031] S2: Initial population: randomly generate the initial population P 0 ;

[0032] S3: Fitness function: Define the fitness function f(x) to measure the fitness of individual x;

[0033] S4: Selection: Individuals are selected based on their fitness, and individuals with high fitness have a higher probability of being selected for reproduction;

[0034] S5: Crossover: Producing offspring through crossover operations, such as single-point crossover, multi-point crossover, or uniform crossover; crossover probability P c Control the frequency of crossover operations;

[0035] S6: Mutation: with probability P m Make random changes to certain genes of individuals to increase the diversity of the population;

[0036] S7: New generation population: Generate new population through selection, crossover and mutation operations;

[0037] S8: Repeat the above process until the termination condition is met.

[0038] Preferably: the safety control module comprises:

[0039] Authentication and authorization unit: implement user identity authentication and permission control to ensure platform security;

[0040] Data encryption unit: uses AES encryption standard to protect the security of data during transmission and storage;

[0041] Intrusion Detection System Unit: Monitor and prevent malicious access and attacks.

[0042] Preferably: the intelligent decision support module comprises:

[0043] Artificial Intelligence Algorithm Unit: Uses deep learning for image and speech recognition to automatically detect and classify production line problems;

[0044] Optimization and simulation unit: evaluate the impact of different decision-making options through simulation models to assist in making optimal decisions;

[0045] Automated Execution Unit: Integrate automation tools and robotics to implement intelligent decision-making.

[0046] Preferably, the artificial intelligence algorithm unit uses deep learning to perform image and speech recognition, specifically in the following manner:

[0047] S11: Data preprocessing: converting image and speech data into a format suitable for neural network processing;

[0048] S12: Model construction: Build a deep neural network, including an input layer, multiple hidden layers, and an output layer;

[0049] S13: Forward propagation: Input data passes through the network and calculates the output of each layer;

[0050] S14: Loss function: defines the difference between the network output and the true value based on the cross entropy loss;

[0051] S15: Back propagation: calculate the gradient of the loss function for each parameter and update the network parameters;

[0052] S16: Iterative training: repeat forward propagation, loss calculation and parameter update until convergence;

[0053] For each neuron, its output is expressed as y = σ(W x +b), where σ is the activation function, W is the weight matrix, x is the input vector, and b is the bias term;

[0054] Among them, the cross entropy loss formula is: where t i is the true label, s i is the network output, C is the number of categories;

[0055] Among them, the gradient descent formula is: Where η is the learning rate, is the gradient of the loss function with respect to the weights.

[0056] Preferably: the edge computing module comprises:

[0057] Local data processing unit: performs data preprocessing and analysis near the data source to reduce data transmission and latency;

[0058] Real-time response unit: implements real-time data processing and decision-making to support rapid response and immediate control;

[0059] Synchronization and coordination unit: Ensures the collaboration between edge devices and cloud services to achieve optimal use of data and resources.

[0060] Preferably, the artificial intelligence algorithm unit performs image recognition through a convolutional neural network, specifically:

[0061]

[0062] Among them, y ij is the output feature map at position (i, j), w mn is the convolution kernel weight, x is the input feature map, b is the bias term, and σ is the activation function;

[0063] The artificial intelligence algorithm unit, through long short-term memory network speech recognition, is specifically:

[0064] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0065] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0066]

[0067] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0068] h t =o t *tanh(C t )

[0069] Among them, f t ,i t , o t are the vectors of the forget gate, input gate, and output gate respectively. is the candidate cell state, C t is the cell state, h t is the hidden state vector, W is the weight matrix, b is the bias term, and σ and tanh are activation functions.

[0070] The beneficial effects of the present invention are:

[0071] 1. Through methods such as adaptive sampling algorithms and Gaussian mixture models, the platform of the present invention can efficiently collect and pre-process data from a variety of industrial equipment and sensors to ensure the quality and availability of data; the use of distributed databases and enhanced encryption technology ensures data security and scalability, while the backup and recovery mechanism ensures the continuous availability of data.

[0072] 2. The present invention utilizes advanced algorithms such as time series analysis and genetic algorithms for data mining and predictive analysis to help industrial enterprises optimize resource allocation and prevent potential failures; by processing data near the data source, the demand and delay of data transmission are reduced, the processing speed and efficiency are improved, and it is particularly suitable for real-time monitoring and rapid decision-making.

