Application data interaction method based on cloud manufacturing collaboration

Abstract

The invention discloses an application data interaction method based on cloud manufacturing collaboration. The method comprises the following steps: S1, carrying out deep processing on data output by a cloud manufacturing system; s2, various data type files which are imported or newly built after preprocessing are read; s3, the read preprocessed data are displayed in a multi-dimensional matrix mode, concept pairs with high correlation are accurately screened out by calculating conditional entropy and relative mutual information between concepts, and then the information entropy of a sub-mode of each information dominant concept pair is fused; and S4, constructing an omnibearing and accurate cloud manufacturing system information association cognitive map based on the mining result, deeply analyzing the uncertainty of the cloud manufacturing system, and displaying and outputting the analysis result in a visual and clear manner. According to the method, the data quality of the cloud manufacturing system is improved, noise and abnormal data are removed, the data format is standardized, the uncertainty of the cloud manufacturing system is deeply analyzed, and visual, clear and powerful data support and decision basis are provided for cooperative work.

Description

Technical Field

[0001] The invention relates to the technical field of application data interaction, and in particular to an application data interaction method based on cloud manufacturing collaboration. Background Art

[0002] Cloud manufacturing collaboration aims to integrate manufacturing resources distributed in different geographical locations and organizations to achieve efficient collaborative production. The application data interaction method is a key link, which ensures the accurate, timely and effective flow of data throughout the collaborative process. In the field of cloud manufacturing collaboration, the existing application data interaction methods are diverse, but there are many shortcomings. Common data preprocessing methods are usually simple data cleaning and basic standardization operations. Data reading relies on traditional methods and has limited ability to process unstructured data. Data mining mostly uses traditional algorithms, which are unable to cope with complex large-scale data and mine deep associations. System analysis mainly relies on simple statistical analysis and chart display, which makes it difficult to deeply analyze system uncertainty. The existing interaction methods are selected because their technology is relatively mature and stable with low risk; second, the cost is relatively low, which meets the resource constraints of enterprises; third, the technical personnel are familiar with it, the operation threshold is low, and it is easy to promote.

[0003] However, existing methods have obvious problems. Data preprocessing is difficult to effectively remove complex noise and abnormal data, which affects the accuracy of subsequent analysis. Data reading cannot fully handle multiple data types, which limits data utilization. Data mining is not deep enough, making it difficult to discover complex associations and patterns. The value of data is not fully mined, and system analysis is not comprehensive and accurate, which cannot effectively support decision-making. In addition, the overall efficiency and performance are low, and the speed of processing large-scale data is slow, making it difficult to meet real-time requirements.

[0004] The application scenarios of cloud manufacturing collaboration are characterized by wide data sources, complex collaboration needs, high uncertainty and high real-time requirements. Existing data interaction methods are difficult to cope with these complex scenarios and cannot meet the growing needs. Summary of the invention

[0005] The present invention provides an application data interaction method based on cloud manufacturing collaboration, which improves the data quality of the cloud manufacturing system, removes noise and abnormal data and standardizes the data format; realizes efficient and accurate reading and integration of multiple data types, breaks down data type barriers, enhances the depth and accuracy of data mining, and discovers complex data relationships and patterns; deeply analyzes the uncertainty of the cloud manufacturing system, and provides intuitive, clear and powerful data support and decision-making basis for collaborative work.

[0006] In order to solve the above problems, the technical solution adopted by the invention is:

[0007] An application data interaction method based on cloud manufacturing collaboration, characterized by comprising the following steps:

[0008] S1 data preprocessing step, deeply processes the data output by the cloud manufacturing system, removes noise and abnormal data, and standardizes and normalizes the data to obtain high-quality ideal data for system uncertainty analysis;

[0009] S2 data reading step, which efficiently and accurately reads various data types of files that are imported or newly created after preprocessing, including but not limited to structured, semi-structured and unstructured data;

