Optimization system based on data content

Abstract

The invention discloses a data content-based optimization method, which comprises the following steps of: 1, acquiring process variable data associated with an alarm in an industrial system, and performing association analysis on the process variable data; 2, introducing influence factors of an operation user of the operation terminal based on a set strong association rule; 3, calculating the correlation between two variables causing alarm according to the number of historical alarms in the industrial system, and performing variable optimization by quantifying the consistency of process variable data and alarm data; step 4, hierarchical clustering is carried out on the alarm data with large correlation, and the alarm threshold value of each alarm class is optimized by using the predicted probability; 5, the threshold value of the key process variable data is optimized with the false alarm rate and the missing alarm rate as evaluation indexes, the optimized path and the optimized threshold value are output, the problem that the false alarm frequency and the missing alarm frequency are high is reduced, and the method has the advantage that the false alarm rate and the missing alarm rate of an industrial system are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial data analysis and optimization, and in particular to an optimization system based on data content. Background Art

[0002] In the alarm system of the industrial field, reasonable alarm threshold is an important guarantee for the safe operation of production and work systems and the safe operation of personnel. When the alarm threshold changes, the generated alarm sequence will also change. When the alarm threshold is set low, the system will generate a large number of false alarms, which will waste manpower and material resources, seriously affect the work efficiency of operators, and cause operators to be unable to handle high-risk alarms in time. When the alarm threshold is set high, the system will generate a large number of missed alarms, resulting in the omission of abnormal states in the production process, and operators cannot monitor the fault state in time, leading to production accidents or even serious accidents and alarm flooding. Therefore, setting the alarm threshold too high or too low will cause the correlation between alarm data to change, and the deviation of the alarm threshold will lead to alarm flooding, that is, it will propagate through various process units and many variables, causing multiple alarms to appear in parallel. Too many interference alarms will drown out the real important alarms, causing operators to miss signals and cause industrial accidents. Therefore, it is necessary to design an optimization system based on data content to optimize the accuracy of alarm system operation value setting and improve the safety of industrial processes. Summary of the invention

[0003] The object of the present invention is to provide an optimization system based on data content to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a data content-based optimization system, comprising a process variable data analysis module, a process variable data optimization module and an alarm threshold prediction optimization module, characterized in that: the process variable data analysis module is used to obtain process variable data associated with alarms in industrial systems and perform correlation analysis; the process variable data optimization module is used to optimize process variable data associated with alarms; the alarm threshold prediction optimization module is used to optimize the alarm threshold of each alarm class using predicted probabilities.

[0005] According to the above technical solution, the process variable data analysis module includes: a process variable data acquisition module, an association analysis module and a correlation description module. The process variable data acquisition module is used to acquire the process variable data associated with the alarm in the statistical industrial system; the association analysis module is used to establish association rules to perform association analysis on the process variable data associated with the alarm; the correlation description module is used to describe the correlation between the process variable data and the operating behavior.

[0006] According to the above technical solution, the process variable data optimization module includes: a user influence factor introduction module, a strong association rule module, a variable correlation calculation module and a variable optimization module. The user influence factor introduction module is used to introduce the influence factors of the operating terminal operating user based on the set strong association rules; the strong association rule module is used to introduce the influence factors of the operating terminal operating user based on the set strong association rules; the variable correlation calculation module is used to calculate the correlation between two variables that cause the alarm according to the number of historical alarms in the industrial system; the variable optimization module is used to optimize the variables by quantifying the consistency between the process variable data and the alarm data.

[0007] According to the above technical scheme, the alarm threshold prediction optimization module includes: a hierarchical clustering module, an alarm threshold optimization module, and an optimization path calculation output module. The hierarchical clustering module is used to perform hierarchical clustering on alarm data with high correlation; the alarm threshold optimization module is used to optimize the alarm threshold of key process variable data with false alarm rate and missed alarm rate as evaluation indicators; the optimization path calculation output module is used to output the optimized optimization path and optimization threshold.

