Intelligent manufacturing quality data traceability method, system, equipment and storage medium based on blockchain
By obtaining the semantic embedding vector and traceability supervision decision algorithm of intelligent IoT sensing data, the traceability attribute description features are extracted and algorithm debugged, the problem of identifying and indexing feature representation of product quality traceability information in intelligent manufacturing is solved, and efficient and accurate traceability and management.
Patent Information
- Application Number
- CN202411345255.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-26
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-26
AI Technical Summary
The existing technology is difficult to effectively combine intelligent IoT sensing data and blockchain technology to achieve comprehensive, real-time and accurate identification of product quality traceability information, especially when processing large amounts of multi-dimensional data, how to extract key traceability attribute description features and in-district Implementing full-life cycle index feature representation on blockchain regulatory nodes is a challenge.
By obtaining the semantic embedding vectors of blockchain traceability nodes and product quality traceability information in intelligent IoT sensing data, the traceability attribute description features are extracted, and the traceability supervision decision algorithm is used to perform project identification and algorithm debugging, and the debugged algorithm is generated to Implement accurate traceability project identification and index feature representation.
It improves the accuracy and efficiency of product quality traceability, enhances the traceability and manageability of traceability information, and ensures the accurate identification and management of product quality traceability information by blockchain traceability nodes.
Smart Images

Figure CN118863674B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of data processing technology, and specifically relates to a blockchain-based smart manufacturing quality data traceability method, system, device and storage medium. Background Art
[0002] In the current field of product quality traceability, with the rapid development of intelligent and Internet of Things technologies, how to efficiently and accurately obtain and process product quality traceability information has become a technical problem that needs to be solved urgently. Traditional traceability methods often rely on paper documents or centralized databases, which have disadvantages such as information islands and easy data tampering, and cannot meet the high standards of modern product quality traceability. At the same time, with the rise of blockchain technology, although it provides a new solution for product quality traceability, how to effectively combine smart IoT sensor data and blockchain technology to achieve comprehensive, real-time and accurate identification of product quality traceability information is still a technical problem. Especially when processing large amounts of multi-dimensional smart IoT sensor data, how to extract key traceability attribute description features and identify traceability items through intelligent algorithms is a major challenge facing current technology. In addition, how to achieve index feature representation of the entire life cycle of product quality on blockchain supervision nodes is also the key to improving traceability efficiency and accuracy. Summary of the invention
[0003] The present application provides a blockchain-based intelligent manufacturing quality data traceability method, system, device and storage medium, which can solve or partially solve the technical problems involved in the above-mentioned background technology.
[0004] The embodiment of the present application provides a blockchain-based intelligent manufacturing quality data traceability method, which is applied to an intelligent manufacturing quality data traceability system. The method includes:
[0005] Acquire multiple smart IoT sensor data, wherein the smart IoT sensor data includes a first semantic embedding vector of a blockchain traceability node and a second semantic embedding vector of product quality traceability information, wherein the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information;
[0006] For each of the smart IoT sensor data, based on the first semantic embedding vector and the second semantic embedding vector included in the smart IoT sensor data, obtaining a traceability attribute description feature corresponding to the smart IoT sensor data;
[0007] Determine the traceability project identification text corresponding to each of the smart IoT sensor data based on the traceability attribute description features corresponding to each of the smart IoT sensor data through a traceability supervision decision algorithm, wherein the traceability project identification text represents the recognition probability of the traceability project text;
[0008] Based on the traceability attribute description features, traceability item identification texts and traceability item prior texts respectively corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain a debugged traceability supervision decision algorithm, wherein the debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotation of the traceability item text;
[0009] Among them, the life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index characteristics in the entire life cycle corresponding to the blockchain supervision node.
[0010] In one implementation, the traceability attribute description features corresponding to the smart IoT sensor data include the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector;
[0011] The traceability supervision decision algorithm is debugged based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each of the smart IoT sensor data to obtain a debugged traceability supervision decision algorithm, including:
[0012] Based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data, a first traceability attribute description feature and a second traceability attribute description feature corresponding to the smart IoT sensor data are generated, wherein the first traceability attribute description feature represents the traceability attribute description feature of the active debugging example corresponding to the blockchain traceability node, and the second traceability attribute description feature represents the traceability attribute description feature of the passive debugging example corresponding to the blockchain traceability node;
[0013] Based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data, a consistency debugging error of the traceability supervision decision algorithm is obtained, wherein the consistency debugging error is used to indicate the performance of the traceability supervision decision algorithm in mastering the content corresponding to the second traceability attribute description feature;
[0014] Based on the traceability project identification text and the traceability project priori text corresponding to each of the smart IoT sensor data, a discrimination debugging error of the traceability supervision decision algorithm is obtained, and the discrimination debugging error is used to indicate the decision offset weight of the traceability supervision decision algorithm for the traceability project text;
[0015] Based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm.
[0016] In one implementation, the first semantic embedding vector includes a first production line semantic embedding vector and a second production line semantic embedding vector, the life cycle state vector in the first production line semantic embedding vector has a matching relationship with the blockchain traceability node, the second production line semantic embedding vector includes the first production line semantic embedding vector, and the second production line semantic embedding vector contains a life cycle state vector that has a non-matching relationship with the blockchain traceability node; the second semantic embedding vector includes a third production line semantic embedding vector and a fourth production line semantic embedding vector, the life cycle state vector in the third production line semantic embedding vector has a matching relationship with the product quality traceability information, and the fourth production line semantic embedding vector is used to represent the access warning record of the product quality traceability information;
[0017] The generating of the first traceability attribute description feature and the second traceability attribute description feature corresponding to the smart IoT sensor data based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data includes:
[0018] For each of the smart IoT sensor data, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the second production line semantic embedding vector corresponding to the smart IoT sensor data; or, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data;
[0019] Perform feature derivation on the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the second production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description features; or perform feature derivation on the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the third production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description features.
[0020] In one implementation, the traceability attribute description feature corresponding to the smart IoT sensor data includes a first production line traceability attribute description feature and a second production line traceability attribute description feature, the first production line traceability attribute description feature and the second production line traceability attribute description feature are both used to represent the smart IoT sensor data, and the first production line traceability attribute description feature and the second production line traceability attribute description feature correspond to the same traceability feature relationship network;
[0021] The step of obtaining the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data includes:
[0022] Obtain the consistency debugging error based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature or the second production line traceability attribute description feature respectively corresponding to each of the smart IoT sensor data;
[0023] Alternatively, for each of the smart IoT sensor data, a first traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a first traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a first traceability attribute combination feature corresponding to the smart IoT sensor data, and a second traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a second traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a second traceability attribute combination feature corresponding to the smart IoT sensor data; based on the first traceability attribute combination feature and the second traceability attribute combination feature respectively corresponding to each of the smart IoT sensor data, the consistency debugging error is obtained;
[0024] Alternatively, a first local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature corresponding to each of the smart IoT sensor data, and a second local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the second production line traceability attribute description feature corresponding to each of the smart IoT sensor data; the first local consistency debugging error and the second local consistency debugging error are weighted to obtain the consistency debugging error.
[0025] In one implementation, obtaining the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data includes:
[0026] For each of the smart IoT sensor data, a first traceability attribute description feature corresponding to the smart IoT sensor data is fused with a traceability attribute description feature of a third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a first traceability attribute fusion feature;
[0027] Crossing the first traceability attribute fusion feature and the time series vector of the soft threshold to obtain a second traceability attribute fusion feature;
[0028] Activate the feature based on the second traceability attribute fusion feature to obtain a third traceability attribute fusion feature;
[0029] Based on the third traceability attribute fusion characteristics corresponding to each of the smart IoT sensor data, the thermal characteristics of the first production link are obtained;
[0030] The second traceability attribute description feature corresponding to the smart IoT sensor data is feature-fused with the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a fourth traceability attribute fusion feature;
[0031] Crossing the fourth traceability attribute fusion feature and the time series vector to obtain a fifth traceability attribute fusion feature;
[0032] Activate the feature based on the fifth traceability attribute fusion feature to obtain a sixth traceability attribute fusion feature;
[0033] Based on the fusion characteristics of the sixth traceability attribute corresponding to each of the smart IoT sensor data, the thermal characteristics of the second production link are obtained;
[0034] Weighting the thermal characteristics of the first production link and the thermal characteristics of the second production link to obtain the thermal characteristics of the third production link;
[0035] Crossing the thermal characteristics of the first production link and the thermal characteristics of the third production link to obtain a seventh traceability attribute fusion feature;
[0036] Based on the seventh traceability attribute fusion feature, the consistency debugging error of the traceability supervision decision algorithm is obtained.
[0037] In one implementation, debugging the traceability supervision decision algorithm based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain the debugged traceability supervision decision algorithm includes:
[0038] Weighting the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain a global debugging error of the traceability supervision decision algorithm;
[0039] In the direction of converging the global debugging error, the algorithm variables of the traceability supervision decision algorithm are cyclically improved to obtain the debugged traceability supervision decision algorithm.
[0040] In one implementation, the index features of the blockchain traceability node and the first semantic embedding vector are recorded in a preset feature space in the form of feature binary tuples, and the index features of the product quality traceability information and the second semantic embedding vector are recorded in the preset feature space in the form of feature binary tuples;
[0041] The obtaining of multiple intelligent IoT sensor data includes:
[0042] Based on the index feature of the blockchain traceability node, obtaining the first semantic embedding vector from the preset feature space;
[0043] Based on the index feature of the product quality traceability information, acquiring the second semantic embedding vector from the preset feature space;
[0044] The smart IoT sensor data is obtained by combining the first semantic embedding vector and the second semantic embedding vector.
[0045] In one implementation, the method further includes:
[0046] For each of the smart IoT sensor data, obtaining a first life cycle state vector in a first semantic embedding vector included in the smart IoT sensor data;
[0047] Combining the first life cycle state vector and the first semantic embedding vector, obtaining a negative debugging example of the blockchain traceability node corresponding to the smart IoT sensor data;
[0048] Based on the negative debugging examples of the blockchain traceability nodes corresponding to the respective smart IoT sensor data, the second traceability attribute description features corresponding to the respective smart IoT sensor data are obtained.
[0049] In one implementation, the method further includes:
[0050] Using the debugged traceability supervision decision algorithm, obtaining the traceability item identification text of the blockchain traceability node for the product quality traceability information;
[0051] Furthermore, based on the blockchain traceability node, the traceability item identification text of the product quality traceability information is output to output quality traceability opinion comments.
[0052] An embodiment of the present application provides an intelligent manufacturing quality data traceability system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.
[0053] The present application embodiment provides a smart manufacturing quality data traceability device, which is used to:
[0054] Acquire multiple smart IoT sensor data, wherein the smart IoT sensor data includes a first semantic embedding vector of a blockchain traceability node and a second semantic embedding vector of product quality traceability information, wherein the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information;
[0055] For each of the smart IoT sensor data, based on the first semantic embedding vector and the second semantic embedding vector included in the smart IoT sensor data, obtaining a traceability attribute description feature corresponding to the smart IoT sensor data;
[0056] Determine the traceability project identification text corresponding to each of the smart IoT sensor data based on the traceability attribute description features corresponding to each of the smart IoT sensor data through a traceability supervision decision algorithm, wherein the traceability project identification text represents the recognition probability of the traceability project text;
[0057] Based on the traceability attribute description features, traceability item identification texts and traceability item prior texts respectively corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain a debugged traceability supervision decision algorithm, wherein the debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotation of the traceability item text;
[0058] Among them, the life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index characteristics in the entire life cycle corresponding to the blockchain supervision node.
[0059] An embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.
[0060] In an embodiment of the present application, by acquiring multiple smart IoT sensor data and deeply analyzing the first semantic embedding vector and the second semantic embedding vector therein, efficient processing and accurate identification of product quality traceability information are achieved. Specifically, the embodiment of the present application can extract comprehensive traceability attribute description features based on smart IoT sensor data, and then accurately determine the traceability project identification text corresponding to each smart IoT sensor data through the traceability supervision decision algorithm. This process not only improves the accuracy and efficiency of product quality traceability, but also makes the traceability supervision decision algorithm more adaptable to actual application scenarios through algorithm debugging, and provides strong support for the blockchain traceability node to identify product quality traceability information. In addition, the embodiment of the present application uses the life cycle state vector to achieve accurate representation of index features in the entire life cycle corresponding to the blockchain supervision node, further enhancing the traceability and manageability of the traceability information. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 A flowchart of a blockchain-based smart manufacturing quality data traceability method provided in an embodiment of the present application.