[0010] S3 data mining step, which displays the read pre-processed data in the form of a multi-dimensional matrix, accurately screens out concept pairs with strong correlation by calculating the conditional entropy and relative mutual information between concepts, and deeply mines the complex effective association rules and accurately determines the corresponding fine sub-patterns for information-dominated concept pairs based on the strictly preset high confidence and high support, and then integrates the information entropy of the sub-patterns in each information-dominated concept pair;

[0011] In the S4 system analysis step, a comprehensive and accurate cloud manufacturing system information association cognitive map is constructed based on the above mining results, the uncertainty of the cloud manufacturing system is deeply analyzed, and the analysis results are displayed and output in an intuitive and clear manner, providing strong data support and decision-making basis for cloud manufacturing collaboration.

[0012] The beneficial effects of this program are:

[0013] Beneficial effects:

[0014] By deeply preprocessing the output data of the cloud manufacturing system, removing noise and abnormal data and standardizing and normalizing it, high-quality ideal data is provided for subsequent analysis. At the same time, multi-dimensional matrix system analysis technology constructs an information association cognitive map, which can accurately mine the association rules and hidden patterns in the data, and deeply analyze the system uncertainty, so that decision makers can make wise decisions based on more accurate, comprehensive and in-depth information, reduce decision-making errors and improve decision accuracy.

[0015] It has the ability to efficiently and accurately read files of various data types, including structured, semi-structured and unstructured data, breaking the limitations of data types and ensuring the integrity and comprehensiveness of data. This helps to have a more comprehensive understanding of the system status, reasonably allocate manufacturing resources, and reduce collaboration barriers caused by data incompatibility and poor circulation, thereby improving the collaborative efficiency of the cloud manufacturing system and reducing production costs.

[0016] Advanced data mining technology can discover complex effective association rules and potential innovation patterns, providing new ideas and directions for the development and optimization of cloud manufacturing collaboration models. Comprehensive and in-depth system analysis helps stimulate innovative thinking, enhance innovation capabilities, and promote continuous progress in the field of cloud manufacturing. During the entire data interaction process, data compression, encryption and other processing measures are taken to ensure data security and privacy, reduce the risk of data leakage, and provide a reliable environment for the stable operation of cloud manufacturing collaboration.

[0017] Furthermore, the S1 adopts a convolutional neural network algorithm based on deep learning. By learning a large number of noise and abnormal data samples, it can accurately identify and remove noise and outliers in the data. At the same time, it uses an automatic encoder to reconstruct the data, accurately remove noise and outliers, and reconstruct the data, which significantly improves the accuracy and completeness of the data, and provides a more reliable and high-quality data foundation for subsequent analysis and decision-making. The processed data is purer, making the models trained based on these data or the analysis conducted more robust and able to better cope with various complex situations and interferences.

[0018] Furthermore, the S2 utilizes lexical analysis, syntactic analysis, and semantic understanding algorithms in natural language processing technology to deeply analyze the text content in unstructured data, extract key information, and convert it into a structured or semi-structured data format, converting unstructured text data into a structured or semi-structured format to facilitate integration and unified processing with other types of data, thereby improving the efficiency of data management. Through the comprehensive use of lexical analysis, syntactic analysis, and semantic understanding algorithms, key information can be extracted more accurately, information omissions and errors can be reduced, and unstructured text data that was originally difficult to process and utilize becomes easy to analyze and use, thereby expanding the scope of data availability.

[0019] Furthermore, the S3 introduces an optimization mechanism based on an evolutionary algorithm, which dynamically adjusts the preset values ​​of confidence and support by simulating the inheritance, mutation and selection operations in the biological evolution process, and continuously optimizes these two parameters according to the distribution characteristics of the data and the mining objectives. By optimizing the confidence and support according to the data distribution characteristics and the mining objectives, it is possible to more accurately mine truly valuable association rules, reduce misjudgments and omissions, and dynamically adjust parameters so that the data mining process can adapt to the ever-changing data situation and maintain a good mining effect, rather than relying on fixed preset values, thereby avoiding invalid mining or over-mining caused by unreasonable initial preset values, saving computing resources and time, and improving overall mining efficiency.

