Intelligent engineering automatic production process quality tracing system and tracing method

Abstract

The invention discloses an intelligent engineering automatic production process quality tracing system and method, and relates to the technical field of intelligent engineering automatic production. According to the method, the data is preprocessed by using the edge computing node and the Kalman filtering algorithm, noise interference is effectively removed, and the accuracy and the real-time performance of the data are ensured; meanwhile, based on a distributed storage system and a block chain technology, data is subjected to fragmentation storage, digital signatures are added, signature consistency of data blocks is verified through an intelligent contract, a trusted data set is generated, and the problems that in a traditional tracing system, data is prone to being tampered, and safety is insufficient are solved; besides, feature extraction and quality fluctuation prediction are performed on the data by using a deep learning model and a time sequence analysis algorithm, and production parameters are dynamically optimized through a reinforcement learning model, so that early warning and process optimization of quality problems are realized, and the whole-process efficient management and accurate control capability of enterprises in a complex production environment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent engineering automated production, and in particular to an intelligent engineering automated production process quality tracing system and tracing method. Background Art

[0002] Intelligent engineering automated production is a core area of transformation and upgrading in the modern manufacturing industry. Its importance lies in significantly improving production efficiency, product quality, and resource utilization, and is directly related to the competitiveness and sustainable development of enterprises. In this field, accurate data collection, process traceability, and real-time quality monitoring are key to promoting technological progress. However, existing methods still have significant limitations in practical applications. Traditional production data collection relies heavily on manual recording or low-precision equipment, which can easily lead to data distortion and lags. Traceability systems often use centralized databases, which are prone to data tampering and lack security. At the same time, the ability to analyze massive amounts of data is limited, making it difficult to achieve early warning of quality issues and process optimization. These shortcomings often make it difficult for enterprises to achieve efficient management and precise control of the entire process when faced with complex production environments. Summary of the Invention

[0003] The purpose of the present invention is to provide a quality traceability system and traceability method for the automated production process of intelligent engineering, which solves the problems of traditional methods in data distortion, insufficient security and limited analysis capabilities, and significantly improves the quality traceability and management efficiency of the automated production process of intelligent engineering.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] This application provides an intelligent engineering automated production process quality traceability system, including:

[0006] The data acquisition and preprocessing module uses a sensor network to acquire parameters such as temperature, pressure, and speed during the production process. It uses high-frequency sampling technology to record the raw data in real time, generating a first data set containing a timestamp. Edge computing nodes are then used to preprocess the data in the first data set, removing noise interference through a filtering algorithm to generate a cleaned second data set.

[0007] The data storage and verification module, based on a distributed storage system, uploads the data shards in the second data set to the blockchain network. During the upload process, a digital signature is added to each data block to generate a third data set with tamper-proof characteristics. At the same time, the smart contract is used to verify the consistency of the signatures of the data blocks. When the consistency is consistent, it is confirmed that the data has not been tampered with, and a reliable fourth data set is obtained.

[0008] The data feature extraction module inputs the fourth data set, extracts features of parameter change trends through a deep learning model, obtains key patterns related to quality, and obtains a fifth data set containing potential rules;

[0009] The quality fluctuation prediction module uses a time series analysis algorithm to analyze the quality-related features in the fifth data set and predict the quality fluctuation. When the predicted value exceeds a preset threshold, an abnormal flag is triggered, and a sixth data set with warning information is obtained;

[0010] The parameter adjustment strategy optimization module uses a reinforcement learning model to iteratively optimize the production parameter adjustment strategy based on the sixth data set to generate a seventh data set that is updated in real time;

[0011] a production equipment dynamic adjustment module, which obtains the optimized parameter configuration from the seventh data set, dynamically adjusts the production equipment through the industrial control system, and obtains an eighth data set reflecting the adjustment results;

[0012] The quality monitoring report generation module uses data visualization technology to process the eighth data set and generate an intuitive quality monitoring report. At the same time, it determines whether the production process meets the preset standards based on the report content to obtain the final ninth data set.

[0013] Furthermore, the noise interference is removed by a filtering algorithm to obtain a cleaned second data set, which specifically includes:

[0014] The Kalman filter algorithm is used to construct a mathematical model based on the dynamic characteristics and statistical laws of the data. For each incoming data point, the reasonable value range of the data point is predicted through iterative calculation with reference to the data of several time points before and after it. The actual value of the current data point is then compared with the predicted value. If the value of the current data point deviates from the predicted range by more than the preset noise threshold, the data point is determined to be noise data and is removed from the data set to remove noise interference. The valid data with noise removed is reorganized and classified through edge computing nodes, classified by parameter type, and a cleaned second data set is generated according to specific data format and storage rules.

[0015] Furthermore, the data storage and verification module includes:

[0016] Performing a data sharding operation on the second data set through a distributed storage system to generate multiple data blocks, and then uploading the sharded data blocks to the blockchain network to obtain the storage location of each data block;

[0017] When uploading, a digital signature is added to each data block to obtain a third data set with tamper-proof characteristics;

[0018] Then, a smart contract is used to verify the signature of the data block in the third data set to determine the consistency of the signature. If the signature consistency is verified, it is confirmed that the data block has not been tampered with, and a fourth data set is generated;

[0019] The data blocks in the fourth data set are integrity checked by a hash algorithm to determine the trusted data. Based on the verification results, the trusted data is obtained and recorded in the blockchain network to obtain the final trusted data set.

