A real-time alarm tracing device and method in process industry production

By designing offline process models to build and online alarm traceability devices in the process industry, real-time traceability of faults in the process industry is realized, and the problem of poor real-time alarm traceability in the existing technology is solved, and fault handling efficiency and accuracy are improved.

CN114529247BActive Publication Date: 2024-11-05SUPCON TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111642035.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-29
Publication Date
2024-11-05
Estimated Expiration
2041-12-29

Smart Images

  • Figure CN114529247B_ABST
    Figure CN114529247B_ABST
Patent Text Reader

Abstract

The present invention discloses an online alarm tracing device and method in a process industrial production process, which includes: an offline process model building device builds an offline process model; an online alarm tracing device and a real-time alarm collection unit collect bit number data and process alarm data; a detection and triggering unit detects alarm flooding information and generates an alarm flooding event; an alarm tracing unit receives the alarm flooding event, and calls an alarm tracing model building unit to obtain a tracing model and then performs alarm tracing; the alarm tracing model building unit builds and outputs alarm tracing model data. The present invention can realize real-time and automatic alarm tracing. When an alarm flood is detected, the alarm source of this alarm flood can be provided to the operator in the first time, thereby assisting the operator to handle the fault faster and better.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of digital signal processing, and in particular relates to a real-time alarm tracing device in a process industrial production process. Background Art

[0002] Terminology explanation:

[0003] Alarm flooding: Today’s complex process industries have complex characteristics such as large-scale uncertainty, strong correlation, and high nonlinearity. When a fault occurs in one part of the system, it will spread along the process path, causing a large number of alarms in other parts in a short period of time, resulting in “alarm flooding”.

[0004] Alarm tracing refers to finding the source of the alarm flood, that is, the alarm event or bit number, through various analysis methods.

[0005] The position number refers to the data monitoring point on site in the industrial production process.

[0006] The complexity of the process industry means that when a fault occurs somewhere in the generation system, the fault will spread along the process path, causing a large number of alarms in other parts within a short period of time, resulting in alarm flooding. At this time, the amount of alarm data is far greater than the operator's processing capacity, and key effective alarms will also be submerged in a large number of invalid alarms, resulting in the operator being unable to find the source of the alarm in a short period of time and take correct measures in time.

[0007] There are two main types of mainstream alarm tracing solutions on the market.

[0008] The first type is to trace the alarm offline, which requires the user to import historical data and mark the fault time period, and then the system will output the traceability results within the fault time period after tracing analysis. The disadvantage of this solution is that users are required to manually import and mark data, and the real-time performance is poor. When an alarm flood occurs on site, the root cause information of the alarm flood cannot be provided to the operator in time to assist the operator in rapid fault handling. For example, the Wanfang Database document "Complex Industrial Alarm Traceability Analysis Based on Bayesian Networks" (Ding Rui, 2020) introduces the use of manual offline analysis to verify the feasibility of the algorithm for alarm tracing using Bayesian networks.

[0009] The second type is to predefine faults. When a fault occurs, if it matches the predefined fault, the root cause configured in the fault will be immediately prompted. The disadvantage of this solution is that it can only predefine known faults. When a fault that does not match is encountered, the root cause of this fault cannot be prompted. In addition, facing increasingly complex industrial sites, experts cannot predefine all faults. Summary of the invention

[0010] The present invention proposes an online alarm tracing device and method in the process industry production process, which can realize real-time and automatic alarm tracing. When alarm flooding is detected, the alarm source of this alarm flooding can be provided to the operator in the first time, thereby assisting the operator to handle the fault faster and better.