[0020] Furthermore, the S4 uses visual analysis tools to present the analysis results in the form of charts, graphs and interactive dashboards. Through data visualization layout design, color coding and animation effects, the intuitiveness and readability of the analysis results are enhanced. Complex data and analysis results are presented in the form of charts, graphs, etc., and abstract data is converted into intuitive visual elements, so that users can see the main trends and key features of the data at a glance. Users can quickly grasp key information and quickly understand the meaning behind the data, so as to make more informed decisions in a short period of time, which is especially advantageous when time is tight.

[0021] Furthermore, S1 also includes compressing the data, using a lossless compression algorithm to reduce data storage space, and at the same time, encrypting the data using an advanced encryption standard encryption algorithm. The use of a lossless compression algorithm can significantly reduce the storage space occupied by the data and reduce storage costs, especially for large-scale cloud manufacturing data, which can save a lot of hardware resources. Using an advanced encryption standard encryption algorithm to encrypt the data keeps the data in an encrypted state during storage and transmission, making it difficult to decipher even if it is illegally obtained, thereby effectively preventing data leakage and unauthorized access.

[0022] Furthermore, the S3 adopts a distributed computing framework to decompose large-scale data processing tasks into multiple computing nodes for parallel execution. Through data partitioning, task scheduling and fault-tolerant mechanisms, it fully utilizes computing resources and significantly accelerates the data mining process. Large-scale data processing tasks are parallelized, and multiple computing nodes work simultaneously, which greatly shortens the data mining time and enables rapid acquisition of analysis results. The fault-tolerant mechanism ensures that when some computing nodes fail, the tasks can continue to be executed and the entire data mining process will not be interrupted due to failures of individual nodes.

[0023] Furthermore, the S4 combines discrete event simulation with simulation technology to simulate and verify the uncertainty of the cloud manufacturing system multiple times. By adjusting the input parameters and model structure, the system performance is optimized. Through multiple simulations and verifications of uncertainty, potential risks can be assessed in advance, possible problems can be predicted, and a basis can be provided for enterprises to formulate response strategies. The input parameters and model structure can be adjusted based on the simulation results to help decision makers find the best solutions and strategies before actual operations, thereby improving the scientificity and accuracy of decision-making. Simulation and verification are performed in a virtual environment to avoid the high costs and potential losses that may be caused by conducting experiments in actual systems.

[0024] Furthermore, the S2 adopts data caching technology to store frequently accessed data in high-speed memory. By optimizing the cache strategy and setting an appropriate cache expiration time and update mechanism, the speed and efficiency of data reading are improved. Storing frequently accessed data in high-speed memory can significantly reduce the time for data reading, achieve fast response, improve user experience, reduce frequent access to the back-end storage system, and reduce I / O operations, thereby reducing the overall burden of the system and improving system stability.

[0025] Furthermore, the S4 establishes a prediction model based on the historical data analysis results using a time series prediction algorithm or an LSTM neural network. The model can consider factors such as seasonality, trend, and periodicity to predict future system status and performance. Based on the prediction of future system status and performance, enterprises can formulate strategies and plans in advance and make forward-looking decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of the principle of the present invention; DETAILED DESCRIPTION

[0027] The following is further described in detail through specific implementation methods:

[0028] Example 1

[0029] First, the data output by the cloud manufacturing system is collected, and a convolutional neural network algorithm based on deep learning is used to train the algorithm in advance using a large amount of sample data containing noise and outliers. When actually processing the data, the algorithm can quickly and accurately identify the noise and outliers in the current data and remove them. At the same time, the automatic encoder is used to reconstruct the processed data to further ensure the accuracy and completeness of the data.

[0030] After completing the above operations, the data is standardized and normalized. The Z-score standardization method is used to convert the data into a distribution with a mean of 0 and a standard deviation of 1. Then, Min-Max normalization is used to map the data to a specific interval.