[0020] Furthermore, the digital signature is generated using an asymmetric encryption algorithm, and each data block is encrypted and signed using the data owner's private key, and decrypted and verified using the corresponding public key during the verification phase; wherein, the verification phase automatically traverses and checks the validity of the signatures of all data blocks through smart contracts, and records the verification results in the blockchain non-tampering log. When inconsistent signatures are found, an alarm mechanism is triggered and abnormal data blocks are marked.

[0021] Furthermore, the data feature extraction module includes:

[0022] Processing the fourth data set through a deep learning model to extract trend features of parameter changes and obtain a feature set;

[0023] Use convolutional neural networks to analyze feature sets, obtain key patterns related to quality, and determine the pattern feature set;

[0024] For the pattern feature set, when the number of key patterns exceeds the preset threshold, the clustering algorithm is used to group them and obtain a grouped pattern set;

[0025] Based on the grouping pattern set, the distribution characteristics of each group's potential rules are calculated to obtain the rule feature set. Through the rule feature set, an intermediate data set containing the potential rules is generated to determine the coverage of the rules.

[0026] A fully connected neural network is used to process the intermediate data set, optimize the regular feature set, and generate the fifth data set.

[0027] Furthermore, the quality fluctuation prediction module includes:

[0028] The fifth dataset is used to obtain time series data, and the ARIMA algorithm is used to analyze quality fluctuations to obtain fluctuation prediction values. When the fluctuation prediction value exceeds the preset threshold, an abnormal flag is triggered, and a marked temporary dataset is obtained.

[0029] Extract outliers from the marked temporary data set, use clustering algorithm to determine the distribution of outliers, and obtain the outlier distribution characteristics;

[0030] According to the abnormal distribution characteristics and the early warning information template matching, an updated data set with early warning information is generated.

[0031] By updating the data set and calculating the changing trend of quality fluctuations, we can obtain trend analysis results. If the trend analysis results show persistent anomalies, we can adjust the time series parameters to obtain optimized forecast values.

[0032] The optimized prediction value is used to update the sixth data set to obtain the final data set with warning information.

[0033] Furthermore, the parameter adjustment strategy optimization module includes: obtaining production parameter information through the sixth data set, performing model training using a reinforcement learning model to obtain a preliminary adjustment strategy, extracting feature data from the sixth data set, performing iterative optimization through reinforcement learning, and determining an optimized adjustment strategy;

[0034] When the optimized adjustment strategy reaches the preset threshold, the production parameters are adjusted according to the adjustment strategy to obtain an adjusted parameter set;

[0035] Based on the adjusted parameter set and combined with the historical information of the sixth data set, real-time updated data content is generated through data generation technology;

[0036] Obtain real-time updated data content and use pre-established verification mechanisms to determine whether the data content meets production parameter requirements;

[0037] When the data content meets the requirements, the data content is integrated into the seventh data set through a real-time update mechanism to complete data generation. Then, information processing technology is used to extract change trends and determine the adjustment direction of subsequent production parameters.

[0038] Furthermore, the production equipment dynamic adjustment module includes:

[0039] Extracting optimized parameters from the seventh data set, determining parameter integrity using a preset threshold, and obtaining a usable parameter configuration. The parameter configuration is then loaded through the industrial control system, dynamically adjusting the production equipment and generating equipment status data.

[0040] Obtaining device status data, using a support vector machine algorithm to analyze the adjustment process to determine whether the adjustment is stable. If the adjustment is stable, generating an eighth data set from the device status data to reflect the adjustment results. If the adjustment is unstable, reloading the optimization parameters through system control to obtain new device status data;

[0041] The extraction results of the eighth data set reflect the characteristics, and the K-means algorithm is used for cluster analysis to determine the adjustment trend. The seventh data set is updated through the adjustment trend to generate new optimization parameters.

[0042] Furthermore, the quality monitoring report generation module includes:

[0043] By loading the eighth data set, using data cleaning technology to remove outliers, obtaining processed data, and using data visualization technology to graphically display the processed data, monitoring information is generated;

[0044] Based on the monitoring information, key indicators of the production process are obtained and the indicator change trend is determined. When the indicator change trend exceeds the preset standard, the abnormal point is marked through logical judgment to obtain abnormal identification data;

[0045] Based on the abnormal identification data, a clustering algorithm is used to divide the data categories and generate classification results. By comparing the classification results with the preset standards, it is determined whether the production process meets the standards and the standard status is obtained. At the same time, a data set that meets the standard status conditions is obtained to generate the ninth data set.

[0046] The present invention provides a quality tracing method for an intelligent engineering automated production process, which is applied to an intelligent engineering automated production process quality tracing system, and includes the following steps:

[0047] The temperature, pressure, speed, and other parameters of the production process are acquired through a sensor network. High-frequency sampling technology is used to record the raw data in real time to obtain a first data set containing a timestamp. The data is then pre-processed using edge computing nodes and noise interference is removed through filtering algorithms to obtain a cleaned second data set.

[0048] Based on the second data set, the data is uploaded to the blockchain network in shards using a distributed storage system. A digital signature is added to each data block during upload to generate a third data set with tamper-proof properties. The full process parameters are obtained from the third data set, and the signature consistency of the data blocks is verified using a smart contract. If consistent, it is confirmed that the data has not been tampered with, resulting in a trusted fourth data set.

[0049] Based on the fourth data set, a deep learning model is used to extract features of parameter change trends, obtain key quality-related patterns, and generate a fifth data set containing potential rules;

[0050] Based on the fifth data set, a time series analysis algorithm is used to predict quality fluctuations. If the predicted value exceeds a preset threshold, an abnormal flag is triggered, resulting in a sixth data set with warning information.

