Industrial Internet of Things gateway data processing method and system based on large language model, and electronic equipment

Abstract

The invention discloses an industrial Internet of Things gateway data processing method and system based on a large language model, and electronic equipment, and relates to the technical field of industrial Internet of Things, and the method comprises the steps: 1, an industrial Internet of Things gateway collects data from industrial equipment in real time, and builds a corpus as a training data set; 2, constructing a large language model, inputting the training data set into the large language model, and performing deep processing on the training data set through the large language model; and step 3, performing feedback adjustment on the working state of the industrial Internet of Things gateway according to a result after large model processing, and optimizing a subsequent data processing flow. According to the method, the intelligence of data processing is realized by utilizing the large language model, and the efficiency of data processing is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial Internet of Things, and in particular to an industrial Internet of Things gateway data processing method, system and electronic equipment based on a large language model. Background Art

[0002] With the rapid development of Internet of Things technology, gateways serve as bridges connecting different devices and networks, and undertake functions such as data transmission and protocol conversion.

[0003] Industrial IoT gateways face numerous challenges when processing data from a large number of industrial devices, including diverse data types, widely varying formats, strict real-time requirements, and large data volumes. Traditional industrial IoT gateway data processing methods often rely on preset rules, lacking flexibility and intelligence. These methods struggle to cope with increasingly complex network environments and volatile data traffic patterns, often suffering from slow processing speeds and low accuracy, making them unable to meet the data processing needs of large-scale, complex scenarios.

[0004] In view of this, it is necessary to provide an industrial Internet of Things gateway data processing method, system and electronic equipment based on a large language model. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems in the prior art to a certain extent. By proposing an industrial Internet of Things gateway data processing method, system and electronic equipment based on a large language model, the large language model is used to realize intelligent data processing and improve the efficiency of data processing.

[0006] To achieve the above objectives, in the first aspect, the present application provides, including A method for processing data of an industrial Internet of Things gateway based on a large language model, characterized by comprising the following steps: Step 1: The IIoT gateway collects data from industrial equipment in real time and builds a corpus as a training dataset; Step 2: Build a large language model, input the training data set into the large language model, and perform deep processing on the training data set through the large language model; Step 3: Use the results of large model processing to feed back and adjust the working status of the industrial Internet of Things gateway to optimize the subsequent data processing process.

[0007] Furthermore, the step 1 includes: The IIoT gateway receives information from different industrial devices and data packets from the network; Preprocess the received data, including format unification, outlier detection and removal, and duplicate removal; A corpus is constructed based on the preprocessed data, and the corpus is used as a training data set.

[0008] Furthermore, the corpus is constructed based on the preprocessed data and the corpus is used as a training data set, including: Use word element analysis algorithm to perform word segmentation on the preprocessed data; Perform word frequency statistics on the word segmentation results and perform deduplication processing based on the word frequency statistics results; Convert the deduplicated word segments into a word sequence that the model can recognize; Perform word frequency statistics on the converted word sequence, remove duplicates based on the word frequency statistics results, and generate a corpus; The corpus is used as a training dataset.

[0009] Furthermore, the word element analysis algorithm includes: byte pair encoding BPE or WordPiece word segmentation.

[0010] Furthermore, the large language model includes: a Bi-LSTM network, an attention mechanism, and a multi-layer perceptron MLP; Among them, the Attention similarity matrix between the word sequence and the word vector of the current industrial Internet of Things gateway node is calculated; According to the Attention similarity matrix, the word vectors of each industrial Internet of Things gateway node are weighted and summed; Concatenate the weighted word vector with the word vector of the next network node; The concatenated result is mapped into a fixed-length representation vector through a feedforward neural network, and then mapped into a fixed-dimensional feature vector through an MLP. Use the multi-layer perceptron MLP to map the feature vector and output the processing results.

[0011] Furthermore, the expression of the Attention similarity matrix is: ; In the formula, Softmax represents the normalization function, Q represents the word vector of the current gateway node; K represents the word vector set in the word sequence, and T represents the transpose of the matrix. Represents dimension.

[0012] Furthermore, according to the Attention similarity matrix, the word vectors of each industrial Internet of Things gateway node are weighted and summed. The expression of the weighted sum is: ; In the formula, Node_Vector represents the word vector of the current gateway node, Attention Matrix represents the similarity matrix, and Weighted_Vector represents the weighted word vector.

[0013] Furthermore, the weighted word vector is concatenated with the word vector of the next network node, expressed as: ; Where V w Represents the weighted word vector, V n The word vector representing the next node.

