Intelligent monitoring method and system for functional ceramic production workshop based on Internet of Things

Abstract

The invention discloses an intelligent monitoring method and system for a functional ceramic production workshop based on the Internet of Things, and belongs to the field of intelligent monitoring of the Internet of Things, and the method comprises the following steps: carrying out the preprocessing of workshop real-time data collected by a sensor, and obtaining a data set with consistent granularity; acquiring communication protocol information of production equipment, and performing protocol conversion on the granularity consistent data set based on the communication protocol information of the production equipment to obtain unified protocol format data; performing real-time processing on the unified protocol format data by adopting a streaming computing framework to obtain a distributed processing result; carrying out modeling analysis on the distributed processing result by adopting a production process decision tree model to obtain production process information; and performing anomaly identification and decision optimization on the production process information to obtain a final optimization decision result.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring of the Internet of Things, and in particular relates to an intelligent monitoring method and system for a functional ceramic production workshop based on the Internet of Things. Background Art

[0002] The production environment of functional ceramic production workshops is complex and changeable. The operating status and process parameters of various equipment are difficult to control in real time. Data collection and transmission in the production process also face many technical challenges. First, there are a large number of sensors and acquisition equipment distributed in the workshop. They work in harsh environments for a long time, which is prone to data loss, error accumulation and other problems, resulting in distorted collected data. Secondly, the data formats generated by different devices are inconsistent, and the timestamps and data granularity are also different, making it difficult to directly perform correlation analysis. Furthermore, production equipment often has its own communication protocols and interface specifications, which makes integration difficult. At the same time, the massive data generated by the production process places high demands on the performance and scalability of the data processing platform. Finally, the production process is complex, and there are differences in the production parameters of different batches of products. There is a lack of a unified data model to describe the entire production process. The value of data is difficult to fully explore, and production optimization and predictive maintenance are out of the question. In summary, functional ceramic production workshops are in urgent need of an intelligent monitoring system to solve key technical problems in data collection, transmission, processing, modeling and analysis, and to achieve real-time insight and optimization decisions in the production process. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes an intelligent monitoring method and system for a functional ceramic production workshop based on the Internet of Things to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above object, the present invention provides an intelligent monitoring method for a functional ceramic production workshop based on the Internet of Things, comprising:

[0005] Preprocess the real-time workshop data collected by sensors to obtain a data set with consistent granularity;

[0006] Acquire communication protocol information of the production equipment, and perform protocol conversion on the granularity consistent data set based on the communication protocol information of the production equipment to obtain unified protocol format data;

[0007] Using a streaming computing framework to process the unified protocol format data in real time to obtain a distributed processing result;

[0008] Using a production process decision tree model to model and analyze the distributed processing results to obtain production process information;

[0009] Abnormal identification and decision optimization are performed on the production process information to obtain the final optimized decision result.

[0010] Optionally, the process of preprocessing the real-time workshop data collected by the sensor to obtain a data set with consistent granularity includes:

[0011] Cleaning the real-time workshop data collected by the sensor to obtain a first data set;

[0012] Converting the first data set into a second data set;

[0013] Aligning the timestamps of the data in the first data set and the second data set using a time synchronization algorithm to obtain third data;

[0014] A data interpolation algorithm is used to perform granularity unification processing on the third data set to obtain a data set with consistent granularity.

[0015] Optionally, the process of cleaning the real-time workshop data collected by the sensor includes:

[0016] Use data cleaning algorithm to detect and remove outliers from real-time data of workshop to obtain data with outliers removed;

[0017] Performing noise elimination and normalization processing on the outlier-removed data to obtain preprocessed data of each sensor;

[0018] A correlation analysis algorithm is used to calculate a correlation coefficient matrix between preprocessed data of each sensor, and based on the correlation coefficient matrix, sensor data with high redundancy is removed to obtain a first data set.

[0019] Optionally, the process of converting the data format of the first data set to obtain the second data set includes:

[0020] S1, identifying the data format of each piece of data in the first data set, and directly adding the data to the second data set when the data is in a standard format, and entering S2 when the data is not in a standard format;

[0021] S2, converting the data in non-standard format into standard format based on data format conversion rules;

[0022] S3, inspect the standard format data converted in S2, and directly add it to the second data set if it meets the inspection requirements; if it does not meet the inspection requirements, proceed to S4;

[0023] S4, repeating S1-S3 until all data in the first data set are processed to obtain a complete second data set;

[0024] S5. Remove duplicate data and invalid data from the complete second data set to obtain a second data set.