[0051] Using the sixth data set, the production parameter adjustment strategy is iteratively optimized using a reinforcement learning model to generate a seventh data set that is updated in real time. The optimized parameter configuration is obtained from the seventh data set, and the production equipment is dynamically adjusted through the industrial control system to obtain an eighth data set reflecting the adjustment results.

[0052] The eighth data set is processed using data visualization technology to generate an intuitive quality monitoring report. The content of the report is then used to determine whether the production process meets the preset standards, resulting in the final ninth data set.

[0053] The beneficial effects of the present invention are:

[0054] Through sensor networks and high-frequency sampling technology, the system can acquire key parameters such as temperature, pressure, and speed during the production process in real time, generating a first data set with a timestamp. Using edge computing nodes and a Kalman filter algorithm, the system preprocesses the first data set to remove noise interference and generate a cleaned second data set. This effectively solves the data distortion and lag problems caused by traditional production data collection relying on manual recording or low-precision equipment, ensuring data accuracy and real-time availability, and providing high-quality basic data for subsequent quality traceability and analysis.

[0055] A distributed storage system is used to shard the second data set, upload the data blocks to the blockchain network, and digitally sign each data block to generate a tamper-proof third data set. The smart contract verifies the signature consistency of the data blocks, confirming that the data has not been tampered with, and then generates a trusted fourth data set. Furthermore, the system uses a hash algorithm to verify the integrity of the data blocks and records the trusted data on the blockchain network. This overcomes the issues of centralized database data being easily tampered with and lacking security in traditional traceability systems, ensuring the data's immutability and credibility, and providing a solid technical guarantee for quality traceability.

[0056] Based on deep learning models and time series analysis algorithms, the system extracts features from parameter change trends in the fourth data set and predicts quality fluctuations. When the predicted value exceeds the preset threshold, an anomaly mark is triggered and a data set with early warning information is generated. The production parameter adjustment strategy is further iteratively optimized through the reinforcement learning model, and the production equipment is dynamically adjusted through the industrial control system. This process not only achieves early warning of quality problems, but also improves production efficiency and product quality by dynamically optimizing production parameters. It solves the shortcomings of traditional methods in terms of limited massive data analysis capabilities and difficulty in achieving early warning of quality problems and process optimization, and enhances the company's full-process efficient management and precise control capabilities in complex production environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0058] Figure 1 A schematic diagram of the structure of the intelligent engineering automated production process quality traceability system provided in Example 1 of the present application;

[0059] Figure 2A flow chart of the method for tracing quality of an automated production process in intelligent engineering provided in Example 2 of the present application. DETAILED DESCRIPTION

[0060] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present application. Rather, they are merely examples of methods and systems consistent with certain aspects of the present application, as detailed in the appended claims.

[0061] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0062] The following describes in detail the specific implementation methods, features and effects of the present invention in conjunction with the accompanying drawings and preferred embodiments.

[0063] Example 1

[0064] See also Figure 1 This embodiment provides an intelligent engineering automated production process quality traceability system, including:

[0065] The data acquisition and preprocessing module uses a sensor network to acquire parameters such as temperature, pressure, and speed during the production process. The sensor network is distributed at key locations on production equipment to sense the production environment and equipment operating status in real time. High-frequency sampling technology is used to record raw data in real time to ensure data integrity and timeliness. After data acquisition is completed, a first data set containing a timestamp is generated, providing a basic data source for subsequent data processing. The edge computing node is then used to preprocess the data in the first data set, removing noise interference through a filtering algorithm, effectively reducing the noise component in the data and improving data quality. After preprocessing, a cleaned second data set is obtained, providing more accurate and reliable data for further analysis and storage.

[0066] Furthermore, the noise interference is removed by a filtering algorithm to obtain a cleaned second data set, which specifically includes:

[0067] The Kalman filter algorithm is used to construct a mathematical model based on the dynamic characteristics and statistical laws of the data. For each incoming data point, the reasonable value range of the data point is predicted through iterative calculation with reference to the data of several time points before and after it. The actual value of the current data point is then compared with the predicted value. If the value of the current data point deviates from the predicted range by more than the preset noise threshold, the data point is determined to be noise data and is removed from the data set to remove noise interference. The valid data with noise removed is reorganized and classified through edge computing nodes, and classified by parameter type (temperature, pressure, speed, etc.). A cleaned second data set is generated according to specific data format and storage rules.

[0068] The data storage and verification module, based on a distributed storage system, uploads the data shards in the second dataset to the blockchain network. During the upload process, a digital signature is added to each data block to ensure the uniqueness and immutability of the data, generating a third dataset with tamper-proof properties. At the same time, the signature consistency of the data blocks is verified using smart contracts. When consistent, it is confirmed that the data has not been tampered with, resulting in a reliable fourth dataset, which provides a reliable data foundation for subsequent quality analysis.

[0069] Furthermore, the data storage and verification module includes:

[0070] Performing a data sharding operation on the second data set through a distributed storage system to generate multiple data blocks, and then uploading the sharded data blocks to the blockchain network to obtain the storage location of each data block;

[0071] When uploading, a digital signature is added to each data block to obtain a third data set with tamper-proof characteristics;

[0072] Then, a smart contract is used to verify the signature of the data block in the third data set to determine the consistency of the signature. If the signature consistency is verified, it is confirmed that the data block has not been tampered with, and a fourth data set is generated;

[0073] The data blocks in the fourth data set are integrity checked by a hash algorithm to determine the trusted data. Based on the verification results, the trusted data is obtained and recorded in the blockchain network to obtain the final trusted data set.

