RTO real-time data processing and analyzing system fused with edge calculation

Abstract

The invention discloses an RTO real-time data processing and analysis system fused with edge computing, and relates to the technical field of RTO.Industrial production real-time data including process parameters and key index time sequence data are acquired through a data acquisition module, an analysis module extracts trend term and periodic term time sequence data by using an MSTL decomposition algorithm, initial standard indexes are determined through comprehensive analysis, and the initial standard indexes are analyzed and analyzed; a final standard index is obtained based on local data fluctuation and numerical feature correction, a construction module classifies data according to process parameters and screens reference data to construct a case library, an edge calculation module preprocesses data on site and improves efficiency, and a dynamic optimization decision module combines the case library and real-time data modeling prediction, provides optimization suggestions and updates the case library. The visual module visually displays information and supports interaction, and the system enhances data processing and analysis capability, optimizes industrial production decisions, and improves efficiency and quality.

Description

Technical Field

[0001] The present invention relates to the technical field of RTO, and more specifically, to an RTO real-time data processing and analysis system integrating edge computing. Background Art

[0002] With the advancement of Industry 4.0 and intelligent manufacturing, the industrial production process increasingly relies on the collection, analysis, and decision-making of real-time data to achieve improvements in production efficiency, cost reduction, and quality optimization. As a key technology to achieve this goal in the process industry, the real-time optimization platform software (RTO) has been widely used in recent years. By tracking changes in raw material properties, on-site parameters, etc. in real time, and on the premise of meeting process and equipment constraints, the RTO uses mathematical models, mechanism models, and fast and efficient optimization and control technologies to adjust the operating parameters of production devices in real time, so as to achieve the best economic benefit operating point for device production.

[0003] However, traditional RTO systems have some limitations in data processing and analysis. On the one hand, with the expansion of industrial production scale and the improvement of automation level, the amount of data generated in industrial sites has increased explosively. Traditional RTO systems often have difficulty in efficiently processing these data in the data collection and preprocessing stages, resulting in data transmission delays and waste of computing resources. On the other hand, for complex and variable industrial production processes, traditional RTO systems also have insufficient capabilities in data analysis and modeling, and it is difficult to accurately capture trends and periodic changes in data, thus affecting the accuracy and timeliness of optimization decisions. In addition, there is also room for improvement in data classification, screening, and case base construction in traditional RTO systems. Since various process parameters and equipment types are involved in industrial production processes, how to effectively classify and screen data according to these parameters in order to provide targeted optimization suggestions for different production scenarios is an urgent problem to be solved. At the same time, with the dynamic changes in the production process and the continuous improvement of processes, the optimization model and case base also need to be updated in a timely manner to maintain their effectiveness and adaptability.

[0004] Therefore, existing RTO systems have problems such as inefficient data collection and preprocessing, insufficient analysis depth, inaccurate classification and screening, slow model update, and weak exception handling. Summary of the Invention

[0005] In order to overcome the problems of inefficient data collection and preprocessing, insufficient analysis depth, inaccurate classification and screening, slow model update, and weak exception handling in existing RTO systems, the present invention discloses an RTO real-time data processing and analysis system integrating edge computing, which can effectively solve the above technical problems.

[0006] To solve the above technical problems, the technical solution of the present invention is as follows:

[0007] An RTO real-time data processing and analysis system integrating edge computing, comprising:

[0008] A data acquisition module for obtaining real-time data in the industrial production process, where the real-time data includes time-series data of multiple process parameters and key indicators;

[0009] An analysis module for decomposing the corresponding trend-term time-series data and periodic-term time-series data from the time-series data of each key indicator; comprehensively analyzing the overall change trend of the trend-term time-series data corresponding to the time-series data of each key indicator and the change situation of the periodic-term time-series data to determine the initial standard indicator corresponding to each real-time data; based on the local data fluctuation situation and numerical characteristics of the data values in each trend-term time-series data, correcting the initial standard indicator to determine the final standard indicator of each real-time data;

[0010] A construction module for classifying all real-time data based on the process parameters in the real-time data to obtain a classification result; screening reference data from the real-time data based on the final standard indicator in each category, and constructing a reference case library based on the reference data;

