RTU real-time monitoring method and system in industrial intelligent scenarios

Abstract

The present application provides a method and system for real-time monitoring of RTUs in an industrial intelligent scenario, which relates to the field of wireless communication technology, including: inputting production line load information, environmental prediction information and monitoring indicator sets into a data trend prediction channel, and outputting an indicator fluctuation prediction trend set; based on the indicator fluctuation prediction trend set, time-segmenting a predetermined period to determine multiple identification monitoring interval sets; generating multiple optimized monitoring frequency sets based on multiple frequency correction coefficient sets and a reference monitoring frequency table; generating an expected monitoring plan based on multiple identification monitoring interval sets and multiple optimized monitoring frequency sets. This application can solve the technical problem that the existing RTU monitoring method applied to industrial production lines has poor data monitoring accuracy and quality due to the inability to set a matching data monitoring frequency according to actual monitoring needs, and can achieve the technical effect of improving data monitoring accuracy and quality and reducing the waste of monitoring resources.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a method and system for real-time monitoring of an RTU in an industrial intelligent scenario. Background Art

[0002] RTU (Remote Terminal Unit) is a device used to remotely monitor and control industrial processes. It usually includes sensors, processors, communication interfaces and user interfaces. The main function of RTU is to convert the data collected by field sensors into a readable format and transmit the data to the central monitoring system through the communication network. Existing RTU monitoring methods usually use a static monitoring frequency, that is, a fixed time interval. This method cannot adapt to the dynamic changes of the production line and has great limitations.

[0003] At present, the existing RTU monitoring methods used in industrial production lines have technical problems such as the inability to set a matching data monitoring frequency according to actual monitoring needs, resulting in poor data monitoring accuracy and quality, and causing waste of monitoring resources. Summary of the invention

[0004] The purpose of this application is to provide an RTU real-time monitoring method and system in an industrial intelligent scenario, so as to solve the technical problem that the existing RTU monitoring method applied to industrial production lines cannot set a matching data monitoring frequency according to actual monitoring needs, resulting in poor data monitoring accuracy and monitoring quality, and causing waste of monitoring resources.

[0005] In view of the above problems, the present application provides an RTU real-time monitoring method and system in an industrial intelligent scenario.

[0006] In the first aspect, the present application provides an RTU real-time monitoring method in an industrial intelligent scenario, which is implemented by an RTU real-time monitoring system in an industrial intelligent scenario, including: obtaining a monitoring indicator set of a target production line, performing a change speed analysis and an impact evaluation on the monitoring indicator set, and determining a change rate set and an impact coefficient set; determining a benchmark monitoring frequency based on the change rate and the impact coefficient, and constructing a benchmark monitoring frequency table based on the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the impact coefficient; collecting production line load information and environmental prediction information within a predetermined period, inputting the production line load information, environmental prediction information and the monitoring indicator set into a data trend prediction channel, and outputting an indicator fluctuation prediction trend set; based on the indicator fluctuation prediction trend set, performing time division and type identification on the predetermined period according to a predetermined interval type, and determining multiple identified monitoring interval sets; obtaining multiple frequency correction coefficient sets based on matching the multiple identified monitoring interval sets, and generating multiple optimized monitoring frequency sets based on the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; generating an expected monitoring plan based on the multiple identified monitoring interval sets and the multiple optimized monitoring frequency sets, and performing target production line monitoring within a predetermined period according to the expected monitoring plan.

[0007] In the second aspect, the present application also provides an RTU real-time monitoring system in an industrial intelligent scenario, which is used to execute an RTU real-time monitoring method in an industrial intelligent scenario as described in the first aspect, including: a monitoring indicator analysis module, which is used to obtain a monitoring indicator set of a target production line, perform change speed analysis and impact assessment on the monitoring indicator set, and determine a change rate set and an impact coefficient set; a benchmark monitoring frequency determination module, which is used to determine a benchmark monitoring frequency based on the change rate and the impact coefficient, and construct a benchmark monitoring frequency table based on the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the impact coefficient; an indicator fluctuation trend prediction module, which is used to collect production line load information and environmental prediction information within a predetermined period, and convert the said The production line load information, environmental prediction information and monitoring indicator set are input into the data trend prediction channel, and the indicator fluctuation prediction trend set is output; the identification monitoring interval determination module is used to determine multiple identification monitoring interval sets by time segmentation and type identification of the predetermined period according to the predetermined interval type based on the indicator fluctuation prediction trend set; the benchmark monitoring frequency optimization module is used to obtain multiple frequency correction coefficient sets based on the matching of the multiple identification monitoring interval sets, and generate multiple optimized monitoring frequency sets based on the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; the target production line monitoring module is used to generate an expected monitoring plan based on the multiple identification monitoring interval sets and the multiple optimized monitoring frequency sets, and perform target production line monitoring within the predetermined period according to the expected monitoring plan.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By analyzing the change speed and impact evaluation of the monitoring indicators of the target production line, and setting the benchmark monitoring frequency based on the change rate and impact coefficient, the adaptability of the benchmark monitoring frequency to the monitoring indicator type can be improved; then, based on the data trend prediction channel constructed by the BP neural network, the production line load information, environmental prediction information and monitoring indicator set within the predetermined period are input into the data trend prediction channel for prediction, and the indicator fluctuation prediction trend set is obtained; further based on the indicator fluctuation prediction trend set, the predetermined period is time-divided and type-identified according to the predetermined interval type, and multiple identified monitoring interval sets are determined, and then multiple frequency correction coefficient sets are obtained according to the matching of the multiple identified monitoring interval sets, and multiple optimized monitoring frequency sets are generated according to the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; finally, the expected monitoring plan is generated based on the multiple identified monitoring interval sets and the multiple optimized monitoring frequency sets, and the target production line monitoring within the predetermined period is executed; by automatically adjusting the RTU monitoring frequency according to the production line equipment status and operation trend, the adaptability of the data monitoring frequency to the production line monitoring needs can be improved, thereby achieving the technical goal of improving the accuracy and quality of data monitoring, reducing the waste of monitoring resources, and achieving the technical effect of ensuring the safe and stable operation of the industrial production line.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 A flowchart of a RTU real-time monitoring method in an industrial intelligent scenario is provided for this application;

[0013] Figure 2 A schematic diagram of a process for determining an influence coefficient set in a RTU real-time monitoring method in an industrial intelligent scenario in this application;

[0014] Figure 3 This application provides a structural diagram of an RTU real-time monitoring system in an industrial intelligent scenario.