[0074] Specifically, a distributed storage system performs data sharding on the second data set, using data blocks of a fixed size of 64MB to generate multiple data blocks. The sharded data blocks are uploaded to the blockchain network, and the IPFS protocol is used to obtain the storage location of each data block, for example, the CID (content identifier) is QmXoypizjW3WknFiJnKLwHCnL72vedxjQkDDP1mXWo6uco. A digital signature is added to the uploaded data block, and a 2048-bit digital signature is generated using the RSA algorithm to obtain a third data set with tamper-proof properties. The signatures of the data blocks in the third data set are verified using a smart contract. The contract code written in Solidity calls the RSA verification function to determine the consistency of the signatures. If the signature consistency is verified, the verification result confirms that the data block has not been tampered with, and a fourth data set is generated. The data blocks in the fourth data set are integrity checked using a hash algorithm. The hash value is calculated using the SHA-256 algorithm and compared with the original hash value to determine the trusted data. Based on the verification results, trusted data is extracted and recorded on the blockchain network. Ethereum smart contracts are used to store the data on-chain, resulting in a final trusted dataset. Information processing techniques are used to extract features from the final trusted dataset, and principal component analysis (PCA) is used for dimensionality reduction to obtain correlation information for the data blocks. This correlation information is combined with the storage location of the distributed storage system, such as the mapping between IPFS CIDs and blockchain addresses, to determine the complete storage path of the data blocks.

[0075] Furthermore, the digital signature is generated using an asymmetric encryption algorithm, and each data block is encrypted and signed using the data owner's private key, and decrypted and verified using the corresponding public key during the verification phase; wherein, the verification phase automatically traverses and checks the validity of the signatures of all data blocks through smart contracts, and records the verification results in the blockchain non-tampering log. When inconsistent signatures are found, an alarm mechanism is triggered and abnormal data blocks are marked.

[0076] The data feature extraction module takes the fourth data set as input and uses a deep learning model to extract features of parameter change trends. The deep learning model can automatically learn complex patterns and regularities in the data, thereby obtaining key quality-related patterns and obtaining a fifth data set containing potential regularities, providing key feature information for quality fluctuation prediction.

[0077] Furthermore, the data feature extraction module includes:

[0078] Processing the fourth data set through a deep learning model to extract trend features of parameter changes and obtain a feature set;

[0079] Use convolutional neural networks to analyze feature sets, obtain key patterns related to quality, and determine the pattern feature set;

[0080] For the pattern feature set, when the number of key patterns exceeds the preset threshold, the clustering algorithm is used to group them and obtain a grouped pattern set;

[0081] Based on the grouping pattern set, the distribution characteristics of each group's potential rules are calculated to obtain the rule feature set. Through the rule feature set, an intermediate data set containing the potential rules is generated to determine the coverage of the rules.

[0082] A fully connected neural network is used to process the intermediate data set, optimize the regular feature set, and generate the fifth data set.

[0083] Specifically, the fourth data set is processed using a deep learning model to extract trend features of parameter changes. For example, a long short-term memory network (LSTM) is used to train time series data to extract the trend of parameters such as temperature and pressure, thereby obtaining an initial feature set containing trend features. A convolutional neural network is used to analyze the initial feature set, such as by performing a convolution operation on the features using a two-dimensional convolutional layer, to extract key patterns related to quality, such as abnormal fluctuations or periodic changes, and determine a pattern feature set. Based on the pattern feature set, the distribution characteristics of the key patterns are calculated using statistical methods, such as calculating the frequency of occurrence and distribution range of each pattern to obtain pattern distribution data. If the number of key patterns in the pattern distribution data exceeds a preset threshold (e.g., 10), the patterns are grouped using a clustering algorithm, such as using K-means clustering to classify similar patterns, to obtain a grouped pattern set. Based on the grouped pattern set, the distribution characteristics of each group of potential patterns are calculated, such as calculating the mean and variance of each group of patterns to obtain a pattern feature set. Using the pattern feature set, an intermediate data set containing the potential patterns is generated, such as by mapping the pattern features to the original data to determine whether the pattern coverage exceeds 80%. A fully connected neural network is used to process the intermediate dataset, for example, using three fully connected layers to optimize features and generate an optimized feature set. This optimized feature set is then combined with a time series analysis algorithm to extract trend information, such as using an ARIMA model to predict future quality fluctuations, generating a fifth dataset. Information processing techniques are then used to verify data integrity using this fifth dataset, for example, by checking that the missing data rate is less than 5%, before finalizing the final dataset.

[0084] The quality fluctuation prediction module uses a time series analysis algorithm to analyze the quality-related features in the fifth data set and predict quality fluctuations. When the predicted value exceeds the preset threshold, an abnormal flag is triggered, identifying potential quality issues in a timely manner. The sixth data set with early warning information is generated to provide a basis for adjusting production parameters.

[0085] Furthermore, the quality fluctuation prediction module includes:

[0086] The fifth dataset is used to obtain time series data, and the ARIMA algorithm is used to analyze quality fluctuations to obtain fluctuation prediction values. When the fluctuation prediction value exceeds the preset threshold, an abnormal flag is triggered, and a marked temporary dataset is obtained.