[0031] The reason for doing this is that deep learning algorithms can learn complex data patterns, thereby removing noise and outliers more accurately, while standardization and normalization help eliminate the impact of data dimensions, making data with different characteristics comparable, providing a high-quality foundation for subsequent analysis, improving the accuracy and reliability of subsequent data processing and analysis, enhancing the robustness of the model, and reducing the error rate.

[0032] In the data reading stage, when the imported or newly created data file is received after preprocessing, it is processed using the lexical analysis, syntactic analysis and semantic understanding algorithms in natural language processing technology. For unstructured text data, lexical analysis is first performed to divide the text into basic elements such as words and punctuation marks. Then, the structure and grammatical relationship of the sentence are determined through syntactic analysis. Finally, the semantic understanding algorithm is used to extract key information.

[0033] Convert the extracted key information into a structured or semi-structured data format, convert the product specification information described in the text into a table, or convert user comments into key-value pairs. The reason for adopting this approach is that natural language processing technology can deeply understand the meaning of the text, accurately extract key information, and convert unstructured data into a structured or semi-structured format to facilitate subsequent storage, query and analysis, improve the efficiency of data management, reduce information omissions and errors, expand the available scope of data, and promote the integration and unified processing of different types of data.

[0034] The read preprocessed data is presented in the form of a multidimensional matrix, and the conditional entropy and relative mutual information between concepts are calculated to screen out concept pairs with strong correlation. For information-dominated concept pairs, an optimization mechanism based on evolutionary algorithms is introduced according to strict high confidence and high support thresholds set in advance. In the evolutionary algorithm, the inheritance, mutation and selection operations of biological evolution are simulated. According to the distribution characteristics of the data and the mining objectives, the preset values ​​of confidence and support are dynamically adjusted, and these two parameters are continuously optimized to mine complex and effective association rules and determine the corresponding fine sub-patterns. Then, the information entropy of the sub-patterns in each information-dominated concept pair is integrated. Because the evolutionary algorithm can adaptively optimize parameters, it can better adapt to the characteristics of the data and mining needs, more accurately mine valuable association rules, reduce misjudgments and omissions, improve mining efficiency, avoid invalid or excessive mining, and save computing resources and time.

[0035] Based on the results of data mining, a comprehensive and accurate cognitive map of cloud manufacturing system information association is constructed. Visual analysis tools such as PowerBI or Tableau are used to present the analysis results in the form of charts, graphs and interactive dashboards. The intuitiveness and readability of the analysis results are enhanced through carefully designed layouts, color coding and animation effects.

[0036] At the same time, combined with simulation technology and discrete event simulation, the uncertainty of the cloud manufacturing system is simulated and verified multiple times. Based on the results of historical data analysis, a prediction model is established using time series prediction algorithm or LSTM neural network. This model fully considers factors such as seasonality, trend and periodicity, and predicts the future system status and performance. Visual analysis tools can intuitively present complex data relationships, simulation technology can assess risks and optimize systems in advance, and time series prediction algorithms can provide a basis for future decision-making, helping users quickly understand and gain insight into system uncertainties, optimize system performance, formulate strategies and plans in advance, make forward-looking decisions, and improve the efficiency and effectiveness of cloud manufacturing collaboration.

[0037] In addition, in S1, data is compressed and encrypted, the LZ77 algorithm is used to reduce data storage space, and the Advanced Encryption Standard (AES) algorithm is used to encrypt data to ensure data security.

[0038] In the data mining step, the distributed computing framework Hadoop or Spark is used to decompose large-scale data processing tasks into multiple computing nodes for parallel execution. Through data partitioning, task scheduling and fault-tolerant mechanisms, computing resources are fully utilized to accelerate the data mining process.

[0039] In the data reading step, Redis cache is used to store frequently accessed data in high-speed memory. By optimizing the cache strategy and setting appropriate cache expiration time and update mechanism, the speed and efficiency of data reading are improved.