[0008] According to the above technical solution, the data content-based optimization system executes a data content-based optimization method, which includes the following steps: Step 1: Based on the process variable data associated with the alarm in the industrial system, the process variable data associated with the alarm is acquired, and the correlation analysis is performed on the process variable data associated with the alarm, and the correlation between the process variable data and the operation behavior is described; Step 2: Introduce the influencing factors of the operation terminal operation users based on the set strong association rules; Step 3: Calculate the correlation between the two variables that cause the alarm according to the number of historical alarms in the industrial system, and optimize the variables by quantifying the consistency between the process variable data and the alarm data; Step 4: Perform hierarchical clustering on the alarm data with high correlation, and use the predicted probability to optimize the alarm threshold of each alarm class; Step 5: Optimize the alarm threshold of key process variable data using false alarm rate and missed alarm rate as evaluation indicators, and output the optimized optimization path and optimization threshold.

[0009] According to the above technical solution, the step of performing correlation analysis on the process variable data associated with the alarm and describing the correlation between the process variable data and the operation behavior includes: The acquired process variable data are counted into a process variable data set, and the process variable data set is used as the input item set associated with the system alarm. An association rule base is established. The operation behavior data of the operation end after the system generates an alarm corresponding to the process variable data are counted and defined as the output item set of the association analysis. The support is calculated based on each input item set, and then the association rule relationship between each data variable in the process variable data set in the input item set and each operation behavior data in the output item set is determined by setting the confidence. The association relationship between the input item set and the output item set includes satisfying the strong association rule and not satisfying the strong association rule. When a process variable data in the input item set corresponds to an operation behavior in the output item set, that is, when the strong association rule is satisfied, the process variable data in the input item set is associated with the operation behavior in the output item set and stored in the association database, and description information is added to the association relationship, that is, the correlation between the process variable data and the operation behavior is described using the explanatory structure model. When the process variable data in the input item set causes an associated alarm, the operation behavior in the output item set is used to process the system alarm, and the iterative calculation output is performed in sequence.

[0010] According to the above technical solution, the step of introducing the influencing factors of the user operating the operating terminal includes: The setting is based on the influencing factors of the operating users of the operation terminal, and two influencing situations are set. The first situation is that the operating users of the operation terminal make more adjustments to the process. The alarm variables in the industrial system are greatly affected by the operation interference of the operating users, that is, alarm interference will occur, resulting in alarm flooding. Its generation characteristics are that the alarm terminal of the system exceeds the normal range of the set alarm quantity, and the alarm threshold changes, resulting in changes in the generated alarm sequence, resulting in more alarms than the normal alarm quantity range and a short duration of a single alarm, resulting in a large number of useless alarms affecting the normal judgment of the operating users of the operation terminal; the second situation is that the operating users of the operation terminal make fewer adjustments to the process in different time periods. The time period is divided according to the working hours of the operating users during the operation of the system, including normal working hours during the day and working hours for checking and maintenance at night. In the case of checking and maintenance at night, the alarm variables of the industrial system are not affected by the frequent operation disturbances of the operating users. The alarm threshold set by the system is large, and the number of system alarms generated is small or even no alarm is issued; According to the two influencing conditions, the process variable alarm thresholds are set during the operation of the industrial system, and the upper and lower limits of the adjustment variables adjusted by the operating terminal users are set. When the alarm threshold changes, the generated alarm sequence will also change, which will cause the correlation between the alarm data to change. Therefore, the difference in the correlation between the process measurement data and the alarm data is used to set the alarm threshold of the industrial system, that is, the threshold is adjusted in a small range within the upper and lower ranges according to the upper and lower limits of the adjustment variables adjusted by the operating terminal users.

[0011] According to the above technical solution, the step of calculating the correlation between the two variables causing the alarm includes: Combine process variable data with historical alarm data to extract process variable data pairs with correlation. When the alarm sequences of two process variable data change with the changes of their respective alarm thresholds, it is defined that the similarity of the two process variable data is also changing. The correlation consistency of the two process variable data at a specific moment is expressed as the difference between the two. The smaller the difference between the two, the closer the alarm correlation of the process variable is to the correlation of the process parameter. Calculate the consistency of the alarm data correlation and the process data correlation in turn, classify the difference between the two, and set the threshold point of the difference by calculating and analyzing the historical alarm data in the industrial system. When the two process variable data When the difference is greater than the threshold point, it is judged that the alarm data cannot represent the correlation of the process variable data, that is, its correlation consistency is weak. The process variable data with weak correlation consistency are uniformly optimized by principal component analysis, that is, the process variable data are reduced in dimension by principal component analysis, the main characteristic components of the process variable data are extracted, the relative importance of the process variable data after characteristic analysis is calculated, and the priority sorting is used for correction. According to the alarm number standard set by the industrial system, the threshold point of the alarm number is obtained, and the variables of important alarms are obtained by combining the priority method. The variables with strong correlation consistency are optimized according to the correlation to construct the objective function.