[0073] 3. The present invention integrates deep learning and other artificial intelligence technologies to provide automated image and voice recognition, as well as optimization and simulation tools to help enterprises improve production quality and maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A flowchart of using genetic algorithm to perform parameter optimization and resource scheduling in a data management platform for industrial Internet proposed by the present invention;

[0075] Figure 2 This is a flowchart of using deep learning for image and speech recognition in an industrial Internet data management platform proposed by the present invention. DETAILED DESCRIPTION

[0076] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0077] Embodiment 1:

[0078] An industrial Internet data management platform, comprising:

[0079] Data acquisition module: collects raw data from various industrial equipment and sensors; uses adaptive sampling algorithm to dynamically adjust sampling frequency and accuracy according to equipment status and network conditions;

[0080] Data preprocessing module: cleans, formats and preliminarily analyzes the collected data; this module detects and corrects outliers based on the outlier detection method of the Gaussian mixture model:

[0081]

[0082] Among them, p(x) is the probability distribution of the observed data, π i is the mixing coefficient, N(x|μ i ,Σi is a product with mean μ i and covariance Σ i Normal distribution of

[0083] Data storage module: Store the processed data in a distributed database to ensure high availability and scalability of the data; use encrypted storage and access control mechanisms to ensure data security;

[0084] Data analysis module: In-depth analysis of stored data, including machine learning model training, trend prediction, and pattern recognition; use time series analysis models to predict the probability of equipment failure:

[0085]

[0086] Where F(t) is the cumulative failure distribution function, S(t) is the survival function, and h(u) is the failure rate function;

[0087] Security control module: responsible for the implementation of security policies for the entire platform, including data encryption, access control, and intrusion detection.

[0088] Wherein, the data acquisition module includes:

[0089] Equipment interface unit: responsible for connecting with different types of industrial equipment and sensors to collect real-time data;

[0090] Network communication protocol unit: supports various communication standards, including OPC-UA, MQTT, AMQP, to ensure reliable data transmission;

[0091] Adaptive sampling unit: adjusts sampling frequency and accuracy according to device status and network conditions to optimize data flow and quality.

[0092] Wherein, the data preprocessing module includes:

[0093] Data cleaning unit: responsible for removing noise, correcting errors and processing missing values;

[0094] Data formatting unit: converts the collected data into a unified format for subsequent processing;

[0095] Preliminary Analysis Unit: Use statistical methods and pattern recognition techniques to conduct preliminary data analysis and extract key features.

[0096] Based on principal component analysis, key features are extracted by finding the main change direction of the data, specifically:

[0097]

[0098] Where S is the covariance matrix of the data, and the eigenvalues ​​and eigenvectors of S are solved.

[0099] Wherein, the data storage module includes:

[0100] Distributed database unit: Use NoSQL database to store large-scale data to ensure scalability and high performance;

[0101] Data encryption unit: encrypt sensitive data to protect data privacy;

[0102] Backup and recovery unit: implements data backup strategies and supports rapid data recovery.

[0103] Wherein, the data analysis module includes:

[0104] Machine Learning Unit: Apply supervised and unsupervised learning algorithms for pattern recognition and predictive analysis;

[0105] Time Series Analysis Unit: Use ARIMA and State Space Model models for trend prediction and anomaly detection;

[0106] Optimization algorithm unit: Use genetic algorithms for parameter optimization and resource scheduling.

[0107] The specific method of the optimization algorithm unit using the genetic algorithm to perform parameter optimization and resource scheduling is as follows:

[0108] S1: Representation: Each individual (solution) is encoded as a chromosome, represented by a binary string or a real-valued vector;

[0109] S2: Initial population: randomly generate the initial population P 0 ;

[0110] S3: Fitness function: Define the fitness function f(x) to measure the fitness of individual x;

[0111] S4: Selection: Individuals are selected based on their fitness, and individuals with high fitness have a higher probability of being selected for reproduction;

[0112] S5: Crossover: Producing offspring through crossover operations, such as single-point crossover, multi-point crossover, or uniform crossover; crossover probability P c Control the frequency of crossover operations;

[0113] S6: Mutation: with probability P m Make random changes to certain genes of individuals to increase the diversity of the population;

[0114] S7: New generation population: Generate new population through selection, crossover and mutation operations;

[0115] S8: Repeat the above process until the termination condition is met.

[0116] Wherein, the safety control module includes:

[0117] Authentication and authorization unit: implement user identity authentication and permission control to ensure platform security;

[0118] Data encryption unit: uses AES encryption standard to protect the security of data during transmission and storage;

[0119] Intrusion Detection System (IDS) Unit: Monitors and prevents malicious access and attacks.

[0120] Embodiment 2:

[0121] A data management platform for industrial Internet, based on embodiment 1, this embodiment further includes:

[0122] Intelligent decision support module: Use machine learning and artificial intelligence algorithms to further interpret the analysis results and provide decision support; automatically adjust production parameters, optimize resource allocation, and make maintenance recommendations; use deep learning algorithms for image recognition to automatically detect product quality issues on the production line;

[0123] Edge computing module: implements data computing on devices near the data collection source to reduce data transmission volume and latency; this module uses Apache Edgent to achieve local data processing and rapid response.