[0025] Optionally, the process of aligning the timestamps of the data in the first data set and the second data set includes:

[0026] Acquire timestamp information of the first data set and the second data set, and convert the timestamp information into a unified time format;

[0027] Determine the time point corresponding to each data in the first data set according to the timestamp information of the second data set, and calculate the time difference between the first data set and the second data set;

[0028] Performing offset correction on the timestamp based on the time difference to obtain a corrected first data set and a second data set;

[0029] The corrected first data set and the second data set are combined to generate a time-synchronized third data set.

[0030] Optionally, the process of performing protocol conversion on the granularity consistent data set based on the communication protocol information of the production equipment includes:

[0031] Analyze the differences in data granularity in the third data set to determine a data subset that needs to be processed with uniform granularity;

[0032] A data interpolation algorithm is used to perform interpolation calculations between data points in the data subset that needs to be uniformly processed to generate new data points;

[0033] Determine the consistency between the new data point and the original data point. If the consistency meets the preset threshold, the new data point is added to the data subset.

[0034] The quality of the data subset after interpolation processing is evaluated by calculating the statistical characteristics of the data subset. When the quality evaluation result of the data subset meets the requirement of uniform granularity, the data subset after interpolation processing is merged with other unprocessed data in the third data set to obtain a fourth data set with uniform granularity.

[0035] Optionally, the process of using a pre-established production process decision tree model to model and analyze the distributed processing results to obtain production process information includes:

[0036] Cleaning, converting and integrating the distributed processing results to obtain standardized production process data;

[0037] Use decision tree algorithm to build production process decision tree model;

[0038] Based on the production process decision tree model, the distributed processing results are predicted and analyzed to obtain the optimal decision path and prediction results of each production link;

[0039] Performing comparative analysis based on the predicted result and the actual value of the distribution processing result to obtain a comparative result;

[0040] The distribution processing result is reclassified based on the comparison result, the production data with similar key attribute values ​​are classified into the same category, and cluster analysis is performed on the production data of each category to obtain an optimization plan for the production process.

[0041] The present invention also discloses an intelligent monitoring system for a functional ceramic production workshop based on the Internet of Things, which is used to implement an intelligent monitoring method for a functional ceramic production workshop based on the Internet of Things. The system includes:

[0042] A data acquisition module, used for acquiring real-time data of all sensors in the workshop, and cleaning the real-time data to obtain a cleaned first data set;

[0043] A data cleaning module, used for performing data format conversion on the first data set to obtain a second data set;

[0044] A data format conversion module, used to align the timestamps of all data using a time synchronization algorithm according to the timestamp information in the second data set, to obtain a third data set after time synchronization;

[0045] A time synchronization module, used for using a data interpolation algorithm to unify the granularity of the third data set according to the data granularity difference in the third data set, so as to obtain a fourth data set with consistent granularity;

[0046] A data interpolation module, used to convert the fourth data set into a unified protocol format based on the communication protocol information of the production equipment to obtain a fifth data set;

[0047] A protocol conversion module, used for inputting the fifth data set into the distributed data processing platform, and using a streaming computing framework to process the massive data in real time to obtain a sixth data set;

[0048] A distributed data processing module, used for performing modeling analysis on the sixth data set according to a pre-established unified data model to obtain a seventh data set;

[0049] A data modeling and analysis module, configured to identify abnormalities in the seventh data set by using a cluster analysis method to obtain an eighth data set;

[0050] The abnormality identification and optimization decision module is used to optimize the production process in real time based on the eighth data set to obtain the final optimization decision result.