[0062] Figure 2 A schematic diagram of the structure of an intelligent manufacturing quality data traceability system provided in an embodiment of the present application. DETAILED DESCRIPTION
[0063] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0064] The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are usually a class, and the number of objects is not limited. For example, the first object can be one or more. In addition, in this application, "and / or" represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.
[0065] Figure 1 A blockchain-based smart manufacturing quality data traceability method is shown, which is applied to a smart manufacturing quality data traceability system. The method includes the following steps 110-140.
[0066] In the field of modern manufacturing and supply chain management, product quality traceability is a vital task. In order to ensure that every link of the product from raw material procurement to final delivery to consumers can be accurately tracked, this application embodiment combines smart IoT sensor technology and blockchain technology to introduce a technical solution based on a smart manufacturing quality data traceability system. This technical solution (steps 110 to 140 below) uses smart IoT sensor data and blockchain technology to achieve effective traceability of product quality.
[0067] Step 110: The intelligent manufacturing quality data traceability system obtains multiple intelligent IoT sensor data, wherein the intelligent IoT sensor data includes a first semantic embedding vector of a blockchain traceability node and a second semantic embedding vector of product quality traceability information, wherein the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information.
[0068] In the embodiment of the present application, the intelligent manufacturing quality data traceability system first obtains data from multiple intelligent IoT sensors (intelligent IoT sensor data). The intelligent IoT sensor data not only includes traditional sensor readings, but also covers the first semantic embedding vector of the blockchain traceability node and the second semantic embedding vector of the product quality traceability information.
[0069] The first semantic embedding vector contains at least one life cycle state vector associated with the blockchain traceability node. The life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index feature of the node in the entire life cycle. For example, the life cycle of a product may include stages such as raw material procurement, production and processing, quality inspection, packaging and logistics. Each stage will have a corresponding state vector to represent its position and status in the entire life cycle.
[0070] The second semantic embedding vector contains at least one life cycle state vector associated with product quality traceability information. Similar to the first semantic embedding vector, the second semantic embedding vector also provides product status information at each stage, but it focuses more on product quality data, such as quality test results of raw materials, quality control parameters in the production process, etc.
[0071] Step 120: The intelligent manufacturing quality data traceability system obtains traceability attribute description features corresponding to each of the intelligent IoT sensor data based on the first semantic embedding vector and the second semantic embedding vector included in the intelligent IoT sensor data.
[0072] After acquiring the smart IoT sensor data, the smart manufacturing quality data traceability system will extract the corresponding traceability attribute description features for each smart IoT sensor data based on the first semantic embedding vector and the second semantic embedding vector contained in it. The traceability attribute description features are highly abstract and generalized product quality traceability information, which can reflect the key attributes and status changes of the product throughout its life cycle.
[0073] For example, for a food product, its traceability attribute description features may include the origin of raw materials, production date, production batch, quality inspection report number, logistics track, etc. These features provide basic data for subsequent traceability project identification.
[0074] Step 130, the intelligent manufacturing quality data traceability system determines the traceability item identification text corresponding to each of the intelligent IoT sensor data based on the traceability attribute description characteristics corresponding to each of the intelligent IoT sensor data through the traceability supervision decision algorithm, and the traceability item identification text represents the recognition probability of the traceability item text.
[0075] The intelligent manufacturing quality data traceability system will then use the traceability supervision decision algorithm to determine the traceability item identification text corresponding to each data point based on the traceability attribute description characteristics corresponding to each intelligent IoT sensor data. The traceability item identification text represents the recognition probability of the traceability item text, that is, it reflects the possibility of a certain traceability item (such as "source of raw materials") under a given traceability attribute description feature.
[0076] For example, if the traceability attribute description feature of a product includes the information that "the raw materials come from a certain farm", then the traceability supervision decision algorithm may give a higher recognition probability, indicating that the traceability item "origin of raw materials" is highly relevant to the product.
[0077] To facilitate understanding of step 130, some examples are provided below for illustration.
[0078] On the one hand, the specific implementation of the traceability supervision decision algorithm is introduced as follows.
[0079] 1. Background of the selection of traceability supervision decision-making algorithm
[0080] In the product quality traceability system, traceability item identification is essentially a text classification problem. Given the traceability attribute description features of a product, it is necessary to determine which traceability items these features are highly correlated with. Deep learning performs well in text classification tasks, especially convolutional neural networks (CNNs), recurrent neural networks (RNNs) and their variants (such as LSTM, GRU), and the recent Transformer model.
[0081] 2. Algorithm Framework
[0082] The embodiment of this application selects a text classification model based on LSTM (Long Short-Term Memory Network) as the core of the traceability supervision decision algorithm. LSTM can process sequence data and capture long-distance dependencies in text, which is crucial for understanding the complex relationships in traceability attribute description features.
[0083] (1) Data preprocessing
[0084] Text vectorization: First, the traceability attribute description features and traceability project text are converted into numerical vectors. This is usually achieved through word embedding (such as Word2Vec, GloVe) or character embedding (such as FastText).
[0085] Sequence padding: Since LSTM requires the length of the input sequence to be fixed, the text needs to be padded or truncated to ensure that all input sequences have the same length.
[0086] (2) Model structure
[0087] Input layer: accepts a sequence of preprocessed text vectors.
[0088] LSTM layer: One or more LSTM layers to capture temporal dependencies in text.
[0089] Fully connected layer: maps the output of the LSTM layer to the category space of the traceability items.
[0090] Output layer: Use the softmax function to generate the recognition probability of each traceability item.
[0091] (3) Model training
[0092] Loss function: The cross entropy loss function is used to measure the difference between the probability distribution predicted by the model and the true distribution.
[0093] Optimizer: Select an appropriate optimizer (such as Adam, RMSprop) to update the model parameters and minimize the loss function.
[0094] Training process: Use labeled training data (i.e., traceability attribute description features and their corresponding traceability items) for iterative training until the model converges.
[0095] 3. Algorithm Implementation Details
[0096] Feature selection: Extract features useful for identifying traceability items from smart IoT sensor data, such as keywords, phrases, etc. These features can be obtained through natural language processing (NLP) technologies such as word segmentation, part-of-speech tagging, and named entity recognition.
[0097] Model tuning: Optimize the performance of the model by adjusting hyperparameters such as the number of LSTM layers, number of hidden units, learning rate, batch size, etc.
[0098] Regularization: Use techniques such as dropout and weight decay to prevent the model from overfitting.
[0099] On the other hand, the specific implementation for a higher recognition probability is as follows.
[0100] In the LSTM-based text classification model, a higher recognition probability is output through the softmax function. The softmax function maps the output of the LSTM layer to a probability distribution, with each traceability item corresponding to a probability value. These probability values represent the likelihood of each traceability item appearing under given traceability attribute description features.
[0101] 1. Probability Threshold Setting
[0102] In order to determine whether a traceability item has a "higher identification probability", a probability threshold needs to be set. This threshold can be adjusted according to actual conditions and is usually determined in the following ways:
[0103] Experience setting: Based on historical data and expert experience, a reasonable threshold is set. For example, if the recognition probability of a traceability item exceeds 0.7, the item is considered to be highly relevant to the product.
[0104] Cross-validation: Use cross-validation techniques to test different thresholds on different training sets and validation sets, and select the threshold with the best performance as the final setting.
[0105] 2. Sorting and filtering
[0106] After obtaining the recognition probability of each traceability item, these probabilities can be sorted and the traceability items with the highest probability can be selected as the final recognition results. This method is particularly suitable for situations where it is necessary to find the traceability items that are most relevant to the product.
[0107] On the other hand, the specific implementation of the "highly relevant" standard is as follows.
[0108] "Highly relevant" is a relatively vague concept that needs to be defined based on specific application scenarios and requirements. In the product quality traceability system, "highly relevant" can be defined from the following aspects:
[0109] 1. Probability value size
[0110] As mentioned above, by setting a probability threshold, traceability items with an identification probability exceeding the threshold can be considered "highly relevant" to the product. This method is intuitive and easy to implement, but it should be noted that the threshold should be set with caution. If it is too high, it may lead to missed identification, while if it is too low, it may lead to misidentification.
[0111] 2. Business Logic Verification
[0112] In addition to the judgment based on probability value, the relevance of traceability items and products can also be verified in combination with business logic. For example, for food products, "origin of raw materials" and "production date" are usually crucial traceability items. If the specific information of these two items is clearly mentioned in the traceability attribute description characteristics of a product, and the corresponding recognition probability is high, then it can be more confident that these items are "highly relevant" to the product.
[0113] 3. Expert Evaluation and Feedback
[0114] In practical applications, domain experts can be invited to evaluate the identification results of traceability items. Experts can judge which traceability items are truly "highly relevant" to the product based on their own experience and expertise. By collecting feedback from experts, the traceability supervision decision-making algorithm can be continuously optimized to improve the accuracy and reliability of identification.
[0115] IV. Historical Data and Comparative Analysis
[0116] The use of historical data and comparative analysis is also an effective way to define "high correlation". A large amount of product data of known traceability items can be collected as training sets and validation sets, and the performance of the comparative analysis model on these data can be used to evaluate which traceability items are more likely to be "highly correlated" with the product. For example, if a traceability item is found to have a high recognition probability in multiple products, and these products are indeed closely related to the traceability item in actual applications, then the item can be considered an important traceability item.
[0117] Through the above analysis, the text classification algorithm based on deep learning has broad application prospects in the product quality traceability system. By reasonably selecting the algorithm framework, carefully designing the model structure, scientifically setting the probability threshold, and combining business logic, expert evaluation and historical data, etc., efficient and accurate identification of traceability items can be achieved, providing strong support for product quality traceability.
[0118] Step 140, the intelligent manufacturing quality data traceability system debugs the traceability supervision decision algorithm based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each of the intelligent IoT sensor data, and obtains a debugged traceability supervision decision algorithm. The debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotation of the traceability item text.
[0119] Among them, the life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index characteristics in the entire life cycle corresponding to the blockchain supervision node.
[0120] In order to further improve the accuracy and reliability of the traceability supervision decision algorithm, the intelligent manufacturing quality data traceability system will debug the algorithm based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each intelligent IoT sensor data. The debugged algorithm can better obtain the traceability item identification text of the blockchain traceability node for product quality traceability information.
[0121] The traceability project prior text represents the prior learning annotations of the traceability project text. It is built based on a large amount of historical data and expert knowledge, and is used to guide the learning and optimization process of the traceability supervision decision algorithm. For example, for food products, "production date" and "quality inspection report number" are usually considered to be very important traceability items, so they are given higher weights in the prior learning annotations.
[0122] It is worth mentioning that in order to explain the weight assignment in more detail, the following content is a further explanation of the relevant technical details of the weight assignment.
[0123] 1. Background and Purpose of Weighting
[0124] In the traceability supervision decision-making algorithm, weight is a quantitative indicator used to measure the importance of different traceability items. By giving higher weights to important traceability items, the algorithm can pay more attention to these items during the learning and optimization process, thereby improving the accuracy and efficiency of traceability.
[0125] 2. Specific steps for weight assignment
[0126] (1) Data collection and preprocessing
[0127] Historical data collection: Collect a large amount of historical traceability data about food products, including production date, quality inspection report number, raw material source, manufacturer information, etc.
[0128] Data cleaning: Remove noise and outliers from data to ensure data accuracy and reliability.
[0129] Feature extraction: Extract features related to the traceability project from the cleaned data, such as date format, report number format, etc.
[0130] (2) Expert knowledge integration
[0131] Expert consultation: Invite experts in the fields of food safety, quality management, etc. to consult and understand their views on the importance of traceability projects.
[0132] Knowledge base construction: Based on the opinions and suggestions of experts, an expert knowledge base containing information such as the importance ranking and key features of traceability projects is constructed.