[0015] Description of reference numerals:

[0016] Monitoring indicator analysis module 11, benchmark monitoring frequency determination module 12, indicator fluctuation trend prediction module 13, identification monitoring interval determination module 14, benchmark monitoring frequency optimization module 15, target production line monitoring module 16. DETAILED DESCRIPTION

[0017] This application provides an RTU real-time monitoring method and system in an industrial intelligent scenario, which solves the technical problem that the existing RTU monitoring method applied to industrial production lines cannot set a matching data monitoring frequency according to actual monitoring needs, resulting in poor data monitoring accuracy and quality, and waste of monitoring resources. By automatically adjusting the RTU monitoring frequency according to the production line equipment status and operation trend, the adaptability of the data monitoring frequency to the production line monitoring needs can be improved, thereby achieving the technical goal of improving data monitoring accuracy and quality, reducing the waste of monitoring resources, and achieving the technical effect of ensuring the safe and stable operation of the industrial production line.

[0018] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0019] For example, please refer to the attached Figure 1 The present application provides an RTU real-time monitoring method in an industrial intelligent scenario, which is applied to an RTU real-time monitoring system in an industrial intelligent scenario, and specifically includes the following steps:

[0020] Step 1: Obtain a monitoring indicator set of a target production line, perform a change speed analysis and impact assessment on the monitoring indicator set, and determine a change rate set and an impact coefficient set.

[0021] Specifically, first, determine the target production line, where the target production line is an industrial intelligent production line, which can be set according to actual conditions, such as food processing production lines, automobile manufacturing production lines, chemical production lines, etc. Then conduct a monitoring demand analysis on the target production line, determine the key parameters and potential failure modes that need to be monitored, and analyze the process flow of the production line to determine the key process steps and links that may affect product quality, and obtain the monitoring indicator set of the target production line, where the monitoring indicators include temperature, pressure, flow, speed, vibration and other parameters.

[0022] After obtaining the monitoring indicator set, the multiple monitoring indicators in the monitoring indicator set are respectively subjected to change speed analysis and impact assessment. Change speed analysis refers to calculating the change speed of each monitoring indicator, that is, the ratio of the change amount of data to the change amount of time. Impact assessment refers to assessing the degree of influence of each monitoring indicator on production line performance and product quality. Then, the change rate set is determined according to the change speed analysis results, and the impact coefficient set is determined according to the impact assessment results. Each monitoring indicator includes a change rate and an impact coefficient. The greater the change rate, the faster the change speed of the indicator. The greater the impact coefficient, the greater the impact of the indicator.

[0023] Get the set of change rates and the set of influence coefficients.

[0024] Step 2: Determine the benchmark monitoring frequency based on the change rate and the impact coefficient, and construct a benchmark monitoring frequency table based on the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the impact coefficient.

[0025] Specifically, first, set the initial monitoring frequency benchmark value, such as every minute, every three minutes, etc.; then adjust the initial monitoring frequency benchmark value according to the change rate and impact coefficient to determine the benchmark monitoring frequency, where the benchmark monitoring frequency is positively correlated with the change rate and impact coefficient, that is, the faster the change rate and the higher the impact coefficient, the higher the benchmark monitoring frequency should be; for example, the change rate and impact coefficient can be mapped to the benchmark monitoring frequency by constructing a benchmark monitoring frequency adjustment model, where the model can be a simple linear relationship or a more complex nonlinear relationship, such as a machine learning model; then use the established model to calculate the adjusted benchmark monitoring frequency for each monitoring indicator; and obtain multiple benchmark monitoring frequencies for multiple monitoring indicators. Finally, construct a benchmark monitoring frequency table based on the mapping relationship between multiple monitoring indicators and multiple benchmark monitoring frequencies.

[0026] Step 3: Collect the production line load information and environmental prediction information within a predetermined period, input the production line load information, environmental prediction information and monitoring indicator set into the data trend prediction channel, and output the indicator fluctuation prediction trend set.

[0027] Specifically, first, configure a predetermined period, which is a period of time in the future and can be set according to actual scenarios and needs; then collect production line load information and environmental forecast information within the predetermined period, and the production line load information can be set according to the expected production plan. For example, according to the expected production plan, set the key load parameters of the production line, such as current, voltage, power, etc.; environmental forecast information can be obtained through weather forecasts and other means, including multiple environmental parameters that affect the equipment or products of the target production line, such as ambient temperature, ambient humidity, etc.

[0028] Then, a data trend prediction channel is constructed based on the BP neural network. The data trend prediction channel is a BP neural network model that can be iteratively optimized in machine learning. It is obtained through supervised training with sample data and is used to predict the fluctuation trend of monitoring indicators within a predetermined period. The data trend prediction channel includes multiple trend prediction channels, and each monitoring indicator corresponds to a trend prediction channel; the input data of the trend prediction channel is the production line load information and environmental prediction information, and the output data is the data fluctuation trend.

[0029] Then, the monitoring indicator set is input into the data trend prediction channel, and multiple trend prediction channels corresponding to the monitoring indicators are matched based on the monitoring indicator set. Then, the production line load information and environmental prediction information are input into multiple trend prediction channels for data fluctuation trend prediction, and the indicator fluctuation prediction trend set is output. By obtaining the indicator fluctuation prediction trend set, a basis is provided for the next step to set a matching monitoring frequency according to the indicator fluctuation prediction trend, thereby improving the adaptability of the data monitoring frequency to the production line monitoring needs.