[0087] Extract outliers from the marked temporary data set, use clustering algorithm to determine the distribution of outliers, and obtain the outlier distribution characteristics;

[0088] According to the abnormal distribution characteristics and the early warning information template matching, an updated data set with early warning information is generated.

[0089] By updating the data set and calculating the changing trend of quality fluctuations, we can obtain trend analysis results. If the trend analysis results show persistent anomalies, we can adjust the time series parameters to obtain optimized forecast values.

[0090] The optimized prediction value is used to update the sixth data set to obtain the final data set with warning information.

[0091] Production parameter information is extracted from the sixth data set, and preliminary training is performed using a reinforcement learning model to obtain a preliminary adjustment strategy. Feature data is extracted from the sixth data set based on the preliminary adjustment strategy, and iterative optimization is performed through reinforcement learning to determine the optimized adjustment strategy.

[0092] Specifically, the fifth data set is used to obtain time series data, and the ARIMA algorithm is used to analyze quality fluctuations. The parameter combination of p=2, d=1, and q=1 is set to calculate the quality fluctuation forecast value for the next 24 hours. The fluctuation forecast value is compared with the preset threshold value (such as ±0.5). If the forecast value exceeds the threshold, the abnormality mark is triggered, and a temporary data set containing the abnormal timestamp and fluctuation value is generated. The abnormal points are extracted from the temporary data set, and the K-means clustering algorithm is used. With k=3, the spatial distribution characteristics of the abnormal points are analyzed to obtain high-density and low-density areas of the abnormal points. According to the abnormal distribution characteristics, the warning information template is matched, and the high-density areas are marked as "high risk" and the low-density areas are marked as "low risk". An updated data set with warning information is generated. The changing trend of the quality fluctuation is calculated through the updated data set, and the slope and intercept of the fluctuation trend are obtained by linear regression analysis to determine whether the trend continues to rise or fall. If the trend analysis results show a slope greater than 0.1 and persist for three time periods, the ARIMA model parameters are adjusted, the p-value is adjusted to 3, and the optimized forecast value is recalculated. The optimized forecast value is used to update the updated dataset with warning information, generating a sixth dataset that ultimately contains the optimized forecast value and warning information. Production parameter information is extracted from the sixth dataset, and preliminary training is performed using a Q-learning reinforcement learning model with a learning rate of α = 0.1 and a discount factor of γ = 0.9 to obtain a preliminary adjustment strategy. Based on the preliminary adjustment strategy, feature data is extracted from the sixth dataset, and iterative optimization is performed using a deep Q network (DQN) with 1000 iterations to determine the optimized adjustment strategy.

[0093] The parameter adjustment strategy optimization module uses a reinforcement learning model to iteratively optimize the production parameter adjustment strategy based on the sixth data set. The reinforcement learning model generates a real-time updated seventh data set through continuous learning and optimization.

[0094] Furthermore, the parameter adjustment strategy optimization module includes: obtaining production parameter information through the sixth data set, performing model training using a reinforcement learning model to obtain a preliminary adjustment strategy, extracting feature data from the sixth data set, performing iterative optimization through reinforcement learning, and determining an optimized adjustment strategy;

[0095] When the optimized adjustment strategy reaches the preset threshold, the production parameters are adjusted according to the adjustment strategy to obtain an adjusted parameter set;

[0096] Based on the adjusted parameter set and combined with the historical information of the sixth data set, real-time updated data content is generated through data generation technology;

[0097] Obtain real-time updated data content and use pre-established verification mechanisms to determine whether the data content meets production parameter requirements;

[0098] When the data content meets the requirements, the data content is integrated into the seventh data set through a real-time update mechanism to complete data generation. Then, information processing technology is used to extract change trends and determine the adjustment direction of subsequent production parameters.

[0099] Specifically, the production parameter information is extracted from the sixth data set, and the deep Q network (DQN) algorithm is used for model training. The initial learning rate is set to 0.001, the batch size is 64, and a preliminary adjustment strategy is obtained after 1000 iterations. According to the preliminary adjustment strategy, key feature data are screened from the sixth data set, including parameters such as temperature, pressure, and speed. The proximal policy optimization (PPO) algorithm is used for policy iteration optimization, and the reward function is set as the percentage of production yield improvement. The yield improvement threshold of the optimized strategy is set to 5%. If the optimized adjustment strategy reaches the threshold, the production parameters are adjusted based on the strategy output, for example, the temperature setting value is adjusted from 150°C to 155°C, and the pressure is adjusted from 2.5MPa to 2.7MPa, to generate an adjusted parameter set. Combined with the sixth data set, Historical information, using a generative adversarial network (GAN) model to generate real-time updated data content, inputting 1,000 historical data and outputting 500 simulated data; through a preset verification mechanism, the mean square error (MSE) of the generated data is calculated to be less than 0.01, and the data is judged to meet the requirements, and the data is integrated into the seventh data set; principal component analysis (PCA) is used to extract the change trend from the seventh data set, and the first three principal components are selected to explain 85% of the variance, and the subsequent adjustment direction is determined to be to reduce the speed by 5%; the optimized parameter configuration is loaded from the seventh data set, and the missing rate threshold is set to be less than 1% when checking the parameter integrity, and the parameters are sent to the equipment through the industrial control system; based on the sixth data set, the asynchronous advantage actor-critic (A3C) algorithm is used for strategy optimization to update the parameter configuration in the seventh data set.