[0011] To this end, the first object of the present invention is to provide a real-time alarm tracing device in a process industry production process, which includes an offline process model building device and an online alarm tracing device, wherein:

[0012] The offline process model building device includes: a process configuration database for storing process configuration data; a process model database for storing process model data; a process model building unit for acquiring process configuration data from the process configuration database and building a corresponding process model;

[0013] The online alarm tracing device includes: a real-time data collection unit for collecting real-time data of bit numbers, and storing the received data in a historical trend database; a real-time alarm collection unit for collecting process alarm data, and storing the received data in a historical alarm database; a detection and triggering unit for triggering an alarm tracing unit to perform alarm tracing by detecting alarm flooding information and generating an alarm flooding event; an alarm tracing unit for receiving an alarm flooding event from the detection and triggering unit, and calling an alarm tracing model construction unit to obtain a tracing model and then perform alarm tracing; an alarm tracing model construction unit for receiving process models, historical trends, historical alarms, and current fault information from an industrial model database, a historical trend database, a historical alarm database, and an alarm tracing unit, respectively, and constructing and outputting alarm tracing model data.

[0014] Furthermore, the process configuration data includes bit numbers, process alarms, and control loops. The process model constructed by the process model construction unit includes building relationships between bit numbers and bit numbers and between bit numbers and process alarms based on bit numbers and control loops.

[0015] Furthermore, it also includes a tracing model database for storing alarm tracing model data, and can be output to the alarm tracing unit for final alarm tracing.

[0016] Furthermore, it also includes a monitoring interface display unit for displaying the traceability results.

[0017] Furthermore, the detection and triggering unit is provided with a detection cycle threshold, and is configured to output at least two alarm flooding events according to the number of alarms within the detection cycle threshold, and the alarm tracing unit performs different actions after receiving the events.

[0018] Furthermore, the detection and triggering unit is set to output three alarm flood events, including an alarm flood start event, an alarm flood continuation event, and an alarm flood end event; the alarm tracing unit includes three sub-modules: a tracing creation sub-module, which is used to create a tracing environment for the fault after receiving the alarm flood start event and start the first alarm tracing; a tracing update sub-module, which is used to perform alarm tracing again and update the tracing results after receiving the alarm flood continuation event; and a tracing end sub-module, which is used to perform the last alarm tracing after receiving the alarm flood end event, update the tracing results, and terminate this tracing.

[0019] The second object of the present invention is to provide a real-time alarm tracing method in a process industry production process, which uses the real-time alarm tracing device according to any of the preceding claims to perform alarm tracing, comprising the following steps:

[0020] 1) The process model building unit of the offline process model building device obtains process configuration data from the process configuration database, builds a process model building unit of the corresponding process model, and stores it in the process model database;

[0021] 2) The real-time data acquisition unit collects the real-time data of the position number and stores it in the historical trend database;

[0022] 3) The real-time alarm collection unit collects process alarm data and stores it in the historical alarm database;

[0023] 4) The detection and triggering unit detects alarm flooding information and generates an alarm flooding event, which is transmitted to the alarm tracing unit to trigger the alarm tracing;

[0024] 5) The alarm tracing unit receives the alarm flooding event from the detection and triggering unit, calls the alarm tracing model building unit to obtain the alarm tracing model, and then performs alarm tracing;

[0025] 6) The alarm tracing model construction unit constructs and outputs a tracing model based on the current process model, historical trends, historical alarms and current fault information.

[0026] Furthermore, the detection and triggering unit sets a detection cycle threshold, and outputs three alarm flooding events according to the number of alarms within the detection cycle threshold. After receiving the events, the alarm tracing unit performs different actions:

[0027] 1) After receiving the alarm flood start event, create the fault tracing environment and start the first alarm tracing;

[0028] 2) After receiving the alarm flooding event, the alarm source is traced again and the traceability results are updated;

[0029] 3) After receiving the alarm flood end event, the last alarm tracing will be performed, the tracing results will be updated, and the tracing will be terminated.

[0030] Furthermore, the detection and triggering unit also sets the following rule: when a continuous event is received again and the current traceability calculation has not yet ended, after the alarm traceability request enters the queue, if there are multiple traceability requests in the queue, only the most recent one is retained.