[0133] (3) Weight calculation
[0134] Weight calculation based on historical data: Use statistical methods (such as frequency analysis, correlation analysis, etc.) to calculate the frequency of each traceability item in historical data and its correlation with product quality issues, so as to obtain a preliminary weight distribution.
[0135] Weight adjustment based on expert knowledge: Adjust the initially calculated weights based on the information in the expert knowledge base to ensure that important traceability items receive higher weights.
[0136] (4) Weight verification and optimization
[0137] Experimental verification: Experimental verification is performed using a portion of traceability data with known results to evaluate the impact of weight distribution on traceability accuracy.
[0138] Weight optimization: Based on the experimental results, the weights are fine-tuned to further improve the accuracy of traceability.
[0139] 3. Specific Examples of Weight Assignment
[0140] Taking food products as an example, there are the following traceability items: production date, quality inspection report number, origin of raw materials, and manufacturer information.
[0141] (1) Weight calculation based on historical data
[0142] Analyzing the historical data, it is found that “production date” and “quality inspection report number” often appear as key information in product quality issues, so their initial weights are higher.
[0143] Although “raw material origin” and “manufacturer information” are also important, they have less direct correlation with product quality issues in historical data, so their initial weights are relatively low.
[0144] (2) Weight adjustment based on expert knowledge
[0145] Experts pointed out that the "production date" is an important basis for judging whether food is expired and should be given the highest weight.
[0146] The "quality inspection report number" can be directly linked to the specific quality inspection record and is also very important for tracing product quality issues. Its weight should be second only to the "production date".
[0147] "Origin of raw materials" and "manufacturer information" may become key information in certain circumstances (such as raw material contamination, manufacturer's illegal production, etc.), so their weight should be appropriately increased.
[0148] (3) Final weight distribution
[0149] Based on the above analysis, the following weights are finally assigned to each traceability item: production date (0.4), quality inspection report number (0.3), origin of raw materials (0.2), and manufacturer information (0.1).
[0150] Through the above steps and examples, we can see that the process of assigning weights to traceability projects is not simply set arbitrarily, but is based on a comprehensive analysis of a large amount of historical data and expert knowledge. Through scientific methods and rigorous processes, we can reasonably assign weights to traceability projects, thereby improving the accuracy and efficiency of the traceability supervision decision-making algorithm.
[0151] Combined with the above content, through the debugging process, the intelligent manufacturing quality data traceability system can continuously adjust and optimize the parameters and model structure of the traceability supervision decision algorithm to make it more suitable for the actual product quality traceability task. Ultimately, the debugged algorithm will be able to more accurately identify key items related to product quality traceability and provide detailed and reliable traceability information to regulators and consumers.
[0152] To illustrate the application of the above technical solutions more specifically, the following is an example scenario of food supply chain supervision.
[0153] For example, there is a large food production company that produces a variety of food products and sells them all over the country. In order to ensure the quality and safety of the products, the company decided to adopt a product quality traceability system based on blockchain and smart IoT sensor technology.
[0154] (1) Data acquisition: The smart manufacturing quality data traceability system was first integrated with the company's production system, logistics system, and quality inspection system. Through integration, the platform can obtain real-time data from smart sensors on the production line, GPS trackers on logistics vehicles, and instruments in the quality inspection laboratory. These data are encapsulated into smart IoT sensor data and include the first semantic embedding vector of the blockchain traceability node and the second semantic embedding vector of the product quality traceability information.
[0155] (2) Feature extraction: For each batch of food products, the intelligent manufacturing quality data traceability system will extract the corresponding traceability attribute description features. These features include the product's raw material origin, production date, production batch, quality inspection report number, logistics track, storage conditions, etc. By extracting these features, the platform can provide a rich data foundation for subsequent traceability project identification.
[0156] (3) Item identification: Using the traceability supervision decision algorithm, the intelligent manufacturing quality data traceability system identifies the traceability items for each batch of products. The algorithm gives the identification probability of each traceability item based on the extracted traceability attribute description features. For example, for a batch of products, the algorithm may consider "raw material origin" and "production date" to be the two most important traceability items and give a higher identification probability.
[0157] (4) Algorithm debugging and optimization: In order to further improve the accuracy of the traceability supervision decision-making algorithm, the intelligent manufacturing quality data traceability system also uses historical data and expert knowledge to construct a priori text of traceability items. Through this text, the platform debugs and optimizes the algorithm. The debugged algorithm is more accurate and reliable in identifying traceability items, and can effectively reduce false positives and false negatives.
[0158] (5) Traceability information query and display: Finally, the intelligent manufacturing quality data traceability system applies the debugged traceability supervision decision algorithm to the actual product quality traceability task. When consumers or regulators need to query the traceability information of a product, the platform can quickly provide a detailed traceability report. The report includes key information such as the origin of the product's raw materials, production date, quality inspection report number, etc., and this information is encrypted and verified by blockchain technology to ensure its authenticity and credibility.
[0159] From the above application scenario examples, it can be seen that the technical solution based on the intelligent manufacturing quality data traceability system can effectively achieve the goal of product quality traceability. By utilizing intelligent IoT sensor data and blockchain technology, the solution can provide detailed and reliable traceability information to help companies and regulators better manage and control product quality and safety. At the same time, through continuous debugging and optimization of traceability supervision decision algorithms, the solution can also continuously improve the accuracy and efficiency of traceability, bringing greater value to the modern manufacturing industry and supply chain management fields.
[0160] Next, detailed explanations of terms and further detailed introduction of technical solutions are given for steps 110 to 140 respectively.
[0161] For step 110, it describes in detail how the intelligent manufacturing quality data traceability system acquires and processes multiple intelligent IoT sensor data to achieve efficient traceability of product quality.
[0162] First, the smart manufacturing quality data traceability system obtains sensor data from the smart Internet of Things. These data are collected by combining smart devices with Internet of Things technology. They not only contain traditional sensor readings, but also incorporate richer information. Smart Internet of Things sensor data is the basis of this step, which provides comprehensive and real-time data support for product quality traceability.
[0163] There are two key concepts in these smart IoT sensor data: blockchain traceability nodes and product quality traceability information. Blockchain traceability nodes are specific nodes established for product quality traceability using blockchain technology. Each node contains key information related to product quality, and this information cannot be tampered with on the blockchain, ensuring the authenticity and credibility of the data. Product quality traceability information covers information on all aspects of the product's life cycle from raw material procurement to final delivery to consumers, including production, processing, quality inspection, logistics, etc.
[0164] Furthermore, these smart IoT sensor data also include two important vectors: the first semantic embedding vector and the second semantic embedding vector. The first semantic embedding vector is closely associated with the blockchain traceability node, and it contains at least one life cycle state vector. This vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index characteristics of the entire life cycle corresponding to the node. In other words, it reveals the various stages and states of the product in the entire life cycle, providing detailed timeline and node information for traceability.
[0165] Similarly, the second semantic embedding vector is also closely related to product quality traceability information, and it also contains at least one life cycle state vector. This vector focuses more on product quality data, such as quality test results of raw materials, quality control parameters in the production process, etc. It provides more specific and detailed information for product quality traceability, making the traceability results more accurate and reliable.
[0166] It can be seen that by acquiring and processing multiple smart IoT sensor data, comprehensive support for product quality traceability is achieved. These data not only contain key information about blockchain traceability nodes and product quality traceability information, but also provide detailed and accurate full life cycle status information through the first semantic embedding vector and the second semantic embedding vector. This provides a solid foundation for the subsequent extraction of traceability attribute description features, identification of traceability items, and debugging and optimization of traceability supervision decision algorithms.
[0167] For step 120, it describes in detail how the intelligent manufacturing quality data traceability system obtains the corresponding traceability attribute description features based on the intelligent Internet of Things sensor data, especially the first semantic embedding vector and the second semantic embedding vector therein.
[0168] First, the smart manufacturing quality data traceability system will conduct in-depth analysis and processing of each smart IoT sensor data. These data not only contain rich sensor readings, but also integrate the full life cycle status of blockchain traceability nodes and product quality traceability information through the first semantic embedding vector and the second semantic embedding vector. This provides a solid foundation for extracting traceability attribute description features.
[0169] The traceability attribute description features are highly abstract and generalized product quality traceability information, which can reflect the key attributes and status changes of products throughout their life cycle. These features are of great significance for subsequent traceability project identification, traceability supervision decision-making and product quality control.
[0170] In the process of obtaining traceability attribute description features, the intelligent manufacturing quality data traceability system will make full use of the information contained in the first semantic embedding vector and the second semantic embedding vector. First, the platform will conduct an in-depth analysis of the two vectors and extract the full life cycle status information. This information includes the status, changes, and key events of the product at each stage, which is the basis for building traceability attribute description features.
[0171] Secondly, the intelligent manufacturing quality data traceability system will further abstract and summarize the traceability attributes of the product based on the extracted full life cycle status information. These attributes may include the source of raw materials, production date, production batch, quality inspection report, etc., which are all key elements of product quality traceability.
[0172] Finally, the intelligent manufacturing quality data traceability system will encode and describe these traceability attributes to form traceability attribute description features. These features represent the product's full life cycle status and key attributes in a structured form, providing rich data support for subsequent traceability project identification and traceability supervision decisions.
[0173] It can be seen that by deeply analyzing the intelligent IoT sensor data, especially the first semantic embedding vector and the second semantic embedding vector, the traceability attribute description features of the product are successfully extracted. These features provide a solid foundation for subsequent traceability work, enabling the intelligent manufacturing quality data traceability system to more accurately and comprehensively identify and trace product quality information.
[0174] Furthermore, steps 130 to 140 constitute the core invention of the present application, which describe in detail how the intelligent manufacturing quality data traceability system utilizes the traceability supervision decision algorithm to determine the traceability item identification text based on the traceability attribute description features corresponding to the intelligent IoT sensor data, and further debugs the algorithm to obtain more accurate traceability item identification results.
[0175] First, in step 130, the intelligent manufacturing quality data traceability system uses the traceability supervision decision algorithm to conduct an in-depth analysis of the traceability attribute description features corresponding to each intelligent IoT sensor data. In this process, the algorithm will comprehensively consider the product's full life cycle status, key attributes, and their changing trends, thereby generating a traceability item identification text that represents the probability of traceability item text recognition. This text is a high-level summary of product quality traceability information, which can help supervisors quickly understand the product's traceability, including key information such as the product's source, production process, and quality control.
[0176] Next, in step 140, the intelligent manufacturing quality data traceability system will debug the traceability supervision decision algorithm based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each intelligent IoT sensor data. The traceability item prior text is an important material that characterizes the prior learning annotations of the traceability item text. It contains a large amount of historical data and empirical knowledge about product quality traceability. By combining this prior knowledge with the traceability attribute description features and traceability item identification text acquired in real time, the intelligent manufacturing quality data traceability system can debug the traceability supervision decision algorithm more accurately and comprehensively.
[0177] In this debugging process, the life cycle state vector plays a key role. It is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index characteristics of the entire life cycle corresponding to the blockchain supervision node. In other words, the life cycle state vector reveals the various stages and states of the product in the entire life cycle, as well as the index information of these stages and states on the blockchain supervision node. This provides detailed and accurate timeline and node information for the traceability supervision decision algorithm, allowing the algorithm to more accurately identify product quality traceability items.
[0178] After debugging, the traceability supervision decision algorithm will have stronger traceability project identification capabilities. It can more accurately obtain the traceability project identification text of the blockchain traceability node for product quality traceability information, and provide more comprehensive and accurate traceability information support for regulators. This will help improve the efficiency and accuracy of product quality traceability and protect the rights and safety of consumers.
[0179] It is worth mentioning that the blockchain traceability node is a specific node established for product quality traceability using blockchain technology. These nodes are distributed on the blockchain network, and each node contains key information related to product quality. Its main function is to record and store product quality traceability information to ensure that this information cannot be tampered with on the blockchain, thereby ensuring the authenticity and credibility of the data. Through the blockchain traceability node, the entire life cycle of the product can be traced, including raw material procurement, production, processing, quality inspection, logistics and other links. Each blockchain traceability node contains specific information related to the product, which is organized in chronological order and logical relationships, forming a complete product quality traceability chain.