[0051] Compared with the prior art, the present invention has the following advantages and technical effects:

[0052] The present invention discloses an intelligent monitoring method for a functional ceramic production workshop based on the Internet of Things. The method proposes a complete set of data processing procedures for the collection, cleaning, format conversion, time synchronization and granularity unification of multi-source heterogeneous data in the workshop. Through protocol conversion and distributed computing technology, real-time processing of massive production data is achieved. Based on a unified data model, the present invention models and analyzes the entire production process, and uses a machine learning algorithm to identify abnormal situations. Finally, the production process is optimized in real time by optimizing the decision-making algorithm. The technical effect of the present invention is that it realizes intelligent processing of the entire process from raw sensor data to optimized decision results, improves production efficiency and quality, reduces the incidence of abnormal situations, and provides strong support for intelligent manufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0054] Figure 1 This is a flow chart of an intelligent monitoring method for a functional ceramic production workshop based on the Internet of Things according to an embodiment of the present invention;

[0055] Figure 2 This is a structural diagram of an intelligent monitoring system for a functional ceramic production workshop based on the Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0057] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0058] Embodiment 1

[0059] like Figure 1 As shown, this embodiment provides an intelligent monitoring method and system for a functional ceramic production workshop based on the Internet of Things, including the following steps:

[0060] S101. Acquire real-time data from all sensors in the workshop, remove outliers and noises through a preset data cleaning algorithm, and obtain cleaned data, i.e., a first data set.

[0061] Furthermore, the raw data collected in real time by all sensors in the workshop is obtained and stored in the database. According to the preset data cleaning algorithm, outliers are detected and removed for the raw data. Noise reduction algorithms such as wavelet transform are used to eliminate noise from the data after outliers are removed. Through data standardization, the denoised data is converted to a uniform scale and distribution to obtain the preprocessed data of each sensor. Using the correlation analysis algorithm, the correlation coefficient matrix between the preprocessed data of each sensor is calculated. According to the correlation coefficient matrix, the sensor data with high redundancy is removed to reduce the data dimension. The processed sensor data is sorted in time series to obtain the first data set after the final cleaning.

[0062] Furthermore, the raw data is collected in real time and stored in a database, involving multiple sensors such as temperature, pressure, and vibration. As a specific implementation of this embodiment, the blast furnace temperature sensor collects data once per second, and the pressure sensor collects data once every 0.5 seconds. These data are transmitted to the central database via industrial Ethernet. The preset outlier detection algorithm is based on statistical methods, such as the triple standard deviation method. In the injection molding workshop, if the mold temperature suddenly jumps from 180°C to 300°C, which is obviously beyond the normal range, the system will mark it as abnormal and eliminate it. Noise reduction processing is crucial to improving data quality. Wavelet transform is a commonly used noise reduction algorithm that can analyze signals simultaneously in the time and frequency domains. In precision machine tool workshops, the data of cutting force sensors are often interfered by high-frequency noise. Wavelet transform can effectively filter out these noises and retain useful cutting force signal characteristics. Data standardization processing enables data of different dimensions to be compared and comprehensively analyzed. Commonly used methods include minimum-maximum normalization and Z-score normalization. For example, temperature (°C), pressure (MPa), flow (m 3 / h) and other data in different units are unified into the interval [0, 1] to facilitate subsequent modeling and analysis.

[0063] S102 . For different data formats in the first data set, a data format conversion module is used to uniformly convert the data into a standard format to obtain a second data set.

[0064] Further, step 1: obtain the first data set, identify the data format of each data in the first data set, and determine whether it is in a standard format. If it is not in a standard format, proceed to step 2, otherwise the data is directly added to the second data set. Step 2: According to the preset data format conversion rules, the corresponding conversion method is used to convert the data in a non-standard format into a standard format. The conversion rules include field mapping, data type conversion, data format unification, etc. Step 3: Verify the converted standard format data to determine whether it meets the standard format requirements. If it meets the requirements, add it to the second data set, otherwise perform exception processing. Step 4: Repeat steps 1-3 until all data in the first data set are processed and a complete second data set is obtained. Step 5: Clean the second data set to remove duplicate data, invalid data, etc. to improve data quality. Use data deduplication algorithms and invalid data identification rules to achieve automatic data cleaning. Step 6: Extract and select features from the cleaned second data set to obtain a feature subset that is valuable for data analysis and mining. Select the optimal feature combination through feature importance evaluation and feature correlation analysis. Step 7: Store the second data set in a specific format and create a data index to facilitate subsequent data retrieval and analysis. Use a distributed storage architecture and indexing technology suitable for big data processing to improve data access efficiency.

[0065] S103: Align the timestamps of all data using a time synchronization algorithm according to the timestamp information in the second data set to obtain a third data set after time synchronization.