[0030] Step 4: Based on the indicator fluctuation prediction trend set, the predetermined period is divided into time periods and type-identified according to the predetermined interval type, and a plurality of identified monitoring interval sets are determined.

[0031] Specifically, a set of predetermined interval types is obtained, wherein the set of predetermined interval types is obtained based on the state characteristics of the monitoring indicators, and the predetermined interval types correspond to the monitoring indicators one by one, including steady-state intervals, fluctuation intervals and abnormal intervals, and the abnormal intervals contain fluctuation intervals, and the fluctuation intervals contain steady-state intervals. The steady-state interval refers to the monitoring indicator remaining relatively stable for a period of time without significant fluctuations. In this interval, the indicator value fluctuates within a certain range, but the fluctuation amplitude is small, and usually does not affect the production process. The steady-state interval is part of normal operation and usually requires a lower monitoring frequency; the fluctuation interval refers to the monitoring indicator value fluctuating within a certain range, but the fluctuation amplitude is greater than the steady-state interval. The indicator value in this interval may have a certain impact on the production process, but usually within an acceptable range, the fluctuation interval requires a higher monitoring frequency than the steady-state interval; the abnormal interval refers to the monitoring indicator value exceeding the normal fluctuation range, with abnormal fluctuations or significant declines or increases. The indicator value in this interval may have a serious impact on the production process, and even cause equipment failure or production accidents, so the abnormal interval requires the highest monitoring frequency.

[0032] Then, based on the indicator fluctuation prediction trend set, the predetermined period is divided into time and type-identified according to the predetermined interval type, that is, according to the fluctuation prediction trend, the time within the predetermined period is divided into multiple small time periods, and each small time period corresponds to one or more fluctuation prediction trends; then each small time period is type-identified to determine whether it belongs to a steady-state interval, a fluctuation interval or an abnormal interval; finally, according to the type identification, multiple identified monitoring interval sets are determined, each monitoring interval set contains one or more small time periods belonging to the same type, and multiple identified monitoring interval sets are obtained.

[0033] Step 5: Acquire multiple frequency correction coefficient sets according to the matching of the multiple identified monitoring interval sets, and generate multiple optimized monitoring frequency sets based on the multiple frequency correction coefficient sets and the reference monitoring frequency table.

[0034] Specifically, first, according to the type of the identified monitoring interval set, the corresponding frequency correction coefficient is matched from the frequency correction coefficient table. The frequency correction coefficient table is determined based on the change rate and the influence coefficient, and is used to adjust the baseline monitoring frequency. For example, the steady-state monitoring interval requires a lower monitoring frequency, and the correction coefficient is set to 0.4. The abnormal monitoring interval requires a higher monitoring frequency, and the correction coefficient is set to 3; multiple frequency correction coefficient sets are obtained. Then, for each identified monitoring interval set, the baseline monitoring frequency is adjusted according to the matched frequency correction coefficient, that is, the optimized monitoring frequency is obtained by multiplying the baseline monitoring frequency by the frequency correction coefficient, and multiple optimized monitoring frequency sets are obtained.

[0035] Step six: Generate an expected monitoring plan based on the multiple identification monitoring interval sets and the multiple optimized monitoring frequency sets, and perform target production line monitoring within a predetermined period according to the expected monitoring plan.

[0036] Specifically, multiple identification monitoring interval sets and multiple optimized monitoring frequency sets are integrated to form a complete expected monitoring plan, which should include the monitoring frequency and duration of each monitoring interval set. Then, based on the expected monitoring plan, an expected monitoring plan is generated, which should list in detail each monitoring indicator, the corresponding monitoring interval set, the monitoring frequency, and the monitoring time point. Finally, according to the expected monitoring plan, the RTU is controlled to perform the target production line monitoring within the predetermined period.

[0037] The RTU real-time monitoring method in an industrial intelligent scenario is applied to an RTU real-time monitoring system in an industrial intelligent scenario, which can solve the technical problems in the existing RTU monitoring method applied to industrial production lines, such as the inability to set a matching data monitoring frequency according to actual monitoring needs, resulting in poor data monitoring accuracy and monitoring quality, and causing waste of monitoring resources. By analyzing the change speed and impact evaluation of the monitoring indicators of the target production line, and setting the benchmark monitoring frequency based on the change rate and impact coefficient, the adaptability of the benchmark monitoring frequency to the monitoring indicator type can be improved; then, based on the data trend prediction channel constructed by the BP neural network, the production line load information, environmental prediction information and monitoring indicator set within the predetermined period are input into the data trend prediction channel for prediction, and the indicator fluctuation prediction trend set is obtained; further based on the indicator fluctuation prediction trend set, the predetermined period is time-divided and type-identified according to the predetermined interval type, and multiple identified monitoring interval sets are determined, and then multiple frequency correction coefficient sets are obtained according to the matching of the multiple identified monitoring interval sets, and multiple optimized monitoring frequency sets are generated according to the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; finally, the expected monitoring plan is generated based on the multiple identified monitoring interval sets and the multiple optimized monitoring frequency sets, and the target production line monitoring within the predetermined period is executed; by automatically adjusting the RTU monitoring frequency according to the production line equipment status and operation trend, the adaptability of the data monitoring frequency to the production line monitoring needs can be improved, thereby achieving the technical goal of improving the accuracy and quality of data monitoring, reducing the waste of monitoring resources, and achieving the technical effect of ensuring the safe and stable operation of the industrial production line.