[0100] The production equipment dynamic adjustment module obtains the optimized parameter configuration from the seventh data set and dynamically adjusts the production equipment through the industrial control system. Based on the optimized parameter configuration, the operating parameters of the production equipment are adjusted in real time to ensure the stability of the production process and product quality. The eighth data set reflecting the adjustment results is obtained to provide data support for the generation of quality monitoring reports.

[0101] Furthermore, the production equipment dynamic adjustment module includes:

[0102] Extracting optimized parameters from the seventh data set, determining parameter integrity using a preset threshold, and obtaining a usable parameter configuration. The parameter configuration is then loaded through the industrial control system, dynamically adjusting the production equipment and generating equipment status data.

[0103] Obtaining device status data, using a support vector machine algorithm to analyze the adjustment process to determine whether the adjustment is stable. If the adjustment is stable, generating an eighth data set from the device status data to reflect the adjustment results. If the adjustment is unstable, reloading the optimization parameters through system control to obtain new device status data;

[0104] The extraction results of the eighth data set reflect the characteristics, and the K-means algorithm is used for cluster analysis to determine the adjustment trend. The seventh data set is updated through the adjustment trend to generate new optimization parameters.

[0105] Obtain new optimization parameters, extract change trends through information processing technology, and determine the direction of parameter adjustment. Based on the parameter adjustment direction and combined with the historical information of the sixth data set, data generation technology is used to generate real-time updated data content.

[0106] Specifically, the optimized parameter configuration is extracted from the seventh data set, such as key parameters such as the temperature set point (200°C ± 5) and the pressure threshold (1.2 MPa). Parameter integrity is determined using preset integrity rules (e.g., missing rate <5%), and a complete parameter set is selected. The parameter configuration is loaded into the industrial control system, such as writing the optimized PID control parameters (proportional coefficient Kp = 2.5, integral time Ti = 10s) to the PLC, driving the production line to perform dynamic adjustments and collecting real-time status data such as equipment vibration (0.05 mm / s) and current (15 A ± 0.3 A). A support vector machine algorithm (with RBF kernel function and penalty coefficient C = 1.0) is used to analyze the equipment status data and calculate the stability index of the adjustment process (e.g., standard deviation < 0.1). If the stability meets the standard, the equipment status data is integrated to generate the eighth data set, which includes indicators such as the adjusted efficiency (95% ± 2%) and energy consumption (120 kWh). If the stability does not meet the standard, the optimized parameters are reloaded, such as adjusting the PID parameters (Kp = 3.0, Ti = 8s) and collecting new status data. Based on the eighth data set, features (such as the efficiency change rate and energy consumption slope) were extracted, and the K-means algorithm (cluster number k=3, Euclidean distance) was used to analyze the adjustment trend, and three modes were divided into "rapid convergence", "fluctuating convergence" and "divergence". The seventh data set was updated according to the clustering results. For example, the parameters corresponding to the "rapid convergence" mode (Kp=2.8, Ti=9s) were marked as priority recommendations; the parameter change trend was analyzed through information processing technology (such as the sliding window mean method) to determine the direction of the next round of adjustments (such as lowering the Kp value). Combined with the historical parameters of the sixth data set (the last 10 adjustment records), the LSTM model (hidden layer units=64) was used to generate a real-time updated parameter combination, for example, predicting that the optimal temperature setting value for the next cycle is 205°C.

[0107] The quality monitoring report generation module uses data visualization technology to process the eighth data set and generate an intuitive quality monitoring report. It displays the quality data and adjustment results of the production process in the form of charts and curves, allowing operators to quickly understand the production status. At the same time, based on the report content, it can determine whether the production process meets the preset standards. The final ninth data set is obtained to provide a basis for production decision-making;

[0108] Furthermore, the quality monitoring report generation module includes:

[0109] By loading the eighth data set, using data cleaning technology to remove outliers, obtaining processed data, and using data visualization technology to graphically display the processed data, monitoring information is generated;

[0110] Based on the monitoring information, key indicators of the production process are obtained and the indicator change trend is determined. When the indicator change trend exceeds the preset standard, the abnormal point is marked through logical judgment to obtain abnormal identification data;

[0111] Based on the abnormal identification data, a clustering algorithm is used to divide the data categories and generate classification results. By comparing the classification results with the preset standards, it is determined whether the production process meets the standards and the standard status is obtained. At the same time, a data set that meets the standard status conditions is obtained to generate the ninth data set.

[0112] Specifically, the eighth data set was loaded and a data cleaning technique based on the 3σ principle was used to remove outliers exceeding ±3 standard deviations from the mean. This generated processed data. The Matplotlib library was used to plot line and box plots of the processed data, generating monitoring information for temperature, pressure, and speed parameters. Based on this monitoring information, temperature fluctuation amplitude, pressure peak, and speed stability were extracted as key indicators, and their 5-minute mean trends were calculated using a sliding window. If the temperature fluctuation exceeded ±2°C or the pressure peak exceeded 10 MPa, the outlier was marked using the isolation forest algorithm, generating identification data with the outlier timestamp. The K-means clustering algorithm was used to classify the anomaly identification data into three categories based on Euclidean distance, generating high / medium / low risk classification results. The classification results were compared with pre-set standards (e.g., the proportion of high-risk points was less than 5%) to determine the compliance status of the production process. Based on the compliance status, data in the temperature range of 20-30°C and the pressure range of 5-8 MPa were selected to generate the ninth data set. The sensor data of the ninth data set was collected through the Modbus protocol. The temperature, pressure, and speed were recorded at a sampling frequency of 100 Hz to obtain a validation data set with millisecond timestamps. The ARIMA model was used to perform time series analysis on the validation data set. The autocorrelation coefficient was calculated to determine whether it was stable within the range of ±0.2 at a lag of 10 orders, and the stability of the quality monitoring report was determined.