[0066] Furthermore, the timestamp information of the first data set and the second data set is obtained, and the timestamp is converted into a unified time format, such as UTC time. According to the timestamp information of the second data set, the time point corresponding to each data in the first data set is determined, and the time difference is calculated. For each data in the first data set, the timestamp is offset corrected according to the calculated time difference so that it is aligned with the timestamp of the second data set. The corrected first data set and the second data set are sorted according to the timestamp to ensure that the data is arranged in chronological order. The corrected first data set and the second data set are merged to generate a third data set after time synchronization. The third data set is cleaned and preprocessed to remove duplicate data and outliers to ensure data quality. The dynamic time warping algorithm is used to further optimize the time synchronization of the third data set to improve the accuracy of time alignment and obtain the final time-synchronized data set.

[0067] Furthermore, unifying timestamps in different formats into UTC time can eliminate the impact of time zone differences. For example, the timestamp in the first dataset may be "2024-06-2210:30:00+0800", and the timestamp in the second dataset is "1687401000". Through conversion, both can be unified into "2024-06-2202:30:00UTC". Time difference calculation is crucial for data alignment. Assuming that the timestamp of a data in the first dataset is "2024-06-2202:35:00UTC", and the closest time point in the second dataset is "2024-06-2202:30:00UTC", the time difference can be calculated to be 5 minutes. This time difference will be used for subsequent offset correction. Offset correction is an important means to ensure that the time of two datasets is consistent. The timestamp of the data in the first dataset is offset forward by 5 minutes to align it with the second dataset. The purpose of this is to eliminate the time error between the two data sets and provide a more accurate time benchmark for subsequent data analysis. Time sorting is the basis of data processing. Sorting the two corrected data sets by timestamp can clearly show the time series of the data. For example, the sorted data may be in the following order: "2024-06-22 02:30:00UTC", "2024-06-22 02:35:00UTC", "2024-06-22 02:40:00UTC", etc. This sorting not only facilitates the visualization of the data, but also lays the foundation for subsequent data analysis. By merging the two sorted data sets, a time-synchronized third data set can be obtained. This new data set contains all the information of the two original data sets, and all data has been aligned in time. The time-aligned data set may have duplicate timestamps, such as "2024-06-22 02:35:00UTC" appears twice in the merged data set. It is necessary to decide whether to keep two data or only one. In addition, some outliers may appear, such as temperature data suddenly jumping from 25°C to 100°C, which may be caused by sensor failure and needs to be identified and processed. The application of dynamic time warping algorithm can further optimize time synchronization. This algorithm can handle nonlinear time deviations and is suitable for more complex scenarios. For example, if it is found that the time difference between two data sets is not fixed but changes over time, the dynamic time warping algorithm can align these data more accurately. This algorithm achieves more accurate time synchronization by finding the best time correspondence and minimizing the distance between the two time series.

[0068] S104 . With respect to the data granularity differences in the third data set, a data interpolation algorithm is used to unify the data granularity to obtain a data set with consistent granularity, namely, a fourth data set.

[0069] Furthermore, a third data set is obtained, the differences in data granularity in the data set are analyzed, and a data subset that needs to be processed with unified granularity is determined; for the data subset that needs to be processed, a data interpolation algorithm is used to perform interpolation calculations between data points based on data characteristics and distribution patterns to generate new data points; after processing by the interpolation algorithm, the consistency between the newly generated data points and the original data points is determined, and if the consistency meets a preset threshold, the new data points are added to the data set; a quality assessment is performed on the data subset after interpolation processing, and by calculating the statistical characteristics of the data subset, it is determined whether it meets the requirements of unified granularity; if the quality assessment result of the data subset meets the requirements, it is merged with other unprocessed data in the third data set to obtain a fourth data set with unified granularity; a statistical analysis is performed on the fourth data set, and by calculating indicators such as the mean and variance of the data set, the overall consistency and availability of the data set are evaluated; the fourth data set is applied to subsequent business analysis and modeling tasks to give full play to the value of the data and support business decision-making and optimization.

[0070] S105. Acquire communication protocol information of the production equipment, and convert data of different protocols into unified protocol format data, i.e., the fifth data set, through a protocol conversion module.