[0180] Blockchain supervision nodes are specific nodes in the blockchain network used to supervise and manage product quality traceability information. These nodes are usually set up by regulatory agencies or authoritative agencies to supervise and manage product quality. Its main function is to receive, store and process product quality traceability information from smart IoT sensor data. Through blockchain supervision nodes, regulators can monitor the status of the entire life cycle of products in real time, including the status, changes and key events of each stage. At the same time, blockchain supervision nodes can also debug and optimize the traceability supervision decision algorithm to improve the accuracy and efficiency of traceability project identification. Blockchain supervision nodes have a special status and authority in the blockchain network, and can comprehensively supervise and manage product quality traceability information. At the same time, due to the immutability of blockchain technology, the information on blockchain supervision nodes is also authentic and reliable, providing strong data support for regulators.
[0181] Therefore, blockchain traceability nodes and blockchain supervision nodes play different roles in the blockchain network. Blockchain traceability nodes are mainly used to record and store product quality traceability information, while blockchain supervision nodes are used to supervise and manage this information to ensure product quality and safety. Together, the two constitute the application system of blockchain technology in the field of product quality traceability.
[0182] The embodiment of the present application achieves efficient processing and accurate identification of product quality traceability information by acquiring multiple smart IoT sensor data and deeply analyzing the first semantic embedding vector and the second semantic embedding vector therein. Specifically, the embodiment of the present application can extract comprehensive traceability attribute description features based on smart IoT sensor data, and then accurately determine the traceability project identification text corresponding to each smart IoT sensor data through the traceability supervision decision algorithm. This process not only improves the accuracy and efficiency of product quality traceability, but also makes the traceability supervision decision algorithm more adaptable to actual application scenarios through algorithm debugging, and provides strong support for the blockchain traceability node to identify product quality traceability information. In addition, the embodiment of the present application uses the life cycle state vector to achieve accurate representation of index features in the entire life cycle corresponding to the blockchain supervision node, further enhancing the traceability and manageability of the traceability information.
[0183] In some examples, the traceability attribute description features corresponding to the smart IoT sensor data include the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector. Then, based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm, including: based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector corresponding to each of the smart IoT sensor data, the first traceability attribute description feature represents the traceability attribute description features of the active debugging example corresponding to the blockchain traceability node, and the second traceability attribute description feature represents the traceability attribute description features of the passive debugging example corresponding to the blockchain traceability node; based on The consistency debugging error of the traceability supervision decision algorithm is obtained based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data, and the consistency debugging error is used to indicate the performance of the traceability supervision decision algorithm in mastering and mining the content corresponding to the second traceability attribute description feature; based on the traceability item identification text and the traceability item prior text respectively corresponding to each of the smart IoT sensor data, the discrimination debugging error of the traceability supervision decision algorithm is obtained, and the discrimination debugging error is used to indicate the decision offset weight of the traceability supervision decision algorithm for the traceability item text; based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm.
[0184] When discussing the debugging process of the traceability supervision decision algorithm based on smart IoT sensor data, it is first necessary to understand the core purpose of the algorithm: to achieve accurate supervision of product or event traceability by analyzing and processing data from smart IoT sensors. In order to achieve this goal, the algorithm needs to be able to accurately identify and parse the traceability attributes in the sensor data, and then make reasonable supervision decisions based on these attributes. The technical solution of this embodiment aims to optimize the performance of the traceability supervision decision algorithm through a series of debugging steps, so that it can process smart IoT sensor data more accurately and efficiently.
[0185] First, considering the multidimensionality and complexity of smart IoT sensor data, the smart manufacturing quality data traceability system decomposes the traceability attribute description features corresponding to these data into two parts: the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector. This decomposition process is based on the semantic embedding technology in deep learning, which can convert high-dimensional and sparse raw data into low-dimensional and dense vector representations, so as to facilitate efficient processing of the algorithm. Among them, the first semantic embedding vector mainly captures the attribute features related to positive debugging examples (i.e. successful traceability cases), while the second semantic embedding vector focuses on the attribute features of negative debugging examples (i.e. cases with traceability failures or problems).
[0186] Then, using these two types of semantic embedding vectors, the intelligent manufacturing quality data traceability system generates the first traceability attribute description feature and the second traceability attribute description feature corresponding to the intelligent IoT sensor data. These two feature sets represent the different traceability attribute patterns exhibited by the blockchain traceability node when processing positive and negative debugging examples. Through such feature extraction, the algorithm can capture the subtle differences in the data more carefully, thereby providing a rich information basis for the subsequent debugging process.
[0187] Subsequently, based on the extracted first traceability attribute description features and the second traceability attribute description features, the intelligent manufacturing quality data traceability system calculates the consistency debugging error of the traceability supervision decision algorithm. This error index measures the consistency of the output results of the algorithm when processing data with different traceability attribute description features. Specifically, it reflects the performance level of the algorithm in identifying the second traceability attribute description features (i.e., the features of the negative debugging example). The smaller the consistency debugging error, the more stable the algorithm can maintain performance when processing various traceability attributes, which is crucial to improving the accuracy and reliability of traceability supervision.
[0188] At the same time, the intelligent manufacturing quality data traceability system also uses the traceability project identification text and traceability project prior text corresponding to the intelligent IoT sensor data to calculate the discrimination and debugging error of the traceability supervision decision algorithm. This error indicator focuses on the algorithm's ability to distinguish different traceability project texts, that is, the algorithm's decision offset weight for the traceability project text. In practical applications, different traceability projects may involve different regulatory requirements and standards, so the algorithm needs to be able to accurately distinguish these projects and make corresponding regulatory decisions. The smaller the discrimination and debugging error, the stronger the decision-making ability of the algorithm when facing different traceability projects, which helps to improve the flexibility and adaptability of traceability supervision.
[0189] Finally, considering the consistency debugging error and the discrimination debugging error, the intelligent manufacturing quality data traceability system debugs the traceability supervision decision algorithm in order to optimize the performance. The debugging process may involve multiple aspects such as the adjustment of algorithm parameters, the improvement of model structure, or the change of training strategy. Through continuous iteration and optimization, the intelligent manufacturing quality data traceability system finally obtains the debugged traceability supervision decision algorithm. While maintaining high consistency, the algorithm also has stronger discrimination ability, which can more effectively deal with complex and changeable intelligent IoT sensor data and realize accurate traceability supervision.
[0190] In this way, by deeply analyzing the traceability attribute description characteristics of smart IoT sensor data and combining the information of traceability project identification text and prior text, the traceability supervision decision algorithm was fully debugged and optimized. This process not only improves the accuracy and stability of the algorithm when processing various traceability data, but also enhances its decision-making ability when facing different traceability projects, providing strong technical support for building a more efficient and reliable traceability supervision system.
[0191] In some other possible embodiments, the first semantic embedding vector includes a first production line semantic embedding vector and a second production line semantic embedding vector, the life cycle state vector in the first production line semantic embedding vector has a matching relationship with the blockchain traceability node, the second production line semantic embedding vector includes the first production line semantic embedding vector, and the second production line semantic embedding vector contains a life cycle state vector that has a non-matching relationship with the blockchain traceability node; the second semantic embedding vector includes a third production line semantic embedding vector and a fourth production line semantic embedding vector, the life cycle state vector in the third production line semantic embedding vector has a matching relationship with the product quality traceability information, and the fourth production line semantic embedding vector is used to represent the access warning record of the product quality traceability information. Then, based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data, the first traceability attribute description features and the second traceability attribute description features corresponding to the smart IoT sensor data are generated, including: for each of the smart IoT sensor data, based on the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the second production line semantic embedding vector corresponding to the smart IoT sensor data, the first traceability attribute description features are generated; or based on the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the second production line semantic embedding vector corresponding to the smart IoT sensor data, the first traceability attribute description features are generated. The first traceability attribute description feature is generated by combining the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the third production line semantic embedding vector; the second traceability attribute description feature is generated by deriving the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the second production line semantic embedding vector corresponding to the smart IoT sensor data; or the second traceability attribute description feature is generated by deriving the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the third production line semantic embedding vector corresponding to the smart IoT sensor data.
[0192] It can be understood that when it comes to the generation of traceability attribute description features of smart IoT sensor data, the smart manufacturing quality data traceability system faces a complex and multi-dimensional task. Especially considering the different production line semantic embedding vectors contained in the first semantic embedding vector and the second semantic embedding vector, how to effectively extract and generate meaningful traceability attribute description features from these vectors has become the key to technical implementation.
[0193] Based on this, first of all, the intelligent manufacturing quality data traceability system needs to understand the composition of the first semantic embedding vector and the second semantic embedding vector and their connection with the blockchain traceability node and product quality traceability information. The first semantic embedding vector includes the first production line semantic embedding vector and the second production line semantic embedding vector. Among them, there is a matching connection between the life cycle state vector in the first production line semantic embedding vector and the blockchain traceability node, which means that these vectors can directly reflect the traceability status of the product on the blockchain. The second production line semantic embedding vector not only contains all the information of the first production line semantic embedding vector, but also additionally contains life cycle state vectors with non-matching connections with the blockchain traceability node. These vectors may represent some special situations or anomalies in the product traceability process.
[0194] The second semantic embedding vector includes the third production line semantic embedding vector and the fourth production line semantic embedding vector. There is a matching relationship between the life cycle state vector in the third production line semantic embedding vector and the product quality traceability information, indicating that these vectors can reveal the quality status of the product. The fourth production line semantic embedding vector is used to represent the access warning record of the product quality traceability information, which may contain access patterns or warning signals about product quality issues.
[0195] With this understanding, the intelligent manufacturing quality data traceability system can begin to generate the first traceability attribute description features and the second traceability attribute description features corresponding to the intelligent IoT sensor data.
[0196] There are two main strategies for the generation of the first traceability attribute description feature in the intelligent manufacturing quality data traceability system. The first strategy is to generate the traceability attribute description feature based on the first production line semantic embedding vector corresponding to the intelligent IoT sensor data and the traceability attribute description feature of the second production line semantic embedding vector. Since the first production line semantic embedding vector has a matching relationship with the blockchain traceability node, and the second production line semantic embedding vector contains all the information of the first production line semantic embedding vector and adds additional information, this strategy can fully reflect the traceability status of the product on the blockchain, including normal status and special situations or exceptions.
[0197] The second strategy is based on the traceability attribute description features of the first production line semantic embedding vector corresponding to the smart IoT sensor data and the traceability attribute description features of the third production line semantic embedding vector. Since the third production line semantic embedding vector has a matching relationship with the product quality traceability information, this strategy can combine the product's traceability status on the blockchain with the product quality status to generate a comprehensive description feature that contains both traceability information and quality information.
[0198] Next, for the generation of the second traceability attribute description feature, the intelligent manufacturing quality data traceability system also has two strategies. The first strategy is to derive features from the traceability attribute description features of the first production line semantic embedding vector and the traceability attribute description features of the second production line semantic embedding vector corresponding to the intelligent IoT sensor data. Feature derivation is a method of creating new features through existing features, which can help the intelligent manufacturing quality data traceability system to dig out more information hidden in the data. By deriving features from the traceability attribute description features of the first production line semantic embedding vector and the second production line semantic embedding vector, the intelligent manufacturing quality data traceability system can generate a richer and more detailed second traceability attribute description feature.
[0199] The second strategy is to derive the traceability attribute description features of the first production line semantic embedding vector and the third production line semantic embedding vector corresponding to the smart IoT sensor data. This strategy can also use the feature derivation method to combine the product's traceability status on the blockchain with the product quality status to generate a comprehensive derived description feature that contains both traceability information and quality information.
[0200] With this design, the first traceability attribute description feature and the second traceability attribute description feature corresponding to the smart IoT sensor data are generated through two strategies. These features not only fully reflect the traceability status of the product on the blockchain, but also combine the information on the product quality status, providing a rich and accurate data basis for subsequent traceability supervision decisions. Through this feature generation method, the smart manufacturing quality data traceability system can better explore and utilize the value of smart IoT sensor data and improve the accuracy and efficiency of traceability supervision.
[0201] In some other examples, the traceability attribute description features corresponding to the smart IoT sensor data include a first production line traceability attribute description feature and a second production line traceability attribute description feature, the first production line traceability attribute description feature and the second production line traceability attribute description feature are both used to represent the smart IoT sensor data, and the first production line traceability attribute description feature and the second production line traceability attribute description feature correspond to the same traceability feature relationship network. Then, the consistency debugging error of the traceability supervision decision algorithm is obtained based on the first traceability attribute description feature and the second traceability attribute description feature corresponding to each of the smart IoT sensor data, including any of the following.