[0124] Wherein, the intelligent decision support module includes:

[0125] Artificial Intelligence Algorithm Unit: Uses deep learning for image and speech recognition to automatically detect and classify production line problems;

[0126] Optimization and simulation unit: evaluate the impact of different decision-making options through simulation models to assist in making optimal decisions;

[0127] Automated Execution Unit: Integrate automation tools and robotics to implement intelligent decision-making.

[0128] The artificial intelligence algorithm unit uses deep learning to perform image and speech recognition in the following ways:

[0129] S11: Data preprocessing: converting image and speech data into a format suitable for neural network processing;

[0130] S12: Model construction: Build a deep neural network, including an input layer, multiple hidden layers, and an output layer;

[0131] S13: Forward propagation: Input data passes through the network and calculates the output of each layer;

[0132] S14: Loss function: defines the difference between the network output and the true value based on the cross entropy loss;

[0133] S15: Back propagation: calculate the gradient of the loss function for each parameter and update the network parameters;

[0134] S16: Iterative training: repeat forward propagation, loss calculation and parameter update until convergence;

[0135] For each neuron, its output is expressed as y = σ(W x +b), where σ is the activation function, W is the weight matrix, x is the input vector, and b is the bias term;

[0136] Among them, the cross entropy loss formula is: where t i is the true label, s i is the network output, C is the number of categories;

[0137] Among them, the gradient descent formula is: Where η is the learning rate, is the gradient of the loss function with respect to the weights.

[0138] Wherein, the edge computing module includes:

[0139] Local data processing unit: performs data preprocessing and analysis near the data source to reduce data transmission and latency;

[0140] Real-time response unit: implements real-time data processing and decision-making to support rapid response and immediate control;

[0141] Synchronization and coordination unit: Ensures the collaboration between edge devices and cloud services to achieve optimal use of data and resources.

[0142] Embodiment 3:

[0143] A data management platform for the industrial Internet. This embodiment is based on Embodiment 1:

[0144] The artificial intelligence algorithm unit performs image recognition through a convolutional neural network, specifically:

[0145]

[0146] Among them, y ij is the output feature map at position (i, j), w mn is the convolution kernel weight, x is the input feature map, b is the bias term, and σ is the activation function.

[0147] The artificial intelligence algorithm unit, through long short-term memory network speech recognition, is specifically:

[0148] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0149] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0150]

[0151] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0152] h t =o t *tanh(C t )

[0153] Among them, f t ,i t , o t are the vectors of the forget gate, input gate, and output gate respectively. is the candidate cell state, C t is the cell state, h t is the hidden state vector, W is the weight matrix, b is the bias term, and σ and tanh are activation functions.

[0154] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A data management platform for industrial Internet, characterized in that: include: Data acquisition module: collects raw data from various industrial equipment and sensors; Adopt adaptive sampling algorithm to dynamically adjust sampling frequency and accuracy according to device status and network conditions; Data preprocessing module: cleans, formats and preliminarily analyzes the collected data; this module detects and corrects outliers based on the outlier detection method of the Gaussian mixture model: Among them, p(x) is the probability distribution of the observed data, π i is the mixing coefficient, N(x|μ i ,Σ i is a product with mean μ i and covariance Σ i Normal distribution of Data storage module: Store the processed data in a distributed database to ensure high availability and scalability of the data; use encrypted storage and access control mechanisms to ensure data security; Data analysis module: In-depth analysis of stored data, including machine learning model training, trend prediction, and pattern recognition; use time series analysis models to predict the probability of equipment failure: Where F(t) is the cumulative failure distribution function, S(t) is the survival function, and h(u) is the failure rate function; Security control module: responsible for the implementation of security policies for the entire platform, including data encryption, access control, and intrusion detection.

2. The data management platform for industrial Internet according to claim 1, characterized in that: Also includes: Intelligent decision support module: Use machine learning and artificial intelligence algorithms to further interpret the analysis results and provide decision support; automatically adjust production parameters, optimize resource allocation, and make maintenance recommendations; use deep learning algorithms for image recognition to automatically detect product quality issues on the production line; Edge computing module: implements data computing on devices near the data collection source to reduce data transmission volume and latency; this module uses Apache Edgent to achieve local data processing and rapid response.