[0071] Furthermore, according to the type and communication protocol of the production equipment, the protocol information corresponding to each device is obtained, and a device protocol information database is established. For data of different protocols, the corresponding protocol parsing module is used to parse and extract key data fields and parameters. Through the protocol mapping rules, the data fields and parameters of different protocols are mapped to the corresponding fields and parameters in the unified protocol format. The mapped data is format converted and the data type converted to ensure that the data meets the requirements of the unified protocol format. The converted data is encapsulated according to the unified protocol format to generate a standardized data packet. The data merging algorithm is used to merge standardized data packets from multiple sources into a complete data set. The merged data set is quality checked and cleaned to eliminate abnormal data to obtain a high-quality fifth data set.

[0072] Furthermore, the establishment of a device protocol information database is the basis of the entire data processing process. Taking factory automation as an example, there may be multiple types of devices, such as PLCs, sensors, actuators, etc. Each device may use a different communication protocol, such as Modbus, Profinet, or OPCUA. When establishing a database, it is necessary to record information such as the device ID, type, protocol used, and its version. The design of the protocol parsing module needs to take into account the characteristics of different protocols. Taking the Modbus protocol as an example, the parsing module needs to be able to identify function codes, register addresses, and data values. For the OPCUA protocol, it is necessary to parse information such as node ID, data type, and timestamp. During the parsing process, the key is to extract core data such as device status and measurement values. Map the register address in the Modbus protocol to the "parameter name" field in the unified format, and map the data value to the "parameter value" field. For the OPCUA protocol, it may be necessary to map the node ID to the "parameter name", and map the data value and quality attribute to the "parameter value" and "data quality" fields respectively. Convert the timestamps in different protocols to UTC format and convert the temperature data to Celsius. Data type conversion involves multiple situations such as from strings to floating point numbers, integers to enumeration values. The data encapsulation process needs to follow a predefined unified protocol format. This includes adding metadata such as device identification, timestamps, and data type tags. The encapsulated data packet should be self-descriptive and convenient for subsequent processing and analysis. The data merging algorithm needs to consider multiple factors. For example, for different parameters of the same device, they need to be aligned according to the timestamp. For data from multiple devices, the logical relationship between devices needs to be considered, such as the data association of upstream and downstream devices on the production line. The final data quality check and cleaning are the key to ensuring the reliability of the fifth data set. This includes detecting outliers (such as temperature sensors suddenly outputting extremely high values), processing missing data (such as through interpolation or previous value filling), removing duplicate records, etc. Through these steps, a standardized, high-quality data set can be obtained, providing a reliable basis for subsequent data analysis and decision-making. This series of processing not only achieves the unification and standardization of data, but also improves the quality and availability of data. The standardized data format enables data from different sources to be compared horizontally and analyzed comprehensively, providing possibilities for equipment performance evaluation and production efficiency optimization. High-quality data sets can improve the accuracy of subsequent data analysis, reduce misjudgments caused by data problems, and thus support more reliable decision-making.

[0073] S106. Input the fifth data set into the distributed data processing platform, and use the streaming computing framework to process the massive data in real time to obtain a distributed processing result, namely, the sixth data set.

[0074] Furthermore, according to the data format and data volume of the fifth data set, a suitable distributed data processing platform and streaming computing framework are determined; the fifth data set is divided into multiple data slices according to certain rules, and allocated to different data processing nodes; on each data processing node, the streaming computing framework is used to perform real-time processing and calculation on the allocated data slices; according to the business logic of data processing, a computing topology graph of data processing is constructed in the streaming computing framework; the data processing operators in the computing topology graph are deployed to different computing nodes of the streaming computing framework; the data processing node sends the processed intermediate result data to the next data processing node according to the data distribution rules; after all data processing nodes complete the calculation, their respective intermediate result data are aggregated to obtain the final sixth data set.