[0012] According to the above technical solution, the step of hierarchically clustering the alarm data with high correlation and optimizing the alarm threshold of each alarm class by using the predicted probability includes: Step 41: Calculate the similarity measurement index according to the correlation coefficient between the alarm data, and use the hierarchical clustering algorithm to cluster and group all the alarm data; Step 42: taking each alarm data as a system alarm class, and calculating the similarity measurement index between each alarm data; Step 43: By predicting the probability of occurrence of each alarm type alarm, the threshold value at the next moment is optimized according to the alarm probability, that is, the alarm data at the previous moment and the current moment are associated and predicted, and the probability value of each process variable data alarming at the next moment is calculated; Step 44: Since the probability value of the alarm corresponding to each process variable data obtained at each moment is different, the optimized alarm threshold is set according to the alarm variable classification ratio divided according to the historical alarm data in the actual industrial system. When the calculated probability value exceeds the set alarm threshold, before each alarm of the specified alarm level occurs, it will trigger the optimization calculation of all alarm thresholds.

[0013] According to the above technical solution, the method for outputting the optimized optimization path and the optimized threshold includes: The false alarm rate and missed alarm rate of the industrial system are used to evaluate the alarm performance of the industrial system. The false alarm rate and missed alarm rate of the industrial system correspond to two state values: normal state and abnormal state. The initial threshold is set based on the industrial system. When a variable is always in the alarm state, that is, the control effect in the alarm process cannot restore the variable to the normal state, the process variable is marked as the process variable to be optimized. The propagation path of the process variable data under the state value is determined according to the process variable that triggers the alarm. The causal path of abnormal propagation is used to determine the optimization order of the marked process variables. Then, the consistency between the calculated process variable data and the alarm data is used as the objective function. The optimization algorithm is used to minimize the difference between the correlation coefficient of the alarm data and the correlation coefficient of the process variable data to obtain the optimal threshold.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. By performing correlation analysis and describing the correlation relationship of the acquired process variable data associated with the alarm in the industrial system, and introducing the influencing factors of the operating terminal user based on the set strong correlation rules, the threshold is adjusted in a small range of upper and lower limits according to the upper and lower limits of the adjustment variables adjusted by the operating terminal user, thereby reducing the missed alarms and false alarms of the alarm system caused by human factors in the industrial system, and optimizing the setting accuracy of the alarm system operation value; 2. Calculate the correlation between the two variables that cause the alarm based on the number of historical alarms in the industrial system, optimize the variables by quantifying the consistency between the process variable data and the alarm data, hierarchically cluster the alarm data with high correlation, and use the predicted probability to optimize the alarm threshold of each alarm class, ensuring that every level of alarm caused by process variable data in the industrial system is taken into account, reducing the number of false alarms and missed alarms overall, lowering the false alarm rate and missed alarm rate of alarms, and having stronger adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1A flowchart of a data content-based optimization method system provided in Embodiment 1 of the present invention; Figure 2 A schematic diagram of the module composition of a data content-based optimization system provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Embodiment 1: Figure 1 This is a flowchart of a data content-based optimization method system provided in Embodiment 1 of the present invention. This embodiment can be applied to the scenario of industrial alarm threshold setting optimization. This method can be executed by a data content-based optimization system provided in this embodiment, such as Figure 1 As shown, the method specifically comprises the following steps: Step 1: Based on the process variable data associated with the alarm in the industrial system, the process variable data associated with the alarm is acquired, and the correlation analysis is performed on the process variable data associated with the alarm, and the correlation between the process variable data and the operation behavior is described; In an embodiment of the present invention, the acquired process variable data are counted into a process variable data set, and the process variable data set is used as an input item set associated with the system alarm, an association rule base is established, the operation behavior data of the operation end after the system generates an alarm corresponding to the process variable data are counted, and defined as an output item set of the association analysis, the support is calculated based on each input item set, and then the association rule relationship between each data variable in the process variable data set in the input item set and each operation behavior data in the output item set is determined by setting the confidence, the association relationship between the input item set and the output item set includes satisfying the strong association rule and not satisfying the strong association rule, when a certain process variable data in the input item set corresponds to a certain operation behavior in the output item set, that is, when the strong association rule is satisfied, the process variable data in the input item set is associated with the operation behavior in the output item set, stored in the association database, and description information is added to the association relationship, that is, the correlation between the process variable data and the operation behavior is described using the explanatory structure model, when the process variable data in the input item set causes an associated alarm, the operation behavior in the output item set is used to process the system alarm, and iterative calculation and output are performed in sequence; Exemplarily, the support is calculated based on each input item set, and then the method of setting the confidence level includes: establishing a mathematical model using the correlation between the process variable data characteristics and the operating behavior of the industrial system alarm process, setting the threshold using the data correlation value of the input item and the output item, establishing a local weighted linear regression model with the process variable data of the input item for prediction, obtaining the confidence interval of the predicted value, and using the upper and lower lines of the confidence interval of the predicted value as the dynamic high and low alarm thresholds of the process variable data, and obtaining the process alarm limit to control the number of false alarms.