[0040] S1 data preprocessing; specifically:

[0041] Collect data: Collect a large amount of raw data from various production links, equipment sensors, quality inspection systems, etc. in the cloud manufacturing system, including production process parameters, equipment operating status, product quality indicators, etc., remove noise and outliers based on deep learning, train convolutional neural network algorithms based on deep learning, and prepare a large number of sample data sets containing noise and outliers. These samples should cover various possible data patterns and abnormal situations.

[0042] Construct a convolutional neural network model, which includes an input layer, multiple convolutional layers, a pooling layer, and a fully connected layer.

[0043] The model is trained using optimization algorithms such as back-propagation and stochastic gradient descent to adjust the weights and biases of the network so that the model can accurately identify noise and outliers. Specific embodiment 2

[0045] S01 Data preparation: Collect a large amount of sample data containing noise and outliers, and divide it into training set, validation set and test set, and perform data normalization, data enhancement and other operations to increase data diversity and robustness.

[0046] S02 Model initialization: Build a convolutional neural network model, including determining the number of network layers, the number of neurons in each layer, the size of the convolution kernel and other parameters.

[0047] S03 forward propagation: Input a sample data in the training set into the model, calculate through the convolution layer, pooling layer and fully connected layer, and get the output of the model.

[0048] S04 Calculate loss: Compare the output of the model with the true label of the sample and calculate the loss value using a predefined loss function (such as mean square error, cross entropy, etc.).

[0049] S05 Backpropagation: Starting from the output layer, according to the gradient of the loss function, the gradient of each neuron is calculated by the chain rule, and the gradient is backpropagated layer by layer until the input layer to calculate the gradient of each weight and bias in the model.

[0050] S06 Weight and bias update: Use the stochastic gradient descent algorithm to update the weights and biases according to the calculated gradients. The learning rate determines the step size of each update and usually needs to be adjusted appropriately.

[0051] Repeat steps S03-S06: Repeat the process of forward propagation, loss calculation, back propagation, and weight update for each sample in the training set until a round of training is completed. After each round of training, use the validation set to evaluate the performance of the model. According to the results of the validation set, adjust the learning rate, network structure and other hyperparameters to improve the generalization ability of the model.

[0052] Use the test set to perform a final evaluation on the trained model to determine whether the model can accurately identify noise and outliers. If the model performance meets the requirements, save the trained model parameters for use in practical applications. During the training process, the parameters need to be continuously adjusted and optimized to obtain the best model performance.

[0053] The specific method for data compression and data encryption in the above S1 is:

[0054] Data compression: Use the LZ77 algorithm: maintain a sliding window and a forward buffer, search for matching strings with the data in the forward buffer in the sliding window, use pointers and lengths to represent the matching strings, encode the unmatched parts, and achieve data compression.

[0055] Data encryption: The Advanced Encryption Standard (AES) algorithm is used to generate encryption keys, group plaintext data into fixed block sizes, and perform encryption operations on each data block, including round key addition, byte replacement, row shift, and column confusion.

[0056] S2 actual data processing: The raw data collected by S1 is input into the trained convolutional neural network model. The model outputs data points marked as noise and outliers through feature extraction and pattern recognition. These marked data points are removed to obtain the preliminarily processed data. The specific steps of data reconstruction of the autoencoder are:

[0057] S201 Autoencoder Construction: Design the autoencoder structure, including the encoder and decoder parts, usually implemented using a multi-layer neural network, determine the number of neurons and the connection method of the encoder and decoder,

[0058] S202 Data Reconstruction: The data after removing noise and outliers is input into the encoder to obtain compressed feature representation. The decoder restores the feature representation to reconstructed data, compares the original data and the reconstructed data, and further ensures the accuracy and completeness of the data.

[0059] S203 Data standardization and normalization, the specific steps are:

[0060] Z-score normalization:

[0061] Calculate the mean and standard deviation of the data and for each data point, transform it using the formula (x-μ) / σ, where x is the original data point, μ is the mean, and σ is the standard deviation.