[0011] An edge computing module deployed at the edge of industrial field devices, preliminarily filtering and preprocessing the collected raw data in real time, extracting key feature information, and transmitting the processed data to the analysis module;

[0012] A dynamic optimization decision-making module for modeling and simulating the industrial production process based on the reference case library and real-time data using intelligent algorithms, predicting the production results under different production parameter adjustments, evaluating the optimization space of the process flow, and providing optimization decision-making suggestions for users including adjusting production parameters, optimizing the process flow, and scheduling production resources; and dynamically updating the reference case library according to the actual production feedback;

[0013] A visualization display module for presenting the real-time data, analysis results, and optimization decision-making suggestions to the user in an intuitive visualization interface and supporting user interaction operations.

[0014] Preferably, in the analysis module: in the trend-term time-series data corresponding to the time-series data of each key indicator, performing a linear fit on all data values, obtaining the slope value of the fitted line, normalizing and negatively correlating the slope value to obtain a first standard coefficient;

[0015] In the periodic-term time-series data corresponding to the time-series data of each key indicator, obtaining all minimum value points, and taking the data between two adjacent minimum value points as a change data segment;

[0016] In all change data segments, determining a second standard coefficient according to the change situation between the data values in the change data segment;

[0017] Calculate the initial standard index corresponding to each real-time data according to the first standard coefficient and the second standard coefficient corresponding to the time-series data of each key index, and both the first standard coefficient and the second standard coefficient are positively correlated with the initial standard index;

[0018] In each change data segment, take the difference between the maximum data value and the minimum data value as the fluctuation amplitude of each change data segment;

[0019] Arrange the fluctuation amplitudes of all change data segments according to the time series of the change data segments to obtain a sorting sequence, and perform backward difference on the values in the sorting sequence to obtain a first-order difference sequence. Take the proportion of the number of positive numbers in the first-order difference sequence as the second standard coefficient.

[0020] Preferably, the method for obtaining the final standard index includes:

[0021] Based on the local data fluctuation situation in the trend item time-series data of each key index time-series data, preliminarily correct the initial standard index to obtain the corrected standard index corresponding to each real-time data;

[0022] Based on the numerical characteristics of the data values in the trend item time-series data of each key index time-series data, determine the abnormal coefficient corresponding to each real-time data;

[0023] Multiply the value obtained by performing negative correlation mapping and normalization on the abnormal coefficient corresponding to each real-time data by the corresponding corrected standard index, and perform normalization processing to obtain the final standard index corresponding to each real-time data.

[0024] Preferably, the method for obtaining the corrected standard index includes:

[0025] In each trend item time-series data, take the data between two adjacent minimum value points as a fluctuation data segment;

[0026] Analyze the fluctuation continuity situation between all fluctuation data segments to determine the fluctuation continuity coefficient;

[0027] In each fluctuation data segment, take the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment as the fluctuation coefficient; perform mean processing on the fluctuation coefficients of all fluctuation data segments to obtain the fluctuation characteristic coefficient;

[0028] According to the fluctuation continuity coefficient and the fluctuation characteristic coefficient, calculate the correction coefficient corresponding to each real-time data, and both the fluctuation continuity coefficient and the fluctuation characteristic coefficient are negatively correlated with the correction coefficient;

[0029] The value obtained by normalizing the product of the correction coefficient corresponding to each real-time data and the initial standard index is used as the corrected standard index of each real-time data.

[0030] Preferably, the method for obtaining the fluctuation continuity coefficient includes:

[0031] The difference between the corresponding times of the maximum values of two adjacent fluctuation data segments is used as the time difference coefficient. Among all the fluctuation data segments, the value obtained by performing a negative correlation mapping on the mean value of all the time difference coefficients is used as the fluctuation continuity coefficient.

[0032] Preferably, the method for obtaining the anomaly coefficient includes:

[0033] In the trend item time series data of each key index time series data, the data values greater than the preset threshold are used as target values, and the target values that are continuous in time series are screened out to obtain all subsequences;

[0034] The proportion of the number of target values in each trend item time series data is used as the quantity coefficient, and the difference between the maximum value of all target values and the preset threshold is used as the deviation coefficient; the ratio of the maximum length of all subsequences to the numerical value of the earliest time of all target values is used as the time coefficient;

[0035] The value obtained by normalizing the product of the quantity coefficient, the deviation coefficient, and the time coefficient is used as the anomaly coefficient.