[0018] Step 2: Introduce the influencing factors of the operation terminal operation users based on the set strong association rules; In the embodiment of the present invention, the setting is based on the influencing factors of the operating user of the operating terminal, and two influencing situations are set. The first situation is that the operating user of the operating terminal makes more adjustments to the process, and the alarm variables in the industrial system are greatly affected by the operation interference of the operating user, that is, alarm interference will occur, resulting in alarm flooding. Its generation characteristics are that the alarm terminal of the system exceeds the normal range of the set alarm quantity, and the alarm threshold changes, resulting in changes in the generated alarm sequence, resulting in the number of alarms exceeding the normal alarm quantity range and the alarm time of a single alarm is short, resulting in a large number of useless alarms affecting the normal judgment of the operating user of the operating terminal; the second situation is that the operating user of the operating terminal makes fewer adjustments to the process in different time periods, and the time period is divided according to the working hours of the operating user during the operation of the system, including normal working hours during the day and working hours for checking and maintenance at night. In the case of checking and maintenance at night, the alarm variables of the industrial system are not affected by the frequent operation disturbances of the operating user, the alarm threshold set by the system is large, and the number of system alarms generated is small or even no alarm is issued; Exemplarily, the process variable alarm threshold during the operation of the industrial system is set according to the two influencing conditions, and the upper and lower limits of the adjustment variables adjusted by the operating terminal user are set. When the alarm threshold changes, the generated alarm sequence will also change accordingly, thereby causing the correlation between the alarm data to change. Therefore, the difference in the correlation between the process measurement data and the alarm data is used to set the alarm threshold of the industrial system, that is, the threshold is adjusted in a small range up and down according to the upper and lower limits of the adjustment variables adjusted by the operating terminal user, so as to reduce underreporting and false alarms caused by human factors.

[0019] Step 3: Calculate the correlation between the two variables that cause the alarm according to the number of historical alarms in the industrial system, and optimize the variables by quantifying the consistency between the process variable data and the alarm data; In an embodiment of the present invention, process variable data is combined with historical alarm data to extract process variable data pairs with an associated relationship. When the alarm sequences of two process variable data change with the change of their respective alarm thresholds, it is defined that the similarity of the two process variable data is also changing. The correlation consistency of the two process variable data at a specific moment is expressed as the difference between the two. The smaller the difference between the two, the closer the alarm correlation of the process variable is to the correlation of the process parameter. The size of the consistency of the alarm data correlation and the process data correlation is calculated in turn, and the difference between the two is classified. By calculating and analyzing the historical alarm data in the industrial system, the threshold point of the difference is set. When the two process When the difference of variable data is greater than the threshold point, it is judged that the alarm data cannot represent the correlation of process variable data, that is, its correlation consistency is weak. These process variable data with weak correlation consistency are uniformly optimized by principal component analysis, that is, the process variable data is reduced in dimension by principal component analysis, the main characteristic components of the process variable data are extracted, the relative importance of the process variable data after characteristic analysis is calculated, and the priority sorting is used for correction. According to the alarm number standard set by the industrial system, the threshold point of the alarm number is obtained, and the variables of important alarms are obtained by combining the priority method. The variables with strong correlation consistency are optimized by constructing the objective function according to the correlation.