[0202] (1) Obtaining the consistency debugging error based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature or the second production line traceability attribute description feature corresponding to each of the smart IoT sensor data.
[0203] (2) For each of the smart IoT sensor data, a first traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a first traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a first traceability attribute combination feature corresponding to the smart IoT sensor data; and a second traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a second traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a second traceability attribute combination feature corresponding to the smart IoT sensor data; based on the first traceability attribute combination feature and the second traceability attribute combination feature corresponding to each of the smart IoT sensor data, the consistency debugging error is obtained.
[0204] (3) Alternatively, a first local consistency debugging error is obtained based on a first traceability attribute description feature and a second traceability attribute description feature generated from first production line traceability attribute description features corresponding to each of the smart IoT sensor data, and a second local consistency debugging error is obtained based on a first traceability attribute description feature and a second traceability attribute description feature generated from second production line traceability attribute description features corresponding to each of the smart IoT sensor data; the first local consistency debugging error and the second local consistency debugging error are weighted to obtain the consistency debugging error.
[0205] In the embodiment of the present application, the core task of determining the consistency debugging error of the traceability supervision decision algorithm of the intelligent IoT sensor data is to evaluate the stability and consistency of the algorithm when processing different production line traceability attribute description features. In order to achieve this goal, the embodiment of the present application proposes several specific technical solutions, each of which revolves around how to effectively utilize and combine the first production line traceability attribute description feature and the second production line traceability attribute description feature.
[0206] First, the intelligent manufacturing quality data traceability system can consider a direct method, that is, to obtain the consistency debugging error based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature or the second production line traceability attribute description feature corresponding to each intelligent IoT sensor data. The core of this method is that it directly takes the traceability attribute description features of different production lines as input, and generates the corresponding first traceability attribute description features and second traceability attribute description features through algorithm processing. Then, by comparing the difference or similarity between these two features, the intelligent manufacturing quality data traceability system can evaluate the consistency of the algorithm when processing data from different production lines. If the difference is small, it means that the algorithm has good stability when processing data from different production lines; conversely, if the difference is large, it means that the consistency of the algorithm needs to be improved.
[0207] Secondly, the intelligent manufacturing quality data traceability system can adopt a more complex but more comprehensive method, that is, the first traceability attribute description feature generated by the first production line traceability attribute description feature of the intelligent IoT sensor data is combined with the first traceability attribute description feature generated by the second production line traceability attribute description feature of the same intelligent IoT sensor data to obtain the first traceability attribute combination feature corresponding to the intelligent IoT sensor data. Similarly, the intelligent manufacturing quality data traceability system can also obtain the second traceability attribute combination feature. The core of this method is that it generates richer and more comprehensive traceability attribute combination features by combining the traceability attribute description features of different production lines. Then, based on these combined features, the intelligent manufacturing quality data traceability system can more accurately evaluate the consistency of the algorithm when processing different production line data. Because this method takes into account the interaction and influence between different production line data, it can provide more comprehensive and accurate evaluation results.
[0208] Finally, the intelligent manufacturing quality data traceability system can also consider a compromise method, that is, based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature corresponding to each intelligent IoT sensor data, the first local consistency debugging error is obtained; at the same time, based on the first traceability attribute description feature and the second traceability attribute description feature generated by the second production line traceability attribute description feature corresponding to each intelligent IoT sensor data, the second local consistency debugging error is obtained. Then, the intelligent manufacturing quality data traceability system can weight these two local consistency debugging errors to obtain the final consistency debugging error. The core of this method is that it considers the impact of different production line data on algorithm consistency separately, and merges these two impacts in a weighted manner. In this way, the intelligent manufacturing quality data traceability system can obtain detailed information about the impact of different production line data on algorithm consistency, and can also obtain an overall consistency evaluation result that comprehensively considers different production line data.
[0209] In this way, three different technical solutions can be proposed to evaluate the consistency of the traceability supervision decision algorithm when processing smart IoT sensor data. Each method has its unique advantages and applicable scenarios. In practical applications, the intelligent manufacturing quality data traceability system can select the most appropriate method for evaluation based on specific needs and data characteristics. Through the application of these evaluation methods, the intelligent manufacturing quality data traceability system can better understand the performance and stability of the algorithm when processing data from different production lines, thereby providing strong support for the improvement and optimization of the algorithm. At the same time, these evaluation methods can also serve as an important basis for algorithm selection and comparison, helping the intelligent manufacturing quality data traceability system to select the most suitable traceability supervision decision algorithm in practical applications.
[0210] In some preferred embodiments, the consistency debugging error of the traceability supervision decision algorithm is obtained based on the first traceability attribute description feature and the second traceability attribute description feature corresponding to each of the smart IoT sensor data, including: for each of the smart IoT sensor data, the first traceability attribute description feature corresponding to the smart IoT sensor data is fused with the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a first traceability attribute fusion feature; the first traceability attribute fusion feature and the time series vector of the soft threshold are crossed to obtain a second traceability attribute fusion feature; the second traceability attribute fusion feature is used as a reference for feature activation to obtain a third traceability attribute fusion feature; the thermal characteristics of the first production link are obtained based on the third traceability attribute fusion feature corresponding to each of the smart IoT sensor data; the traceability attribute fusion feature corresponding to the smart IoT sensor data is activated. The second traceability attribute description feature is fused with the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a fourth traceability attribute fusion feature; the fourth traceability attribute fusion feature and the time series vector are crossed to obtain a fifth traceability attribute fusion feature; feature activation is performed based on the fifth traceability attribute fusion feature to obtain a sixth traceability attribute fusion feature; based on the sixth traceability attribute fusion feature corresponding to each smart IoT sensor data, the thermal feature of the second production link is obtained; the thermal feature of the first production link and the thermal feature of the second production link are weighted to obtain the thermal feature of the third production link; the thermal feature of the first production link and the thermal feature of the third production link are crossed to obtain the seventh traceability attribute fusion feature; based on the seventh traceability attribute fusion feature, the consistency debugging error of the traceability supervision decision algorithm is obtained.
[0211] It can be understood that this embodiment involves how to obtain the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature corresponding to each smart IoT sensor data. This process involves multi-level feature fusion, crossover and activation, as well as the extraction and weighting of thermal features. The ultimate goal is to accurately evaluate the consistency of the algorithm when processing data from different production lines.
[0212] First, for each smart IoT sensor data, the smart manufacturing quality data traceability system fuses the corresponding first traceability attribute description features with the traceability attribute description features of the third production line semantic embedding vector. The purpose of this step is to organically combine the traceability attribute description features from different sources to form a more comprehensive and rich first traceability attribute fusion feature. Feature fusion can be achieved through various methods, such as vector splicing, weighted summation, etc. The specific method to be selected depends on the characteristics of the data and actual needs.
[0213] Next, the intelligent manufacturing quality data traceability system crosses the first traceability attribute fusion feature with the time series vector of the soft threshold. The time series vector is a vector containing time information, which can reflect the changing trend of data over time. By crossing with the time series vector, the intelligent manufacturing quality data traceability system can extract the time information in the first traceability attribute fusion feature to form the second traceability attribute fusion feature. This step is crucial to capturing the temporal dynamics of the data.
[0214] Then, feature activation is performed based on the second traceability attribute fusion feature. Feature activation is a method to enhance feature representation capabilities, which can be achieved through nonlinear transformation, attention mechanism, etc. Through feature activation, the intelligent manufacturing quality data traceability system can further mine the useful information in the second traceability attribute fusion feature to form a more expressive third traceability attribute fusion feature.
[0215] Based on the third traceability attribute fusion characteristics corresponding to each smart IoT sensor data, the smart manufacturing quality data traceability system can obtain the thermal characteristics of the first production link. Thermal characteristics are a kind of characteristics that can reflect the distribution and density of data, which is of great significance for evaluating the consistency of the algorithm. By calculating the thermal value of the third traceability attribute fusion characteristics, the smart manufacturing quality data traceability system can obtain the thermal characteristics of the first production link, and then understand the processing performance of the algorithm in this link.
[0216] Similarly, the intelligent manufacturing quality data traceability system can also fuse the second traceability attribute description feature corresponding to the intelligent IoT sensor data with the traceability attribute description feature of the third production line semantic embedding vector to obtain the fourth traceability attribute fusion feature. Then, through intersection and feature activation with the time series vector, the intelligent manufacturing quality data traceability system can obtain the sixth traceability attribute fusion feature, and based on it, obtain the thermal characteristics of the second production link.
[0217] After obtaining the thermal characteristics of the first production link and the thermal characteristics of the second production link, the intelligent manufacturing quality data traceability system can weight them to obtain the thermal characteristics of the third production link. The weighting process can be adjusted according to actual needs to reflect the degree of influence of different production links on the consistency of the algorithm.
[0218] Finally, the intelligent manufacturing quality data traceability system crosses the thermal characteristics of the first production link and the thermal characteristics of the third production link to obtain the seventh traceability attribute fusion feature. Based on the seventh traceability attribute fusion feature, the intelligent manufacturing quality data traceability system can calculate the consistency debugging error of the traceability supervision decision algorithm. This error reflects the consistency of the algorithm when processing data from different production lines and is an important indicator for evaluating algorithm performance.
[0219] With this design, the consistency debugging error of the traceability supervision decision algorithm is obtained through multi-level feature fusion, crossover and activation, as well as the extraction and weighting of thermal features. This process not only considers the impact of different production line data on the consistency of the algorithm, but also fully explores the time dynamics and useful information in the data. Through this method, the intelligent manufacturing quality data traceability system can more accurately evaluate the performance and stability of the algorithm when processing intelligent IoT sensor data, providing strong support for the improvement and optimization of the algorithm. At the same time, this method can also serve as an important basis for algorithm selection and comparison, helping the intelligent manufacturing quality data traceability system to select the most suitable traceability supervision decision algorithm in practical applications.
[0220] In detail, the following are formula examples of the relevant steps in the above embodiment.
[0221] (1) Feature fusion: The first traceability attribute description feature A and the second traceability attribute description feature (here replaced by the traceability attribute description feature E of the third production line semantic embedding vector) are fused to obtain the first traceability attribute fusion feature F1.
[0222] F1_i=Fuse(A_1_i,E_3_i)
[0223] (2) Cross operation and time series vector: The fused first traceability attribute fusion feature F1 is cross-operated with the time series vector T to obtain the second traceability attribute fusion feature F2.
[0224] F2_i=Cross(F1_i,T_i)
[0225] (3) Feature activation: Activate the second traceability attribute fusion feature F2 to obtain the activated third traceability attribute fusion feature F3.
[0226] F3_i=Activate(F2_i)
[0227] (4) Thermal feature extraction (first production link): Thermal feature extraction is performed on all activated third traceability attribute fusion features F3 to obtain the thermal feature H1 of the first production link.
[0228] H1=Extract Heat({F3_i})
[0229] Repeat steps 1-4 to obtain the thermal characteristics of the second production link. Similarly, through fusion, crossover, activation and thermal characteristics extraction, the thermal characteristics H2 of the second production link are obtained.
[0230] H2=Extract Heat({Activate(Cross(Fuse(A_2_i,E_3_i),T_i))})
[0231] (5) Weighted and comprehensive thermal characteristics: The weighted sum of the thermal characteristics H1 and H2 of the first and second production links is performed to obtain the comprehensive thermal characteristic H3.
[0232] H3=w1*H1+w2*H2
[0233] (6) Cross-over and debugging error calculation: The thermal characteristic H1 of the first production link is cross-operated with the comprehensive thermal characteristic H3 to obtain the cross-over characteristic F7.
[0234] F7_i=Cross(H1,H3)
[0235] (7) Then, the debugging error between the cross feature F7 and the expected thermal feature H_expected is calculated.
[0236] Error=Calculate Error(F7_i,H_expected)
[0237] The above formula describes the calculation process of the consistency debugging error of the traceability supervision decision algorithm based on smart IoT sensor data. In practical applications, the functions and parameters in these steps can be adjusted according to specific data and business requirements.