[0036] Preferably, the method for obtaining the classification result includes:

[0037] The process parameters at least include the production equipment type, the production stage, and the product specifications;

[0038] Among all the real-time data, the real-time data with the same process parameters are regarded as one category, so as to obtain the classification result of all the real-time data.

[0039] Preferably, the method for obtaining the reference data includes:

[0040] In each category, all the real-time data are sorted in descending order according to the final standard index to obtain a descending sequence;

[0041] In the descending sequences corresponding to each category, the first preset number of real-time data are used as the reference data.

[0042] Preferably, decomposing the corresponding trend item time series data and periodic item time series data from each key index time series data includes:

[0043] Based on the MSTL decomposition algorithm, each key index time series data is decomposed to obtain the corresponding trend item time series data and periodic item time series data of each key index time series data.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows: By deploying an edge computing module at the edge of industrial field devices, the present invention preliminarily filters and preprocesses the collected raw data in real time, extracts key feature information, thereby solving the problem of insufficient data acquisition and preprocessing capabilities of traditional RTO systems, improving the real-time performance of data transmission and reducing waste of computing resources; The analysis module uses the MSTL decomposition algorithm to decompose the trend-term time-series data and periodic-term time-series data from the time-series data of each key index, and comprehensively analyzes to determine the initial standard index. At the same time, based on the local data fluctuations and numerical characteristics of the data values in the trend-term time-series data, the initial standard index is corrected to obtain the final standard index, solving the problem of insufficient data analysis depth in existing RTO systems, being able to more accurately capture potential laws and change characteristics in the data, and improving the scientificity and accuracy of optimization decisions; The construction module classifies all real-time data based on various process parameters in the real-time data to obtain a classification result, and screens reference data from the real-time data based on the final standard index in each category to construct a reference case library, solving the problem of inaccurate data classification and screening, making the optimization suggestions more targeted and practical; The dynamic optimization decision-making module, based on the reference case library and real-time data, uses intelligent algorithms to model and simulate the industrial production process, predict the production results under different production parameter adjustments, evaluate the optimization space of the process flow, and dynamically update the reference case library according to the actual production feedback, not only solving the problems of insufficient model update and adaptability, but also enabling the system to continuously adapt to changes in the production process and maintain the effectiveness and adaptability of optimization decisions; In addition, by determining the anomaly coefficient corresponding to each real-time data, and performing normalization processing on the product of the value obtained by performing negative correlation mapping and normalization on it and the corrected standard index to obtain the final standard index, the system solves the problem of limited ability to process abnormal data, can effectively identify and correct abnormal fluctuations and noises in the data, and avoid biases and mistakes in optimization decisions; The visualization display module presents the real-time data, analysis results, and optimization decision suggestions to the user through an intuitive visualization interface and supports user interaction operations, solving the problem of lack of a user interface, and improving the user's operation experience and decision-making efficiency. Description of the Drawings

[0045] **Detailed Implementation Modes** The following will clearly and completely describe the technical solutions in this application in conjunction with the accompanying drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. The components of this application usually described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of this application to be protected, but only represents the selected embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of this application. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0046] Figure 1 It is a structural diagram of an RTO real-time data processing and analysis system integrating edge computing. **Detailed Implementation Modes**

[0047] The accompanying drawings are only for illustrative purposes and cannot be construed as limiting the patent;

[0048] To better illustrate this embodiment, some components in the accompanying drawings will be omitted, enlarged or reduced, which does not represent the size of the actual product;

[0049] For those skilled in the art, it is understandable that some well-known structures and their descriptions in the accompanying drawings may be omitted.

[0050] The following further describes the technical solutions of the present invention in conjunction with the accompanying drawings and embodiments.