[0020] Step 4: Perform hierarchical clustering on the alarm data with high correlation, and use the predicted probability to optimize the alarm threshold of each alarm class; In the embodiment of the present invention, the specific clustering method and optimization method include: Step 41: Calculate the similarity measurement index according to the correlation coefficient between the alarm data, and use the hierarchical clustering algorithm to cluster and group all the alarm data; Step 42: taking each alarm data as a system alarm class, and calculating the similarity measurement index between each alarm data; Step 43: By predicting the probability of occurrence of each alarm type alarm, the threshold value at the next moment is optimized according to the alarm probability, that is, the alarm data at the previous moment and the current moment are associated and predicted, and the probability value of each process variable data alarming at the next moment is calculated; Step 44: Since the probability value of the alarm corresponding to each process variable data obtained at each moment is different, the optimized alarm threshold is set according to the alarm variable classification ratio divided according to the historical alarm data in the actual industrial system. When the calculated probability value exceeds the set alarm threshold, all alarm thresholds will be triggered to be optimized before each alarm of the specified alarm level occurs, ensuring that every level of alarm caused by process variable data is taken into account, reducing the possibility of underreporting and having stronger adaptability.

[0021] Step 5: Optimize the alarm threshold of key process variable data using false alarm rate and missed alarm rate as evaluation indicators, and output the optimized optimization path and optimization threshold.

[0022] In an embodiment of the present invention, a false alarm is an alarm generated under normal operation of an industrial system, and a missed alarm is an abnormal situation in the industrial system but no alarm is triggered. The false alarm rate and missed alarm rate statistically obtained by the industrial system are used to evaluate the alarm performance of the industrial system. The false alarm rate and missed alarm rate of the industrial system both correspond to two state values, a normal state and an abnormal state. An initial threshold is set based on the industrial system. When a certain variable is always in an alarm state, that is, the control effect in the alarm process cannot restore the variable to a normal state, the process variable is marked as a process variable to be optimized, and the propagation path of the process variable data under the state value is determined according to the process variable that triggers the alarm. The causal path of abnormal propagation is used to determine the optimization order of the marked process variables. Then, the consistency between the calculated process variable data and the alarm data is used as the objective function, and the optimization algorithm is used to minimize the difference between the alarm data correlation coefficient and the process variable data correlation coefficient to obtain the optimal threshold, thereby reducing the number of false alarms and missed alarms as a whole, and reducing the false alarm rate and missed alarm rate of the alarm.

[0023] Embodiment 2: Embodiment 2 of the present invention provides an optimization system based on data content. Figure 2 A schematic diagram of the module composition of a data content-based optimization system provided in Embodiment 2 of the present invention, such as Figure 2 As shown, the system includes: Process variable data analysis module, used to obtain process variable data associated with alarms in industrial systems and perform correlation analysis; Process variable data optimization module, used to optimize the process variable data associated with the alarm; The alarm threshold prediction and optimization module is used to optimize the alarm threshold of each alarm class using the predicted probability.

[0024] In some embodiments of the present invention, the process variable data analysis module includes: A process variable data acquisition module is used to acquire process variable data associated with alarms in a statistical industrial system; The association analysis module is used to establish association rules to perform association analysis on the process variable data associated with the alarm; The correlation description module is used to describe the correlation between process variable data and operation behavior.

[0025] In some embodiments of the present invention, the process variable data optimization module includes: A user influencing factor introduction module is used to introduce influencing factors of operating terminal operation users based on set strong association rules; A strong association rule module is used to introduce influencing factors of operation terminal operation users based on the set strong association rules; A module for calculating the correlation between variables, used to calculate the correlation between two variables causing the alarm according to the number of historical alarms in the industrial system; The variable optimization module is used to optimize the variables by quantifying the consistency between process variable data and alarm data.

[0026] In some embodiments of the present invention, the alarm threshold prediction optimization module includes: Hierarchical clustering module, used to perform hierarchical clustering on alarm data with high correlation; An alarm threshold optimization module is used to optimize the alarm threshold of key process variable data using false alarm rate and missed alarm rate as evaluation indicators; The optimization path calculation output module is used to output the optimized optimization path and optimization threshold.

[0027] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0028] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data content-based optimization system, comprising a process variable data analysis module, a process variable data optimization module and an alarm threshold prediction optimization module, characterized in that: The process variable data analysis module is used to obtain process variable data associated with alarms in the industrial system and perform correlation analysis; the process variable data optimization module is used to optimize process variable data associated with alarms; and the alarm threshold prediction optimization module is used to optimize the alarm threshold of each alarm class using the predicted probability.