[0031] Furthermore, the alarm tracing unit performs reverse reasoning based on the tracing model to obtain possible alarm propagation paths, and finally selects the path with the highest average posterior probability among these propagation paths to complete the alarm tracing.

[0032] Compared with the prior art, the real-time alarm tracing device and method of the present invention has the following effects:

[0033] 1) It improves the real-time and automation of alarm tracing in process industry production and enhances the ability to trace the alarm of unknown faults.

[0034] 2) When a large number of alarms occur in a short period of time due to a fault in the on-site production process, the alarm flood can be realized, allowing the operator to quickly locate the source of the alarm, thereby speeding up the troubleshooting.

[0035] 3) Since the process control configuration does not change frequently, an offline process model is used, and its construction time is not chosen to be temporarily generated when a fault occurs, but after each process control configuration is modified and takes effect, which can reduce the operating load and tracing time of the online traceability device after a fault occurs.

[0036] 4) In the online alarm tracing unit, by configuring the alarm flooding detection cycle, during the alarm flooding process, as time goes by and the alarm changes, the alarm can be traced multiple times to make the tracing result closer to reality, while taking into account the timeliness and accuracy of the alarm. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solution of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0038] Figure 1 This is a schematic diagram of the online alarm tracing principle of an embodiment of the present invention;

[0039] Figure 2 It is a principle block diagram of a real-time alarm tracing device and an online alarm tracing data flow diagram according to an embodiment of the present invention;

[0040] Figure 3 An example of a control loop in an embodiment of the present invention is given;

[0041] Figure 4 and5 This is an embodiment of the process model of the present invention;

[0042] Figure 6 This is a schematic diagram of the principle of the alarm tracing model construction unit according to an embodiment of the present invention;

[0043] Figure 7 This is an example diagram of an alarm tracing result set according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following is a clear and complete description of the technology in the embodiments of the present invention in conjunction with the accompanying drawings of the embodiments of the present invention. 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 making creative work are within the scope of protection of the present invention.

[0045] Example 1

[0046] An embodiment of the present invention proposes a real-time alarm tracing device in a process industry production process, and its application scope is the production device level. That is to say, if online alarm tracing is to be performed on multiple production devices on site, each production device includes its own independent alarm tracing device to avoid mutual influence of data between multiple production devices during tracing.

[0047] like Figure 1 As shown, when a failure occurs in the on-site production system, it is possible that many alarms will be generated in a short period of time due to the complexity of the system itself, causing alarm flooding. This system detects alarm flooding and triggers online alarm tracing to find the source of the failure.

[0048] The alarm source tracing in the embodiment of the present invention refers to finding out the source of the alarm flooding, such as the alarm event or the bit number, through comprehensive calculation and analysis of the alarm event, bit number data, process model and other data.

[0049] like Figure 2 The real-time alarm tracing device of this embodiment includes two parts: an online alarm tracing device 100 and an offline process model building device 200.

[0050] The offline process model building device 200 obtains the process configuration from the process configuration database 201 , and builds the corresponding process model through the process model building unit 202 , and stores it in the industrial model database 203 .

[0051] The process configuration includes configuration data such as bit number, process alarm, control loop, etc. The process model includes the relationship between bit numbers and the relationship between bit numbers and process alarms.

[0052] like Figure 3The figure below is an example of a simple temperature control loop. In the figure: the temperature setting refers to the set value of this control loop, such as setting 80 degrees Celsius, and water vapor refers to the temperature control medium, which is heated by water vapor to control the valve opening so that the target reaches the set value. FL1, FV1, and TT1 are bit numbers, which usually refer to monitoring data points obtained from field equipment such as the temperature of a temperature instrument, the flow rate of a flow meter, or control data points used to control field equipment such as adjusting the opening of a valve. A loop contains multiple bit numbers, and the relationship between these bit numbers is directed. Figure 3 Changes in FV1 will lead to changes in TT1.