[0062] Min-Max Normalization:

[0063] Determine the minimum and maximum values ​​of the data, and for each data point, use the formula (x-min) / (max-min) to convert the data to the [0,1] interval.

[0064] S204, data reading; the specific steps include S2040 to S2042:

[0065] S2040 Receive data files: Receive pre-processed, imported or newly created data files, including structured production plan data, semi-structured equipment maintenance records, unstructured customer feedback texts, etc.

[0066] S2041 Application of natural language processing technology: Use lexical analysis tools, such as the lexical analysis module in NLTK (Natural Language Toolkit), to segment unstructured text into basic elements such as words and punctuation marks, build a vocabulary, record word frequency and part of speech, and then use syntactic analysis algorithms to generate a sentence tree, identify subject, predicate, object and other components, extract key information from the text based on a semantic understanding network based on deep learning, and understand the meaning and intent of the text through context analysis and semantic association.

[0067] S2042. Data format conversion: For the extracted key information, convert it into a structured or semi-structured data format, convert the product specification information described in text into a table format, clarify the column names and corresponding data values, and convert user comments into key-value pairs, such as "Problem type: quality issue" and "Severity: high".

[0068] S3, data mining; specifically including steps S31 to S35:

[0069] Step S31, multi-dimensional matrix presentation of data: construct the pre-processed and read data into a multi-dimensional matrix, for example, the data of different production links are used as different dimensions of the matrix, and each data point is used as an element in the matrix.

[0070] Step S32, conditional entropy and relative mutual information calculation:

[0071] S321 Conditional entropy calculation: For two random variables X and Y, calculate the conditional entropy H(X|Y) of X given Y. The conditional entropy is calculated by the joint probability distribution and conditional probability distribution of X and Y in the statistical data.

[0072] S322 Relative mutual information calculation: Calculate the mutual information I(X; Y) between two random variables, and then obtain the relative mutual information through normalization to measure the strength of the correlation between the variables.

[0073] Step S33, screening strongly correlated concept pairs: setting a correlation threshold, screening out concept pairs whose conditional entropy and relative mutual information exceed the threshold, and considering these concept pairs to have strong correlation.

[0074] Step S34, detailed specific steps of parameter optimization based on evolutionary algorithm are as follows:

[0075] S341 Evolutionary Algorithm Initialization: Randomly generate an initial population of confidence and support parameter values, where each individual represents a set of confidence and support values.

[0076] S342 Fitness evaluation: Use the current parameter values ​​to perform association rule mining, and evaluate the fitness of each individual based on the quality of the mining results, such as the number of valid rules mined, the accuracy of the rules, etc.

[0077] S343 Genetic operation: select individuals with high fitness as parents, generate parameter values ​​for offspring through crossover operation, perform mutation operation on some offspring parameter values, and introduce new parameter combinations.

[0078] Repeat optimization: Continuously iterate the process of S341-S343 until the optimal or satisfactory confidence and support parameter values ​​are found.

[0079] Step S35, association rule mining and sub-pattern determination: Based on the optimized confidence and support, complex effective association rules are mined, and for information-dominated concept pairs, corresponding fine sub-patterns are determined, and the information entropy of the sub-patterns is integrated to comprehensively evaluate the uncertainty and information content of the data.

[0080] S4 system analysis; specifically includes steps S40 to S42:

[0081] S40. Construction of information association cognitive map: Based on the results of data mining, construct a cognitive map that shows the association relationship between data, using a graph database or network analysis tool to represent nodes (data elements) and edges (association relationships).

[0082] S41. Visual analysis: Choose the visual analysis tool Tableau to present cognitive maps and analysis results in the form of charts (such as bar charts, line charts), graphics (such as flow charts, relationship diagrams) and interactive dashboards. In the design layout, the design includes the arrangement of charts, color matching and label setting to highlight the key points and enhance readability. Color coding is applied to distinguish different data categories or importance levels. Animation effects are added to mark dynamic updates and interactive responses of data to enhance user experience.