[0113] Example 2

[0114] See also Figure 2 This embodiment provides a quality tracing method for an intelligent engineering automated production process, which is applied to an intelligent engineering automated production process quality tracing system, and includes the following steps:

[0115] S1. Parameters such as temperature, pressure, and speed during the production process are acquired through a sensor network. High-frequency sampling technology is used to record the raw data in real time to obtain a first data set containing a timestamp. Edge computing nodes are then used to preprocess the data, and a filtering algorithm is used to remove noise interference to obtain a cleaned second data set.

[0116] S2. Based on the second dataset, upload the data in shards to the blockchain network using a distributed storage system. Add a digital signature to each data block during upload to generate a third dataset with tamper-proof properties. Obtain full-process parameters from the third dataset and use smart contracts to verify the consistency of the signatures of the data blocks. If consistent, confirm that the data has not been tampered with, thereby obtaining a trusted fourth dataset.

[0117] S3. Based on the fourth data set, extract the characteristics of the parameter change trend through the deep learning model, obtain the key patterns related to quality, and generate a fifth data set containing potential rules;

[0118] S4. Based on the fifth data set, a time series analysis algorithm is used to predict quality fluctuations. If the predicted value exceeds a preset threshold, an abnormal flag is triggered, thereby obtaining a sixth data set with warning information.

[0119] S5. Using the sixth data set, utilize a reinforcement learning model to iteratively optimize the production parameter adjustment strategy to generate a seventh data set that is updated in real time. The optimized parameter configuration is obtained from the seventh data set, and the production equipment is dynamically adjusted through the industrial control system to obtain an eighth data set reflecting the adjustment results.

[0120] S6. Use data visualization technology to process the eighth data set to generate an intuitive quality monitoring report. Then, based on the content of the report, determine whether the production process meets the preset standards to obtain the final ninth data set.

[0121] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. Intelligent engineering automated production process quality traceability system, characterized by: include: The data acquisition and preprocessing module acquires the temperature, pressure and speed parameters of the production process through the sensor network, uses high-frequency sampling technology to record the raw data in real time, and generates a first data set with a timestamp; The edge computing nodes are then used to preprocess the data of the first data set, and noise interference is removed through a filtering algorithm to obtain a cleaned second data set; The data storage and verification module, based on a distributed storage system, uploads the data shards in the second data set to the blockchain network. During the upload process, a digital signature is added to each data block to generate a third data set with tamper-proof characteristics. At the same time, the smart contract is used to verify the consistency of the signatures of the data blocks. When the consistency is consistent, it is confirmed that the data has not been tampered with, and the fourth data set is obtained. The data feature extraction module inputs the fourth data set, extracts features of parameter change trends through a deep learning model, obtains key patterns related to quality, and obtains a fifth data set containing potential rules; The quality fluctuation prediction module uses a time series analysis algorithm to analyze the quality-related features in the fifth data set and predict the quality fluctuation. When the predicted value exceeds a preset threshold, an abnormal flag is triggered, and a sixth data set with warning information is obtained; The parameter adjustment strategy optimization module uses a reinforcement learning model to iteratively optimize the production parameter adjustment strategy based on the sixth data set to generate a seventh data set that is updated in real time; a production equipment dynamic adjustment module, which obtains the optimized parameter configuration from the seventh data set, dynamically adjusts the production equipment through the industrial control system, and obtains an eighth data set reflecting the adjustment results; The quality monitoring report generation module uses data visualization technology to process the eighth data set and generate an intuitive quality monitoring report. At the same time, it determines whether the production process meets the preset standards based on the report content to obtain the final ninth data set.

2. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The noise interference is removed by filtering algorithm to obtain a cleaned second data set, which specifically includes: The Kalman filter algorithm is used to construct a mathematical model based on the dynamic characteristics and statistical laws of the data. For each incoming data point, the reasonable value range of the data point is predicted through iterative calculation with reference to the data of several time points before and after it. The actual value of the current data point is then compared with the predicted value. If the value of the current data point deviates from the predicted range by more than the preset noise threshold, the data point is determined to be noise data and is removed from the data set to remove noise interference. The valid data with noise removed is reorganized and classified through edge computing nodes, classified by parameter type, and a cleaned second data set is generated according to specific data format and storage rules.

3. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The data storage and verification module includes: Performing a data sharding operation on the second data set through a distributed storage system to generate multiple data blocks, and then uploading the sharded data blocks to the blockchain network to obtain the storage location of each data block; When uploading, a digital signature is added to each data block to obtain a third data set with tamper-proof characteristics; Then, a smart contract is used to verify the signature of the data block in the third data set to determine the consistency of the signature. If the signature consistency is verified, it is confirmed that the data block has not been tampered with, and a fourth data set is generated; The data blocks in the fourth data set are integrity checked using a hash algorithm to determine the trusted data. Based on the verification results, the trusted data is obtained and recorded in the blockchain network to obtain the final data set.

4. The intelligent engineering automated production process quality traceability system according to claim 3 is characterized by: The digital signature is generated using an asymmetric encryption algorithm, using the data owner's private key to encrypt and sign each data block, and decrypted and verified using the corresponding public key during the verification phase. The verification phase automatically checks the signature validity of all data blocks through smart contracts, and records the verification results in the blockchain non-tampering log. When inconsistent signatures are found, an alarm mechanism is triggered and abnormal data blocks are marked.

5. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The data feature extraction module includes: Processing the fourth data set through a deep learning model to extract trend features of parameter changes and obtain a feature set; Use convolutional neural networks to analyze feature sets, obtain key patterns related to quality, and determine the pattern feature set; For the pattern feature set, when the number of key patterns exceeds the preset threshold, the clustering algorithm is used to group them and obtain a grouped pattern set; Based on the grouping pattern set, the distribution characteristics of each group's potential rules are calculated to obtain the rule feature set. Through the rule feature set, an intermediate data set containing the potential rules is generated to determine the coverage of the rules. A fully connected neural network is used to process the intermediate data set, optimize the regular feature set, and generate the fifth data set.

6. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The quality fluctuation prediction module includes: The fifth dataset is used to obtain time series data, and the ARIMA algorithm is used to analyze quality fluctuations to obtain fluctuation prediction values. When the fluctuation prediction value exceeds the preset threshold, an abnormal flag is triggered, and a marked temporary dataset is obtained. Extract outliers from the marked temporary data set, use clustering algorithm to determine the distribution of outliers, and obtain the outlier distribution characteristics; According to the abnormal distribution characteristics and the early warning information template matching, an updated data set with early warning information is generated. By updating the data set and calculating the changing trend of quality fluctuations, we can obtain trend analysis results. If the trend analysis results show persistent anomalies, we can adjust the time series parameters to obtain optimized forecast values. The optimized prediction value is used to update the sixth data set to obtain the final data set with warning information.

7. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The parameter adjustment strategy optimization module includes: obtaining production parameter information through the sixth data set, using a reinforcement learning model to perform model training to obtain a preliminary adjustment strategy, extracting feature data from the sixth data set, performing iterative optimization through reinforcement learning, and determining an optimized adjustment strategy; When the optimized adjustment strategy reaches the preset threshold, the production parameters are adjusted according to the adjustment strategy to obtain an adjusted parameter set; Based on the adjusted parameter set and combined with the historical information of the sixth data set, real-time updated data content is generated through data generation technology; Obtain real-time updated data content and use pre-established verification mechanisms to determine whether the data content meets production parameter requirements; When the data content meets the requirements, the data content is integrated into the seventh data set through a real-time update mechanism to complete data generation. Then, information processing technology is used to extract change trends and determine the adjustment direction of subsequent production parameters.

8. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The production equipment dynamic adjustment module includes: Extracting optimized parameters from the seventh data set, determining parameter integrity using a preset threshold, and obtaining parameter configurations, which are then loaded through the industrial control system to dynamically adjust production equipment and generate equipment status data; Obtaining device status data, using a support vector machine algorithm to analyze the adjustment process to determine whether the adjustment is stable. If the adjustment is stable, generating an eighth data set from the device status data to reflect the adjustment results. If the adjustment is unstable, reloading the optimization parameters through system control to obtain new device status data; The extraction results of the eighth data set reflect the characteristics, and the K-means algorithm is used for cluster analysis to determine the adjustment trend. The seventh data set is updated through the adjustment trend to generate new optimization parameters.

9. The intelligent engineering automated production process quality traceability system according to claim 1 is characterized by: The quality monitoring report generating module includes: By loading the eighth data set, using data cleaning technology to remove outliers, obtaining processed data, and using data visualization technology to graphically display the processed data, monitoring information is generated; Based on the monitoring information, key indicators of the production process are obtained and the indicator change trend is determined. When the indicator change trend exceeds the preset standard, the abnormal point is marked through logical judgment to obtain abnormal identification data; Based on the abnormal identification data, a clustering algorithm is used to divide the data categories and generate classification results. By comparing the classification results with the preset standards, it is determined whether the production process meets the standards and the standard status is obtained. At the same time, a data set that meets the standard status conditions is obtained to generate the ninth data set.

10. A method for tracing the quality of an automated production process of an intelligent engineering project, applied to a system for tracing the quality of an automated production process of an intelligent engineering project as claimed in any one of claims 1 to 9, characterized in that: The steps include: The temperature, pressure, and speed parameters of the production process are acquired through a sensor network. High-frequency sampling technology is used to record the raw data in real time to obtain a first data set containing a timestamp. The data is then preprocessed using edge computing nodes and noise interference is removed through a filtering algorithm to obtain a cleaned second data set. Based on the second data set, the data is uploaded to the blockchain network in shards using a distributed storage system. A digital signature is added to each data block during upload to generate a third data set with tamper-proof properties. The full process parameters are obtained from the third data set, and the signature consistency of the data blocks is verified using a smart contract. If consistent, it is confirmed that the data has not been tampered with, resulting in a fourth data set. Based on the fourth data set, a deep learning model is used to extract features of parameter change trends, obtain key quality-related patterns, and generate a fifth data set containing potential rules; Based on the fifth data set, a time series analysis algorithm is used to predict quality fluctuations. If the predicted value exceeds a preset threshold, an abnormal flag is triggered, resulting in a sixth data set with warning information. Using the sixth data set, the production parameter adjustment strategy is iteratively optimized using a reinforcement learning model to generate a seventh data set that is updated in real time. The optimized parameter configuration is obtained from the seventh data set, and the production equipment is dynamically adjusted through the industrial control system to obtain an eighth data set reflecting the adjustment results. The eighth data set is processed using data visualization technology to generate an intuitive quality monitoring report. The content of the report is then used to determine whether the production process meets the preset standards, resulting in the final ninth data set.

Images