[0014] In a second aspect, the present application also provides an industrial Internet of Things gateway data processing system based on a large language model, comprising the following modules: Data acquisition module: The industrial IoT gateway collects data from industrial equipment in real time and builds a corpus as a training data set; Large language model construction module: constructs a large language model, inputs the training data set into the large language model, and performs deep processing on the training data set through the large language model; Output feedback module: The results of large model processing are used to adjust the working status of the industrial Internet of Things gateway and optimize the subsequent data processing process.

[0015] In a third aspect, the present application also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, an industrial Internet of Things gateway data processing method based on a large language model is implemented.

[0016] Beneficial effects of the present invention: Compared with existing technologies, the present invention provides an industrial Internet of Things gateway data processing method, system, and electronic device based on a large language model. By integrating advanced large model technology, the data received by the industrial Internet of Things gateway is intelligently analyzed and processed to achieve rapid data response, accurate classification, and efficient data transmission. Specific beneficial effects are as follows; (1) Improve data processing efficiency: By integrating large model technology, rapid response and efficient processing of data can be achieved.

[0017] (2) Improve data accuracy: Utilize the intelligent analysis capabilities of large models to improve data classification and recognition accuracy.

[0018] (3) Reduce resource consumption: Reduce unnecessary data transmission and resource waste by optimizing data transmission strategies and control decisions.

[0019] (4) Enhance system stability: Improve the overall stability and reliability of the Industrial Internet of Things gateway through real-time feedback and dynamic adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of the steps of the method of the present invention; Figure 2 is a functional block diagram of the system of the present invention; DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] Example 1, please refer to Figure 1 As shown, the present application provides an industrial Internet of Things gateway data processing method based on a large language model, comprising the following steps: Step 1: The IIoT gateway collects data from industrial equipment in real time and builds a corpus as a training dataset; Step 2: Build a large language model, input the training data set into the large language model, and perform deep processing on the training data set through the large language model; Step 3: Use the results of large model processing to feed back and adjust the working status of the industrial Internet of Things gateway to optimize the subsequent data processing process.

[0023] In the specific implementation process, step one includes: The IIoT gateway receives information from different industrial devices and data packets from the network; Preprocess the received data, including format unification, outlier detection and removal, and deduplication to ensure data quality; A corpus is constructed based on the preprocessed data, and the corpus is used as a training dataset.

[0024] Specifically, based on the above steps, ensure that the text data has been deduplicated and tokenized. Deduplication can avoid redundancy when the model processes repeated information, and tokenization breaks down the text into meaningful units (words or subwords).

[0025] In the specific implementation process, a corpus is constructed based on the preprocessed data and used as a training dataset, including: Use word element analysis algorithm to perform word segmentation on the preprocessed data; Perform word frequency statistics on the word segmentation results and perform deduplication processing based on the word frequency statistics results; Convert the deduplicated word segments into a word sequence that the model can recognize for subsequent processing; Perform word frequency statistics on the converted word sequence, remove duplicates based on the word frequency statistics results, and generate a corpus; The corpus is used as a training dataset.

[0026] In the specific implementation process, the word element analysis algorithm includes: byte pair encoding BPE or WordPiece word segmentation.

[0027] Specifically, constructing a corpus of word sequences is an important step in natural language processing, which is related to the training and effects of various subsequent models.

[0028] More specifically, you can use lists, sets, or specific data types to store word sequences. Choosing the right data structure will help with subsequent processing. Specifically, you can save the corpus as a text file, CSV, JSON, or database, depending on your subsequent usage requirements.

[0029] Specifically, after building the corpus, you can perform vectorization processing and use word embedding models (such as Word2Vec, GloVe, etc.) to convert word units into vector representations. Specifically, the following steps are involved: 1. Build a vocabulary: You need to create a vocabulary that contains all possible words and their indexes. This process usually includes the following steps: Collect corpus: Get word segmentation results from all preprocessed texts.

[0030] Term frequency statistics: Calculate the frequency of each word and set a threshold to decide which words to keep in the vocabulary.

[0031] Build an index: Give each word a unique index.

[0032] 2. Mapping tokens to indexes: Using the constructed vocabulary, map the deduplicated tokens to corresponding token indices. Assume there is a token result ["Token 1", "Token 2", "Token 3"], whose indices in the vocabulary are 1, 2, and 3 respectively. The index sequence for this token can be expressed as: ; 3. Create word sequence vector In one embodiment, the model requires vector representations of word units, such as: One-Hot Encoding: Creates a one-hot vector for each token, which is suitable for small vocabularies.

[0033] Word Embeddings: Use pre-trained word vector models (such as Word2Vec and GloVe) or word embeddings obtained through model training (such as BERT and GPT) to represent word units. This method can better capture the semantic relationship of word units in high-dimensional space.

[0034] Generate the final input vector: Convert the index or embedding representation of the word unit into an input form acceptable to the model (such as a tensor), and adjust the input dimension according to the specific model requirements.