2. The data content-based optimization system according to claim 1, characterized in that: The process variable data analysis module includes: a process variable data acquisition module, an association analysis module and a correlation description module. The process variable data acquisition module is used to acquire process variable data associated with an alarm in a statistical industrial system; the association analysis module is used to establish association rules to perform association analysis on process variable data associated with the alarm; and the correlation description module is used to describe the correlation between process variable data and operating behavior.

3. The data content-based optimization system according to claim 2, characterized in that: The process variable data optimization module includes: a user influence factor introduction module, a strong association rule module, a variable correlation calculation module and a variable optimization module. The user influence factor introduction module is used to introduce the influence factors of the operation terminal operation user based on the set strong association rules; the strong association rule module is used to introduce the influence factors of the operation terminal operation user based on the set strong association rules; the variable correlation calculation module is used to calculate the correlation between two variables causing the alarm according to the number of historical alarms in the industrial system; the variable optimization module is used to optimize the variables by quantifying the consistency between the process variable data and the alarm data.

4. The data content-based optimization system according to claim 3, characterized in that: The alarm threshold prediction and optimization module includes: a hierarchical clustering module, an alarm threshold optimization module, and an optimization path calculation output module. The hierarchical clustering module is used to perform hierarchical clustering on alarm data with high correlation; the alarm threshold optimization module is used to optimize the alarm threshold of key process variable data with false alarm rate and missed alarm rate as evaluation indicators; the optimization path calculation output module is used to output the optimized optimization path and optimization threshold.

5. According to the data content-based optimization system of claim 4, a data content-based optimization method is implemented, characterized in that: The method comprises the following steps: Step 1: Based on the process variable data associated with the alarm in the industrial system, the process variable data associated with the alarm is acquired, and the correlation analysis is performed on the process variable data associated with the alarm, and the correlation between the process variable data and the operation behavior is described; Step 2: Introduce the influencing factors of the operation terminal operation users based on the set strong association rules; Step 3: Calculate the correlation between the two variables that cause the alarm according to the number of historical alarms in the industrial system, and optimize the variables by quantifying the consistency between the process variable data and the alarm data; Step 4: Perform hierarchical clustering on the alarm data with high correlation, and use the predicted probability to optimize the alarm threshold of each alarm class; Step 5: Optimize the alarm threshold of key process variable data using false alarm rate and missed alarm rate as evaluation indicators, and output the optimized optimization path and optimization threshold.

6. The data content-based optimization method according to claim 5, characterized in that: The step of performing correlation analysis on the process variable data associated with the alarm and describing the correlation between the process variable data and the operation behavior includes: The acquired process variable data are counted into a process variable data set, and the process variable data set is used as the input item set associated with the system alarm. An association rule base is established. The operation behavior data of the operation end after the system generates an alarm corresponding to the process variable data are counted and defined as the output item set of the association analysis. The support is calculated based on each input item set, and then the association rule relationship between each data variable in the process variable data set in the input item set and each operation behavior data in the output item set is determined by setting the confidence. The association relationship between the input item set and the output item set includes satisfying the strong association rule and not satisfying the strong association rule. When a process variable data in the input item set corresponds to an operation behavior in the output item set, that is, when the strong association rule is satisfied, the process variable data in the input item set is associated with the operation behavior in the output item set and stored in the association database, and description information is added to the association relationship, that is, the correlation between the process variable data and the operation behavior is described using the explanatory structure model. When the process variable data in the input item set causes an associated alarm, the operation behavior in the output item set is used to process the system alarm, and the iterative calculation output is performed in sequence.