[0038] Further, to determine the change rate set, step 1 of the present application includes:

[0039] A first monitoring indicator is randomly selected from the monitoring indicator set, and a sample change rate set is obtained based on industrial Internet retrieval with the target production line and the first monitoring indicator as constraints, wherein the change rate is the ratio of the indicator change amount to the time change amount; a mean is calculated for the sample change rate set, and the sample change rate set is screened based on the sample change rate mean to determine a standard change rate set, and the standard change rate mean is set as the first indicator change rate, and added to the change rate set.

[0040] Specifically, first, any one monitoring indicator in the monitoring indicator set is randomly selected as the first monitoring indicator, for example, temperature is selected as the first monitoring indicator; then the target production line and the first monitoring indicator are used as retrieval constraints, and positive sample retrieval is performed based on the industrial Internet. Positive sample retrieval is an information retrieval technology that aims to retrieve positive sample data that meet specific conditions from a large number of data sets. In the field of industrial monitoring, positive samples usually refer to sample data that meet the constraints and meet normal operating conditions or expectations, and a set of sample change rates is obtained, where the change rate is the ratio of the indicator change to the time change.

[0041] Next, the sample change rate set is averaged to obtain the sample change rate mean, and then the sample change rate set is screened according to the sample change rate mean, that is, the sample change rate mean and the multiple sample change rates in the sample change rate set are respectively calculated for deviation, and the sample change rate whose deviation calculation result is less than the predetermined rate deviation threshold is selected and set as the standard change rate to obtain the standard change rate set. Finally, the multiple standard change rates in the standard change rate set are averaged, and the standard change rate mean is set as the first indicator change rate; the same method is used to sequentially obtain multiple indicator change rates of multiple monitoring indicators, and a change rate set is constructed based on the multiple indicator change rates. By screening the sample change rate set, determining the standard change rate set, and setting the indicator change rate according to the standard change rate set, the accuracy and reliability of the indicator change rate setting can be improved, thereby indirectly improving the accuracy of the benchmark monitoring frequency setting.

[0042] Further, determine the influence coefficient set, as shown in the attached Figure 2 As shown, step one of this application includes:

[0043] Taking the target production line as a constraint, retrieve and obtain a set of sample production line performance abnormality events and a set of sample product quality defect events; based on the set of sample production line performance abnormality events and the set of sample product quality defect events, perform correlation tracing analysis on the monitoring indicator set, count the number of occurrences of related indicators, and obtain a monitoring indicator frequency set; perform an impact assessment on the monitoring indicator set based on the monitoring indicator frequency set, and determine the impact coefficient set, wherein the impact coefficient is positively correlated with the monitoring indicator frequency.

[0044] Specifically, taking the target production line as a constraint, positive sample retrieval is performed based on the industrial Internet to obtain a set of abnormal performance events of the sample production line and a set of quality defect events of the sample product. Then, based on the set of abnormal performance events of the sample production line and the set of quality defect events of the sample product, the monitoring indicator set is subjected to correlation traceability analysis, that is, the monitoring indicator values ​​related to each abnormal event and defect event are extracted from the data set to determine which monitoring indicators are associated with the occurrence of abnormal events and defect events; the number of occurrences of each associated indicator in abnormal events and defect events is further analyzed, and the frequency of occurrence of each associated indicator in the event set is counted. The statistical results can reflect the degree of association between each indicator and abnormal events and defect events, and a monitoring indicator frequency set is obtained. Then, the monitoring indicator set is evaluated for its influence according to the monitoring indicator frequency set, where the influence coefficient is positively correlated with the monitoring indicator frequency, that is, the indicator with a higher frequency of occurrence has a stronger correlation with abnormal events and defect events, and the higher the frequency of occurrence, the greater the influence coefficient, and multiple influence coefficients of multiple monitoring indicators are obtained to construct an influence coefficient set.

[0045] Further, to construct a data trend prediction channel, step three of this application includes:

[0046] A first monitoring indicator is randomly selected from the monitoring indicator set, and the historical monitoring records of the first monitoring indicator are queried to obtain a sample prediction data set, wherein the sample prediction data includes sample production line load information, sample environmental information and sample indicator fluctuation trend, and the sample indicator fluctuation trend is a fluctuation curve of the first monitoring indicator within a predetermined historical period; the sample prediction data set is equally divided into K parts to obtain K data sets, and K data sets are selected with replacement K times in the K data sets to construct a first training set, and K training sets are selected iteratively K times to obtain K training sets; the BP neural network is supervised and trained using the K training sets to obtain K converged first prediction units, and a first trend prediction channel is built based on the K converged first prediction units, wherein the output of the first trend prediction channel is a fitting curve of the output results of the K converged first prediction units; multiple trend prediction channels for multiple monitoring indicators are sequentially constructed, and the data trend prediction channel is generated based on the multiple trend prediction channels.

[0047] Specifically, first, randomly select any one monitoring indicator from the monitoring indicator set and set it as the first monitoring indicator; then query the historical monitoring records of the first monitoring indicator, and obtain a sample prediction data set based on the historical monitoring records, wherein the sample prediction data includes sample production line load information, sample environmental information, and sample indicator fluctuation trend, and the sample indicator fluctuation trend is the fluctuation curve of the first monitoring indicator within a historical predetermined period. Then, the sample prediction data set is divided into K parts to obtain K data sets, and K data sets are selected with replacement K times to construct a first training set, and iteratively selected K times to obtain K training sets.

[0048] In the field of industrial monitoring and prediction, BP neural network is a commonly used supervised learning algorithm, which is used to learn patterns from input data and used for prediction; with sample production line load information and sample environmental information as input, and sample indicator fluctuation trend as supervision, the K training sets are used to supervise the BP neural network training, that is, each training set is used to train the BP neural network, and the network parameters are adjusted through forward propagation and back propagation algorithms to minimize the error between the predicted value and the actual value. The training process is repeated until the network converges or reaches a preset number of training iterations, and K converged first prediction units are obtained; then, a first trend prediction channel is built based on the K converged first prediction units, wherein the output of the first trend prediction channel is a fitting curve of the output results of the K converged first prediction units. Multiple trend prediction channels for multiple monitoring indicators are constructed in sequence using the same method, and the data trend prediction channel is generated based on the multiple trend prediction channels.