[0035] Specifically, a large language model is trained based on the constructed corpus.

[0036] In the specific implementation process, the large language model includes: Bi-LSTM network, attention mechanism and multi-layer perceptron MLP; Among them, the Attention similarity matrix between the word sequence and the word vector of the current industrial Internet of Things gateway node is calculated; According to the Attention similarity matrix, the word vectors of each industrial IoT gateway node are weighted and summed to achieve attention alignment; Concatenate the weighted word vector with the word vector of the next network node; The concatenated result is mapped into a fixed-length representation vector through a feedforward neural network, and then mapped into a fixed-dimensional feature vector through an MLP. Use the multi-layer perceptron MLP to map the feature vector and output the processing results.

[0037] In the specific implementation process, the expression of the Attention similarity matrix is: ; In the formula, Softmax represents the normalization function, Q represents the word vector of the current gateway node; K represents the word vector set in the word sequence, and T represents the transpose of the matrix. Represents dimension.

[0038] In the specific implementation process, according to the Attention similarity matrix, the word vector of each industrial Internet of Things gateway node is weighted and summed. The expression of the weighted sum is: ; In the formula, Node_Vector represents the word vector of the current gateway node, Attention Matrix represents the similarity matrix, and Weighted_Vector represents the weighted word vector.

[0039] In the specific implementation process, the weighted word vector is spliced ​​with the word vector of the next network node, which is expressed as: ; Where V w Represents the weighted word vector, V n The word vector representing the next node.

[0040] Specifically, the feedforward neural network mapping is an implicit step that maps the concatenated vector to a fixed-length representation through a series of linear transformations and nonlinear activation functions.

[0041] A multi-layer perceptron (MLP) consists of multiple fully connected layers and activation functions, and the size of the final output depends on the specific task requirements (such as classification or regression).

[0042] Through the above steps, an attention-based alignment process is implemented, and a multilayer perceptron is used for feature extraction and mapping. In practical applications, the model structure, hyperparameters, and training process can be flexibly adjusted according to needs. This invention helps improve the understanding of complex relationships and achieve better performance in tasks such as the Industrial Internet of Things.

[0043] See also Figure 2 As shown, the present application also provides 9. an industrial Internet of Things gateway data processing system based on a large language model, characterized in that it includes the following modules: Data acquisition module: The industrial IoT gateway collects data from industrial equipment in real time and builds a corpus as a training data set; Large language model construction module: Build a large language model, input the training data set into the large language model, and perform in-depth processing on the training data set through the large language model; Output feedback module: The results of large model processing are used to adjust the working status of the industrial Internet of Things gateway and optimize the subsequent data processing process.

[0044] The present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, an industrial Internet of Things gateway data processing method based on a large language model is implemented.

[0045] Example 2, using a large language model-based industrial Internet of Things gateway data processing method provided by this application for specific industrial Internet of Things gateway data, and applying it to the scenario of power outage and power supply transfer strategy optimization in power scheduling, includes: Step 1: The IIoT gateway collects data from industrial equipment in real time and builds a corpus as a training dataset; Step 2: Build a large language model, input the training data set into the large language model, and perform deep processing on the training data set through the large language model; Step 3: Use the results of large model processing to feed back and adjust the working status of the industrial Internet of Things gateway to optimize the subsequent data processing process.

[0046] During the specific implementation process, the industrial Internet of Things gateway collects data from various types of power equipment in real time, including power outage information, grid status, equipment status, etc.

[0047] The collected data is cleaned to remove noise, outliers and redundant information.

[0048] Formatting: Formatting the cleaned data into a unified format, including: format unification, outlier detection and removal, and deduplication, to facilitate subsequent processing.

[0049] During the specific implementation process, part-of-speech tagging and named entity recognition are performed on power outage entries to identify key information such as the power outage area and power outage time.

[0050] Corpus construction: The annotated power outage entries are constructed into a corpus as training data.

[0051] To obtain word vectors, we use word embedding models (such as Word2Vec and GloVe) to convert word units into vector representations. These word embeddings are then fed into a Bi-LSTM network to further learn the contextual information of the words and generate richer word vectors.

[0052] In the specific implementation process, the power outage node and the power supply transfer node are processed, including: Node identification: Extract power outage nodes and potential power transfer nodes from power outage information.

[0053] Word embedding conversion: The power outage nodes and power supply transfer nodes are converted into corresponding word embedding vectors and input into the same Bi-LSTM network to obtain the word embedding vectors to be processed that contain context information.

[0054] In the specific implementation process, result splicing and attention mechanism: Attention similarity calculation: Calculate the attention similarity matrix between the word vector of the power outage node and the word vector to be processed for each power supply node.