[0075] Furthermore, Apache Hadoop is selected as a distributed data processing platform, combined with Apache Flink as a streaming computing framework. Hadoop provides powerful distributed storage and computing capabilities, while Flink is good at processing real-time data streams. Data sharding is the key to improving processing efficiency. Data can be sharded according to the timestamp or device ID of the data to ensure that the data is evenly distributed on each processing node. As a specific implementation of this embodiment, the data of a day is divided into 24 shards by hour and assigned to different processing nodes. In this way, cluster resources can be fully utilized to improve parallel processing capabilities. In the streaming computing framework, the core of data processing is to build a computing topology map. Taking the energy consumption analysis of production equipment as an example, a topology map including steps such as data cleaning, feature extraction, energy consumption calculation and anomaly detection is designed. Each step corresponds to one or more data processing operators, which can be flexibly deployed on different computing nodes. The deployment of data processing operators needs to consider load balancing and data locality. As a specific implementation of this embodiment, data cleaning and feature extraction operators are deployed on the same node to reduce network transmission overhead. And computing-intensive energy consumption calculation operators can be deployed on nodes with stronger performance to increase processing speed. A message queue system (such as Apache Kafka) is used to achieve efficient data distribution. Each processing node sends the calculation results to the specified Kafka topic, and the downstream nodes consume data from the corresponding topic. This method not only ensures the reliable transmission of data, but also provides good decoupling and scalability. The aggregation process of the final result can adopt a hierarchical aggregation strategy. First, local aggregation is performed within each data center, and then the aggregation results of each data center are transmitted to the central node for global aggregation. This method can significantly reduce the network transmission volume and improve the aggregation efficiency. Through the above process, the fifth data set can be efficiently processed to generate a sixth data set containing rich information. As a specific implementation method of this embodiment, for the energy consumption analysis of production equipment, the sixth data set may contain real-time energy consumption data, energy efficiency indicators, abnormal event records, etc. of each device. This information provides an important basis for subsequent production optimization and equipment maintenance. The implementation of this process not only improves data processing efficiency, but also brings many technical effects. Distributed processing greatly improves the scalability of the system and can cope with the growing amount of data. Real-time streaming computing enables enterprises to detect production anomalies in a timely manner and respond quickly. The unified data processing process provides a standardized processing solution for data from different sources and in different formats, laying the foundation for the integration and utilization of enterprise data assets.

[0076] S107. Based on the sixth data set, the entire production process is modeled and analyzed using a pre-established unified data model to obtain a seventh data set.

[0077] Furthermore, according to the pre-established production process decision tree model, the full production process data in the sixth data set is obtained, and the standardized production process data is obtained by cleaning, converting and integrating the data; the standardized production process data is modeled and analyzed by using the decision tree algorithm, and a decision tree model of the production process is constructed according to the key attributes of each production link, and the decision tree model is optimized by pruning to obtain an optimized production process decision tree model; according to the optimized production process decision tree model, the full production process data in the sixth data set is predicted and analyzed, the key attribute values ​​of each production link are judged, and the optimal decision path of each production link is determined according to the branch conditions of the decision tree model; the prediction results of the production process decision tree model are compared and analyzed with the actual production data in the sixth data set to obtain a comparison result, and the prediction accuracy, recall rate and F1 value and other evaluation indicators of the model are calculated. The evaluation index is obtained, and the decision tree model is further optimized according to the comparison results to obtain the final production process optimization model; according to the final production process optimization model, the full production process data in the sixth data set is re-divided, and the production data with similar key attribute values ​​are divided into the same category, and the production data of each category are clustered to obtain the optimization plan of the production process; the optimization plan of the production process is applied to the actual production process, and the production process is dynamically optimized by collecting production data in real time. According to the changes in production data, the key parameters of the production process are dynamically adjusted to ensure that the production process is always in the optimal state; the optimized production process data is merged with the sixth data set to obtain the production process information, that is, the seventh data set, which contains the optimization data and key decision information of the whole production process, which can be used to guide subsequent production optimization and decision analysis.

[0078] S108. For the seventh data set, a cluster analysis method in a machine learning algorithm is used to identify abnormal conditions in the production process to obtain an eighth data set.

[0079] Furthermore, the original data in the production process is obtained according to the seventh data set, and a feature data set suitable for cluster analysis is obtained through data preprocessing and feature extraction; the K-means clustering algorithm is used to perform cluster analysis on the feature data set, and similar data points are divided into the same category by calculating the similarity between data points; according to the results of the cluster analysis, it is judged whether each category is an abnormal situation, and if the data points of a certain category are greatly different from the normal situation, the category is marked as an abnormal situation; the abnormal situation category obtained by the cluster analysis is obtained, the data points in the category are extracted, and the feature vector of the abnormal situation is constructed; the support vector machine (SVM) algorithm is used to train the feature vector of the abnormal situation to obtain an abnormal situation recognition model; the new data in the production process is input into the abnormal situation recognition model, and it is judged whether the data belongs to an abnormal situation through model prediction; according to the results of the abnormal situation recognition, the identified abnormal data is separated from the original data to obtain the eighth data set for subsequent abnormal cause analysis and formulation of improvement measures.