[0051] **Embodiment**

[0052] For the RTO real-time data processing and analysis system integrating edge computing, please refer to Figure 1 , including:

[0053] A data acquisition module for obtaining real-time data in the industrial production process, where the real-time data contains time-series data of multiple process parameters and key indicators;

[0054] An analysis module, which is used to decompose the corresponding trend item time series data and periodic item time series data from the time series data of each key indicator; comprehensively analyze the overall change trend of the trend item time series data corresponding to each key indicator time series data and the change of the periodic item time series data, and determine the initial standard indicator corresponding to each real-time data; based on the local data fluctuation situation and the numerical characteristics of the data values in each trend item time series data, correct the initial standard indicator, so as to determine the final standard indicator of each real-time data;

[0055] A construction module, which is used to classify all real-time data based on various process parameters in the real-time data to obtain a classification result; screen reference data from the real-time data based on the final standard indicator in each category, and construct a reference case library based on the reference data;

[0056] An edge computing module, which is deployed at the edge of industrial field devices, preliminarily filters and preprocesses the collected raw data in real time, extracts key feature information, and transmits the processed data to the analysis module;

[0057] A dynamic optimization decision-making module, which is used to model and simulate the industrial production process based on the reference case library and real-time data by using intelligent algorithms, predict the production results under different production parameter adjustments, evaluate the optimization space of the process flow, and provide optimization decision-making suggestions for users, including adjusting production parameters, optimizing the process flow, and scheduling production resources; and dynamically update the reference case library according to the actual production feedback;

[0058] A visualization display module, which is used to present the real-time data, analysis results, and optimization decision-making suggestions to users in an intuitive visualization interface and support user interaction operations.

[0059] Taking the production process optimization of a large chemical enterprise as an example, a variety of sensors, such as temperature sensors, pressure sensors, flow meters, on-line component analyzers, etc., are installed on various devices at the production site, such as reaction kettles, pumps, heat exchangers, etc. These sensors collect real-time data in the production process at a high frequency, such as once per second, and transmit the collected data to the edge computing server deployed in the workshop through a wired or wireless communication network. For example, in a large chemical plant area, multiple temperature sensors, multiple pressure sensors, and multiple flow meters are installed, and they jointly form a data acquisition network to obtain key process parameters such as temperature, pressure, and flow in the production process in real time.

[0060] The edge computing server is deployed in the workshop computer room, close to the data acquisition source. The server is installed with edge computing software, which has functions of data filtering, preprocessing and feature extraction. When the original data arrives at the edge computing server, the software first cleans the data, removing obvious error or abnormal data points, such as temperature values beyond the physical range. For example, the temperature of a reactor suddenly shows -200°C while the normal operating temperature is 50 - 150°C. Then, data normalization is carried out to convert data with different dimensions and ranges into a unified numerical range (such as 0 - 1). At the same time, key feature information is extracted, such as calculating statistical features of parameters like temperature, pressure and flow rate, including mean, standard deviation, peak value, etc., as well as the correlation coefficient between them, to reduce the data dimension and transmission volume. The processed data is transmitted to the analysis server in the central control room through the enterprise internal network to ensure the timeliness and reliability of data transmission.

[0061] The analysis server is installed with data analysis software, which uses the MSTL decomposition algorithm to decompose the time-series data of each key indicator. Taking the product quality indicator as an example, the software decomposes the time-series data of the product quality indicator into trend-term time-series data (reflecting the long-term change trend of product quality, such as gradually increasing or decreasing) and periodic-term time-series data (reflecting the periodic fluctuations of product quality, such as regular changes every day or every week). In the analysis of trend-term time-series data, a straight-line fit is performed on all data values to obtain the slope value of the fitted straight line. This slope value reflects the rate of change of product quality. For example, a slope of 0.5 means that the product quality indicator increases by 0.5 units per unit time. The slope value is normalized (mapped to the 0 - 1 interval) and undergoes a negative correlation mapping process to obtain the first standard coefficient. Assuming the normalized slope is 0.8, after the negative correlation mapping, such as subtracting the normalized value from 1, the first standard coefficient is 0.2. This indicates that the trend change of the product quality indicator has a relatively small positive contribution to the initial standard indicator. Because the negative correlation mapping means that the larger the slope (i.e., the faster the change rate), the smaller the first standard coefficient, indicating that this rapid change may not be conducive to the stability of product quality, so a lower weight is given in the calculation of the initial standard indicator.