[0055] Threshold Attention Alignment and Concatenation: Based on the Attention Similarity Matrix, the word vectors for each power supply node are weighted and summed to achieve attention alignment. The weighted word vectors are then concatenated with the word vectors for the power outage node to form the final representation vector.

[0056] During the specific implementation process, the selection and optimization of the supply transfer strategy includes: Feature mapping: The concatenated result is mapped into a vector of fixed length through a feedforward neural network, and then mapped into a feature vector of fixed dimension through MLP.

[0057] Prediction and Decision-Making: Another MLP layer maps the feature vectors and outputs a predicted value. A matching threshold is preset, and the node with the highest predicted value exceeding the threshold is selected as the optimal power transfer strategy.

[0058] Strategy execution: Send the optimal power transfer strategy to the power grid control system and execute the corresponding power transfer operation.

[0059] During the specific implementation process, monitoring and feedback include: Real-time monitoring: Real-time monitoring of power grid status and power transfer effects through the Industrial Internet of Things gateway.

[0060] Feedback optimization: Adjust model parameters and matching thresholds based on actual results, and optimize the power transfer strategy selection algorithm.

[0061] Through the above method, the industrial Internet of Things gateway can realize the intelligent processing of power outage and power supply transfer strategies in power dispatching, and improve the reliability and operation efficiency of the power grid.

[0062] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of systems, modules and units can be electrical, mechanical or other forms.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A data processing method for industrial Internet of Things gateway based on large language model, It is characterized in that The following steps are involved: Step 1: The IIoT gateway collects data from industrial equipment in real time and builds a corpus as a training data set; Step 2: construct a large language model, input the training data set into the large language model, and perform deep processing on the training data set through the large language model; Step 3: Use the processing results of the large model to adjust the working status of the industrial Internet of Things gateway and optimize the subsequent data processing process.

2. According to claim 1, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that The step one comprises: The IIoT gateway receives information from different industrial devices and data packets from the network; Preprocess the received data, including: format unification, outlier detection and removal, and deduplication; A corpus is constructed based on the preprocessed data, and the corpus is used as a training data set.

3. According to the method for processing data of an industrial Internet of Things gateway based on a large language model according to claim 1, It is characterized in that The method of constructing a corpus based on the preprocessed data and using the corpus as a training data set includes: The word element analysis algorithm is used to perform word segmentation on the preprocessed data; Perform word frequency statistics on the word segmentation results and perform deduplication processing based on the word frequency statistics results; Convert the deduplicated segmented words into a word sequence that the model can recognize; Perform word frequency statistics on the converted word sequence, perform deduplication processing based on the word frequency statistics results, and generate a corpus; The corpus is used as a training data set.

4. According to claim 3, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that The word element analysis algorithm includes: byte pair encoding BPE or WordPiece word segmentation.

5. According to claim 3, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that The large language model includes: a Bi-LSTM network, an attention mechanism, and a multi-layer perceptron MLP; Among them, the Attention similarity matrix between the word sequence and the word vector of the current industrial Internet of Things gateway node is calculated; According to the Attention similarity matrix, the word vectors of each industrial IoT gateway node are weighted and summed; Concatenate the weighted word vector with the word vector of the next network node; The concatenated result is mapped into a representation vector of fixed length through a feedforward neural network, and then mapped into a feature vector of fixed dimension through MLP; Use multi-layer perceptron MLP to map the feature vector and output the processing result.

6. According to claim 5, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that The expression of the Attention similarity matrix is: ； In the formula, Softmax represents the normalization function, Q represents the word vector of the current gateway node; K represents the word vector set in the word sequence, and T represents the transpose of the matrix. Represents dimension.

7. According to claim 6, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that According to the Attention similarity matrix, the word vectors of each industrial Internet of Things gateway node are weighted and summed. The expression of the weighted sum is: ； In the formula, Node_Vector represents the word vector of the current gateway node, Attention Matrix represents the similarity matrix, and Weighted_Vector represents the weighted word vector.

8. According to claim 7, a method for processing data of an industrial Internet of Things gateway based on a large language model, It is characterized in that The weighted word vector is concatenated with the word vector of the next network node, expressed as: ； Where V w Represents the weighted word vector, V n The word vector representing the next node.

9. An industrial Internet of Things gateway data processing system based on a large language model, It is characterized in that Includes the following modules: Data acquisition module: The industrial IoT gateway collects data from industrial equipment in real time and builds a corpus as a training data set; Large language model construction module: construct a large language model, input the training data set into the large language model, and perform deep processing on the training data set through the large language model; Output feedback module: Feedback the results of large model processing to the working status of the industrial Internet of Things gateway to optimize the subsequent data processing process.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, the industrial Internet of Things gateway data processing method based on a large language model is implemented as described in any one of claims 1 to 8.

Images