[0080] S109. According to the eighth data set, the production process is optimized in real time by using an optimization decision algorithm to obtain a final optimization decision result.

[0081] Furthermore, according to the real-time data of the production process, the relevant data in the eighth data set is obtained as the input of the optimization decision algorithm. The support vector machine algorithm is adopted to establish the production process optimization decision model by training the eighth data set. The real-time data of the production process is input into the optimization decision model to obtain the optimization decision for the current production status. If the optimization decision does not meet the preset decision threshold, an early warning is triggered, indicating that manual intervention in the production process is required. The relevant parameters of the production process are dynamically adjusted according to the optimization decision to achieve real-time optimization of the production process. The convolutional neural network algorithm is adopted to extract features of the optimized production process data to judge the optimization effect. The optimization decision results are compared with the historical optimization cases in the eighth data set, and the optimization decision model is continuously improved to improve the accuracy and reliability of the decision.

[0082] Embodiment 2

[0083] like Figure 2 As shown, the present invention provides an intelligent monitoring system for functional ceramic production workshops based on the Internet of Things, which mainly includes:

[0084] The data acquisition module is used to obtain the real-time data of all sensors in the workshop, remove outliers and noise through a preset data cleaning algorithm, and obtain the first cleaned data set;

[0085] A data cleaning module, for converting the data in different data formats in the first data set into a standard format using a data format conversion module to obtain a second data set;

[0086] A data format conversion module, used to align the timestamps of all data using a time synchronization algorithm according to the timestamp information in the second data set, to obtain a third data set after time synchronization;

[0087] A time synchronization module, used to use a data interpolation algorithm to unify the granularity of the data in the third data set according to the data granularity difference, so as to obtain a fourth data set with consistent granularity;

[0088] A data interpolation module is used to obtain communication protocol information of the production equipment, and convert data of different protocols into a unified protocol format through a protocol conversion module to obtain a fifth data set;

[0089] A protocol conversion module, used for inputting the fifth data set into the distributed data processing platform, and using a streaming computing framework to process the massive data in real time to obtain a sixth data set;

[0090] A distributed data processing module, used for modeling and analyzing the entire production process according to the sixth data set through a pre-established unified data model to obtain a seventh data set;

[0091] A data modeling and analysis module is used to identify abnormal conditions in the production process using a cluster analysis method in a machine learning algorithm for the seventh data set, thereby obtaining an eighth data set;

[0092] The abnormality identification and optimization decision module is used to optimize the production process in real time according to the eighth data set through the optimization decision algorithm to obtain the final optimization decision result.

[0093] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent monitoring method for a functional ceramic production workshop based on the Internet of Things, characterized in that: The following steps are involved: Preprocess the real-time workshop data collected by sensors to obtain a data set with consistent granularity; Acquire communication protocol information of the production equipment, and perform protocol conversion on the granularity consistent data set based on the communication protocol information of the production equipment to obtain unified protocol format data; Using a streaming computing framework to process the unified protocol format data in real time to obtain a distributed processing result; Using a production process decision tree model to model and analyze the distributed processing results to obtain production process information; Abnormal identification and decision optimization are performed on the production process information to obtain the final optimized decision result.

2. The method according to claim 1, characterized in that The process of preprocessing the real-time workshop data collected by sensors to obtain a data set with consistent granularity includes: Cleaning the real-time workshop data collected by the sensor to obtain a first data set; Converting the first data set into a second data set; Aligning the timestamps of the data in the first data set and the second data set using a time synchronization algorithm to obtain third data; A data interpolation algorithm is used to perform granularity unification processing on the third data set to obtain a data set with consistent granularity.