[0062] In the periodic item time series data, all minimum points are obtained, and the data between two adjacent minimum points is used as a change data segment. For example, in the periodic item time series data of a certain product quality index, a minimum point appears every 24 hours, so the data is divided into multiple 24-hour change data segments. In each change data segment, the difference between the maximum data value and the minimum data value is calculated as the fluctuation amplitude of this change data segment. The fluctuation amplitudes of all change data segments are arranged in chronological order to form a sorting sequence, and backward differences are performed on the values in the sorting sequence to obtain a first-order difference sequence. The proportion of the number of positive numbers in the first-order difference sequence is calculated as the second standard coefficient. For example, if there are 10 fluctuation amplitude values in the sorting sequence and 6 positive numbers in the first-order difference sequence, the proportion is 0.6, that is, the second standard coefficient is 0.6. This reflects the proportion of the upward trend in the periodic fluctuation and has a positive effect on the calculation of the initial standard index because a higher upward proportion may mean more improvement opportunities for product quality in the periodic fluctuation.

[0063] According to the first standard coefficient and the second standard coefficient, with a certain weight distribution, such as each accounting for 50%, the initial standard index is calculated. Assuming the first standard coefficient is 0.2 and the second standard coefficient is 0.6, then the initial standard index is (0.2×0.5 + 0.6×0.5) = 0.4. Then, based on the local data fluctuation situation in the trend item time series data, such as calculating the mean of the absolute values of the differences between adjacent data points and the numerical characteristics of the data values, such as the mean and standard deviation of the data, the initial standard index is corrected. For example, if the local data fluctuation is large, indicating poor data stability and possibly having an adverse impact on product quality, the initial standard index is corrected downward; if the numerical characteristics of the data values show that the data is concentrated in the high-quality interval, the initial standard index is corrected upward, and finally the final standard index of each real-time data is determined.

[0064] The construction module runs on the analysis server. According to the process parameters in the real-time data, such as the production equipment types including reactor A, B, and C, the production stages are divided into raw material input, reaction progress, product separation, etc., and the product specifications are high purity, ordinary purity, etc., all real-time data are classified. For example, all real-time data that occur in reactor A, are in the reaction progress stage, and produce high-purity products are classified into one category. In each category, the top 100 real-time data are selected as reference data in descending order of the final standard index to construct a reference case library. The reference data covers high-quality production cases under different process conditions and provides a basis for subsequent optimization decisions.

[0065] The dynamic optimization decision-making module is integrated into the optimization software of the analysis server. Based on the reference case library and real-time data, it uses intelligent algorithms such as genetic algorithms and neural networks to model and simulate the chemical production process. For example, it establishes the kinetic model of chemical reactions in the reactor and simulates the changes in product quality and output under the adjustment of production parameters such as different temperatures, pressures, and raw material ratios. By comparing the simulation results with the actual production goals, such as maximizing product quality and minimizing production costs, it evaluates the optimization space of the process flow. According to the evaluation results, it provides optimization decision-making suggestions for users, such as adjusting the temperature setting value of a certain reactor and optimizing the raw material feed flow. At the same time, according to the actual production feedback, such as the production result data after implementing the optimization suggestions, it dynamically updates the reference case library, incorporates new high-quality cases into the library, and replaces old or ineffective cases to maintain the timeliness and effectiveness of the case library.

[0066] Install visualization display software on the monitoring large screen in the central control room and the workstations of the operators. This software presents real-time data, analysis results, and optimization decision-making suggestions in intuitive charts, such as line charts, bar charts, radar charts, etc., and in tabular form. For example, it uses a line chart to display the real-time change trends of the reactor temperature, pressure, and product quality indicators, uses a bar chart to compare the final standard indicators in different production stages, and uses a radar chart to present the comprehensive impact of the changes in process parameters before and after optimization on multiple production goals, such as quality, cost, and efficiency. At the same time, it supports user interaction operations. Operators can click on chart elements to view detailed data or simulate adjustments to the optimization suggestions to view the expected effects after adjustment, so as to make better production decisions.