[0238] Under some preferred design ideas, the traceability supervision decision algorithm is debugged based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain the debugged traceability supervision decision algorithm, including: weighting the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain the global debugging error of the traceability supervision decision algorithm; and cyclically improving the algorithm variables of the traceability supervision decision algorithm in the direction of converging the global debugging error to obtain the debugged traceability supervision decision algorithm.
[0239] When it comes to how to optimize and debug the traceability supervision decision algorithm, this embodiment aims to comprehensively debug and improve the algorithm by comprehensively considering the consistency debugging error and the discrimination debugging error.
[0240] First of all, the core goal of the traceability supervision decision algorithm is to accurately track and supervise various traceability information, so as to provide strong support for decision-making. However, in practical applications, due to the influence of various factors such as data complexity and algorithm design, the algorithm often has certain debugging errors. These errors may be caused by data consistency problems or by the inaccuracy of the algorithm in the judgment process. Therefore, in order to improve the accuracy and reliability of the algorithm, the intelligent manufacturing quality data traceability system must conduct in-depth analysis and processing of these debugging errors.
[0241] In the process of handling debugging errors, the intelligent manufacturing quality data traceability system puts forward a key concept: global debugging error. The core of this concept is to weight the consistency debugging error and the discrimination debugging error to obtain a comprehensive indicator that can fully reflect the debugging status of the algorithm. Specifically, the intelligent manufacturing quality data traceability system can assign different weights to the consistency debugging error and the discrimination debugging error according to actual needs and application scenarios. These weights can be set according to factors such as the severity of the error and the degree of impact on the algorithm performance. Through weighted synthesis, the intelligent manufacturing quality data traceability system can obtain a more comprehensive and accurate debugging error indicator, providing strong support for subsequent algorithm debugging.
[0242] With the comprehensive indicator of global debugging error, the intelligent manufacturing quality data traceability system can cyclically improve the traceability supervision decision algorithm in the direction of converging the error. The core of this process is to gradually reduce the global debugging error by constantly adjusting the variables and parameters of the algorithm, thereby achieving the purpose of optimizing the algorithm performance. Specifically, the intelligent manufacturing quality data traceability system can use various optimization algorithms and machine learning methods, such as gradient descent, stochastic gradient descent, Newton's method, etc., to iteratively update the variables of the algorithm. In each iteration process, the intelligent manufacturing quality data traceability system will adjust the variables accordingly according to the changes in the global debugging error, so that the algorithm gradually approaches the optimal solution.
[0243] In the process of cyclic improvement, the intelligent manufacturing quality data traceability system also needs to pay attention to some issues. First, the intelligent manufacturing quality data traceability system needs to ensure the convergence of the algorithm. In other words, the intelligent manufacturing quality data traceability system must ensure that the global debugging error can be gradually reduced during the iteration process and finally converge to a stable value. In order to achieve this goal, the intelligent manufacturing quality data traceability system can use various convergence judgment criteria, such as gradient norm, function value change, etc., to monitor the convergence of the algorithm. Secondly, the intelligent manufacturing quality data traceability system needs to consider the computational complexity and time complexity of the algorithm. In practical applications, the intelligent manufacturing quality data traceability system often needs to obtain the optimal solution or approximate optimal solution within a limited time. Therefore, when selecting the optimization algorithm and setting the number of iterations, the intelligent manufacturing quality data traceability system needs to fully consider the computational efficiency and time cost of the algorithm. Finally, the intelligent manufacturing quality data traceability system also needs to fully test and verify the algorithm to ensure its accuracy and reliability in various application scenarios.
[0244] Through the process of cyclic improvement, the intelligent manufacturing quality data traceability system can obtain a debugged traceability supervision decision algorithm. While retaining the advantages of the original algorithm, the algorithm further reduces the debugging error and improves the accuracy and reliability of the algorithm. In practical applications, the algorithm can better track and supervise various traceability information and provide stronger support for decision-making. At the same time, due to the introduction of global debugging errors and the application of optimization algorithms, the algorithm also has better generalization ability and robustness, and can better adapt to various complex application scenarios and data changes.
[0245] In addition, it is worth mentioning that this technical solution can also be combined with other optimization technologies and algorithms to further improve the performance of the traceability supervision decision-making algorithm. For example, the intelligent manufacturing quality data traceability system can consider introducing advanced technologies such as deep learning and reinforcement learning to optimize and improve the algorithm more deeply. At the same time, the intelligent manufacturing quality data traceability system can also customize the design and development of the algorithm in combination with actual application scenarios and needs to meet more specific and complex application needs.
[0246] It can be seen that by comprehensively considering the consistency debugging error and the discrimination debugging error, and making cyclic improvements in the direction of converging the global debugging error, the intelligent manufacturing quality data traceability system can obtain a debugged traceability supervision decision algorithm. The algorithm has been significantly improved in accuracy and reliability, and can better meet the needs of actual application scenarios.
[0247] In an optional embodiment, the index features of the blockchain traceability node and the first semantic embedding vector are recorded in a preset feature space in the form of feature binary tuples, and the index features of the product quality traceability information and the second semantic embedding vector are recorded in the preset feature space in the form of feature binary tuples; then the acquisition of multiple smart IoT sensor data includes: acquiring the first semantic embedding vector from the preset feature space based on the index features of the blockchain traceability node; acquiring the second semantic embedding vector from the preset feature space based on the index features of the product quality traceability information; and acquiring the smart IoT sensor data by combining the first semantic embedding vector and the second semantic embedding vector.
[0248] It is worth mentioning that this embodiment makes full use of the index features of blockchain traceability nodes, the index features of product quality traceability information, and their respective corresponding semantic embedding vectors to achieve accurate acquisition of smart IoT sensor data in an efficient and orderly manner.
[0249] First, in the infrastructure of this embodiment, the index features and the first semantic embedding vector of the blockchain traceability node are recorded in the preset feature space in the form of feature binary tuples. Similarly, the index features and the second semantic embedding vector of the product quality traceability information are also stored in the preset feature space in the form of feature binary tuples. This storage method not only makes the data more orderly, but also provides great convenience for subsequent rapid retrieval and matching.
[0250] Next, the intelligent manufacturing quality data traceability system elaborates on how to obtain multiple intelligent IoT sensor data based on this architecture. The whole process can be divided into several closely connected steps:
[0251] 1. Obtain the first semantic embedding vector based on the index features of the blockchain traceability node: In this step, the intelligent manufacturing quality data traceability system first focuses on the index features of the blockchain traceability node. These index features are like the "identity cards" of the blockchain traceability nodes, which uniquely and accurately identify each blockchain traceability node. By using these index features as query keywords, the intelligent manufacturing quality data traceability system can quickly locate the first semantic embedding vector that matches these index features in the preset feature space. These first semantic embedding vectors contain rich semantic information, which can provide the intelligent manufacturing quality data traceability system with in-depth understanding and insights about the blockchain traceability nodes;
[0252] 2. Obtain the second semantic embedding vector based on the index features of product quality traceability information: Next, the intelligent manufacturing quality data traceability system shifts its attention to the index features of product quality traceability information. These index features also play a key role. They uniquely and accurately identify each type of product quality traceability information. Using these index features as a guide for queries, the intelligent manufacturing quality data traceability system can easily find the second semantic embedding vectors corresponding to them in the preset feature space. These second semantic embedding vectors are also rich in profound semantic connotations. They reveal the core elements and key features of product quality traceability information for the intelligent manufacturing quality data traceability system;
[0253] 3. Combining the first semantic embedding vector and the second semantic embedding vector to obtain smart IoT sensor data: After successfully obtaining the first semantic embedding vector and the second semantic embedding vector, the smart manufacturing quality data traceability system cleverly combines these two types of semantic embedding vectors. The combination process is not a simple splicing or superposition, but is based on a deep semantic matching and fusion mechanism. Through this mechanism, the smart manufacturing quality data traceability system can accurately mine smart IoT sensor data that is closely related to both types of semantic embedding vectors. These data not only cover the key information of the blockchain traceability node, but also contain important details of product quality traceability. They provide a comprehensive and in-depth perspective for the smart manufacturing quality data traceability system to examine the entire traceability process.
[0254] It is worth mentioning that the technical solution demonstrated by this embodiment has extremely high flexibility and scalability. In practical applications, the intelligent manufacturing quality data traceability system can adjust and optimize the solution according to specific needs and scenarios. For example, the intelligent manufacturing quality data traceability system can customize the dimensions and types of index features and semantic embedding vectors according to actual business needs to ensure that they can most accurately reflect the traceability information that the intelligent manufacturing quality data traceability system is concerned about. At the same time, the intelligent manufacturing quality data traceability system can also be combined with other advanced technical means, such as machine learning, deep learning, etc., to further improve the performance and accuracy of the solution.
[0255] In addition, this embodiment also fully considers the security and privacy protection of data. Since all index features and semantic embedding vectors are stored in the preset feature space in the form of feature binary tuples, they are effectively encrypted and protected. This can not only prevent data leakage and abuse, but also ensure the integrity and consistency of data. This is undoubtedly a huge advantage for industries and application scenarios that have extremely high requirements for data security.
[0256] In this way, the intelligent manufacturing quality data traceability system can obtain multiple intelligent IoT sensor data in an efficient, accurate and secure way. These data not only provide the intelligent manufacturing quality data traceability system with comprehensive insights into blockchain traceability nodes and product quality traceability information, but also provide a solid data foundation for subsequent data analysis, decision support, etc. of the intelligent manufacturing quality data traceability system.
[0257] In another alternative embodiment, the method also includes: for each of the smart IoT sensor data, obtaining a first life cycle state vector in a first semantic embedding vector included in the smart IoT sensor data; combining the first life cycle state vector and the first semantic embedding vector, obtaining a negative debugging example of the blockchain traceability node corresponding to the smart IoT sensor data; based on the negative debugging examples of the blockchain traceability nodes corresponding to each of the smart IoT sensor data, obtaining a second traceability attribute description feature corresponding to each of the smart IoT sensor data.
[0258] It should be noted that the core of this embodiment is that it not only focuses on the smart IoT sensor data itself, but also deeply explores the life cycle status information implied behind these data, and how this information is associated with the blockchain traceability node, and ultimately affects the extraction of traceability attribute description features.
[0259] First, the intelligent manufacturing quality data traceability system extracts the first life cycle state vector in the first semantic embedding vector from each intelligent IoT sensor data. This step is crucial because it enables the intelligent manufacturing quality data traceability system to understand the changes and evolution of sensor data at a more fine-grained level. The first life cycle state vector, as the name suggests, depicts the different states of intelligent IoT sensor data during its life cycle, which may include various aspects such as data generation, transmission, storage, and processing. By extracting this vector, the intelligent manufacturing quality data traceability system can more clearly grasp the dynamic changes of data during its life cycle, providing a richer information basis for subsequent analysis and decision-making.
[0260] Next, the intelligent manufacturing quality data traceability system organically combines the first life cycle state vector with the first semantic embedding vector. This combination is not a simple superposition or splicing, but is based on a deep semantic matching and fusion mechanism. Through this mechanism, the intelligent manufacturing quality data traceability system can generate a new representation: a negative debugging example of a blockchain traceability node. This representation not only contains the semantic information of the smart IoT sensor data, but also incorporates the state information of the data life cycle, thereby providing a more comprehensive and in-depth perspective for the intelligent manufacturing quality data traceability system to examine the behavior of the blockchain traceability node.
[0261] After obtaining the negative debugging examples of the blockchain traceability nodes corresponding to each smart IoT sensor data, the smart manufacturing quality data traceability system further extracts the second traceability attribute description features based on these examples. This step is the key to the entire embodiment because it is directly related to the understanding and application of traceability information by the smart manufacturing quality data traceability system. The second traceability attribute description feature is a richer and more comprehensive feature representation obtained by further integrating the life cycle state information and negative debugging information on the basis of the first traceability attribute description feature. These features not only include the basic attributes of smart IoT sensor data, but also reflect the state changes of the data in its life cycle and the behavioral characteristics of the blockchain traceability nodes.