[0053] The process model building unit analyzes the data in the configuration to establish the relationship between all the bit numbers, forming a directed graph. Figure 3 For example, the control loop example will generate Figure 4 An example of a process model for a single loop.

[0054] When the process model building unit has parsed all the information in the configuration, a complete Figure 5 As shown in the figure, X1-X10 represent the position numbers in the loop.

[0055] There is also an association between the bit number and the process alarm. For example, if a process alarm configuration is configured to generate a high limit alarm when TT1 (steam temperature) > 100, then an association is established between this high limit alarm and the FV1 bit number. Similarly, the process model building unit will also establish an association between the bit number and the alarm by parsing the alarm configuration.

[0056] When the process model is built, it will be stored in the process model database 203. It can be used by the online alarm tracing device. For example, the relationship between the alarm and the bit number can be used to deduce which bit number directly causes an alarm, and then the relationship between the bit numbers can be used to deduce which bit numbers affect the bit number data anomaly.

[0057] Users can also view all established relationships through a visual interface and manually correct incorrect relationships or add relationships that cannot be established through automatic construction.

[0058] The online alarm tracing device 100 of this embodiment is used to trace the source of real-time alarms. The device mainly includes:

[0059] 1) Real-time data acquisition unit 106. This unit is used to collect real-time data of all bit numbers in the on-site production control device, such as the current value of the temperature, flow or valve opening bit number, and store the received data in the historical trend database 107.

[0060] 2) Real-time alarm collection unit 105 . This unit is used to collect all process alarm data in the on-site production control device and store the received data in the historical alarm database 108 .

[0061] 3) Detection and triggering unit 101. This unit triggers subsequent alarm tracing by detecting the occurrence of alarm flooding. The definition of alarm flooding can be configured by the user, for example: setting the detection cycle threshold to 30 seconds, when the current state is non-alarm flooding, and the total number of alarms generated within the current cycle threshold of 30 seconds exceeds a certain number of alarms, such as 10, it is defined as an alarm flooding start event; when the current state is alarm flooding, and the number of alarms generated within the current cycle threshold is less than or equal to a certain number of alarms, such as 2, it is defined as an alarm flooding end event.

[0062] Furthermore, this embodiment also adds an alarm flooding continuous event, such as: when the current state is alarm flooding, and the total number of alarms generated within the current cycle threshold is greater than 2, it is defined as an alarm flooding continuous event. Because during the alarm flooding continuous process, the relevant alarms of this fault may not all be generated in the first detection cycle, but new relevant alarms will continue to be generated in the subsequent process. The definition of the alarm flooding continuous event can be used to notify the alarm tracing unit to re-perform alarm tracing according to the current fault information, so as to make the alarm tracing result more accurate.

[0063] Within each detection cycle threshold, such as 30 seconds, the detection and triggering unit will detect the alarm flooding state according to the configuration, and generate the following events and push them to the alarm tracing unit 102. After receiving the event, the alarm tracing unit will perform different actions. The following Table 1 shows the event classification, generation conditions and corresponding actions of the alarm tracing unit after receiving the event.

[0064] Table 1 Classification of events generated by detection and triggering units

[0065]

[0066] 4) Alarm tracing unit 102. The alarm tracing unit manages the current fault and status. It receives events from the detection and trigger unit 101 and regards the alarm flooding from the beginning to the end as a fault. The alarm tracing unit 102 receives events from the detection and trigger unit 101. After receiving the events, the alarm tracing unit starts alarm tracing according to the process of creating a tracing environment - obtaining a tracing model - performing alarm tracing.

[0067] The alarm tracing unit comprises:

[0068] The tracing creation submodule is used to create a tracing environment for the fault and start the first alarm tracing after receiving the alarm flood start event;

[0069] The source tracing update submodule is used to trace the alarm again and update the tracing results after receiving the alarm flooding continuous event;

[0070] The tracing end submodule is used to perform the last alarm tracing and update the tracing results after receiving the alarm flood end event;

[0071] The tracing model calling submodule is used to obtain the tracing model through the alarm tracing model construction unit.