[0067] In the analysis module: In the trend item time series data corresponding to each key indicator time series data, perform linear fitting on all data values and obtain the slope value of the fitting line. Normalize and perform negative correlation mapping on the slope value to obtain the first standard coefficient;

[0068] In the periodic item time series data corresponding to each key indicator time series data, obtain all the minimum value points, and take the data between two adjacent minimum value points as a change data segment;

[0069] In all change data segments, determine the second standard coefficient according to the change situation between the data values in the change data segment;

[0070] According to the first standard coefficient and the second standard coefficient corresponding to each key indicator time series data, calculate the initial standard index corresponding to each real-time data, and both the first standard coefficient and the second standard coefficient are positively correlated with the initial standard index;

[0071] In each change data segment, take the difference between the maximum data value and the minimum data value as the fluctuation range of each change data segment;

[0072] Arrange the fluctuation amplitudes of all variable data segments in the time sequence of the variable data segments to obtain a sorted sequence, perform backward difference on the values in the sorted sequence to obtain a first-order difference sequence, and take the proportion of the number of positive numbers in the first-order difference sequence as the second standard coefficient.

[0073] The method for obtaining the final standard index includes:

[0074] Based on the local data fluctuation conditions in the trend item time series data of each key index time series data, preliminarily correct the initial standard index to obtain the corrected standard index corresponding to each real-time data;

[0075] Based on the numerical characteristics of the data values in the trend item time series data of each key index time series data, determine the anomaly coefficient corresponding to each real-time data;

[0076] Multiply the value obtained by performing negative correlation mapping and normalization on the anomaly coefficient corresponding to each real-time data by the corresponding corrected standard index, and perform normalization processing to obtain the final standard index of each real-time data.

[0077] The method for obtaining the corrected standard index includes:

[0078] In each trend item time series data, take the data between two adjacent minimum points as a fluctuation data segment;

[0079] Analyze the fluctuation continuity between all fluctuation data segments to determine the fluctuation continuity coefficient;

[0080] In each fluctuation data segment, take the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment as the fluctuation coefficient; perform averaging processing on the fluctuation coefficients of all fluctuation data segments to obtain the fluctuation characteristic coefficient;

[0081] According to the fluctuation continuity coefficient and the fluctuation characteristic coefficient, calculate the correction coefficient corresponding to each real-time data, and both the fluctuation continuity coefficient and the fluctuation characteristic coefficient are negatively correlated with the correction coefficient;

[0082] Take the value obtained by normalizing the product of the correction coefficient corresponding to each real-time data and the initial standard index as the corrected standard index of each real-time data.

[0083] The method for obtaining the fluctuation continuity coefficient includes:

[0084] Take the difference between the corresponding times of the maximum values of two adjacent fluctuation data segments as the time difference coefficient, and in all fluctuation data segments, take the value obtained by performing negative correlation mapping on the average value of all time difference coefficients as the fluctuation continuity coefficient.

[0085] The method for obtaining the anomaly coefficient includes:

[0086] In the trend item time series data of each key index time series data, the data values greater than the preset threshold are used as target values, and the target values that are continuous in time series are filtered out to obtain all subsequences;

[0087] The proportion of the number of target values in each trend item time series data is used as the quantity coefficient, and the difference between the maximum value of all target values and the preset threshold is used as the deviation coefficient; the ratio of the maximum length of all subsequences to the numerical value of the earliest time of all target values is used as the time coefficient;

[0088] The value obtained by normalizing the product of the quantity coefficient, the deviation coefficient, and the time coefficient is used as the anomaly coefficient.

[0089] The method for obtaining the classification result includes:

[0090] The process parameters at least include the production equipment type, the production stage, and the product specifications;

[0091] Among all the real-time data, the real-time data with the same process parameters is regarded as one category, so as to obtain the classification result of all the real-time data.

[0092] The method for obtaining the reference data includes:

[0093] In each category, all the real-time data is sorted in descending order according to the final standard index to obtain a descending sequence;

[0094] In the descending sequence corresponding to each category, the first preset number of real-time data is used as the reference data.

[0095] The decomposition of the corresponding trend item time series data and periodic item time series data from each key index time series data includes:

[0096] Based on the MSTL decomposition algorithm, each key index time series data is decomposed to obtain the corresponding trend item time series data and periodic item time series data of each key index time series data.