3. The data management platform for industrial Internet according to claim 1, characterized in that: The data preprocessing module comprises: Data cleaning unit: responsible for removing noise, correcting errors and processing missing values; Data formatting unit: converts the collected data into a unified format for subsequent processing; Preliminary analysis unit: Use statistical methods and pattern recognition techniques to conduct preliminary data analysis and extract key features; Based on principal component analysis, key features are extracted by finding the main change direction of the data, specifically: Where S is the covariance matrix of the data, and the eigenvalues ​​and eigenvectors of S are solved.

4. The data management platform for industrial Internet according to claim 1, characterized in that: The data storage module comprises: Distributed database unit: Use NoSQL database to store large-scale data to ensure scalability and high performance; Data encryption unit: encrypt sensitive data to protect data privacy; Backup and recovery unit: implement data backup strategy and support rapid data recovery; The data analysis module comprises: Machine Learning Unit: Apply supervised and unsupervised learning algorithms for pattern recognition and predictive analysis; Time Series Analysis Unit: Use ARIMA and State Space Model models for trend prediction and anomaly detection; Optimization algorithm unit: Use genetic algorithms for parameter optimization and resource scheduling.

5. The data management platform for industrial Internet according to claim 4, characterized in that: The specific method of the optimization algorithm unit using the genetic algorithm to perform parameter optimization and resource scheduling is as follows: S1: Representation: Each individual is encoded as a chromosome, represented by a binary string or a real-valued vector; S2: Initial population: randomly generate the initial population P0; S3: Fitness function: Define the fitness function f(x) to measure the fitness of individual x; S4: Selection: Individuals are selected based on their fitness, and individuals with high fitness have a higher probability of being selected for reproduction; S5: Crossover: Produce offspring through crossover operation, crossover probability P c Control the frequency of crossover operations; S6: Mutation: with probability P m Make random changes to certain genes of individuals to increase the diversity of the population; S7: New generation population: Generate new population through selection, crossover and mutation operations; S8: Repeat the above process until the termination condition is met.

6. The data management platform for industrial Internet according to claim 1, characterized in that: The safety control module comprises: Authentication and authorization unit: implement user identity authentication and permission control to ensure platform security; Data encryption unit: uses AES encryption standard to protect the security of data during transmission and storage; Intrusion Detection System Unit: Monitors and prevents malicious access and attacks.

7. The data management platform for industrial Internet according to claim 2, characterized in that: The intelligent decision support module comprises: Artificial Intelligence Algorithm Unit: Uses deep learning for image and speech recognition to automatically detect and classify production line problems; Optimization and simulation unit: evaluate the impact of different decision-making options through simulation models to assist in making optimal decisions; Automated Execution Unit: Integrate automation tools and robotics to implement intelligent decision-making.

8. The data management platform for industrial Internet according to claim 7, characterized in that: The artificial intelligence algorithm unit uses deep learning to perform image and speech recognition in the following ways: S11: Data preprocessing: converting image and speech data into a format suitable for neural network processing; S12: Model construction: Build a deep neural network, including an input layer, multiple hidden layers, and an output layer; S13: Forward propagation: Input data passes through the network and calculates the output of each layer; S14: Loss function: defines the difference between the network output and the true value based on the cross entropy loss; S15: Back propagation: calculate the gradient of the loss function for each parameter and update the network parameters; S16: Iterative training: repeat forward propagation, loss calculation and parameter update until convergence; For each neuron, its output is expressed as y = σ(W x +b), where σ is the activation function, W is the weight matrix, x is the input vector, and b is the bias term; Among them, the cross entropy loss formula is: where t i is the true label, s i is the network output, C is the number of categories; Among them, the gradient descent formula is: Where η is the learning rate, is the gradient of the loss function with respect to the weights.

9. The data management platform for industrial Internet according to claim 7, characterized in that: The edge computing module includes: Local data processing unit: performs data preprocessing and analysis near the data source to reduce data transmission and latency; Real-time response unit: implements real-time data processing and decision-making to support rapid response and immediate control; Synchronization and coordination unit: Ensures the collaboration between edge devices and cloud services to achieve optimal use of data and resources.

10. The data management platform of the industrial Internet according to claim 7, characterized in that: The artificial intelligence algorithm unit performs image recognition through a convolutional neural network, specifically: Among them, y ij is the output feature map at position (i, j), w mn is the convolution kernel weight, x is the input feature map, b is the bias term, and σ is the activation function; The artificial intelligence algorithm unit, through long short-term memory network speech recognition, is specifically: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) the t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t *tanh(C t ) Among them, f t ,i t , o t are the vectors of the forget gate, input gate, and output gate respectively. is the candidate cell state, C t is the cell state, h t is the hidden state vector, W is the weight matrix, b is the bias term, and σ and tanh are activation functions.

Images