[0049] Further, multiple identification monitoring interval sets are determined, and step 4 of this application includes:

[0050] A set of predetermined interval types is configured, wherein the predetermined interval types correspond to monitoring indicators one-to-one, including steady-state intervals, fluctuation intervals and abnormal intervals, and the abnormal intervals include fluctuation intervals, and the fluctuation intervals include steady-state intervals; a first indicator fluctuation prediction trend is randomly selected from the indicator fluctuation prediction trend set, and based on the first indicator fluctuation prediction trend, a predetermined period is time-divided and type-identified according to the first predetermined interval type to determine a first identified monitoring interval set, and added to the multiple identified monitoring interval sets.

[0051] Specifically, first, configure the set of predetermined interval types, where the predetermined interval types correspond to the monitoring indicators one by one, including steady-state intervals, fluctuation intervals and abnormal intervals, and the abnormal intervals contain fluctuation intervals, and the fluctuation intervals contain steady-state intervals. The steady-state interval refers to the monitoring indicator remaining relatively stable for a period of time without significant fluctuations. In this interval, the indicator value fluctuates within a certain range, but the fluctuation amplitude is small, and usually does not affect the production process. The steady-state interval is part of normal operation and usually requires a lower monitoring frequency; the fluctuation interval refers to the monitoring indicator value fluctuating within a certain range, but the fluctuation amplitude is greater than the steady-state interval. The indicator value in this interval may have a certain impact on the production process, but usually within an acceptable range, the fluctuation interval requires a higher monitoring frequency than the steady-state interval; the abnormal interval refers to the monitoring indicator value exceeding the normal fluctuation range, with abnormal fluctuations or significant declines or increases. The indicator value in this interval may have a serious impact on the production process, and even cause equipment failure or production accidents, so the abnormal interval requires the highest monitoring frequency.

[0052] Then, a first indicator fluctuation prediction trend is randomly selected from the indicator fluctuation prediction trend set, and based on the first indicator fluctuation prediction trend, a predetermined period is time-divided and type-identified according to a first predetermined interval type, wherein the first predetermined interval type and the first indicator fluctuation prediction trend have a corresponding relationship, and a first identified monitoring interval set is obtained and added to the multiple identified monitoring interval sets.

[0053] Further, the predetermined period is divided into time periods and type-identified according to the first predetermined interval type, and the present application further includes the following steps:

[0054] The first predetermined interval types include a first steady-state interval, a first fluctuation interval and a first abnormal interval; based on the first indicator fluctuation prediction trend, the predetermined period is divided into time and type-identified according to the first steady-state interval, the first fluctuation interval and the first abnormal interval to obtain a first steady-state monitoring interval set, a first fluctuation monitoring interval set and a first abnormal monitoring interval set, wherein the time period corresponding to the first indicator fluctuation prediction trend in the first steady-state interval is identified as a steady-state monitoring interval, the time period corresponding to the first indicator fluctuation prediction trend in the first fluctuation interval and not belonging to the first steady-state interval is identified as a fluctuation monitoring interval, and the time period corresponding to the first indicator fluctuation prediction trend in the first abnormal interval and not belonging to the first fluctuation interval is identified as an abnormal monitoring interval; the first identified monitoring interval set is constructed based on the first steady-state monitoring interval set, the first fluctuation monitoring interval set and the first abnormal monitoring interval set.

[0055] Specifically, the first predetermined interval type includes a first steady-state interval, a first fluctuation interval, and a first abnormal interval; then, based on the first indicator fluctuation prediction trend, the predetermined period is divided into time and type-identified according to the first steady-state interval, the first fluctuation interval, and the first abnormal interval, and the time period corresponding to the first indicator fluctuation prediction trend in the first steady-state interval is identified as a steady-state monitoring interval, the time period corresponding to the first indicator fluctuation prediction trend in the first fluctuation interval and not belonging to the first steady-state interval is identified as a fluctuation monitoring interval, and the time period corresponding to the first indicator fluctuation prediction trend in the first abnormal interval and not belonging to the first fluctuation interval is identified as an abnormal monitoring interval, thereby obtaining a first steady-state monitoring interval set, a first fluctuation monitoring interval set, and a first abnormal monitoring interval set. Finally, a first identified monitoring interval set is constructed according to the first steady-state monitoring interval set, the first fluctuation monitoring interval set, and the first abnormal monitoring interval set.

[0056] Further, according to the matching of the multiple identification monitoring interval sets, multiple frequency correction coefficient sets are obtained. Step five of the present application includes:

[0057] Taking the target production line, the first monitoring indicator and the first steady-state interval as constraints, a sample frequency data set is retrieved and obtained, wherein the sample frequency data includes a sample correction coefficient and a sample event status, and the sample event status includes a normal state and an abnormal state caused by a monitoring frequency deviation; the sample frequency data set is clustered based on a predetermined coefficient step, the proportion of abnormal events in multiple clustering results is counted, and the minimum sample correction coefficient whose abnormal event proportion is less than a predetermined abnormal threshold is selected and set as a first steady-state correction coefficient; a first fluctuation correction coefficient and a first abnormal correction coefficient are obtained in sequence, and a first indicator correction coefficient is obtained in combination with the first steady-state correction coefficient; based on a decision tree, a frequency correction database is constructed according to the mapping association between the first monitoring indicator and the first indicator correction coefficient; the multiple identification monitoring interval sets are matched according to the frequency correction database, and multiple frequency correction coefficient sets are output.