[0097] In a specific implementation, after production starts, the data acquisition module continuously obtains the time series data of process parameters such as temperature, pressure, flow rate, and material composition in the chemical production process, as well as key indicators such as product quality indicators and equipment operating status, and transmits this data to the edge computing module.

[0098] The edge computing module performs preliminary filtering and preprocessing on the original data, extracts key feature information, such as calculating statistical features such as the mean, standard deviation, and peak value of each parameter, as well as the correlation coefficient between them, and then transmits the processed data to the analysis module.

[0099] The analysis module uses the MSTL decomposition algorithm to decompose the time-series data of each key indicator, obtaining the trend-term time-series data and the periodic-term time-series data. The slope value is obtained by performing a linear fit on the trend-term time-series data, and after normalization and negative correlation mapping processing, the first standard coefficient is obtained. The minimum value points are obtained from the periodic-term time-series data, the variable data segments are divided, the fluctuation amplitude and the proportion of positive numbers in the first-order difference sequence are calculated to obtain the second standard coefficient. The initial standard index is calculated based on the first standard coefficient and the second standard coefficient, and the initial standard index is corrected based on the local data fluctuation situation and the numerical characteristics of the data values in the trend-term time-series data to obtain the final standard index.

[0100] The construction module classifies all real-time data based on the process parameters in the real-time data, such as production equipment type, production stage, product specifications, etc., to obtain the classification results. In each category, the reference data is screened according to the descending order of the final standard index, and a reference case library is constructed based on the reference data.

[0101] The dynamic optimization decision-making module, based on the reference case library and real-time data, uses intelligent algorithms to model and simulate the chemical production process, predicts the production results under different production parameter adjustments, evaluates the optimization space of the process flow, provides optimization decision-making suggestions for users, and dynamically updates the reference case library according to the actual production feedback.

[0102] The visualization display module presents the real-time data, analysis results, and optimization decision-making suggestions to the user through an intuitive visualization interface, and supports user interaction operations. The operator can adjust the production parameters and process flow in a timely manner according to the visualization information and optimization suggestions to achieve the optimized control of the production process.

[0103] The same or similar reference numerals correspond to the same or similar components;

[0104] The terms used to describe the positional relationship in the drawings are only for illustrative purposes and should not be construed as a limitation of this patent;

[0105] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limiting the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. An RTO real-time data processing and analysis system integrating edge computing, characterized in that Including: A data acquisition module for obtaining real-time data in the industrial production process, where the real-time data includes time-series data of multiple process parameters and key indicators; An analysis module for decomposing the corresponding trend-term time-series data and periodic-term time-series data from the time-series data of each key indicator; Comprehensively analyzing the overall change trend of the trend-term time-series data corresponding to each key indicator time-series data and the change situation of the periodic-term time-series data to determine the initial standard indicator corresponding to each real-time data; Based on the local data fluctuation situation and numerical characteristics of the data values in each trend-term time-series data, correcting the initial standard indicator to determine the final standard indicator of each real-time data; A construction module for classifying all real-time data based on various process parameters in the real-time data to obtain a classification result; Screening reference data from the real-time data based on the final standard indicator in each category and constructing a reference case library based on the reference data; An edge computing module deployed at the edge of industrial field devices, which preliminarily filters and preprocesses the collected raw data in real time, extracts key feature information, and transmits the processed data to the analysis module; A dynamic optimization decision-making module for modeling and simulating the industrial production process based on the reference case library and real-time data, predicting the production results under different production parameter adjustments, evaluating the optimization space of the process flow, and providing optimization decision-making suggestions for users including adjusting production parameters, optimizing the process flow, and scheduling production resources; And dynamically updating the reference case library according to the actual production feedback; A visualization display module for presenting the real-time data, analysis results, and optimization decision-making suggestions to the user in an intuitive visualization interface and supporting user interaction operations.