[0072] When the alarm tracing unit is to perform alarm tracing, it will first call the alarm tracing model building unit 103 to obtain the tracing model. After obtaining the tracing model, the alarm tracing unit will perform reverse reasoning based on the tracing model to obtain possible alarm propagation paths, and the alarm tracing is completed at this time.

[0073] 5) Alarm tracing model construction unit 103. The process of alarm tracing model construction is an asynchronous process, which is triggered by the alarm tracing unit. The input of this unit is the current process model, historical trends, historical alarms, and current fault information, and the output is the current tracing model. The fault time period and the normal operating time period can be identified based on this information, and the model can be constructed based on this. The construction of the alarm tracing model can be implemented using different algorithms, such as data-driven algorithms such as Bayesian networks. After the model is built, it will be saved in the tracing model database 109.

[0074] like Figure 7 As shown, the alarm tracing model construction unit is implemented in two parts: data preprocessing and algorithm analysis. The following is a simple example of the Bayesian network algorithm (for the specific algorithm, please refer to the algorithm introduced in Ding Rui's 2020 Wanfang Database "Complex Industrial Alarm Tracing Analysis Based on Bayesian Network"):

[0075] In the data preprocessing stage, alarms are obtained from the historical alarm database, and the alarm generation probability and alarm duration distribution table in this fault and under normal operating conditions are compared. Then, combined with the process model, the Bayesian network structure scoring function is used to finally obtain the Bayesian network structure with the highest score.

[0076] In the algorithm analysis phase, the Bayesian network parameters can be obtained by combining the real-time data of the bit number and using the Bayesian network parameter learning algorithm to determine the contribution of each bit number. For example, bit number X6 will eventually produce the following contribution result table 2, where √ indicates that the bit number and X6 are associated. For example, when X4, X5, and X8 are associated, the contribution of X6 to the fault condition is 0.75. The higher the contribution, the stronger the association between the two bit numbers. That is, when X4, X5, and X8 alarm at the same time, the alarm probability of X6 is 0.75.

[0077] Table 2 Example of contribution of bit number correlation

[0078] X4 √ √ √ √ X5 √ √ √ √ X8 √ √ √ √ Contribution 0.75 0.35 0.21 0.12 0.65 0.80 0.58

[0079] The algorithm analysis will save this part of data into the tracing model database 109 and output it to the alarm tracing unit for final alarm tracing.

[0080] The monitoring interface display unit 104 is used to display the traceability results, for example, Figure 3 For example, it is shown that 75% may be due to abnormal feed flow FL1, and 25% may be due to abnormal valve opening FV1.

[0081] Example 2

[0082] This embodiment proposes a real-time alarm tracing method in a process industry production process.

[0083] The offline process model building device 200 builds a process model of the corresponding on-site production control device by parsing the bit number, control loop and alarm configuration after the configuration is released. The process model will be used by the alarm building unit to generate a traceability model.

[0084] The process model construction unit 202 reads the bit number, control loop and alarm configuration in the control configuration in the process configuration database, and generates a process model by analyzing the relationship between the bit number and the bit number in the control loop configuration, as well as the relationship between the alarm and the bit number in the alarm configuration, and stores it in the process model database.

[0085] In the online alarm tracing device 100:

[0086] The real-time data acquisition unit 106 acquires the real-time data of all the bit numbers in the on-site production control device, such as the current value of the temperature, flow or valve opening bit number, and stores the received data in the historical trend database 107.

[0087] The real-time alarm collection unit 105 collects all process alarm data in the on-site production control device and stores the received data in the historical alarm database 108 .