[0083] S42. Simulation and predictive modeling. Specific steps of simulation and predictive modeling:

[0084] Discrete event simulation: define the system model and parameters, including production processes, resource allocation, etc., run multiple simulations, observe the system's behavior and performance under different conditions, analyze simulation results, and evaluate the system's uncertainties and potential risks.

[0085] Specific steps for establishing a time series forecasting model:

[0086] S421 collects historical data, including production indicators, market demand, etc.

[0087] S422 selects ARIMA or LSTM neural network time series prediction algorithm, trains the prediction model, adjusts model parameters to improve prediction accuracy, and uses the trained model to predict future system status and performance.

[0088] The above is only an embodiment of the invention, and the common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the invention, which should also be regarded as the scope of protection of the invention, and these will not affect the effect of the implementation of the invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for application data interaction based on cloud manufacturing collaboration, characterized in that: The following steps are involved: S1 data preprocessing: deeply process the data output by the cloud manufacturing system, remove noise and abnormal data, and standardize and normalize the data to obtain high-quality ideal data for system uncertainty analysis; S2 data reading, which can efficiently and accurately read various data types of files imported or newly created after preprocessing, including but not limited to structured, semi-structured and unstructured data; S3 data mining displays the read pre-processed data in the form of a multi-dimensional matrix. By calculating the conditional entropy and relative mutual information between concepts, it accurately screens out concept pairs with strong correlation. For information-dominated concept pairs, based on the strictly preset high confidence and high support, it deeply mines their complex effective association rules and accurately determines the corresponding fine sub-patterns, and then integrates the information entropy of the sub-patterns in each information-dominated concept pair. S4 system analysis builds a comprehensive and accurate cloud manufacturing system information association cognitive map based on the above mining results, deeply analyzes the uncertainty of the cloud manufacturing system, and displays and outputs the analysis results in an intuitive and clear manner, providing strong data support and decision-making basis for cloud manufacturing collaboration.

2. The method for application data interaction based on cloud manufacturing collaboration according to claim 1 is characterized in that: The S1 adopts a convolutional neural network algorithm based on deep learning. By learning a large number of noise and abnormal data samples, it can accurately identify and remove noise and abnormal values ​​in the data, and reconstruct the data using an automatic encoder.

3. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S2 uses lexical analysis, syntactic analysis and semantic understanding algorithms in natural language processing technology to deeply analyze the text content in unstructured data, extract key information, and convert it into structured or semi-structured data format.

4. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S3 introduces an optimization mechanism based on an evolutionary algorithm, which dynamically adjusts the preset values ​​of confidence and support by simulating the inheritance, mutation and selection operations in the biological evolution process, and continuously optimizes these two parameters according to the distribution characteristics of the data and the mining objectives.

5. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S4 uses visual analysis tools to present the analysis results in the form of charts, graphs and interactive dashboards, and enhances the intuitiveness and readability of the analysis results through data visualization layout design, color coding and animation effects.

6. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S1 also includes compressing the data, using a lossless compression algorithm to reduce data storage space, and at the same time, using an advanced encryption standard encryption algorithm to encrypt the data.

7. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S3 uses a distributed computing framework to decompose large-scale data processing tasks into multiple computing nodes for parallel execution. Through data partitioning, task scheduling and fault-tolerant mechanisms, computing resources are fully utilized to significantly accelerate the data mining process.

8. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: S4 combines discrete event simulation with simulation technology to simulate and verify the uncertainty of cloud manufacturing systems multiple times. By adjusting input parameters and model structure, it optimizes system performance and improves the accuracy and reliability of predictions.

9. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S2 adopts data caching technology to store frequently accessed data in high-speed memory, and sets appropriate cache expiration time and update mechanism through optimization of cache strategy.

10. The method for application data interaction based on cloud manufacturing collaboration according to claim 1, characterized in that: The S4 establishes a prediction model based on the historical data analysis results using a time series prediction algorithm or an LSTM neural network. The model can consider factors such as seasonality, trend, and periodicity to predict future system status and performance.

Images