2. The data processing and analysis system according to claim 1, characterized in that, In the analysis module: In the trend-term time-series data corresponding to each key indicator time-series data, perform linear fitting on all data values, obtain the slope value of the fitting line, normalize and negatively correlate the slope value to obtain the first standard coefficient; In the periodic-term time-series data corresponding to each key indicator time-series data, obtain all minimum value points, and use the data between two adjacent minimum value points as a change data segment; In all change data segments, determine the second standard coefficient according to the change situation between the data values in the change data segment; Calculate the initial standard indicator corresponding to each real-time data according to the first standard coefficient and the second standard coefficient corresponding to each key indicator time-series data, and both the first standard coefficient and the second standard coefficient are positively correlated with the initial standard indicator; In each change data segment, use the difference between the maximum data value and the minimum data value as the fluctuation amplitude of each change data segment; Arrange the fluctuation amplitudes of all change data segments in the order of time series of the change data segments to obtain a sorting sequence, perform backward difference on the values in the sorting sequence to obtain a first-order difference sequence, and use the proportion of the number of positive numbers in the first-order difference sequence as the second standard coefficient.

3. The data processing and analysis system according to claim 1, wherein The method for obtaining the final standard indicator includes: Based on the local data fluctuation in the trend item time series data of each key indicator time series data, the initial standard indicator is preliminarily corrected to obtain the corrected standard indicator corresponding to each real-time data; Based on the numerical characteristics of the data values in the trend item time series data of each key indicator time series data, the anomaly coefficient corresponding to each real-time data is determined; The product of the value obtained by performing negative correlation mapping and normalization on the anomaly coefficient corresponding to each real-time data and the corresponding corrected standard indicator is normalized to obtain the final standard indicator of each real-time data.

4. The data processing and analysis system according to claim 3, wherein The method for obtaining the corrected standard indicator includes: In each trend item time series data, the data between two adjacent minimum value points is used as a fluctuation data segment; Analyze the fluctuation continuity between all fluctuation data segments to determine the fluctuation continuity coefficient; In each fluctuation data segment, the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment is used as the fluctuation coefficient; the fluctuation coefficients of all fluctuation data segments are averaged to obtain the fluctuation characteristic coefficient; According to the fluctuation continuity coefficient and the fluctuation characteristic coefficient, calculate the correction coefficient corresponding to each real-time data, and both the fluctuation continuity coefficient and the fluctuation characteristic coefficient are negatively correlated with the correction coefficient; The value obtained by normalizing the product of the correction coefficient corresponding to each real-time data and the initial standard indicator is used as the corrected standard indicator of each real-time data.

5. The data processing and analysis system according to claim 4, wherein The method for obtaining the fluctuation continuity coefficient includes: The difference between the corresponding times of the maximum values of two adjacent fluctuation data segments is used as the time difference coefficient. Among all fluctuation data segments, the value obtained by performing negative correlation mapping on the average value of all time difference coefficients is used as the fluctuation continuity coefficient.

6. The data processing and analysis system according to claim 3, characterized in that, The method for obtaining the anomaly coefficient includes: In the trend item time series data of each key indicator time series data, the data values greater than the preset threshold are used as target values, and the sequentially continuous target values are screened out to obtain all subsequences; The proportion of the number of target values in each trend item time series data is used as the quantity coefficient, and the difference between the maximum value of all target values and the preset threshold is used as the deviation coefficient; the ratio of the maximum length of all subsequences to the numerical value of the earliest time of all target values is used as the time coefficient; The value obtained by normalizing the product of the quantity coefficient, the deviation coefficient, and the time coefficient is used as the anomaly coefficient.

7. The data processing and analysis system according to claim 1, wherein The method for obtaining the classification result includes: The process parameters at least include the production equipment type, the production stage, and the product specification; Among all real-time data, the real-time data with the same process parameters are grouped into one category, thereby obtaining the classification result of all real-time data.

8. The data processing and analysis system according to claim 1, wherein The method for obtaining the reference data includes: In each category, all real-time data are sorted in descending order according to the final standard indicator to obtain a descending sequence; In the descending sequences corresponding to each category, the first preset number of real-time data are used as the reference data.

9. The data processing and analysis system according to claim 1, wherein The decomposition of the corresponding trend item time series data and periodic item time series data from each key indicator time series data includes: Decompose the time series data of each key indicator based on the MSTL decomposition algorithm to obtain the trend item time series data and the periodic item time series data corresponding to the time series data of each key indicator.

Images