[0058] Specifically, first, with the target production line, the first monitoring indicator and the first steady-state interval as constraints, a positive sample search is performed based on the industrial Internet to obtain a sample frequency data set, wherein the sample frequency data includes a sample correction coefficient and a sample event state, the sample correction coefficient refers to the correction coefficient of the original monitoring frequency, and the sample event state includes a normal state and an abnormal state caused by a monitoring frequency deviation, such as a production line abnormality or a product abnormality caused by a monitoring frequency deviation. Then, the sample frequency data set is clustered based on a predetermined coefficient step size. First, a predetermined coefficient step size is determined, which can be set according to the division accuracy requirements, such as setting the coefficient step size to 0.05; then, the sample frequency data set is clustered based on the predetermined coefficient step size, that is, the sample frequency data with a coefficient deviation within 0.05 is clustered into one category, and multiple clustering results are obtained.

[0059] Further statistics are made on the proportion of abnormal events in multiple clustering results, and the minimum sample correction coefficient whose abnormal event proportion is less than the predetermined abnormal threshold is selected as the first steady-state correction coefficient, that is, under the premise of satisfying the monitoring quality, the monitoring resource utilization rate is improved; then the first fluctuation correction coefficient and the first abnormal correction coefficient are obtained in turn by the same method, and the first indicator correction coefficient is obtained by combining the first steady-state correction coefficient; the decision tree is a tree structure, in which each node represents a feature or attribute, each branch represents a certain value of a feature, and each leaf node represents a decision or category; multiple indicator correction coefficients of multiple monitoring indicators are obtained in turn, and further based on the decision tree principle, the frequency correction database is constructed with the monitoring indicator as the child node and the corresponding indicator correction coefficient as the leaf node of the child node. Finally, the multiple identification monitoring interval sets are matched according to the frequency correction database, and multiple frequency correction coefficient sets are output.

[0060] In summary, the RTU real-time monitoring method in an industrial intelligent scenario provided by this application has the following technical effects:

[0061] By analyzing the change speed and impact evaluation of the monitoring indicators of the target production line, and setting the benchmark monitoring frequency based on the change rate and impact coefficient, the adaptability of the benchmark monitoring frequency to the monitoring indicator type can be improved; then, based on the data trend prediction channel constructed by the BP neural network, the production line load information, environmental prediction information and monitoring indicator set within the predetermined period are input into the data trend prediction channel for prediction, and the indicator fluctuation prediction trend set is obtained; further based on the indicator fluctuation prediction trend set, the predetermined period is time-divided and type-identified according to the predetermined interval type, and multiple identified monitoring interval sets are determined, and then multiple frequency correction coefficient sets are obtained according to the matching of the multiple identified monitoring interval sets, and multiple optimized monitoring frequency sets are generated according to the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; finally, the expected monitoring plan is generated based on the multiple identified monitoring interval sets and the multiple optimized monitoring frequency sets, and the target production line monitoring within the predetermined period is executed; by automatically adjusting the RTU monitoring frequency according to the production line equipment status and operation trend, the adaptability of the data monitoring frequency to the production line monitoring needs can be improved, thereby achieving the technical goal of improving the accuracy and quality of data monitoring, reducing the waste of monitoring resources, and achieving the technical effect of ensuring the safe and stable operation of the industrial production line.

[0062] Embodiment 2, based on the RTU real-time monitoring method in an industrial intelligent scenario in the above embodiment, the same inventive concept, this application also provides an RTU real-time monitoring system in an industrial intelligent scenario, please refer to the attached Figure 3 ,include:

[0063] The monitoring indicator analysis module 11 is used to obtain the monitoring indicator set of the target production line, analyze the change speed and evaluate the impact of the monitoring indicator set, and determine the change rate set and the impact coefficient set; the benchmark monitoring frequency determination module 12 is used to determine the benchmark monitoring frequency based on the change rate and the impact coefficient, and construct a benchmark monitoring frequency table based on the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the impact coefficient; the indicator fluctuation trend prediction module 13 is used to collect the production line load information and environmental prediction information within a predetermined period, input the production line load information, environmental prediction information and monitoring indicator set into the data trend prediction channel, and output the indicator A fluctuation prediction trend set; an identification monitoring interval determination module 14, which is used to determine multiple identification monitoring interval sets by time segmenting and type-identifying a predetermined period according to a predetermined interval type based on the indicator fluctuation prediction trend set; a benchmark monitoring frequency optimization module 15, which is used to obtain multiple frequency correction coefficient sets based on the matching of the multiple identification monitoring interval sets, and generate multiple optimized monitoring frequency sets based on the multiple frequency correction coefficient sets and the benchmark monitoring frequency table; a target production line monitoring module 16, which is used to generate an expected monitoring plan based on the multiple identification monitoring interval sets and the multiple optimized monitoring frequency sets, and perform target production line monitoring within a predetermined period according to the expected monitoring plan.

[0064] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0065] A first monitoring indicator is randomly selected from the monitoring indicator set, and a sample change rate set is obtained based on industrial Internet retrieval with the target production line and the first monitoring indicator as constraints, wherein the change rate is the ratio of the indicator change amount to the time change amount; a mean is calculated for the sample change rate set, and the sample change rate set is screened based on the sample change rate mean to determine a standard change rate set, and the standard change rate mean is set as the first indicator change rate, and added to the change rate set.

[0066] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0067] Taking the target production line as a constraint, retrieve and obtain a set of sample production line performance abnormality events and a set of sample product quality defect events; based on the set of sample production line performance abnormality events and the set of sample product quality defect events, perform correlation tracing analysis on the monitoring indicator set, count the number of occurrences of related indicators, and obtain a monitoring indicator frequency set; perform an impact assessment on the monitoring indicator set based on the monitoring indicator frequency set, and determine the impact coefficient set, wherein the impact coefficient is positively correlated with the monitoring indicator frequency.