7. The data content-based optimization method according to claim 6, characterized in that: The step of introducing the influencing factors of the operation terminal operation user includes: The setting is based on the influencing factors of the operating users of the operation terminal, and two influencing situations are set. The first situation is that the operating users of the operation terminal make more adjustments to the process. The alarm variables in the industrial system are greatly affected by the operation interference of the operating users, that is, alarm interference will occur, resulting in alarm flooding. Its generation characteristics are that the alarm terminal of the system exceeds the normal range of the set alarm quantity, and the alarm threshold changes, resulting in changes in the generated alarm sequence, resulting in more alarms than the normal alarm quantity range and a short duration of a single alarm, resulting in a large number of useless alarms affecting the normal judgment of the operating users of the operation terminal; the second situation is that the operating users of the operation terminal make fewer adjustments to the process in different time periods. The time period is divided according to the working hours of the operating users during the operation of the system, including normal working hours during the day and working hours for checking and maintenance at night. In the case of checking and maintenance at night, the alarm variables of the industrial system are not affected by the frequent operation disturbances of the operating users. The alarm threshold set by the system is large, and the number of system alarms generated is small or even no alarm is issued; According to the two influencing conditions, the process variable alarm thresholds are set during the operation of the industrial system, and the upper and lower limits of the adjustment variables adjusted by the operating terminal users are set. When the alarm threshold changes, the generated alarm sequence will also change, which will cause the correlation between the alarm data to change. Therefore, the difference in the correlation between the process measurement data and the alarm data is used to set the alarm threshold of the industrial system, that is, the threshold is adjusted in a small range within the upper and lower ranges according to the upper and lower limits of the adjustment variables adjusted by the operating terminal users.

8. The data content-based optimization method according to claim 7, characterized in that: The step of calculating the correlation between the two variables causing the alarm comprises: Combine process variable data with historical alarm data to extract process variable data pairs with correlation. When the alarm sequences of two process variable data change with the changes of their respective alarm thresholds, it is defined that the similarity of the two process variable data is also changing. The correlation consistency of the two process variable data at a specific moment is expressed as the difference between the two. The smaller the difference between the two, the closer the alarm correlation of the process variable is to the correlation of the process parameter. Calculate the consistency of the alarm data correlation and the process data correlation in turn, classify the difference between the two, and set the threshold point of the difference by calculating and analyzing the historical alarm data in the industrial system. When the two process variable data When the difference is greater than the threshold point, it is judged that the alarm data cannot represent the correlation of the process variable data, that is, its correlation consistency is weak. The process variable data with weak correlation consistency are uniformly optimized by principal component analysis, that is, the process variable data are reduced in dimension by principal component analysis, the main characteristic components of the process variable data are extracted, the relative importance of the process variable data after characteristic analysis is calculated, and the priority sorting is used for correction. According to the alarm number standard set by the industrial system, the threshold point of the alarm number is obtained, and the variables of important alarms are obtained by combining the priority method. The variables with strong correlation consistency are optimized according to the correlation to construct the objective function.

9. The data content-based optimization method according to claim 8, characterized in that: The step of hierarchically clustering the alarm data with high correlation and optimizing the alarm threshold of each alarm class by using the predicted probability includes: Step 41: Calculate the similarity measurement index according to the correlation coefficient between the alarm data, and use the hierarchical clustering algorithm to cluster and group all the alarm data; Step 42: taking each alarm data as a system alarm class, and calculating the similarity measurement index between each alarm data; Step 43: By predicting the probability of occurrence of each alarm type alarm, the threshold value at the next moment is optimized according to the alarm probability, that is, the alarm data at the previous moment and the current moment are associated and predicted, and the probability value of each process variable data alarming at the next moment is calculated; Step 44: Since the probability value of the alarm corresponding to each process variable data obtained at each moment is different, the optimized alarm threshold is set according to the alarm variable classification ratio divided according to the historical alarm data in the actual industrial system. When the calculated probability value exceeds the set alarm threshold, before each alarm of the specified alarm level occurs, it will trigger the optimization calculation of all alarm thresholds.

10. The data content-based optimization method according to claim 9, characterized in that: The method for outputting the optimization path and the optimization threshold of the optimization includes: The false alarm rate and missed alarm rate of the industrial system are used to evaluate the alarm performance of the industrial system. The false alarm rate and missed alarm rate of the industrial system correspond to two state values: normal state and abnormal state. The initial threshold is set based on the industrial system. When a variable is always in the alarm state, that is, the control effect in the alarm process cannot restore the variable to the normal state, the process variable is marked as the process variable to be optimized. The propagation path of the process variable data under the state value is determined according to the process variable that triggers the alarm. The causal path of abnormal propagation is used to determine the optimization order of the marked process variables. Then, the consistency between the calculated process variable data and the alarm data is used as the objective function. The optimization algorithm is used to minimize the difference between the correlation coefficient of the alarm data and the correlation coefficient of the process variable data to obtain the optimal threshold.

Images