[0088] The detection and triggering unit 101 detects the occurrence of alarm flooding to trigger subsequent alarm tracing. The definition of alarm flooding can be configured by the user. The detection and triggering unit will detect the alarm flooding state once every period according to the configuration, and generate three events: alarm flooding start, alarm flooding continuation and alarm flooding end, and push them to the alarm tracing unit 102. After receiving the event, the alarm tracing unit will perform different actions.

[0089] The alarm tracing unit 102 receives events from the detection and triggering unit 101. After receiving the events, the alarm tracing unit will start tracing according to the process of creating a tracing environment - obtaining a tracing model - tracing. The following is the alarm tracing workflow:

[0090] a. After receiving the alarm flood start event, a tracing environment for the fault will be created and the first alarm tracing will begin.

[0091] b. After receiving the alarm flooding continuous event, perform alarm tracing again and update the tracing results. If the current tracing calculation has not ended after receiving the continuous event again, the alarm tracing request will enter the queue. If there are multiple tracing requests in the queue, only the most recent one will be retained to improve efficiency and avoid unnecessary tracing calculations. During the fault time period, the tracing results may be updated according to the passage of fault time, so it is necessary to trace the source again to find the source of the fault more accurately.

[0092] c. After receiving the alarm flooding end event, the last alarm tracing will be performed and the tracing results will be updated. The tracing will then be terminated and the tracing environment will be cleared.

[0093] When the alarm tracing unit performs tracing, it first calls the alarm tracing model construction unit 103. After obtaining the tracing model, the alarm tracing unit will perform reverse reasoning based on the tracing model to obtain possible alarm propagation paths, and finally select the paths with the highest average posterior probability ranking among these propagation paths, such as the top three. At this point, the alarm tracing is completed, and the root cause of this fault is the starting bit number in these three paths. The possibility of it being the root cause of the alarm is the average posterior probability of each path, and it is output to the monitoring interface display unit 104 for display.

[0094] like Figure 7 The figure shows an example of alarm tracing according to an embodiment of the present invention. After reverse reasoning, N paths and their posterior probabilities are obtained. For example, in path 1, X1, X4, X6, and X8 are all bit numbers, and the root bit number is X1. The numbers on the path between each bit number, such as 0.75, are the posterior probabilities, and the average value of all posterior probabilities on the path, 0.77, is the average posterior probability. As can be seen from the figure, the average posterior probabilities of paths 1, 2, and 3 are the highest, which are 0.77, 0.75, and 0.72, respectively. Then, the root bit numbers X1, X2, and X3 in these three paths are taken as the result bit numbers of the final alarm tracing.

[0095] The alarm tracing model construction unit 103 can be implemented using different algorithms. The process of alarm tracing model construction is an asynchronous process, which is triggered by the alarm tracing unit. The input of this unit is the current process model, historical trends, historical alarms, and current fault information, and the output is the current tracing model. Based on this information, the fault time period and the normal operating time period can be identified, and the model can be constructed based on this. After the model is built, it will be saved in the tracing model database 109, and the monitoring interface display unit 104 is used to display the tracing results.

[0096] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. 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 real-time alarm tracing device in a process industry production process, characterized in that: It includes an offline process model building device and an online alarm tracing device; among which: The offline process model building device includes: a process configuration database for storing process configuration data; a process model database for storing process model data; a process model building unit for acquiring process configuration data from the process configuration database and building a corresponding process model; The online alarm tracing device includes: a real-time data collection unit for collecting real-time data of bit numbers, and storing the received data in a historical trend database; a real-time alarm collection unit for collecting process alarm data, and storing the received data in a historical alarm database; a detection and triggering unit for triggering an alarm tracing unit to perform alarm tracing by detecting alarm flooding information and generating an alarm flooding event; an alarm tracing unit for receiving an alarm flooding event from the detection and triggering unit, and calling an alarm tracing model construction unit to obtain a tracing model and then perform alarm tracing; an alarm tracing model construction unit for receiving process models, historical trends, historical alarms, and current fault information from an industrial model database, a historical trend database, a historical alarm database, and an alarm tracing unit, respectively, and constructing and outputting alarm tracing model data.