[0068] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0069] A first monitoring indicator is randomly selected from the monitoring indicator set, and the historical monitoring records of the first monitoring indicator are queried to obtain a sample prediction data set, wherein the sample prediction data includes sample production line load information, sample environmental information and sample indicator fluctuation trend, and the sample indicator fluctuation trend is a fluctuation curve of the first monitoring indicator within a predetermined historical period; the sample prediction data set is equally divided into K parts to obtain K data sets, and K data sets are selected with replacement K times in the K data sets to construct a first training set, and K training sets are selected iteratively K times to obtain K training sets; the BP neural network is supervised and trained using the K training sets to obtain K converged first prediction units, and a first trend prediction channel is built based on the K converged first prediction units, wherein the output of the first trend prediction channel is a fitting curve of the output results of the K converged first prediction units; multiple trend prediction channels for multiple monitoring indicators are sequentially constructed, and the data trend prediction channel is generated based on the multiple trend prediction channels.

[0070] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0071] A set of predetermined interval types is configured, wherein the predetermined interval types correspond to monitoring indicators one-to-one, including steady-state intervals, fluctuation intervals and abnormal intervals, and the abnormal intervals include fluctuation intervals, and the fluctuation intervals include steady-state intervals; a first indicator fluctuation prediction trend is randomly selected from the indicator fluctuation prediction trend set, and based on the first indicator fluctuation prediction trend, a predetermined period is time-divided and type-identified according to the first predetermined interval type to determine a first identified monitoring interval set, and added to the multiple identified monitoring interval sets.

[0072] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0073] The first predetermined interval types include a first steady-state interval, a first fluctuation interval and a first abnormal interval; based on the first indicator fluctuation prediction trend, the predetermined period is divided into time and type-identified according to the first steady-state interval, the first fluctuation interval and the first abnormal interval to obtain a first steady-state monitoring interval set, a first fluctuation monitoring interval set and a first abnormal monitoring interval set, wherein the time period corresponding to the first indicator fluctuation prediction trend in the first steady-state interval is identified as a steady-state monitoring interval, the time period corresponding to the first indicator fluctuation prediction trend in the first fluctuation interval and not belonging to the first steady-state interval is identified as a fluctuation monitoring interval, and the time period corresponding to the first indicator fluctuation prediction trend in the first abnormal interval and not belonging to the first fluctuation interval is identified as an abnormal monitoring interval; the first identified monitoring interval set is constructed based on the first steady-state monitoring interval set, the first fluctuation monitoring interval set and the first abnormal monitoring interval set.

[0074] Furthermore, the RTU real-time monitoring system in the industrial intelligent scenario is also used for:

[0075] Taking the target production line, the first monitoring indicator and the first steady-state interval as constraints, a sample frequency data set is retrieved and obtained, wherein the sample frequency data includes a sample correction coefficient and a sample event status, and the sample event status includes a normal state and an abnormal state caused by a monitoring frequency deviation; the sample frequency data set is clustered based on a predetermined coefficient step, the proportion of abnormal events in multiple clustering results is counted, and the minimum sample correction coefficient whose abnormal event proportion is less than a predetermined abnormal threshold is selected and set as a first steady-state correction coefficient; a first fluctuation correction coefficient and a first abnormal correction coefficient are obtained in sequence, and a first indicator correction coefficient is obtained in combination with the first steady-state correction coefficient; based on a decision tree, a frequency correction database is constructed according to the mapping association between the first monitoring indicator and the first indicator correction coefficient; the multiple identification monitoring interval sets are matched according to the frequency correction database, and multiple frequency correction coefficient sets are output.

[0076] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The RTU real-time monitoring method and specific examples in the aforementioned embodiment 1 are also applicable to the RTU real-time monitoring system in an industrial intelligent scenario in this embodiment. Through the aforementioned detailed description of the RTU real-time monitoring method in an industrial intelligent scenario, those skilled in the art can clearly know the RTU real-time monitoring system in an industrial intelligent scenario in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0077] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0078] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. The RTU real-time monitoring method in the industrial intelligent scene is characterized by: include: Obtain a monitoring indicator set for a target production line, perform a change speed analysis and impact assessment on the monitoring indicator set, and determine a change rate set and an impact coefficient set; Determine a benchmark monitoring frequency based on the change rate and the impact coefficient, and construct a benchmark monitoring frequency table based on the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the impact coefficient; Collecting production line load information and environmental prediction information within a predetermined period, inputting the production line load information, environmental prediction information and monitoring indicator set into a data trend prediction channel, and outputting an indicator fluctuation prediction trend set; Based on the indicator fluctuation prediction trend set, the predetermined period is divided into time periods and type-identified according to the predetermined interval type, and a plurality of identified monitoring interval sets are determined; Acquire multiple frequency correction coefficient sets according to the matching of the multiple identified monitoring interval sets, and generate multiple optimized monitoring frequency sets based on the multiple frequency correction coefficient sets and the reference monitoring frequency table; Generate an expected monitoring plan based on the multiple identification monitoring interval sets and the multiple optimized monitoring frequency sets, and perform target production line monitoring within a predetermined period according to the expected monitoring plan; Among them, building a data trend prediction channel includes: Randomly selecting a first monitoring indicator from the monitoring indicator set, querying historical monitoring records of the first monitoring indicator, and obtaining a sample prediction data set, wherein the sample prediction data includes sample production line load information, sample environment information, and sample indicator fluctuation trend, and the sample indicator fluctuation trend is a fluctuation curve of the first monitoring indicator within a historical predetermined period; The sample prediction data set is divided into K equal parts to obtain K data sets, and K data sets are selected with replacement K times in the K data sets to construct a first training set, and the selection is iteratively selected K times to obtain K training sets; Performing supervised training on the BP neural network using the K training sets to obtain K convergent first prediction units, and building a first trend prediction channel based on the K convergent first prediction units, wherein the output of the first trend prediction channel is a fitting curve of the output results of the K convergent first prediction units; A plurality of trend prediction channels for a plurality of monitoring indicators are sequentially constructed, and the data trend prediction channel is generated based on the plurality of trend prediction channels.