3. The method according to claim 2, characterized in that The process of data cleaning for real-time workshop data collected by sensors includes: Use data cleaning algorithm to detect and remove outliers from real-time data of workshop to obtain data with outliers removed; Performing noise elimination and normalization processing on the outlier-removed data to obtain preprocessed data of each sensor; A correlation analysis algorithm is used to calculate a correlation coefficient matrix between preprocessed data of each sensor, and based on the correlation coefficient matrix, sensor data with high redundancy is removed to obtain a first data set.

4. The method according to claim 3, characterized in that The process of converting the data format of the first data set to obtain the second data set includes: S1, identifying the data format of each piece of data in the first data set, and directly adding the data to the second data set when the data is in a standard format, and entering S2 when the data is not in a standard format; S2, converting the data in non-standard format into standard format based on data format conversion rules; S3, inspect the standard format data converted in S2, and directly add it to the second data set if it meets the inspection requirements; if it does not meet the inspection requirements, proceed to S4; S4, repeating S1-S3 until all data in the first data set are processed to obtain a complete second data set; S5. Remove duplicate data and invalid data from the complete second data set to obtain a second data set.

5. The method according to claim 4, characterized in that The process of aligning the timestamps of the data in the first data set and the second data set includes: Acquire timestamp information of the first data set and the second data set, and convert the timestamp information into a unified time format; Determine the time point corresponding to each data item in the first data set according to the timestamp information of the second data set, and calculate the time difference between the first data set and the second data set; Performing offset correction on the timestamp based on the time difference to obtain a corrected first data set and a second data set; The corrected first data set and the second data set are combined to generate a time-synchronized third data set.

6. The method according to claim 2, characterized in that The process of performing protocol conversion on the granularity consistent data set based on the communication protocol information of the production equipment includes: Analyze the differences in data granularity in the third data set to determine a data subset that needs to be processed with uniform granularity; A data interpolation algorithm is used to perform interpolation calculations between data points in the data subset that needs to be uniformly processed to generate new data points; Determine the consistency between the new data point and the original data point. If the consistency meets the preset threshold, the new data point is added to the data subset. The quality of the data subset after interpolation processing is evaluated by calculating the statistical characteristics of the data subset. When the quality evaluation result of the data subset meets the requirement of uniform granularity, the data subset after interpolation processing is merged with other unprocessed data in the third data set to obtain a fourth data set with uniform granularity.

7. The method according to claim 6, characterized in that The process of using a pre-established production process decision tree model to model and analyze the distributed processing results to obtain production process information includes: Cleaning, converting and integrating the distributed processing results to obtain standardized production process data; Use decision tree algorithm to build production process decision tree model; Based on the production process decision tree model, the distributed processing results are predicted and analyzed to obtain the optimal decision path and prediction results of each production link; Performing comparative analysis based on the predicted result and the actual value of the distribution processing result to obtain a comparative result; The distribution processing result is reclassified based on the comparison result, the production data with similar key attribute values ​​are classified into the same category, and cluster analysis is performed on the production data of each category to obtain an optimization plan for the production process.

8. An intelligent monitoring system for functional ceramic production workshops based on the Internet of Things, characterized in that: For implementing the method for intelligent monitoring of functional ceramic production workshops based on the Internet of Things according to claim 1, the system comprises: A data acquisition module, used for acquiring real-time data of all sensors in the workshop, and cleaning the real-time data to obtain a cleaned first data set; A data cleaning module, used for performing data format conversion on the first data set to obtain a second data set; A data format conversion module, used to align the timestamps of all data using a time synchronization algorithm according to the timestamp information in the second data set, to obtain a third data set after time synchronization; A time synchronization module, used for using a data interpolation algorithm to unify the granularity of the third data set according to the data granularity difference in the third data set, so as to obtain a fourth data set with consistent granularity; A data interpolation module, used to convert the fourth data set into a unified protocol format based on the communication protocol information of the production equipment to obtain a fifth data set; A protocol conversion module, used for inputting the fifth data set into the distributed data processing platform, and using a streaming computing framework to process the massive data in real time to obtain a sixth data set; A distributed data processing module, used for performing modeling analysis on the sixth data set according to a pre-established unified data model to obtain a seventh data set; A data modeling and analysis module, configured to identify abnormalities in the seventh data set by using a cluster analysis method to obtain an eighth data set; The abnormality identification and optimization decision module is used to optimize the production process in real time based on the eighth data set to obtain the final optimization decision result.

Images