[0262] It is worth noting that the technical solution demonstrated by this embodiment has extremely high flexibility and scalability. In practical applications, the intelligent manufacturing quality data traceability system can adjust and optimize the solution according to specific needs and scenarios. For example, the intelligent manufacturing quality data traceability system can customize the dimensions and types of life cycle state vectors according to actual business needs to ensure that they can most accurately reflect the data status information that the intelligent manufacturing quality data traceability system is concerned about. At the same time, the intelligent manufacturing quality data traceability system can also be combined with other advanced technical means, such as machine learning, deep learning, etc., to further improve the performance and accuracy of the solution.
[0263] In addition, this embodiment also fully considers the security and privacy protection of data. Since all smart IoT sensor data, the first semantic embedding vector, the first life cycle state vector, and the negative debugging examples of the blockchain traceability node are processed and analyzed under strict security control, they are effectively encrypted and protected. This can not only prevent the leakage and abuse of data, but also ensure the integrity and consistency of data. This is undoubtedly a huge advantage for industries and application scenarios that have extremely high requirements for data security.
[0264] In this way, the intelligent manufacturing quality data traceability system can understand and analyze intelligent IoT sensor data in a more comprehensive and in-depth way. These data not only provide the intelligent manufacturing quality data traceability system with rich information about blockchain traceability nodes, but also provide a more solid data foundation for subsequent data analysis, decision support, etc. of the intelligent manufacturing quality data traceability system.
[0265] In other optional embodiments, the method also includes: using the debugged traceability supervision decision algorithm to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information; and outputting quality traceability opinion comments based on the traceability item identification text of the blockchain traceability node for the product quality traceability information.
[0266] Based on this embodiment, the application scope and functional depth of the method are further expanded, aiming to achieve more accurate and comprehensive processing of product quality traceability information on blockchain traceability nodes through the debugged traceability supervision decision algorithm. This expansion not only enhances the understanding and grasp of traceability information, but also provides a novel way to output quality traceability opinion comments.
[0267] First, the intelligent manufacturing quality data traceability system uses the debugged traceability supervision decision algorithm to deeply explore the intrinsic connection between the blockchain traceability node and the product quality traceability information. After debugging and optimization, the algorithm can more accurately identify the traceability items of the blockchain traceability node for specific product quality traceability information. These traceability items may cover multiple links such as product production, processing, circulation, and sales, and each link contains rich traceability information and data. Through the precise identification of the algorithm, the intelligent manufacturing quality data traceability system can effectively connect the traceability information scattered in various links to form a complete and clear traceability chain.
[0268] After successfully obtaining the traceability project identification text of the blockchain traceability node for product quality traceability information, the intelligent manufacturing quality data traceability system further conducts in-depth analysis and mining based on these texts. The key to this step is that the intelligent manufacturing quality data traceability system must not only understand the surface meaning of the text, but also deeply explore the opinions, attitudes and emotions implied behind the text. In order to achieve this goal, the intelligent manufacturing quality data traceability system may use a series of advanced natural language processing technologies, such as sentiment analysis, opinion mining, semantic understanding, etc. Through the comprehensive application of these technologies, the intelligent manufacturing quality data traceability system can more accurately capture the quality traceability opinions contained in the text, and provide a solid data foundation for the subsequent output of quality traceability opinion comments.
[0269] Then, based on the above-mentioned in-depth analysis and mining, the intelligent manufacturing quality data traceability system begins to output quality traceability opinion comments. This step is the core of the entire embodiment, because it is directly related to the final expression and application of traceability information by the intelligent manufacturing quality data traceability system. When outputting opinion comments, the intelligent manufacturing quality data traceability system focuses on maintaining an objective, neutral and comprehensive attitude, and strives to reflect the actual opinions in the text while incorporating the intelligent manufacturing quality data traceability system's in-depth understanding and unique insights into the traceability information. At the same time, the intelligent manufacturing quality data traceability system also pays attention to the readability and understandability of opinion comments, and strives to make the language concise and clear, and the logic clear and coherent, so that more people can easily understand and accept the opinion comments of the intelligent manufacturing quality data traceability system.
[0270] It is worth mentioning that the technical solution demonstrated by this embodiment has extremely high flexibility and scalability. In practical applications, the intelligent manufacturing quality data traceability system can adjust and optimize the solution according to specific needs and scenarios. For example, the intelligent manufacturing quality data traceability system can customize the parameters and models of the traceability supervision decision algorithm according to actual business needs to ensure that they can most accurately identify the traceability items that the intelligent manufacturing quality data traceability system is concerned about. At the same time, the intelligent manufacturing quality data traceability system can also be combined with other advanced technical means, such as machine learning, deep learning, etc., to further improve the performance and accuracy of the solution.
[0271] In this way, the intelligent manufacturing quality data traceability system can not only accurately identify these traceability items, but also output quality traceability opinion comments with depth and breadth based on these items, thereby accurately assisting the product quality monitoring of intelligent manufacturing and the credibility of product quality monitoring.
[0272] In some independent embodiments, the output of quality traceability opinion comments based on the traceability project identification text of the product quality traceability information by the blockchain traceability node includes: performing semantic mining on the traceability project identification text of the product quality traceability information by the blockchain traceability node according to a first opinion semantic mining strategy to obtain a first opinion semantic, wherein the first opinion semantic mining strategy is a feature mining strategy based on the paragraph level; based on the first opinion semantics of the traceability project identification text of the product quality traceability information by the blockchain traceability node, determining at least one comment information matching the traceability project identification text of the product quality traceability information by the blockchain traceability node, and mining the comment information according to a second opinion semantic mining strategy. The blockchain traceability node performs semantic mining on the traceability item identification text and each comment information of the product quality traceability information to obtain a second viewpoint semantics, and the second viewpoint semantic mining strategy is a feature mining strategy based on the word and sentence level; based on the extracted second viewpoint semantics, the blockchain traceability node performs quality traceability event matching processing on the traceability item identification text of the product quality traceability information with each comment information in turn, and the target comment information in the at least one comment information is determined as the quality traceability viewpoint comment, and the number of quality traceability events that match between the target comment information and the traceability item identification text of the blockchain traceability node for the product quality traceability information is greater than a set value.
[0273] It can be understood that this embodiment focuses on how to output quality traceability opinion comments with depth and breadth based on the blockchain traceability node for the traceability item recognition text of product quality traceability information. The realization of this process requires not only the intelligent manufacturing quality data traceability system to accurately understand the semantic content in the recognition text, but also the intelligent manufacturing quality data traceability system to effectively match these contents with relevant comment information, so as to extract the most representative quality traceability opinion.
[0274] First, the intelligent manufacturing quality data traceability system conducts in-depth semantic mining on the traceability project identification text of the blockchain traceability node for product quality traceability information according to the first-view semantic mining strategy. This strategy is based on the feature mining strategy at the paragraph level, which allows the intelligent manufacturing quality data traceability system to capture more macro and overall semantic information in the text. Through this step, the intelligent manufacturing quality data traceability system can obtain the first-view semantics, and these semantic information provide an important basis for the subsequent processing of the intelligent manufacturing quality data traceability system.
[0275] Next, the intelligent manufacturing quality data traceability system determines at least one comment information that matches the traceability item identification text of the blockchain traceability node for product quality traceability information based on the extracted first opinion semantics. The key to this step is that the intelligent manufacturing quality data traceability system needs to find comment information that is highly semantically related to the identification text or has similar opinions. To achieve this goal, the intelligent manufacturing quality data traceability system may use a series of advanced text matching and semantic similarity calculation technologies.
[0276] After determining the matching comment information, the intelligent manufacturing quality data traceability system further performs semantic mining on these comment information according to the second view semantic mining strategy. Unlike the first view semantic mining strategy, the second view semantic mining strategy is a feature mining strategy based on the word and sentence level. This strategy allows the intelligent manufacturing quality data traceability system to dig deeper into the detailed information in the text and capture the subtle semantic differences that may be hidden between words and sentences. Through this step, the intelligent manufacturing quality data traceability system can obtain the second view semantics, which provides a more refined data foundation for the subsequent quality traceability event matching processing of the intelligent manufacturing quality data traceability system.
[0277] Finally, based on the extracted second-view semantics, the intelligent manufacturing quality data traceability system matches the traceability project identification text of the blockchain traceability node for the product quality traceability information with each comment information in turn for quality traceability event matching. The core of this step is that the intelligent manufacturing quality data traceability system needs to find those comment information that are highly consistent with the identification text in the quality traceability event. In order to achieve this goal, the intelligent manufacturing quality data traceability system may define a series of quality traceability event features, such as event type, event time, event location, etc., and perform matching calculations based on these features. Through this step, the intelligent manufacturing quality data traceability system can identify those comment information whose number of quality traceability events matching the identification text is greater than the set value as the target comment information, that is, the quality traceability viewpoint comments of the intelligent manufacturing quality data traceability system.
[0278] With this design, the intelligent manufacturing quality data traceability system can not only accurately mine the semantic information of opinions in these texts, but also output representative and in-depth quality traceability opinion comments based on this information. These opinion comments not only provide the intelligent manufacturing quality data traceability system with rich insights into traceability information, but also provide a more solid data foundation for subsequent data analysis, decision support, etc. of the intelligent manufacturing quality data traceability system.
[0279] Further, Figure 2 The structure diagram of an intelligent manufacturing quality data traceability system 200 provided in an embodiment of the present application is shown in FIG. Figure 2 The intelligent manufacturing quality data traceability system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.
[0280] Alternatively, if Figure 2 As shown, the intelligent manufacturing quality data traceability system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.
[0281] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .
[0282] Alternatively, if Figure 2 As shown, the intelligent manufacturing quality data traceability system 200 may also include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.
[0283] Optionally, the intelligent manufacturing quality data traceability system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.
[0284] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities.
[0285] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.
[0286] The embodiment of the present application provides a smart manufacturing quality data traceability device, which is used to: obtain multiple smart IoT sensor data, the smart IoT sensor data includes a first semantic embedding vector of a blockchain traceability node, and a second semantic embedding vector of product quality traceability information, the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information; for each of the smart IoT sensor data, based on the first semantic embedding vector and the second semantic embedding vector included in the smart IoT sensor data, obtain the traceability attribute description feature corresponding to the smart IoT sensor data; through a traceability supervision decision algorithm, based on each of the smart IoT sensor data, respectively correspond to The traceability attribute description features are used to determine the traceability item identification texts corresponding to each of the smart IoT sensor data, and the traceability item identification texts represent the recognition probability of the traceability item texts; based on the traceability attribute description features, traceability item identification texts and traceability item prior texts corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm, and the debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotations of the traceability item text; wherein the life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index feature in the entire life cycle corresponding to the blockchain supervision node.
[0287] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and the program or instruction implements the steps of the above method when executed by a processor.
[0288] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.
[0289] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.
[0290] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the present application, all of which are within the protection of the present application.