2. The real-time alarm tracing device in the process industry production process as claimed in claim 1 is characterized in that: The process configuration data includes bit numbers, process alarms, and control loops. The process model constructed by the process model construction unit includes building relationships between bit numbers and bit numbers and between bit numbers and process alarms based on bit numbers and control loops.

3. The real-time alarm tracing device in the process industry production process as claimed in claim 1 is characterized in that: It also includes a monitoring interface display unit for displaying traceability results.

4. The real-time alarm tracing device in a process industry production process as described in any one of claims 1 to 3, characterized in that: The detection and triggering unit is provided with a detection cycle threshold, and outputs at least two alarm flooding events according to the number of alarms within the detection cycle threshold. The alarm tracing unit performs different actions after receiving the events.

5. The real-time alarm tracing device in the process industry production process as claimed in claim 4, characterized in that: The detection and triggering unit is configured to output three alarm flooding events, including an alarm flooding start event, an alarm flooding continuation event, and an alarm flooding end event; the alarm tracing unit includes three submodules: a tracing creation submodule, which is used to create a tracing environment for the fault and start the first alarm tracing after receiving the alarm flooding start event; The traceability update submodule is used to perform alarm traceability again and update the traceability results after receiving an alarm flood continuation event; the traceability end submodule is used to perform the last alarm traceability after receiving an alarm flood end event, update the traceability results, and terminate this traceability.

6. A real-time alarm tracing method in a process industry production process, characterized in that: The real-time alarm tracing device according to any one of claims 1 to 5 is used to perform alarm tracing, comprising the following steps: 1) The process model building unit of the offline process model building device obtains process configuration data from the process configuration database, builds a process model building unit of the corresponding process model, and stores it in the process model database; 2) The real-time data acquisition unit collects the real-time data of the position number and stores it in the historical trend database; 3) The real-time alarm collection unit collects process alarm data and stores it in the historical alarm database; 4) The detection and triggering unit detects alarm flooding information and generates an alarm flooding event, which is transmitted to the alarm tracing unit to trigger the alarm tracing; 5) The alarm tracing unit receives the alarm flooding event from the detection and triggering unit, calls the alarm tracing model building unit to obtain the alarm tracing model, and then performs alarm tracing; 6) The alarm tracing model construction unit constructs and outputs a tracing model based on the current process model, historical trends, historical alarms and current fault information.

7. The real-time alarm tracing method in the process industry production process as claimed in claim 6, characterized in that: The detection and triggering unit is provided with a detection cycle threshold, and outputs at least two alarm flooding events according to the number of alarms within the detection cycle threshold. The alarm tracing unit performs different actions after receiving the events.

8. The real-time alarm tracing method in the process industry production process as claimed in claim 7, characterized in that: There are three alarm flooding events, including alarm flooding start event, alarm flooding continuation event, and alarm flooding end event. The alarm tracing unit performs different actions after receiving the event: 1) After receiving the alarm flood start event, create the fault tracing environment and start the first alarm tracing; 2) After receiving the alarm flooding event, the alarm source is traced again and the traceability results are updated; 3) After receiving the alarm flood end event, the last alarm tracing will be performed, the tracing results will be updated, and the tracing will be terminated.

9. The real-time alarm tracing method in the process industry production process as claimed in claim 8, characterized in that: The detection and triggering unit also sets the following rule: when a continuous event is received again and the current traceability calculation has not yet ended, after the alarm traceability request enters the queue, if there are multiple traceability requests in the queue, only the most recent one is retained.

10. The real-time alarm tracing method in a process industry production process according to any one of claims 6 to 9, characterized in that: The alarm tracing unit performs reverse reasoning based on the tracing model to obtain possible alarm propagation paths, and finally selects the path with the highest average posterior probability among these propagation paths to complete the alarm tracing.