2. The RTU real-time monitoring method in the industrial intelligent scenario according to claim 1 is characterized in that: Determine the set of change rates, including: Randomly select a first monitoring indicator from the monitoring indicator set, take the target production line and the first monitoring indicator as constraints, and obtain a sample change rate set based on industrial Internet retrieval, where the change rate is the ratio of the indicator change amount to the time change amount; The sample change rate set is averaged, the sample change rate set is screened based on the sample change rate average, a standard change rate set is determined, and the standard change rate average is set as the first indicator change rate and added to the change rate set.

3. The RTU real-time monitoring method in the industrial intelligent scenario according to claim 2 is characterized in that: Determine the set of influence coefficients, including: Taking the target production line as a constraint, retrieve and obtain a sample production line performance abnormality event set and a sample product quality defect event set; Based on the sample production line performance abnormality event set and the sample product quality defect event set, the monitoring indicator set is subjected to correlation tracing analysis, and the number of occurrences of the correlation indicators is counted to obtain a monitoring indicator frequency set; An impact evaluation is performed on the monitoring indicator set according to the monitoring indicator frequency set to determine the impact coefficient set, wherein the impact coefficient is positively correlated with the monitoring indicator frequency.

4. The RTU real-time monitoring method in an industrial intelligent scenario according to claim 1 is characterized in that: Determine multiple identification monitoring interval sets, including: Configure a set of predetermined interval types, where the predetermined interval types correspond to monitoring indicators one by one, including a steady-state interval, a fluctuation interval, and an abnormal interval, and the abnormal interval includes the fluctuation interval, and the fluctuation interval includes the steady-state interval; A first indicator fluctuation prediction trend is randomly selected from the indicator fluctuation prediction trend set, and based on the first indicator fluctuation prediction trend, a predetermined period is time-divided and type-identified according to a first predetermined interval type, a first identification monitoring interval set is determined, and added to the multiple identification monitoring interval sets.

5. The RTU real-time monitoring method in the industrial intelligent scenario according to claim 4 is characterized in that: The predetermined period is divided into time periods and type-identified according to the first predetermined interval type, including: The first predetermined interval types include a first steady-state interval, a first fluctuation interval, and a first abnormal interval; Based on the first indicator fluctuation prediction trend, the predetermined period is divided into time periods and type-identified according to the first steady-state interval, the first fluctuation interval, and the first abnormal interval, to obtain a first steady-state monitoring interval set, a first fluctuation monitoring interval set, and a first abnormal monitoring interval set, wherein the time period corresponding to the first indicator fluctuation prediction trend in the first steady-state interval is identified as a steady-state monitoring interval, the time period corresponding to the first indicator fluctuation prediction trend in the first fluctuation interval and not belonging to the first steady-state interval is identified as a fluctuation monitoring interval, and the time period corresponding to the first indicator fluctuation prediction trend in the first abnormal interval and not belonging to the first fluctuation interval is identified as an abnormal monitoring interval; The first identification monitoring interval set is constructed based on the first steady-state monitoring interval set, the first fluctuation monitoring interval set and the first abnormality monitoring interval set.

6. The RTU real-time monitoring method in the industrial intelligent scenario according to claim 5 is characterized in that: Acquiring multiple frequency correction coefficient sets according to the matching of the multiple identification monitoring interval sets includes: Taking the target production line, the first monitoring indicator and the first steady-state interval as constraints, a sample frequency data set is retrieved and obtained, wherein the sample frequency data includes a sample correction coefficient and a sample event state, and the sample event state includes a normal state and an abnormal state caused by a monitoring frequency deviation; Clustering the sample frequency data set based on a predetermined coefficient step size, counting the proportion of abnormal events in multiple clustering results, and selecting a minimum sample correction coefficient whose abnormal event proportion is less than a predetermined abnormal threshold value to set as a first steady-state correction coefficient; A first fluctuation correction coefficient and a first abnormality correction coefficient are obtained in sequence, and a first indicator correction coefficient is obtained by combining the first steady-state correction coefficient; Based on the decision tree, a frequency correction database is constructed according to the mapping association between the first monitoring indicator and the first indicator correction coefficient; The multiple identification monitoring interval sets are matched according to the frequency correction database, and multiple frequency correction coefficient sets are output.

7. The RTU real-time monitoring system in the industrial intelligent scenario is characterized by: The steps for implementing the RTU real-time monitoring method in the industrial intelligent scenario according to any one of claims 1 to 6 include: A monitoring indicator analysis module is used to obtain a monitoring indicator set of a target production line, perform change speed analysis and impact assessment on the monitoring indicator set, and determine a change rate set and an impact coefficient set; A benchmark monitoring frequency determination module, used to determine the benchmark monitoring frequency based on the change rate and the influence coefficient, and to construct a benchmark monitoring frequency table according to the monitoring indicator set and the benchmark monitoring frequency set, wherein the benchmark monitoring frequency is positively correlated with the change rate and the influence coefficient; An indicator fluctuation trend prediction module is used to collect production line load information and environmental prediction information within a predetermined period, input the production line load information, environmental prediction information and monitoring indicator set into a data trend prediction channel, and output an indicator fluctuation prediction trend set; A module for determining an identification monitoring interval, for determining a plurality of identification monitoring interval sets by performing time segmentation and type identification on a predetermined period according to a predetermined interval type based on the indicator fluctuation prediction trend set; A reference monitoring frequency optimization module, configured to obtain a plurality of frequency correction coefficient sets according to the matching of the plurality of identified monitoring interval sets, and generate a plurality of optimized monitoring frequency sets based on the plurality of frequency correction coefficient sets and the reference monitoring frequency table; The target production line monitoring module is used to generate an expected monitoring plan based on the multiple identification monitoring interval sets and the multiple optimized monitoring frequency sets, and perform target production line monitoring within a predetermined period according to the expected monitoring plan.

Images