Claims
1. A blockchain-based intelligent manufacturing quality data traceability method, characterized in that: The method is applied to an intelligent manufacturing quality data traceability system, and the method comprises: Acquire multiple smart IoT sensor data, wherein the smart IoT sensor data includes a first semantic embedding vector of a blockchain traceability node and a second semantic embedding vector of product quality traceability information, wherein the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information; For each of the smart IoT sensor data, based on the first semantic embedding vector and the second semantic embedding vector included in the smart IoT sensor data, obtaining a traceability attribute description feature corresponding to the smart IoT sensor data; Determine the traceability project identification text corresponding to each of the smart IoT sensor data based on the traceability attribute description features corresponding to each of the smart IoT sensor data through a traceability supervision decision algorithm, wherein the traceability project identification text represents the recognition probability of the traceability project text; Based on the traceability attribute description features, traceability item identification texts and traceability item prior texts respectively corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain a debugged traceability supervision decision algorithm, wherein the debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotation of the traceability item text; The life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index feature in the entire life cycle corresponding to the blockchain supervision node; The traceability attribute description features corresponding to the smart IoT sensor data include the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector; The traceability supervision decision algorithm is debugged based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each of the smart IoT sensor data to obtain a debugged traceability supervision decision algorithm, including: Based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data, a first traceability attribute description feature and a second traceability attribute description feature corresponding to the smart IoT sensor data are generated, wherein the first traceability attribute description feature represents the traceability attribute description feature of the active debugging example corresponding to the blockchain traceability node, and the second traceability attribute description feature represents the traceability attribute description feature of the passive debugging example corresponding to the blockchain traceability node; Based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data, a consistency debugging error of the traceability supervision decision algorithm is obtained, wherein the consistency debugging error is used to indicate the performance of the traceability supervision decision algorithm in mastering the content corresponding to the second traceability attribute description feature; Based on the traceability project identification text and the traceability project priori text corresponding to each of the smart IoT sensor data, a discrimination debugging error of the traceability supervision decision algorithm is obtained, and the discrimination debugging error is used to indicate the decision offset weight of the traceability supervision decision algorithm for the traceability project text; Based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm; The first semantic embedding vector includes a first production line semantic embedding vector and a second production line semantic embedding vector, the life cycle state vector in the first production line semantic embedding vector has a matching relationship with the blockchain traceability node, the second production line semantic embedding vector includes the first production line semantic embedding vector, and the second production line semantic embedding vector contains a life cycle state vector that has a non-matching relationship with the blockchain traceability node; the second semantic embedding vector includes a third production line semantic embedding vector and a fourth production line semantic embedding vector, the life cycle state vector in the third production line semantic embedding vector has a matching relationship with the product quality traceability information, and the fourth production line semantic embedding vector is used to represent the access warning record of the product quality traceability information; The generating of the first traceability attribute description feature and the second traceability attribute description feature corresponding to the smart IoT sensor data based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data includes: For each of the smart IoT sensor data, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the second production line semantic embedding vector corresponding to the smart IoT sensor data; or, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data; Performing feature derivation on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the second production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description feature; or, performing feature derivation on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description feature; The traceability attribute description feature corresponding to the smart IoT sensor data includes a first production line traceability attribute description feature and a second production line traceability attribute description feature, wherein the first production line traceability attribute description feature and the second production line traceability attribute description feature are both used to represent the smart IoT sensor data, and the first production line traceability attribute description feature and the second production line traceability attribute description feature correspond to the same traceability feature relationship network; The step of obtaining the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data includes: Obtain the consistency debugging error based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature or the second production line traceability attribute description feature respectively corresponding to each of the smart IoT sensor data; Alternatively, for each of the smart IoT sensor data, a first traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a first traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a first traceability attribute combination feature corresponding to the smart IoT sensor data, and a second traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a second traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a second traceability attribute combination feature corresponding to the smart IoT sensor data; based on the first traceability attribute combination feature and the second traceability attribute combination feature respectively corresponding to each of the smart IoT sensor data, the consistency debugging error is obtained; Alternatively, a first local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature corresponding to each of the smart IoT sensor data, and a second local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the second production line traceability attribute description feature corresponding to each of the smart IoT sensor data; the first local consistency debugging error and the second local consistency debugging error are weighted to obtain the consistency debugging error.
2. The method according to claim 1, characterized in that The step of obtaining the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data includes: For each of the smart IoT sensor data, a first traceability attribute description feature corresponding to the smart IoT sensor data is fused with a traceability attribute description feature of a third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a first traceability attribute fusion feature; Crossing the first traceability attribute fusion feature and the time series vector of the soft threshold to obtain a second traceability attribute fusion feature; Activate the feature based on the second traceability attribute fusion feature to obtain a third traceability attribute fusion feature; Based on the third traceability attribute fusion characteristics corresponding to each of the smart IoT sensor data, the thermal characteristics of the first production link are obtained; The second traceability attribute description feature corresponding to the smart IoT sensor data is feature-fused with the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to obtain a fourth traceability attribute fusion feature; Crossing the fourth traceability attribute fusion feature and the time series vector to obtain a fifth traceability attribute fusion feature; Activate the feature based on the fifth traceability attribute fusion feature to obtain a sixth traceability attribute fusion feature; Based on the fusion characteristics of the sixth traceability attribute corresponding to each of the smart IoT sensor data, the thermal characteristics of the second production link are obtained; Weighting the thermal characteristics of the first production link and the thermal characteristics of the second production link to obtain the thermal characteristics of the third production link; Crossing the thermal characteristics of the first production link and the thermal characteristics of the third production link to obtain a seventh traceability attribute fusion feature; Based on the seventh traceability attribute fusion feature, the consistency debugging error of the traceability supervision decision algorithm is obtained.
3. The method according to claim 1, characterized in that The method of debugging the traceability supervision decision algorithm based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain the debugged traceability supervision decision algorithm includes: Weighting the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm to obtain a global debugging error of the traceability supervision decision algorithm; In the direction of converging the global debugging error, the algorithm variables of the traceability supervision decision algorithm are cyclically improved to obtain the debugged traceability supervision decision algorithm.
4. The method according to claim 1, characterized in that The index features of the blockchain traceability node and the first semantic embedding vector are recorded in a preset feature space in the form of feature binary tuples, and the index features of the product quality traceability information and the second semantic embedding vector are recorded in the preset feature space in the form of feature binary tuples; The obtaining of multiple smart IoT sensor data includes: Based on the index feature of the blockchain traceability node, obtaining the first semantic embedding vector from the preset feature space; Based on the index feature of the product quality traceability information, acquiring the second semantic embedding vector from the preset feature space; Combining the first semantic embedding vector and the second semantic embedding vector, acquiring the smart IoT sensor data; The method further comprises: For each of the smart IoT sensor data, obtaining a first life cycle state vector in a first semantic embedding vector included in the smart IoT sensor data; Combining the first life cycle state vector and the first semantic embedding vector, obtaining a negative debugging example of the blockchain traceability node corresponding to the smart IoT sensor data; Based on the negative debugging examples of the blockchain traceability nodes corresponding to the respective smart IoT sensor data, obtaining the second traceability attribute description features corresponding to the respective smart IoT sensor data; The method further comprises: Using the debugged traceability supervision decision algorithm, obtaining the traceability item identification text of the blockchain traceability node for the product quality traceability information; Furthermore, based on the blockchain traceability node, the traceability item identification text of the product quality traceability information is output to output quality traceability opinion comments.
5. An intelligent manufacturing quality data traceability system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 4.
6. An intelligent manufacturing quality data traceability device, characterized in that: The intelligent manufacturing quality data traceability equipment is used for: Acquire multiple smart IoT sensor data, wherein the smart IoT sensor data includes a first semantic embedding vector of a blockchain traceability node and a second semantic embedding vector of product quality traceability information, wherein the first semantic embedding vector includes at least one life cycle state vector associated with the blockchain traceability node, and the second semantic embedding vector includes at least one life cycle state vector associated with the product quality traceability information; For each of the smart IoT sensor data, based on the first semantic embedding vector and the second semantic embedding vector included in the smart IoT sensor data, obtaining a traceability attribute description feature corresponding to the smart IoT sensor data; Determine the traceability project identification text corresponding to each of the smart IoT sensor data based on the traceability attribute description features corresponding to each of the smart IoT sensor data through a traceability supervision decision algorithm, wherein the traceability project identification text represents the recognition probability of the traceability project text; Based on the traceability attribute description features, traceability item identification texts and traceability item prior texts respectively corresponding to each of the smart IoT sensor data, the traceability supervision decision algorithm is debugged to obtain a debugged traceability supervision decision algorithm, wherein the debugged traceability supervision decision algorithm is used to obtain the traceability item identification text of the blockchain traceability node for the product quality traceability information, and the traceability item prior text represents the prior learning annotation of the traceability item text; The life cycle state vector is used to represent the blockchain supervision node corresponding to the product quality traceability information, and the index feature in the entire life cycle corresponding to the blockchain supervision node; The traceability attribute description features corresponding to the smart IoT sensor data include the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector; The traceability supervision decision algorithm is debugged based on the traceability attribute description features, traceability item identification text and traceability item prior text corresponding to each of the smart IoT sensor data to obtain a debugged traceability supervision decision algorithm, including: Based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data, a first traceability attribute description feature and a second traceability attribute description feature corresponding to the smart IoT sensor data are generated, wherein the first traceability attribute description feature represents the traceability attribute description feature of the active debugging example corresponding to the blockchain traceability node, and the second traceability attribute description feature represents the traceability attribute description feature of the passive debugging example corresponding to the blockchain traceability node; Based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data, a consistency debugging error of the traceability supervision decision algorithm is obtained, wherein the consistency debugging error is used to indicate the performance of the traceability supervision decision algorithm in mastering the content corresponding to the second traceability attribute description feature; Based on the traceability project identification text and the traceability project priori text corresponding to each of the smart IoT sensor data, a discrimination debugging error of the traceability supervision decision algorithm is obtained, and the discrimination debugging error is used to indicate the decision offset weight of the traceability supervision decision algorithm for the traceability project text; Based on the consistency debugging error and the discrimination debugging error of the traceability supervision decision algorithm, the traceability supervision decision algorithm is debugged to obtain the debugged traceability supervision decision algorithm; The first semantic embedding vector includes a first production line semantic embedding vector and a second production line semantic embedding vector, the life cycle state vector in the first production line semantic embedding vector has a matching relationship with the blockchain traceability node, the second production line semantic embedding vector includes the first production line semantic embedding vector, and the second production line semantic embedding vector contains a life cycle state vector that has a non-matching relationship with the blockchain traceability node; the second semantic embedding vector includes a third production line semantic embedding vector and a fourth production line semantic embedding vector, the life cycle state vector in the third production line semantic embedding vector has a matching relationship with the product quality traceability information, and the fourth production line semantic embedding vector is used to represent the access warning record of the product quality traceability information; The generating of the first traceability attribute description feature and the second traceability attribute description feature corresponding to the smart IoT sensor data based on the traceability attribute description features of the first semantic embedding vector and the traceability attribute description features of the second semantic embedding vector respectively corresponding to each of the smart IoT sensor data includes: For each of the smart IoT sensor data, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the second production line semantic embedding vector corresponding to the smart IoT sensor data; or, the first traceability attribute description feature is generated based on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data; Performing feature derivation on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the second production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description feature; or, performing feature derivation on the traceability attribute description feature of the first production line semantic embedding vector and the traceability attribute description feature of the third production line semantic embedding vector corresponding to the smart IoT sensor data to generate the second traceability attribute description feature; The traceability attribute description feature corresponding to the smart IoT sensor data includes a first production line traceability attribute description feature and a second production line traceability attribute description feature, wherein the first production line traceability attribute description feature and the second production line traceability attribute description feature are both used to represent the smart IoT sensor data, and the first production line traceability attribute description feature and the second production line traceability attribute description feature correspond to the same traceability feature relationship network; The step of obtaining the consistency debugging error of the traceability supervision decision algorithm based on the first traceability attribute description feature and the second traceability attribute description feature respectively corresponding to each of the smart IoT sensor data includes: Obtain the consistency debugging error based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature or the second production line traceability attribute description feature respectively corresponding to each of the smart IoT sensor data; Alternatively, for each of the smart IoT sensor data, a first traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a first traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a first traceability attribute combination feature corresponding to the smart IoT sensor data, and a second traceability attribute description feature generated by the first production line traceability attribute description feature of the smart IoT sensor data is combined with a second traceability attribute description feature generated by the second production line traceability attribute description feature of the smart IoT sensor data to obtain a second traceability attribute combination feature corresponding to the smart IoT sensor data; based on the first traceability attribute combination feature and the second traceability attribute combination feature respectively corresponding to each of the smart IoT sensor data, the consistency debugging error is obtained; Alternatively, a first local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the first production line traceability attribute description feature corresponding to each of the smart IoT sensor data, and a second local consistency debugging error is obtained based on the first traceability attribute description feature and the second traceability attribute description feature generated by the second production line traceability attribute description feature corresponding to each of the smart IoT sensor data; the first local consistency debugging error and the second local consistency debugging error are weighted to obtain the consistency debugging error.
7. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by a processor, the method according to any one of claims 1 to 4 is implemented.
Citation Information
Patent Citations
Manufacturing industry cross-domain heterogeneous data identity construction and tracing method based on block chain
CN110084070A
Agricultural product quality control traceability platform based on block chain
CN111680865A
Medicine tracing system and method based on data intelligence
CN116777295A
Data credibility traceability method based on block chain
CN117520945A
Quality traceability system and method based on industrial intelligent model
CN117934017A
